Advanced Display Technology: Next Generation Self-Emitting Displays 9813365811, 9789813365810, 9789813365827

Table of contents :
Contents
About the Editors
Phosphorescent OLEDs for Power-Efficient Displays
1 Introduction
2 Design of Highly Efficient OLED Devices
2.1 Device Operation
2.2 Components of an OLED Stack
3 Properties of Phosphorescent Materials
3.1 Photophysical Processes in Phosphorescent Materials
3.2 Excited-State Characteristics for Phosphorescent Materials
3.3 Color Tuning of Phosphorescent Emitters
4 Cyclometallated Complexes for Efficient PHOLEDs
4.1 Ir(III) Complexes
4.2 Pt(II) Complexes
5 Summary and Outlook
References
TADF and Hyperfluorescence
1 Fluorescence and Phosphorescence
2 TADF (Thermally Activated Delayed Fluorescence)
3 Hyperfluorescence
4 TADF Molecules
4.1 Design Principles of TADF Molecules
4.2 Classification of Typical TADF Molecules
4.3 Intramolecular TADF Type
4.4 Soluble TADF Materials
References
Small Molecules in Ink Jet Printed OLEDs—History, Status, and Prospects
1 OLED Deposition Technologies
2 Material Requirements for Ink Jet Printing
2.1 Ink-substrate interaction
2.2 Fluorescent Blue
2.3 Phosphorescence Red and Green
2.4 Top Emission Performance
3 Outlook
References
Solution-Processible OLED Material: Based on Conjugated Polymer Technology
1 Introduction
2 Polymer OLED
2.1 Development History
2.2 Characteristics of Polymer OLED Materials
2.3 Features of Polymer OLED
2.4 Polymer OLED Performance: Emission Efficiency
2.5 Polymer OLED Performance: Emission Color
2.6 Polymer OLED Performance: Device Operating Lifetime
2.7 Features of OLED Printing
2.8 Current Design of Polymer OLED Materials
2.9 Current Performance of Polymer OLEDs
2.10 Next Direction of R&D
2.11 Future Perspectives
3 Conclusion
References
Chemical Mechanisms of Intrinsic Degradation of Emitting Layers in Organic Light-Emitting Devices
1 Introduction
2 Types of Defects
3 Chemical Mechanisms of Defect Formation
3.1 Unimolecular Degradation Mechanism
3.2 Bimolecular Degradation Mechanism
4 Conclusions
References
Encapsulation Technology for Flexible OLEDs
1 Introduction
2 Encapsulation Barrier Technologies
2.1 Dark Spots and Defects in OLEDs
2.2 Thin-Film Encapsulation
2.3 Atomic Layer Deposition (ALD)
2.4 Measurement Methods
2.5 ALD-Based Multi-barrier
3 Flexible and Reliable Thin Film Encapsulation
3.1 TFE Mechanical Failure
3.2 Neutral Axis Engineering for TFE
3.3 Nano-stratified Structure TFE
4 Conclusion
References
Oxide Thin-Film Transistors for OLED Displays
1 High-Performance Metal-Oxide Thin-Film Transistors
1.1 TFT Configuration and Fabrication Process
1.2 TFT Optimization
1.3 Beyond the IGZO and SiO2 Stack
2 Reliability of Amorphous Oxide TFTs
3 Summary and Outlook
References
Pixel Circuits for OLED Displays
1 The Need for OLED Pixel Circuit
1.1 OLED Driving in a Display
1.2 Role of the OLED Pixel Circuit
1.3 TFT-to-OLED Connection Structure
2 Operating Principles of OLED Pixel Circuit
2.1 In-Pixel Compensation
2.2 Ext-Panel Compensation
3 Examples of OLED Pixel Circuit
3.1 7T-1C-6L PMOS Diode-Connection Pixel Circuit
3.2 4T-2C-6L NMOS Source-Follower Pixel Circuit
3.3 6T-1C-7L LTPO Diode-Connection Pixel Circuit
4 Limitations of In-Pixel Compensation Circuits
4.1 Current Uniformity Versus VT Extraction Time
4.2 Low Grey-Level Mura Versus Dimming Method
References
Large-Size OLED TVs with White OLED
1 Introduction
2 White Organic Light-Emitting Devices
2.1 Two-Stack Tandem WOLEDs
2.2 Three-Stack Tandem WOLEDs
3 Display Transformation with OLEDs
4 Top-Emitting White OLEDs and Display Application
4.1 Microcavity Top-Emitting White OLED Devices
4.2 Non-microcavity Top-Emitting White OLED Devices
4.3 Transparent OLED Display
5 Conclusion
References
Quantum Dot-Enabled Displays
1 Introduction
1.1 Commercial History
1.2 Quantum Dots as Emitters
1.3 Benefit of QDs in Displays
2 QD-Enhanced LCD Backlight Displays
2.1 Overview
2.2 Technology Status
2.3 Market for the Quantum Dot LCDs Today
3 QD Color Conversion
3.1 Overview
3.2 Technology Status
3.3 Applications
4 Electroluminescent QD Displays
4.1 Overview
4.2 Status
4.3 Efficiency
4.4 Lifetime
4.5 Challenges
5 Conclusions and Outlook
References
Electroluminescence Devices with Colloidal Quantum Dots
1 Introduction of Quantum Dot Based Display Applications
2 QD-Based Electroluminescence Device (QD-LED)
2.1 General Description of QD-LED
2.2 Operation Mechanism of QD-LED
2.3 QD Emissive Materials
2.4 QD-LED Structure
3 Perspectives on Future Quantum Dot Displays
3.1 Recent Progress in Cd-Free QD-LEDs
3.2 QD Patterning for Display Applications
4 Summary
References
Micro-LED Technology for Display Applications
1 Introduction
1.1 The Micro-LED Market and Application Forecast
1.2 Development of Micro-LED-Based Displays
2 Issues Concerning Micro-LED Development
2.1 Micro-LED Development Status
2.2 Efficiency of Micro-LEDs
2.3 The Current Status of Technological Development for Solving the Problems of Micro-LED Efficiency
3 LED Transfer Technology
3.1 The Pick-and-Place Transfer Method
3.2 Monolithic Transfer (Monolithic Integration and Hybridization)
3.3 Chip Manufacturing—LED Inspection and Replacement
4 Conclusions
References
Display Techniques for Augmented Reality and Virtual Reality
1 Virtual Reality Display
2 Reflective Augmented Reality Display
3 Waveguide-Based Head-Mounted Displays
4 Pin-Light Augmented Reality Display
5 Light Field Display
6 Sensor and Display Techniques
7 Summary
References

Citation preview

Series in Display Science and Technology

In Byeong Kang Chang Wook Han Jae Kyeong Jeong   Editors

Advanced Display Technology Next Generation Self-Emitting Displays

Series in Display Science and Technology Series Editors Karlheinz Blankenbach, FH für Gestaltung, Technik, Hochschule Pforzheim FH für Gestaltung, Technik, Pforzheim, Germany Fang-Chen Luo, Hsinchu Science Park, AU Optronics Hsinchu Science Park, Hsinchu, Taiwan Barry Blundell, University of Derby, Derby, UK Robert Earl Patterson, Human Analyst Augmentation Branch, Air Force Research Laboratory Human Analyst Augmentation Branch, Wright-Patterson AFB, OH, USA Jin-Seong Park, Division of Materials Science and Engineering, Hanyang University, Seoul, Korea (Republic of)

The Series in Display Science and Technology provides a forum for research monographs and professional titles in the displays area, covering subjects including the following: • • • • • • • • • • • • •

optics, vision, color science and human factors relevant to display performance electronic imaging, image storage and manipulation display driving and power systems display materials and processing (substrates, TFTs, transparent conductors) flexible, bendable, foldable and rollable displays LCDs (fundamentals, materials, devices, fabrication) emissive displays including OLEDs low power and reflective displays (e-paper) 3D display technologies mobile displays, projection displays and headworn technologies display metrology, standards, characterisation display interaction, touchscreens and haptics energy usage, recycling and green issues

More information about this series at http://www.springer.com/series/15379

In Byeong Kang · Chang Wook Han · Jae Kyeong Jeong Editors

Advanced Display Technology Next Generation Self-Emitting Displays

Editors In Byeong Kang LG Display Seoul, Korea (Republic of)

Chang Wook Han LG Display Seoul, Korea (Republic of)

Jae Kyeong Jeong Department of Electronic Engineering Hanyang University Seoul, Korea (Republic of)

ISSN 2509-5900 ISSN 2509-5919 (electronic) Series in Display Science and Technology ISBN 978-981-33-6581-0 ISBN 978-981-33-6582-7 (eBook) https://doi.org/10.1007/978-981-33-6582-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Contents

Phosphorescent OLEDs for Power-Efficient Displays . . . . . . . . . . . . . . . . . . Tyler Fleetham and Michael S. Weaver

1

TADF and Hyperfluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Junji Adachi and Hisashi Okada

39

Small Molecules in Ink Jet Printed OLEDs—History, Status, and Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sebastian Meyer, Manuel Hamburger, Sebastian Stolz, Miriam Engel, Anna Hayer, Hsin-Rong Tseng, Rouven Linge, and Rémi Anémian Solution-Processible OLED Material: Based on Conjugated Polymer Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Takeshi Yamada

67

83

Chemical Mechanisms of Intrinsic Degradation of Emitting Layers in Organic Light-Emitting Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Youngmin You Encapsulation Technology for Flexible OLEDs . . . . . . . . . . . . . . . . . . . . . . . 129 Eun Gyo Jeong and Kyung Cheol Choi Oxide Thin-Film Transistors for OLED Displays . . . . . . . . . . . . . . . . . . . . . 151 Hyeon Joo Seul, Min Jae Kim, and Jae Kyeong Jeong Pixel Circuits for OLED Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Kee Chan Park Large-Size OLED TVs with White OLED . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 Chang Wook Han, Yoon Deok Han, Hyun Chul Choi, and In Byeong Kang Quantum Dot-Enabled Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 Charlie Hotz and Jeff Yurek Electroluminescence Devices with Colloidal Quantum Dots . . . . . . . . . . . . 251 Seunghyun Rhee, Jeong Woo Park, and Wan Ki Bae v

vi

Contents

Micro-LED Technology for Display Applications . . . . . . . . . . . . . . . . . . . . . 271 Dong-Seon Lee and Jang-Hwan Han Display Techniques for Augmented Reality and Virtual Reality . . . . . . . . 307 Byoungho Lee and Youngjin Jo

About the Editors

In Byeong Kang received his B.S. and M.S. degrees in Electronic Engineering from Hanyang University, Seoul, Korea, in 1989 and 1991, respectively. He also received his Ph.D. degree in Electronic Engineering from the University of South Australia in 1998. He subsequently joined LG Display, Korea, as a engineer in 1991, where he worked on the design and characterization of display for various applications such as IT, TV, and Mobile. He served as CTO, executive vice president, in LG Display from 2014 to 2020, leading technology research and development of future flat panel display such as TFT-LCD, AMOLED, and flexible display. In particular, he has led development of rollable OLED TV, transparent signage OLED, and 8K UHD OLED TV. He has published more than 30 co-authored international journal papers and 130 international conferences. He has also presented many talks at major display conference including keynote speech. He is a fellow of The Society for Information Display. Now, he is also a regular member of The National Academy of Engineering of Korea. Chang Wook Han is the head of OLED device technology department as a chief research fellow and a vice president in LG Display, Korea. He obtained his B.S. and M.S. degrees in Material Science from Seoul National University, Korea, in 1987 and in 1989, respectively. He also received Ph.D. degree in Electrical Engineering from Seoul National University, in 2007. The title of his Ph.D. thesis was “a-Si:H TFT and pixel structure for AMOLED on flexible metal substrates”. Since joining LG Display in 1990, he has worked on the device and process development on a-Si:H TFT backplanes for AMLCD. He has also focused on enhancing the performance of OLED device and developing new OLED panel structure since 1999. In particular, he has led several projects that have developed tandem white OLED and encapsulation technology for TV. He has experienced the successful application of these technologies to the world’s first 55-inch OLED TV and transparent OLED products. His research achievements were recognized and he was presented the LG Group’s R&D Grand Award, in 2013. He has also developed the innovated white OLED devices to meet the requirement for 4K and 8K OLED TV. Recently, he is in charge of developing TADF, blue phosphorescence, self-emitting QLED, and cultivating new OLED device for AR/VR micro-display. He has published more than 16 authored vii

viii

About the Editors

international journal papers and 30 international patents on flat panel display. He has published 4 authored chapters of book. He has also presented over 20 talks at major display conference including invited talks. He has served The Society for Information Display and International Meeting on Information Display conference as a program committee member on OLED. Jae Kyeong Jeong received his B.S., M.S., and Ph.D. degrees in material science and engineering from Seoul National University, Seoul, Korea, in 1997, 1999, and 2002, respectively. In 2003, he was a postdoctoral researcher at the University of Illinois at Urbana–Champaign. In 2004, he joined Samsung Mobile Display Corp. as a senior engineer, where he had focused on the design and development of Si and amorphous IGZO TFTs for AMOLED display. In 2008, he successfully developed the world’s largest 12.1” Oxide TFT-driven AMOLED display as a project leader. In 2009, he joined Inha University Incheon, Korea, as an assistant professor, where he had continued the oxide semiconductor and related field-effect transistors. Since September 2015, he is with the Department of Electronic Engineering at Hanyang University as a professor. His research group is interested in next-generation display electronics, CMOS TFTs for IoT, stretchable electronics, and emerging semiconductor devices. He has published more than 163 authored (or co-authored) international journal papers and 112 international patents. His total citation number and h-index are 21,498 and 56, respectively, according to the latest scholar google search. Also, he is currently an editorial board member of Scientific Reports and Journal of Information Display.

Phosphorescent OLEDs for Power-Efficient Displays Tyler Fleetham and Michael S. Weaver

1 Introduction Over the past several decades, organic light-emitting devices (OLEDs) have transitioned from the lab to the marketplace and are now the leading technology for a wide range of display applications. OLEDs today can be manufactured at 8 K resolution [1], on a large scale, with remarkable efficiencies, color purity, and long lifetimes. Furthermore, many of the imaginative possibilities for OLED technologies have already found their way into commercial devices including foldable devices [2, 3], transparent displays [4], and rollable televisions [5]. These benefits and possibilities have led to their widespread adoption in a constantly growing number of markets including mobile displays, televisions, monitors, wearables, lighting, and more. While many technological developments have led to this success, none is more crucial than the development of highly efficient phosphorescent OLEDs (PHOLEDs). In this chapter, the physics and chemistry of phosphorescent emitters and their application in PHOLEDs will be discussed. The first practical OLEDs, reported by Tang and VanSlyke in 1987, employed a heterojunction of an aryl amine hole transporting layer and the electron transporting complex tris(8-hydroxyquinolinato)aluminum (Alq3 ), which also served as the fluorescent emissive material [6, 7]. This breakthrough demonstrated the utility of heterojunction interfaces in effectively confining charges enabling them to recombine to form excitons. Many further advances in device design helped improve the efficiency of charge recombination and further development of fluorescent materials improved the emission efficiency of the emitters. Ultimately, the efficiencies for fluorescent OLEDs were limited by the inability to harvest the triplet excitons which T. Fleetham 305 Yardley-Newtown Rd., Yardley, PA 19067, USA M. S. Weaver (B) 23 Governors Lane, Princeton, NJ 08540, USA e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_1

1

2

T. Fleetham and M. S. Weaver

are generated in electroluminescent (EL) devices, limiting their internal electron to photon conversion efficiency ηint ≤ 25% [8]. In 1998, this limitation was overcome with the development of PHOLEDs which incorporated heavy metal organometallic compounds as dopants [9]. The incorporation of a heavy metal center such as Pt or Ir affords efficient harvesting of all the electrogenerated singlets and triplets and can emit efficiently from the lowest excited triplet state through the use of spin–orbit coupling (SOC) [10]. While this initial PHOLED report demonstrated only 4% external quantum efficiency (EQE), the principle of using heavy metal complexes to harvest 100% of the electrogenerated excitons energized the field and ηint approaching 100% were reported in just a few years [11]. In the more than two decades, since the first report of PHOLEDs, thousands of researchers have devoted their efforts to further improvement and discovery in this field leading to tens of thousands of published journal articles and patents. This large body of work has enabled extremely high EQEs exceeding 60% [12], operational lifetimes in the hundreds of thousands of hours, and OLEDs emitting across the spectrum from the ultraviolet to the infrared [13, 14].

2 Design of Highly Efficient OLED Devices The efficiency of an OLED can be most completely described by the power efficacy (η) which is a measure of the perceived light per input power. This metric encompasses the power input, light emission, and the photopic response of the human eye and is given in units of lm/W. For a Lambertian emitting OLED source, the power efficacy for a Lambertian emitting OLED is given by η = ηle π/V

(2.1)

where V is the operating voltage, and ηle is the luminance efficacy (in cd/A) given by the equation: ηle = kηInt · ηOut

(2.2)

where ηInt is the internal quantum efficiency (% excitons to photons), ηOut is the outcoupling efficiency (a measure of how many generated photons are emitted from the device), and k is a constant depending on the photopic response of the human eye. Taken together the power efficacy can be described by η = kη I nt · η Out π/V

(2.3)

This relationship highlights that to optimize device performance steps need to be taken to maximize the internal quantum efficiency, ηInt , and the light extraction efficiency, ηOut , and minimize the voltage, V.

Phosphorescent OLEDs for Power-Efficient Displays

3

Another common metric often reported in published results is the EQE (ηEQE ) which is the percentage of photons emitted from the device per electron injected into the device. η E Q E = η I nt · η Out

(2.4)

For the purposes of separating the analysis for the various steps in EL generation, ηInt can be expanded to be the product of photoluminescent quantum yield (PLQY, ), recombination efficiency, ηr , and the exciton utilization constant χ which is 0.25 for fluorescent materials and 1 for phosphorescent materials as a result of differences triplet harvesting discussed in Sect. 2.2.5. The resulting expression for EQE is η E Q E = χ ·  · ηr · η Out

(2.5)

2.1 Device Operation An OLED consists of one or more organic layers 10–1000 s of nanometers thick between two electrodes. An electric field on the order of 106 Vcm−1 is then applied across the OLED in order to generate EL. To optimize the efficiency parameters mentioned above, it is important to consider each of the steps in converting injected charges into emitted photons. Figure 1 shows the process for generating electroluminescence from a typical OLED. Upon applying a voltage to the device, there is a series of five key processes that take place, namely: (1) charge injection, (2) charge transport, (3) exciton formation, (4) exciton decay, and (5) light extraction.

Fig. 1 Key processes in OLED device operation

4

(1)

(2)

(3)

(4)

(5)

T. Fleetham and M. S. Weaver

Charge injection: An electron is injected from the cathode to the lowest unoccupied molecular orbital (LUMO) of an organic molecule creating an anion. Similarly, an electron is removed from the highest unoccupied molecular orbital (HOMO) of an organic molecule by the anode creating a cation. While these charged species are formally organic molecules in their radical doublet states, it is often more convenient to describe them as “electron” and “hole” charge carriers, respectively, to borrow terminology from analogous charge transport processes in inorganic devices. Charge transport: Electrons and holes move through their respective transport layers under the influence of the applied electric field. In contrast to inorganic devices, these charge carriers are not free carriers, but rather oxidized or reduced molecules. The localized nature of these charged states requires that the charges must hop from the molecular orbitals of one molecule to the next via an intermolecular electron transfer. The spatial distribution in electron density of molecular orbitals, the barriers between potential wells of neighboring molecules, and the random orientation of the molecules in the amorphous film all slow this hopping process compared to charge transport in a crystalline inorganic solid [15]. As a result, the mobility of holes in typical organic materials is on the order of 10−3 cm2 /(V s) while electron mobility can be an order of magnitude or more lower [16, 17]. Exciton formation: The electrons and holes are coulombically attracted and recombine in the emissive portion of the device to form a neutral excited state called an exciton28 . This exciton is a bound excited state which is coulombically stabilized to prevent its dissociation back into free charge carriers. This recombination process generates a statistical mixture of 25% singlets and 75% triplets which for PHOLEDs all rapidly become triplets due to favorable intersystem crossing [18]. A more detailed description of this process follows in Sect. 2.2.5. Charge recombination can also form bimolecular or larger excited states extending over more than one molecule such as excimers [19] or exciplexes [20]. Exciton decay: The exciton in its excited singlet or triplet state can relax to its ground state through either a radiative or non-radiative process. At room temperature, fluorescent materials only radiatively decay from their excited singlet state with a nanosecond lifetime [21]. Since 75% of the electrogenerated excitons are triplets, a large portion of the excited states are either lost through non-radiative pathways [8] or are upconverted back to the singlet state through a delayed fluorescence pathway [22, 23]. In the case of phosphors, all the excited states rapidly intersystem cross into the triplet state which can emit radiatively with a radiative decay lifetime in the microsecond time range [9]. These longer excited-state lifetimes allow for substantial exciton diffusion or other energy transfer processes before relaxing to the ground state. Light extraction: The photon generated in the exciton decay process is emitted from the OLED structure. Many photons are lost in this process due to a number of loss pathways including absorption processes in the organic stack or total internal reflection in the organic layers or substrate [24]. As a result of these

Phosphorescent OLEDs for Power-Efficient Displays

5

light extraction losses, and some losses from undesirable energy transfer events in the exciton decay process, in standard bottom emitting OLEDs less than 20– 40% of the excited states created exit the device out of the front of the substrate [25, 26].

2.2 Components of an OLED Stack The first organic EL devices attempted to create electroluminescence from single layers of organic material [27–29]. This required all the key steps to be optimized within a single layer. The organic heterojunction developed by Tang and Van Slyke afforded the ability to optimize the injection and transport of the holes and electrons separately [6]. Additionally, the barriers created by the heterojunction confined the recombination of charges to near the organic heterojunction interface. This ensured that a large proportion of the injected charges successfully formed an exciton and that this exciton recombination zone was far away from quenching interfaces at the electrodes. Further separation of the functions of charge injection, charge transport, and emission evolved toward the much more complicated device architectures used today which contain many layers [30–32], sometimes including layers of mixed compositions [33] to optimize each of the five key steps in generating efficient electroluminescence. The reason for this increased complexity is the limited capacity to tune the various properties of organic materials independently. It is often convenient to separate all the key steps in OLED operation into separate layers, each with a composition optimized for a specific role or small number of roles. As shown in Fig. 2, a typical device stack for modern OLEDs is depicted that will contain separate layers or materials for charge injection (HIL/EIL), charge transport (HTL/ETL), blocking layers to ensure efficient exciton formation (EBL/HBL), and an emissive layer (EML) optimized for efficient exciton formation and decay. Not depicted in Fig. 2 are the various light outcoupling strategies that will optimize the light extraction out of the device which will be discussed in Sect. 2.2.6. Despite the distribution of roles, it is important to consider the energetic and mobility differences between the layers to ensure the barriers to charge injection are low while also avoiding charge buildup at any interface. It is particularly important to consider the differences in electron and hole injection and transport. A large discrepancy between the delivery of charge carriers to the EML can lead to charge buildup and undesirable quenching or charge leakage effects [80]. Finding this balance can be challenging since the relative mobility of holes is typically an order of magnitude or larger than electrons in many common organic materials. Furthermore, ensuring this balance occurs across the entire range of practical driving conditions adds an additional layer of complexity with many devices showing reduced performances at very high or very low driving currents due to this imbalance [86].

6

T. Fleetham and M. S. Weaver

Fig. 2 a Typical OLED device stack and b schematic energy level diagram for state-of-the-art OLEDs where HIL is the hole injection layer, HTL is the hole transport layer, EBL is the electron blocking layer, EML is the emissive layer (which is typically doped with the emitter), HBL is the hole blocking layer, ETL is the electron transport layer, and EIL is the electron injection layer

2.2.1

Deposition of OLED Stacks

While the photophysical and electrochemical properties of each of the layers can be optimized independently, a common property among all the layers is the need to be compatible with the deposition technique and form high-quality, defect-free films. The most commonly used deposition technique is vacuum thermal evaporation (VTE), where organic layers are deposited sequentially onto the substrate from resistively heated source boats in a high-vacuum environment [34]. The organic material evaporation can take place from either the liquid or the solid state depending on the melting point and the temperatures required to reach an appreciable vapor pressure > 10−3 Torr. For example, the hole transporting material, 4,4 -bis[N-(1-napthyl)-Nphenyl-amino] biphenyl (α-NPD), melts before evaporating under typical growth rates, whereas many of the other organic materials achieve a higher vapor pressure well before their melting point and evaporate via sublimation. The vacuum conditions under which OLEDs are fabricated are extremely important to the efficiency and operational stability of the devices so depositions are typically carried out in high vacuum at a base pressure of around 10−7– 10–8 Torr [35–37]. Evaporation rates of the organic materials are monitored using quartz oscillators and are typically in the range 0.01–0.5 nm/s in research and development tools. In large-scale manufacturing tools, higher deposition rates and longer times of continuous deposition from a source are required wherein the source is kept at elevated temperatures for 24–336 h. Therefore, materials must be able to withstand this environment without degradation which may reduce the quality of the deposited material or impact the ability to adequately access stable deposition conditions.

Phosphorescent OLEDs for Power-Efficient Displays

7

In order to maintain compatibility with the deposition requirements, there are several morphological and thermal considerations to be taken into account including molecular weight, morphological properties, sublimation temperature, decomposition temperature, and many others. Since small molecule organic materials must be able to be cleanly deposited by VTE without decomposing, selecting materials with moderate sublimation temperature in the range of 150 °C–450 °C and high decomposition temperature is important to ensure optimal device performance. Furthermore, during deposition, the material must form high-quality, defect-free films with precisely controlled thicknesses, typically in the region of 5–200 nm. The films must be stable for long periods and have a high glass transition temperature (T g ), e.g., T g > 110 °C, to avoid crystallization over time [95, 96].

2.2.2

Electrodes

As mentioned in Sect. 2.1, the first key step in the operation of an OLED is the injection of charges into the organic layers. Making low resistance contacts requires optimal selection of the electrode materials with good conductivity, appropriate work function, high reflectivity (or transparency depending on the electrode), and deposition and patterning techniques compatible with the rest of the device stack. The anode for many OLED devices is typically indium–tin oxide (ITO) due it its good stability, transparency, and high conductivity. Several alternatives to ITO have been studied including polymers [38], metals [39], and other metal oxides [40]; however, ITO remains the present industry standard for bottom emitting devices (where light exits through the substrate) due to its favorable properties as well as its ability to be easily patterned using standard lithographic techniques to define different emitting regions or pixels on a substrate. Various deposition techniques and surface treatments have been studied to optimize the work function, conductivity, uniformity, and surface roughness of the ITO electrodes [41–44]. Treatment such as oxygen plasma or UV ozone exposure is particularly common since they lead to an enhanced device performance [42, 45], possibly due to the resulting increase in the work function of the ITO surface [43]. The exact stoichiometry of the ITO layer and the thickness can be fine-tuned for the required specifications of transparency, outcoupling constant, and conductivity. Typically, the ITO film is 50–200 nm thick and care has to be taken to minimize the surface roughness of the ITO layer to prevent shorts or nonuniform light emission from the thin organic layers. This can be particularly challenging in the case of flexible substrates and additional planarizing layers must be employed prior to deposition of the anode [46]. For bottom emitting OLEDs, the cathode contact should be highly reflective, have a low work function, and use a deposition technique that is compatible with organic layers underneath. Since the cathodes on standard bottom emitting devices are deposited on top of the organic layers care has to be taken to select materials, deposition rates, deposition source hardware, and patterning techniques which will not damage or contaminate the underlying organic layers. Aluminum is often the first choice as a cathode material chosen for its high reflectivity, moderate work function,

8

T. Fleetham and M. S. Weaver

and can be deposited relatively easily via VTE from resistively heated sources. Often, the cathode materials need to be used in conjunction with an electron injection material or alloyed with a low work function metal such as lithium, calcium, or magnesium in order to achieve an appropriate work function for efficient electron injection [47]. These materials are effective in tuning the work function of the electrode but these reactive metals are highly water and oxygen-sensitive which can make handling and processing challenging [48]. Other device structures such as top-emitting OLEDs or inverted OLEDs will employ different electrodes to satisfy the transparency and work function requirements of those devices. For top-emitting OLEDs the cathode is transparent allowing light to exit through the top of the device, while the anode is usually a highly reflective, high work function metal. Cathodes in top-emitting device architectures can be formed by using a thin metal contact [49], sometimes in conjunction with a conductive metal oxide [50], or a metal-free electrode [51]. The anode for a top-emitting OLED should have a high reflectance and a high work function such as Pt, Ag/ITO, or Al/Ni. If the anode is kept transparent, (e.g., ITO) in this top-emitting structure, the OLEDs can be made transparent and will emit light out of both sides of the device [52]. In an inverted structure, the cathode is in contact with the substrate and the organic stack is built up from the cathode interface to the anode in reverse order [53].

2.2.3

Hole Injection, Hole Transport, and Electron/Exciton Blocking Layers

The hole injection, hole transport, and electron blocking layers are collectively responsible for optimizing the delivery of holes to the EML while preventing the leakage of electrons to the anode where they avoid exciton formation. While these layers may not be directly involved with the emission of a photon, it is important to optimize these layers to reduce the driving voltage while preventing loss mechanisms such as hole trapping, exciton quenching at the EML interface, or electron leakage into these layers. Achieving optimal device performance requires an efficient injection into the HTLs, efficient hole transport through the layers and efficient injection into the EML. This multi-step process can be broken up into injection, transport, and blocking layers so that materials selection can be optimized separately for injection efficiency, high hole mobility, or suitable band offsets with other organic layers within the device. For more details on these processes, see Kalinowski [54] and Greenham and Friend [55]. Since the common anode material, ITO, has a work function < 5 eV [56] and the HOMO level of the hole conducting species in the EML can be as deep as −6 eV in blue OLEDs [57], it is important to make successive steps to ease the injection of holes into the EML or employ any strategies available to reduce this barrier. One strategy is the use of a hole injection layer which can serve several roles including modulate the work function of the anode, reduce the barrier to charge injection, planarize the substrate surface, and suppress recrystallization and delamination of the organic layer

Phosphorescent OLEDs for Power-Efficient Displays

9

at the anode. Films such as copper phthalocyanine (CuPc) have been used to increase the adhesion of the HTLs and increase device lifetime [58], polymer layers such as poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) have been used to planarize the substrate surface to improve film uniformity [59], thin layers of carbon between the ITO and the HTL have been shown to decrease the operating voltage and improve the device characteristics [60], and thin fluorocarbon films on ITO have been used to enhance the stability of the anode interface [61]. Some of the largest performance enhancements from using hole injection materials come from the use of 1, 4, 5, 8, 9, 11-hexaazatriphenylene hexacarbonitrile (HATCN) (shown in Fig. 3) or MoO3 because they have LUMO or conduction band levels close to the HOMO of common transport materials [62]. These extremely deep vacant energy levels allow these materials to freely remove electrons from the neighboring transport layer effectively doping the HTL with holes. In some cases, these materials have been codeposited to more strongly dope the hole transporting layer [63]. The primary layer for hole conduction is the HTL which is optimized for hole mobility. These materials typically contain electron-rich moieties which will readily accept holes from neighboring molecules or layers. Since charge transport occurs via hopping from the HOMO of one molecule to the next, hole transporting materials and their orientation in thin films should be designed to have enough orbital overlap between neighboring molecules to reduce the barrier for charge hopping. Various models have been proposed to describe charge transport in such materials [64, 65].

Fig. 3 Molecular structures of common hole injection, hole transporting, or electron/exciton blocking materials

10

T. Fleetham and M. S. Weaver

One of the most common classes of material used are aryl-amines or carbazolecontaining moieties, such as those shown in Fig. 3, which exhibit high hole mobilities [66, 67]. Furthermore, these materials have energy levels that are well aligned with the work function of ITO (>4.7 eV), giving a relatively small barrier to hole injection. α-NPD, for example, has a hole mobility of 10−3– 10−4 cm2 /(V s) and a HOMO level of 5.4 eV [16]. While the hole mobility in these materials is already quite high, the drive voltage can be further reduced by conductivity doping of the HTL layers [68]. A green-emitting tris(2-phenylpyridine)iridium(III) (Ir(ppy)3 )-doped PHOLED device which used the tetrafluoro-tetracyanoquinodimethane (F4 -TCNQ) as a conductivity dopant in the HTL was observed to have a driving voltage 2.65 V to produce 100 cd/m2 which is close to the minimum voltage possible to create a light emission of that energy [30]. This dopant achieves such a low voltage by removing electrons from the HOMO of the hole transport material to create charge carriers at a very low bias similar to the process described above for HATCN. In order to achieve very high efficiencies, devices also typically contain an electron and/or exciton blocking layer (EBL). Such a layer requires particular attention paid to its energy levels more than the mobility. To control the exciton formation and emission processes, electrons should be blocked from entering the HTLs to ensure charge recombination occurs within close proximity to the desired emitter. Also, after exciton formation on the desired emitter, no nearby materials should have lower excited state energies which would serve as an exciton sink. In PHOLEDs, this means that the LUMO of the exciton blocker should be shallower (closer to vacuum) than the EML components and the triplet energy should be higher than the desired emitter. However, since only a thin layer is required to block electrons and the energy transfer radius is only a few nm, this layer can be kept sufficiently thin so the hole mobility is not as stringently important as in the thicker HTL. Conversely, since this blocking layer separates the HTL and the EML, there is more freedom in selecting an HTL material for high mobility without consideration of its triplet energy. Typically, an EBL can be achieved with molecules containing either carbazole or triarylamine moieties due to their hole mobilities and high bandgap, such as in 4,4 ,4 -tris(Ncarbazolyl)-triphenylamine (TCTA) [69], 9-Phenyl-3,6-bis(9-phenyl-9Hcarbazol-3yl)-9H-carbazole (TrisPCz) [70], 1,3-Bis(N-carbazolyl)benzene (mCP) [71], or 1,1Bis[(di-4-tolylamino)phenyl]cyclohexane (TAPC) [68].

2.2.4

Electron Injection, Electron Transport, and Hole/Exciton Blocking Layers

The general design criteria for efficient electron transport are the same as those discussed above for hole transport, namely: good injection efficiency, high electron mobility, and suitable band offsets with other organic layers within the device. Similarly, electron transport can also be separated into injection, transport, and blocking layers to independently optimize the selection of the materials. One major difference between the hole transporting material selection and those for electron injection is

Phosphorescent OLEDs for Power-Efficient Displays

11

Fig. 4 Molecular structures of common electron injection, electron transporting, and hole/exciton blocking materials

the large barriers common for electron injection and that electron mobilities of most organic materials are orders of magnitude lower than those for holes. The appropriate selection of an electron injection material will decrease the barrier between the work function of the cathode (>4 eV for metals like Ag and Al) and the LUMO of the electron-conducting material with the EML (often shallower than −2.5 eV). Lithium salts such as LiF [72] or complexes such as (8hydroxyquinolinato)lithium (Liq) [73] (shown in Fig. 4) are also common as either thin electron injection layers or as conductivity dopants within the ETL to facilitate charge injection into the organic layers. It has been reported that LiF can reduce the work function for an Al electrode to 3.0 eV. The primary layer for electron conduction is the electron transporting layer (ETL) which is optimized for electron mobility and also serves the function of being kept sufficiently thick to protect against defects introduced by the hot cathode deposition as well as prevent quenching of the emission by the plasmon modes from the cathode interface. These materials typically contain electron-deficient moieties such as aza-substituted aromatic rings, phosphine oxides [74], and boron-containing moieties, all of which will readily accept electrons from neighboring molecules or layers. Among the most commonly used ETL in OLEDs is Alq3 , as shown in Fig. 4 which has a LUMO energy level of 3 eV16 and electron mobility of ~ 5 × 10−5 cm2 /(V s) [17]. More recently, several designs of electron-deficient molecules with

12

T. Fleetham and M. S. Weaver

large aromatic planes such as 2,4,6-tris(biphenyl-3-yl)-1,3,5-triazine (T2T) [75], 2,7Bis(2,2 -bipyridin-5-yl)triphenylene (Bpy-TP2) [76], and 4,6-Bis(3,5-di(pyridin4-yl)phenyl)-2-methylpyrimidine (B4PyMPM) [77] have demonstrated very high mobilities and improved device performance. In addition to selecting materials with high inherent mobilities, the ETL can be doped to further improve the mobility within the layer. One common class of material for this are organolithium complexes such as Liq [30]. In the region of the ETL that contacts the EML, it is common to use a hole and exciton blocking later to effectively confine the charges and excitons to the EML. Materials at this interface need to have several requirements: sufficient triplet energy to avoid quenching of the excitons in the device, good electron mobility and appropriate LUMO level to effectively inject electrons into the EML, and a deep (far from vacuum) HOMO level to prevent hole leakage into the ETL. One of the first materials used for this purpose is bathocuproine (BCP) which has a very deep HOMO level but a relatively low triplet [32]. Since then, several newly developed materials have been used as blocking layers including phosphine oxide-based materials such as dibenzo[b,d]thiophene-2,8diylbis(diphenylphosphine oxide) (PO15) [74, 78], tetraphenylsilane materials such as diphenylbis(4-(pyridin-3-yl)phenyl)silane (DPPS) [79], boron derivatives such as tris(2,4,6-trimethyl-3-(pyridin-3-yl)phenyl)borane (3TPYMB) [80], or triazoles such as 3-([1,1 -biphenyl]-4-yl)-5-(4-(tert-butyl)phenyl)-4-phenyl-4H-1,2,4-triazole (TAZ) which are all good selections for this role due to their high triplets and deep HOMOs.

2.2.5

Emissive Layer

The emissive layer is largely responsible for the key performance metrics of internal quantum efficiency (IQE), color of emitted light, and intrinsic operational lifetime. In this crucial layer, the key steps of charge transport, exciton formation, and exciton decay must all be optimized simultaneously. EMLs can be comprised of either a neat layer of material or a mixture of two or more materials. While the former is appealing from a manufacturing perspective, the simultaneous optimization of all the key processes with a single organic or organometallic compound is extremely challenging. Furthermore, high concentrations or neat films of phosphorescent materials can lead to undesirable self-quenching effects such as triplet–triplet annihilation events or self-absorption that reduce the device efficiency [81]. As a result, most PHOLED EMLs are often comprised of a small amount of emissive material doped into a host matrix comprised of one or more host materials. In these doped layers, both the properties of the individual components of the EML and the compatibility of the various components need to be considered to optimize charge balance, reduce exciton quenching, and avoid emission contamination.

Phosphorescent OLEDs for Power-Efficient Displays

13

Charge Transport in Emissive Layers Charge transport within the EML is primarily concerned with ensuring a balance of electron and hole transport so that the exciton formation occurs uniformly throughout the EML. A poor distribution of charges can lead to high concentrations of electrons or holes at an interface of the EML which in turn creates a concentrated exciton formation zone. High concentrations of charges and excitons can lead to bimolecular interactions such as triplet–polaron annihilation or triplet–triplet annihilation in which the excited triplet states are quenched by a neighboring charge or exciton reducing the internal quantum efficiency ηInt [82–84]. In order to avoid this efficiency loss it is important to select materials, and appropriate concentrations of the EML components that spread the exciton recombination zone across the EML. In a neat layer, achieving this balance would require the emissive material to have nearly identical hole and electron mobilities and would have no mechanism to ensure that the charges recombine. An elegant solution to the problem of achieving charge balance is to dope the emissive material into an organic host matrix. This was first realized by Tang et al. by doping highly luminescent fluorescent dyes into a host matrix [31]. Since then, a wide range of fluorescent dopants has been used in OLEDs where the dopants can be used to easily alter the emission color, in addition to improving the efficiency and the lifetime of devices [85, 86]. In PHOLEDs, doping in an organic host effectively dilutes the concentration of the phosphorescent material preventing aggregation which can lead to an undesirable self-quenching effect [81]. If the excited state energy of the phosphor is lower than that of the host, any excitons formed in the host material will migrate to the dopant. Furthermore, since the relative ratio of host and dopant can be easily controlled in vapor-deposited devices achieving a charge-balanced device that optimizes the charge transport and exciton formation processes can be done with fine-tuning of the deposition rates of the EML components. At low doping concentrations of the phosphorescent material, the majority of charge transport will occur on the host molecules [87]. In typical Ir(III)- or Pt(II)based phosphors, the HOMO level is shallower (closer to vacuum) than that of the host material and can be considered a “hole-trap.” This trapping can slow down charge transport so particular care needs to be taken in the selection of the host materials to ensure that these traps are not too deep of an energetic well for the charges to get out (i.e., a host material should be selected with a HOMO close to that of the dopants). On the other hand, if the host energy levels are such that no charge trapping occurs, charges may transport too easily to the edges of the EML and lead to high concentrations of excitons and charges at those interfaces, which can reduce device performance. One strategy to achieve charge balance is to use a combination of host materials with electron and hole transporting capabilities. In this system, optimization of the relative concentration of the host components can effectively tune the charge balance so that the conduction of holes and electrons within the EML is approximately equal to avoid charge build up at either side of the EML [88]. However, if neither charge is trapped in these cohost systems, it is

14

T. Fleetham and M. S. Weaver

possible that both charges may transport easily through their separate domains to the opposite EML interface, so some degree of charge trapping is typically desired. Once the concentration of the phosphorescent material is raised to ~3% the probability of the nearest neighbor being another dopant concentration increases and percolation pathways begin to form where charge transport can occur along with successive hops from dopant molecule to dopant molecule [89]. In typical phosphor systems at these doping concentrations, hole transport may occur primarily along these percolation pathways. One challenge in designing a balanced EML is ensuring that the charges can meet to eventually recombine. If both charges are carried through percolation pathways, it is possible that the charges may not come close enough for exciton formation to occur and the exciton formation process will be inefficient leading to losses in ηInt or increases in driving voltage. One strategy for achieving charge balance to improve device efficiencies and operational lifetime is using multiple EMLs of varying concentrations or using a gradient of doping concentrations to manipulate the recombination zone toward the center of the EML [90]. The selection of host materials is clearly important from a charge transport perspective, but, several other parameters also need to be considered. The key parameters in host selection include appropriate triplet energy to avoid exciton quenching, appropriate HOMO and LUMO levels to aid in charge transport and avoid exciplex formation [91, 92], high charge mobility, compatibility with the deposition process (e.g., appropriate sublimation temperature for vapor deposition techniques), high glass transition temperature (Tg) [93, 94], good electrochemical and exciton stability [95], and good morphological properties. The ideal host or host mixture should be capable of efficiently transporting both holes and electrons as well as have appropriate energy-level alignment with the adjacent blocking or charge-transporting layers. Since OLEDs are designed for emission of a wide range of spectral regions, the exact requirements of the host material will depend on the spectral demands and the particular emitter combination. Typical host materials contain many of the same moieties as electron and hole transporting moieties but the particular composition will depend on the application. Of note are host materials which contain both electron-transporting and hole-transporting moieties, often called ambipolar hosts, which provide good charge transport for both holes and elections to avoid charge build up at the EML edges. Similarly, using a blend of electron and hole transporting organic molecules in a host mixture, sometimes called exciplex hosts [96], gives the additional freedom to fine-tune the ratio of the electron- and hole-transporting components to achieve charge balance. Due to the wide variety of applications and materials for hosts, the reader should refer to a more comprehensive review [97].

Exciton Formation in Emissive Layers When the anions and cations (electrons and holes) are brought together in the EML to nearby neighboring molecules, they can recombine to form an exciton on one of the molecules through an electron transfer leaving a ground state on the other

Phosphorescent OLEDs for Power-Efficient Displays

15

Fig. 5 Spin states in OLEDs

molecule. This can occur through either an electron transfer from the LUMO of the anion to the LUMO of cation or through the transfer of an electron on the HOMO of the anion to the HOMO of the cation. It is also possible to have an electron transfer from the LUMO of the anion to the HOMO of the cation, but this process is unfavorable due to a large amount of energy needing to be dissipated for such a transfer to occur [98]. This electron transfer results in one molecule returning to its ground state configuration and the other molecule forms an exciton. If this excited state forms between two molecules of offset energy levels an exciplex may form where the excited state is shared between the two molecules [20]. For most PHOLED applications, it is desired to form the exciton on a single emitter molecule. This process of bringing charges together to create excited states generates a statistical mixture of four different electron spin configurations (shown in Fig. 5) with 25% having a net spin of S = 0 (called a singlet) and 75% having a net spin of S = 1 (called a triplet) [8]. Most materials have a singlet ground state since they have all their electrons paired with opposite spin. Since emission of a photon typically requires spin to be conserved, 25% of the excitons created can emit while relaxing from their excited singlet state to the singlet ground state in a process called fluorescence. Decay from the other 75% of the excitons in the triplet (S = 1) excited state is typically a non-radiative process for most organic materials and these excitons are lost as heat. This spin conservation process is the major reason fluorescent OLEDs have limited efficiency. Phosphorescent materials on the other hand incorporate heavy metals which enable them, through strong spin–orbit coupling effects, to undergo a spin–flip process to go between singlet and triplet states [9]. Spin–orbit coupling is a quantum mechanical effect that facilitates the mixing of singlet and triplet character to

16

T. Fleetham and M. S. Weaver

Fig. 6 Exciton formation process

allow conversion between singlet and triplet states. The strength of this interaction is proportional to the atomic number to the fourth power, so it is much more significant for heavy metals [118]. As a consequence of the strong spin–orbit coupling for heavy metal complexes, all the generated excitons formed on the phosphors are rapidly interconverted to triplets due to favorable intersystem crossing. Ultimately, the heavy metal complexes are capable of emitting from these triplet states allowing the excitons to decay radiatively in a process called phosphorescence [9, 99]. Two different mechanisms have been suggested for how excited states can be formed in doped PHOLEDs. As shown in Fig. 6a, charges can be trapped on the dopant molecule and the opposite charge can be transported to this emitter to form the exciton directly on the emitter. This has the advantage of controlling the excited state localization to only those sites where the emission process is preferred to originate. In the second case, Fig. 6b, excitons are formed on the host molecules and the excited state is then energy transferred to the emitter. Singlet excitons can be transferred over several nanometers via a non-radiative dipole coupling called Förster resonance energy transfer (FRET), while triplet excitons must energy transfer by triplet diffusion, hopping molecule by molecule, via a short-range electron exchange mechanism called Dexter energy transfer (Dexter). These two processes can be effective in transferring the energy to the emitter to form an exciton (Fig. 6c) as long as there is sufficient energetic driving force for these energy transfers to occur and high enough doping concentration of the phosphor.

Exciton Decay Once the triplet exciton has been formed on the phosphorescent material, the energy can be transferred to another material, the excited state can relax to the ground state non-radiatively, or, preferably, the excited state can relax radiatively with the emission of a photon. This competition of radiative and non-radiative pathways is a competition

Phosphorescent OLEDs for Power-Efficient Displays

17

of rates. Therefore, important emitter design parameters to consider are PLQY and the emission transient (τ) where ideally the PLQY will be as close to unity as possible and the transient will be as fast as possible as well as tuning the triplet energy to the desired emission color. The competition between the preferred radiative decay from the triplet state of the phosphor and the non-radiative relaxation of the excited state will be discussed in more detail in the following section. Several intermolecular pathways for the exciton to return to the ground state also exist including annihilation events with other excited or charged molecules, energy transfer to another molecule, or charge transfer quenching [81, 100]. One solution to these intermolecular interactions is to add bulky substituents which prevent aggregation or quenching effects, however, these groups can also slow charge transport leading to an increased device operating voltage and lower power efficiency [101, 102].

2.2.6

Light Extraction

Control over the charge injection and transport processes and appropriate selection of a highly efficient phosphor can lead to internal quantum efficiencies which approach 100% [11]. However, these photons created within the device still need to outcouple from the device. The ratio of the photons created to photons emitted is called the outcoupling efficiency (ηOut ). Photons can be lost through self-absorption of the organic layers, waveguiding within the device, surface plasmon modes, and absorption of the photons in the cathode. For bottom-emitting PHOLEDs on glass substrates this results in a typical outcoupling efficiency ηOut = 20–40% [26] with most of the remaining light waveguided in the substrate and the organic layers. The exact outcoupling efficiency requires a sophisticated analysis to account for the coupling of excited states to the device cavity modes [103]. Most of the methods to overcome the outcoupling limitations have focused on expanding the escape cone from the substrate and suppressing the waveguide modes in the organic layers. For example, using rough or textured surfaces [104], mesa structures [105, 106], lenses [107, 108], the use of reflecting surfaces or distributed Bragg reflectors [109, 110], incorporating a film of a scattering medium [111], or even a thin layer of a very low refractive index silica aerogel (ni ~ 1.03) in the device [112]. One of the most promising methods of enhancing the outcoupling efficiency has been the use of an ordered array of microlenses [113]. Despite the enhanced EQE using these outcoupling methods, the higher efficiency is often accompanied by changes in the angular distribution profile of the emission or changes in the emission spectrum with viewing angle. In the case of patterned displays, the blurring of the emitted light between neighboring pixels can significantly distort the image. Nevertheless, these strategies can be highly useful for lighting where pixel definition is less of a concern and the commercialization of lighting technology is driven largely by efficiency performance. The loss of light to waveguiding in the organic layers and substrate is only a portion of the loss pathway. A substantial amount of light is also lost to surface plasmon modes. Surface plasmon polaritons (SPP) arise from oscillations of free

18

T. Fleetham and M. S. Weaver

charges at the cathode–organic interface resulting in an electric field that penetrates into the organic layers of the OLED. The excited states of phosphorescent dopants can also be considered as oscillations in electron density called “transition dipoles.” The excited state can energy transfer into the SPP modes non-radiatively, leading to a loss in emission efficiency [103]. Increasing the ETL thickness can reduce this effect but often leads to increasing losses from waveguiding effects. An elegant solution to reducing this problem is to design emitting molecules that are aligned in the thin film such that their transition dipole moment is poorly aligned for coupling to the SPP modes and can emit efficiently without interacting with this electric field. The molecular design criteria for this strategy focus on both electronic and geometric molecular modifications which can change both the direction of the transition dipole moment and influence the orientation of the molecule within the thin film [114].

3 Properties of Phosphorescent Materials As laid out in Sect. 2, the various layers of an OLED stack can be optimized to reduce the voltage, optimize charge balance, and ensure efficient recombination but ultimately, much of the device performance hinges on the appropriate design of the phosphorescent emitter. The exact design criteria for a phosphor will depend on the specific application, but the important parameters for all phosphor designs focus on color, efficiency, emission transient, electrochemical properties, and stability. Color: One of the most important design parameters is the emission color from the device. For pixelated active matrix displays, this means the design of red, green, and blue monochromatic pixels, whereas for lighting, it is generating devices with broad emission spectrum and uniform emission across the visible spectrum. In a phosphor, the emission color is determined primarily by the energy and molecular orbital parentage of the lowest excited triplet state. Several standards have been established to help quantify and codify the color emitted from the device. From a photophysical standpoint, some common parameters used to describe the emission of the phosphor include emission onset, the peak wavelength emission (λmax ), and the full width at half maximum (FWHM). Due to the simplicity and conciseness of metrics like these, research scientists will often use these parameters to compare properties of materials such as triplet energies or the character of the excited state. The information provided by simple single wavelength metrics is an incomplete representation of the whole emission spectrum and has no consideration for the human perception of the light, both of which are crucial for both display and lighting applications. One of the most common metrics that address these issues have been developed by the Commission Internationale de l’Eclairage which developed a coordinate system and a color space in which perceived color is measured [115]. The metric, which takes into account the entire emission spectrum, uses a set of coordinates (CIEx, CIEy) on a two-dimensional XY plot of all the perceived colors which can be created. This type of standard representation of emission color affords the ability of manufacturing companies or government standard agencies to develop

Phosphorescent OLEDs for Power-Efficient Displays

19

color standards for various technologies. Refinements to the 1931 color standard have been made over the years, for example, the 1976 standard renders a more uniform representation of color. However, the 1931 standard is still the most widely used tool in the display industry. For lighting, additional standards are required since the perception of white lighting is not only the emission from the source (which can be represented by CIE coordinates) but also by the fidelity with which a lighting source reproduces the colors of illuminated objects. The color rendering index (CRI) is the most common scale used to evaluate the ability of white light sources to reproduce the colors of objects in comparison with an ideal or natural light source. The formal standards for CRI are also controlled by the Commission Internationale de l’Eclairage (where the value is called CIE Ra ) and have a scale from 0 to 100 where 100 is a spectrum identical to daylight or any other blackbody radiation source. Another parameter for lighting, often used in lieu of CIE coordinates to describe the color of the emitted light is the correlated color temperature (CCT) which is the temperature of a black body radiation source whose perceived color is the closest to that of the white light source under study. Efficiency: the efficiency of a phosphorescent emitter can be well described by the PLQY which is the ratio of emitted photons to absorbed photons. This metric is based on measurement with optical excitation. However, since the decay from the excited state is roughly the same for an electrically excited molecule, the PLQY is assumed to be a good representation of the efficiency of an EL device. However, since charge-exciton interactions can be deleterious to the efficiency and the exciton densities in an electrically driven device are typically much higher than optically excited films, the PLQY is typically considered the upper bound of ηint . Another major difference between optical and electrical excitation is the exciton formation process where optical excitation could provide a different distribution of excited states among the EML components compared to the excited states formed by direct charge trapping, for example, there could be incomplete energy transfer in optically excited PLQY measurements. Also, the PLQY in dilute solutions or inert polymer matrices could also have very different efficiencies than the PLQY of the EML composition. Emission transient: the emission transient (τ), or excited state lifetime, is a term used to describe the decay time of the excited state on the phosphorescent material. The value for the τ is the time for a population of excited states to decay to 1/e of the initial population. Phosphorescent materials can have τ as low as 1 μs while maintaining high PLQY [133]. It is important to mention here that emission transients are typically inclusive of both non-radiative and radiative decays and are often fast for emitters with high non-radiative decay rates, so it is a good practice to consider PLQY and emission transient together. The emission transients are of crucial importance to both the device efficiency and the device operational lifetime. Since longer τ gives more time for the emitter to decompose, energy transfer, or decay non-radiatively, it is typically desired to reduce the transient to be as low as possible. This is particularly important at high drive currents where the density of excitons and charges can be high resulting in undesirable bimolecular interactions.

20

T. Fleetham and M. S. Weaver

Electrochemical properties: In order to evaluate the charge and exciton dynamics of phosphors in devices, it is important to discuss their energy levels and the relationship of these levels to other molecules in the device. While each molecule has many energy levels, it is often appropriate to simplify the description to just frontier molecular orbitals, the HOMO and LUMO, since these are the most active orbitals for electron transfer processes. These HOMO and LUMO energies can be measured directly in the thin film from methods such as UV photoelectron spectroscopy (UPS) to determine the HOMO level or inverse photoelectron spectroscopy (IPES) to determine the LUMO energies [116]. Alternatively, due to the molecular and chemical nature of these materials, it is often convenient to estimate the HOMO and LUMO indirectly from electrochemical measurements such as cyclic voltammetry or differential pulse voltammetry [57, 117]. These electrochemical measurements can give a precise description of the oxidation and reduction processes occurring on these materials but ignore the effects of interfaces and solid-state effects. Furthermore, the cyclic voltammetry measurements can give some information on the reversibility of the redox reactions which can give insights into electrochemical stability of phosphors. Phosphor stability: There are several methods to probe phosphor stability including chemical stability, electrochemical stability, photostability, etc., each of which can provide some information about the phosphor but ultimately the degradation in a device which contains high local electric fields, large concentrations of charges, and excited states is a complicated system to emulate outside a device. Therefore, the ultimate test of the stability of a phosphor is to carry out operational life testing on test pixels using the phosphorescent emitter. This typically involves fabricating a small pixel or series of pixels and driving them at a constant current density while measuring the luminance loss over time. The metrics used to report operational lifetime are therefore the time to decay to a given percent of initial luminance at a specific driving condition. The percentage luminance loss and device driving conditions used to quantify device lifetime vary wildly in the literature with everything from LT97 (decay to 97% of initial luminance) to LT50 or half-life (decay to 50% of initial luminance) and driving conditions of 1 mA/cm2 to 10 s of mA/cm2 or even initiating the driving condition at the current which has an initial luminance of 1000 cd/m2 . All of these metrics can make comparing device operational lifetimes between reports challenging and often the use of stretched exponential fits or application of acceleration factors is required to interpolate or extrapolate lifetime curves to a comparable metric.

3.1 Photophysical Processes in Phosphorescent Materials One of the key steps in the generation of electroluminescence is the exciton decay process, the step in which a photon may be created from a molecule in its excited state returning to the ground state. Since this process typically concerns neutral molecules in their excited and ground states, it is analogous to the decay processes in photoluminescence and it is easy to visualize with a Jablonski diagram shown in

Phosphorescent OLEDs for Power-Efficient Displays

21

Fig. 7 Jablonski diagram

Fig. 7 [118]. This type of diagram is a simplified schematic representation of the various states of interest and processes of interchanging between them. In a more accurate representation, each state would be a complex potential energy surface all of which contain a large number of vibrational levels. To a first approximation, the photophysical processes of many molecules can be described by a three-state system: the ground state (S0 ), lowest singlet excited state (S1 ), and the lowest triplet excited state (T1 ). In photoluminescence, the excited states are accessed initially by the absorption of a photon (represented by the solid line S0 → S1 ) show in Fig. 7. In an EL process, the population of the S1 will occur by either energy transfer or direct charge recombination as was discussed in Sect. 2.2.4 rather than by the absorption of a photon. The population of higher lying excited states in either photoexcitation or electrical excitation typically thermalize to the lowest excited states S1 . Following the population of the S1 state, the molecule can take several pathways to relax back to the ground state. The excited singlet can radiatively relax to the ground state emitting a photon by fluorescence (represented by the solid line S1 → S0 ). This process can occur quickly, on the order of nanoseconds [118], and many types of fluorescent materials can primarily relax through this process with high PLQY. In some cases, other deactivation pathways are competitive with fluorescence. Non-radiative decay from the S1 (represented by the dashed line S1 → S0 ) is a broad term to describe various thermal deactivation processes, called internal conversion (IC) in which the excited states thermalize through accessing various molecular vibrations to the ground state [118]. One pathway of particular importance to PHOLEDs is the intersystem crossing (ISC) from S1 → T1 . In an EL device, the recombination of charges can also form an excited state in the T1 directly. Relaxation from this excited state can also occur via emission of a photon through phosphorescence (represented by the solid line T1 → S0 ), or by thermal non-radiative deactivation (represented by the dashed line T1 → S0 ). In organic materials, the direct relation from T1 → S0 is symmetry forbidden and consequently very slow with luminescent lifetimes in the range of seconds. Organometallic complexes containing

22

T. Fleetham and M. S. Weaver

a heavy metal such as Ir or Pt, on the other hand, have a more allowed and much faster phosphorescence decay, on the order of microseconds, as a result of strong spin–orbit coupling from these heavy atoms.

3.2 Excited-State Characteristics for Phosphorescent Materials In order to understand the exciton decay process, it is instructive to explore the molecular orbitals involved in the transitions. The excited states of phosphorescent organometallic complexes can have a variety of parentages including ligand-centered (3 LC) transitions, metal-centered (3 MC) transitions, metal-to-ligand charge transfer (1,3 MLCT) transitions, or admixtures between these to name a few. Most phosphorescent materials for OLED purposes, in particular those based on Ir(III) and Pt(II), are comprised of conjugated organic ligands chelated to the metal center through carbon, nitrogen, or oxygen bonds [10]. This typically results in a lowest excited triplet state comprised of 3 LC (from the conjugated ligands) and 1,3 MLCT character involving the metal center in the HOMO as shown schematically in Fig. 8. In Fig. 8, the singlet metal to ligand charge transfer state (1 MLCT) is also shown since these transitions involving the heavy metal have sufficient SOC and are at a low enough in energy to effectively mix with the other triplet admixtures. This mixing of 1 MLCT is largely the reason organometallic phosphors can emit efficiently via phosphorescence with relatively fast exciton decay lifetimes on the order of 1 μs. In the case of 2, 3, 7, 8, 12, 13, 17, 18-octaethyl-12H,23H-porphyrin (PtOEP), the pyrrole rings are relatively easy to oxidize and are hard to reduce. As a result, Fig. 8 Schematic energy level diagram for a typical phosphorescent organometallic complex

Phosphorescent OLEDs for Power-Efficient Displays

23

the 1,3 MLCT transitions are minimally involved relative to the 3 LC which is highly stabilized due to the extended conjugation throughout the ligand. As a result, the exciton decay lifetime of PtOEP is quite slow (90 μs) [119] which is reflective of an excited state with poor spin–orbit coupling that is primarily localized on an organic ligand. Conversely, 3 MC transitions involve primarily heavy metal, which is favorable from an SOC standpoint; however, the radiative transition between two d-orbitals is symmetry forbidden and the population of an 3 MC typically results in the quenching of the emission. Thus, from the standpoint of developing emitters with short excited-state lifetimes and high efficiencies, striking the right balance of involving the metal for high SOC while avoiding accessing the metal-centered states is a key design parameter. Metal complexes with 1,3 MLCT transition involve the d-orbitals in the HOMO (or hole in the excited state) and the π* orbitals of the organic ligands in the LUMO (or electron in the excited state). The natural transition orbitals for Ir(ppy)3 , shown in Fig. 9, illustrate the orbitals involved in an MLCT transition. The hole for the transition has a large density on the metal whereas the electron is primarily localized on the pyridine of the ligand. These d → π* transitions are not symmetry forbidden like 3 MC states but still involve the metal for substantial SOC which allows more admixing of the 1 MLCT into the lowest triplet excited state. The increased involvement of the singlet character can ultimately lead to faster radiative rates and consequently higher PLQY and shorter excited state lifetimes as is the case for Ir(ppy)3 which has a reported transient of 5 μs with nearly 100% PLQY [81]. On the other hand, 1,3 MLCT transitions, have large charge redistributions, are highly solvatochromic, and typically have broad Gaussian emission which are not desirable for certain display applications. As a result, finely tuning the admixture of 3 MLCT into the 3 LC triplet can be an effective strategy to tune the emission wherein the energy, the vibronic structure of the emission spectrum, and the emission decay lifetimes are all interdependent.

Fig. 9 Natural transition orbitals for Ir(ppy)3 [120]

24

T. Fleetham and M. S. Weaver

3.3 Color Tuning of Phosphorescent Emitters In order to span the full visible spectrum, efficient phosphors emitting from the deep blue to deep red portions of the visible spectrum are needed. Lamansky et al., [121] demonstrated that small electronic modifications to the primary chromophoric ligand (where the 3 LC contribution originates) could effectively tune the emission of tris-bidentate Ir(III) complexes from the blue to red spectral regions. In particular, the authors demonstrated color tuning through both the extension of conjugation to modulate the 3 LC and through the use of electron-donating and -withdrawing groups to tune the 1,3 MLCT energies. Phosphorescent-cyclometalated Pt complexes spanning the visible spectrum were also developed by Brooks et al. using a similar color tuning strategy [122]. The strategies for color tuning used by Lamansky and Brooks focuses on either modulating the energy of the ligand (3 LC) itself by breaking or extending the conjugation or by modulating the energy of 3 MLCT transition through the use of heterocycles or substituents which make portions of the molecule either more electron-rich or more electron-deficient. A schematic representation of possible color tuning strategies starting from the prototypical 2-phenylpyridine (ppy) ligand is shown in Fig. 10 where M is the metal center, and only one chromophoric ligand is shown for simplicity since this strategy could apply to complexes of Ir, Pt, or many other metals as part of a heteroleptic or multidentate complex. On the left-hand side of Fig. 10, there are

Fig. 10 Schematic representation of color tuning strategies of phosphorescent materials

Phosphorescent OLEDs for Power-Efficient Displays

25

several strategies to redshift emission from the ppy ligand, which typically forms green-emitting phosphors. The top-left strategy (1) is to lower the LUMO level by making the ligand easier to reduce. Since the MLCT transition typically involves a HOMO localization on the metal and LUMO localization on the ligand, lowering the reduction potential of the pyridine ring will lower the MLCT energy giving a red shift. The structural modifications to accomplish this redshift include benzannulation of the pyridine ring, aza-substitution, or the addition of an electron-withdrawing group. Similarly, the bottom left strategy (2) is to raise the HOMO energy by making either the ligand or the metal atom easier to oxidize, which will lower the energy of the MLCT transition to red shift the emission. This can be achieved by either donating to the metal orbitals (e.g., using a strong donating group para to the metal) or by replacing the six-membered phenyl ring with a five-membered heterocyclic ring such as thiophene since the contraction of the ring destabilizes the energy levels. Finally, the LC energy can also be redshifted by extending the conjugation of the ligand. This can be achieved by either substitution with additional aromatic rings, as shown in the middle left strategy (3), or through fusing aromatic rings to the ligand. The color tuning strategies are similar when blueshifting the emission. The emission can be blueshifted by raising the LUMO level (4) using electron-donating groups or contracting the ring size, lowering the HOMO level (5) using electron-withdrawing groups or aza-substitutions, or by breaking the conjugation (6) by replacing the five-membered chelate with a more weakly conjugated six-membered chelate.

4 Cyclometallated Complexes for Efficient PHOLEDs To overcome the limitations set forth by electrogenerated spin statistics, it is important to develop materials which emit from the lowest excited triplet states. Organometallic materials such as lanthanide complexes [123] or organic phosphors such as benzophenone [124] can emit via phosphorescent pathways, however, the transient decay lifetime of the classes of materials is typically on the order of milliseconds to seconds which make them impractical for EL devices. The first reported PHOLEDs contained the platinum complex PtOEP [9]. This first PHOLED device had an EQE of only 4% which was subsequently increased to 6% with the introduction of a blocking layer within the device structure [32]. These initial reports demonstrated the promise of phosphorescent materials, but, these EQEs are low by today’s standards and this device efficiency could only be realized at low drive currents and luminance levels due to the long-lived nature of the triplet excitons in PtOEP. Following these promising initial reports, a flurry of research began into the use and development of new organometallic complexes containing heavy metal centers. It was soon thereafter that metal complexes using Ir(III) metal centers were first employed demonstrating high EQEs [125, 126]. Due to the ease of synthesis for the bidentate ligands, fast radiative rates, and high PLQYs Ir(III) complexes quickly became the focal point of much of the OLED research for the next several years. Over the past few years, the development of tetradentate ligand designs for Pt(II) complexes have brought this

26

T. Fleetham and M. S. Weaver

class of materials back to the forefront in emitter design. Recently, work on phosphorescent Au(III) [127] and Pd(II) [128] complexes have also started to show some initial promise in efficiency and stability but the emission transients of these classes of materials have slowed their progress compared to their Ir(III) and Pt(II) analogs.

4.1 Ir(III) Complexes Soon after the first PHOLED devices were reported, it was quickly realized that neutral Ir complexes [129], such as Ir(ppy)3 , were capable of emission that was at least an order of magnitude faster and were quickly used in an OLED [99]. As a result of the faster excited-state lifetime for Ir(ppy)3 , EQEs of 30% [130] or more have been realized with less roll-off in efficiency at high current density. In the two decades, since these initial results, a wide variety of Ir(III) complexes have been developed for OLED applications. There are several different ways to arrange ligands around the ocathedral Ir(III) center. Typically for synthetic ease, the six coordination sites to the metal center can be satisfied by three bidentate mono anionic ligands to form a neutral tris bidentate Ir(III) complex. Ligands without anionic bonds like 2,2 bipyridine can also be used. The resulting complex will be charged or will need a dianionic ligand to compensate. From a manufacturing standpoint, neutral complexes are preferred in order to be compatible with thermal evaporation deposition techniques. The tris bidentate Ir(III) complex can contain either three of the same ligand (homoleptic) or have two or three different ligands (heteroleptic). Homoleptic complexes such as those shown in Fig. 11 can be arranged in either a facial geometry, fac-Ir(ppy)3 in Fig. 11e, where the ligands are symmetric about the C3 axis or can be meridional geometry, mer-Ir(ppy)3 in Fig. 11f, where two of the cyclometalated carbons are

Fig. 11 Coordination strategies for neutral Ir(III) complexes

Phosphorescent OLEDs for Power-Efficient Displays

27

trans to each other as are two of the dative bonds. In the facial configuration, all of the ligands are identical by symmetry and this isomer is the thermodynamic product. Heteroleptic complexes, using a monoanionic acetylacetonate (acac) ligand such as (ppy)2 Ir(acac) shown in Fig. 11g, can also be synthesized easily in high yield and can have nearly 100% internal quantum efficiencies [11]. Recently, researchers have also explored multidentate ligands which provide the opportunity for increased rigidity to the coordination of the ligands [131]. Redshifting the emission has been primarily achieved through extending the conjugation and lowering the LUMO relative to the LUMO of pyridine. This strategy is effective in shifting the emission from the green phenyl-pyridine ligands ((ppy)2 Ir(acac) with an emission peak of 520 nm [121]) to yellow using phenylbenzothiophene ligand ((bt)2 Ir(acac) with an emission peak of 557 nm [121]), to orange using phenyl-quinoline ((pq)2 Ir(acac) with an emission peak of 597nm [121]), or red using phenyl-isoquinoline ligands ((piq)2 Ir(acac) with an emission peak of 630 nm [132]). In order to blue shift the emission color several designs of bidentate ligands have been developed as shown in Fig. 12. Iridium complexes using the carbene ligands, 3-methyl-1-phenyl-1H-3λ4 -benzo[d]imidazole (pmb) and 1-methyl-3-phenyl-3H-1 λ4 -imidazo[4,5-b]pyridine (pmp), yielded ultraviolet (390 nm) to deep blue (465 nm) emission due to the strong ligand field of the carbenes which stabilizes the filled Ir(III) d-orbitals and the HOMO level, and the relatively high LUMO levels of the N-heterocyclic carbene. Both the fac and mer isomers of Ir(pmp)3 had high PLQY over 75% in solution at room temperature while the fac-Ir(pmp)3 only had

Fig. 12 Bidentate ligands for Ir(III) complexes spanning the visible spectrum

28

T. Fleetham and M. S. Weaver

a PLQY of 37% likely due to accessible MC states at such high excited state energies [133]. The very high energy triplet state for fac-Ir(pmb)3 would be quenched in typical host materials. However, the slightly lower energy for fac-Ir(pmp)3 and mer-Ir(pmp)3 permitted their use in devices which achieved efficiencies of 9.0 and 13.3% at 1000 cd/m2 respectively [133]. One of the most studied blue Ir(III) complexes are those using difluorophenyl-pyridine pyridine ligands (4,6F2 ppy) such as (4,6F2 ppy)Ir(pic) (FIrpic) with a peak emission of 470 nm and PLQY of nearly 100% [36]. FIrpic is a good candidate for highly efficient blue OLEDs and devices using this emitter have reached over 25% EQE [134]. Nevertheless, reports have demonstrated that FIrpic devices degrade rapidly, due in part to fluorine cleavage from the phenyl ring [135]. Sky-blue devices have also been fabricated with facIr(mpim)3 were also capable of very high EQEs of 29.6% at 100 cd/m2 and 1931 CIE coordinates of (0.18, 0.41) [136]. An alternative strategy to modifying the emissive ligand to tune the emission color is to modify the ligand field strength of the ancillary ligand in heteroleptic complexes. One such study looked at the effect of changing the ancillary ligands for a phenyl-pyridine based emitter with the structures (CˆN)2 Ir(LL’) where CˆN represents 2-phenyl-pyridine-based ligands and LL’ are ancillary ligands [137]. By increasing the ligand field strength of the ancillary ligand, the energetic splitting between the filled and unfilled d-orbitals increases, resulting in a deeper HOMO level on the Ir. This HOMO stabilization results in a blue shift corresponding to a destabilization of the MLCT state with respect to the LC state. As a result, the triplet of the blue-shifted emitters became more ligand centered and consequently had slower excited-state lifetimes. These results illustrate how interconnected all the photophysical properties of phosphorescent emitters can be, and finding the optimal properties is often a delicate balance.

4.2 Pt(II) Complexes In contrast to octahedral Ir(III) complexes, which are mostly focused on tris bidentate coordination schemes, platinum complexes have a square planar coordination geometry with a wide variety of ligand designs. As shown in Fig. 13, the ligands can be monodentate, bidentate, tridentate, tetradentate, or macrocyclic many of which can contain either five- or six-membered chelation rings. This diversity in ligand designs provides structural variables to tune photophysical properties. For the first decade of PHOLED research, Pt complexes were relatively unexplored compared to the wide-ranging research into Ir(III) complexes. Macrocycles and porphyrins, like PtOEP (Fig. 13d) which was used in the first PHOLED [9], are relatively hard to modify synthetically. On the other hand, complexes using bidentate ligands analogous to those used in Ir complexes, are more easily synthesized and modified to span the visible spectrum [121]. However, unlike their Ir(III) analogs, Pt complexes using bidentate ligands often suffered from low PLQY due to potential distortions between the two bidentate ligands in the excited state, particularly

Phosphorescent OLEDs for Power-Efficient Displays

29

Fig. 13 Examples of Pt complex designs

for higher triplet energies where 3 MC states are accessible [121]. Using tridentate ligands rigidifies the molecule affording very high PLQYs and device efficiencies [138]. However, the monodentate ligands, which are typically halides or acetylides, may be cause for concern due to their propensity to be labile or reactive. Thus, from the standpoint of rigidifying the emitter for improved efficiency and operational lifetimes, tetradentate ligands would be preferred. As shown in the bottom half of Fig. 13, there are several design possibilities for how to coordinate tetradentate ligands to the Pt center. These can be classified by the size of their chelation rings which are typically either five- or six-membered chelates. This rich diversity in design possibilities and intense research over the past several years have resulted in very high external quantum efficiencies approaching 40% [25], high spectral purity with narrow emitters [139], and long operational lifetimes [140]. The color tuning strategies for platinum complexes are similar to those discussed for Ir(III) complexes described in Sect. 3.3. For a more in-depth overview of progress in Pt complexes, see reviews by Huo [141], Fleetham [142], or Yam [143] for more details. One unique design constraint relevant to Pt(II) complexes and other square planar emitters is the ability to form aggregate excited states called excimers. Due to their square planar geometry, some Pt(II)s are capable of forming closely spaced pairs or higher order domains of emitting molecules which share sufficient molecular orbital overlap in the excited state giving a redshifted structureless emission [101]. To achieve high color purity red, green, or blue emission this process needs to be discouraged by either the introduction of bulky groups to prevent close packing or through the introduction of a twist into the coordination geometry as is the case for 5–6–6 or 6–6–6 coordination schemes such as those shown in Fig. 13h-i [144, 145]. On the other hand, this broad red-shifted emission could be used as an advantage by providing a complimentary orange or red emission to deep blue-emitting Pt(II) emitters forming a broad white spectrum using only a single emissive dopant. This strategy has resulted in ideal white CIE coordinates with a CRI value over 80 [146], and EQEs exceeding 25% [147]. Similarly, fabricating neat layers of Pt(II) complexes can also provide highly

30

T. Fleetham and M. S. Weaver

efficient emission with excited states that are suspected to extend over larger crystalline domains. Red OLEDs comprising these neat layers of Pt(II) complexes have yielded EQEs approaching 40% as a result of the high PLQY of the neat films and the highly oriented nature of the excited state dipoles [25].

5 Summary and Outlook Progress in phosphorescent OLEDs has advanced tremendously over the past couple of decades. Multilayer device structures with dedicated charge injection, transport, and blocking layers have dramatically improved the efficiency of charge recombination and reduced driving voltages leading to more power-efficient OLEDs. The growing diversity of organic materials for host or transport layers along with a wide variety of designs for efficient Ir(III) and Pt(II) complexes with nearly 100% PLQY have yielded devices with ηInt approaching 100% across the visible spectrum [11, 136, 144]. Furthermore, designs of emitters with reduced energy transfer to SPP modes in the cathode have driven EQEs from −5% [9] for first-generation PHOLEDs to nearly 40% [25] in a standard bottom emitting OLED architecture. Continued efforts to reduce the loss of light to waveguide modes using novel outcoupling strategies have further pushed EQEs to over 60% [12]. This remarkable progress in research and development of PHOLEDs has enabled this technology to become ubiquitous in the marketplace where there is a growing number of applications and commercial products. Despite the remarkable scientific and commercial success for PHOLEDs, thus far, there remains a constant drive for more efficient, brighter, and longer lived devices. Many of the strategies described in this chapter are being used to further improve device efficiencies. Recently, there has been a focus on improving the outcoupling efficiencies, since many state-of-the-art PHOLEDs already exhibit ηInt approaching 100%. Red-, green-, and blue-emitting phosphorescent materials with narrow spectral line shapes are being developed for displays with better color purity and reduced efficiency losses [148]. Meanwhile, broad phosphorescent emitters are needed to develop white lighting with high CRI. To improve operational lifetimes, many research groups are focusing on understanding the intrinsic degradation of the active materials due to inherent charge, exciton, or chemical instabilities. In particular, researchers are studying the interaction between triplets leading to triplet–triplet annihilation (TTA) or triplets and polarons leading to triplet–polaron annihilation (TPA) in devices and how bimolecular events may lead to degradation [82–84]. Useful overviews of the factors affecting device reliability are given by Forrest et al. [149], Popovic et al. [150] and So et al. [151]. Despite the large number of challenges and uncertainties, device operational lifetimes today are orders of magnitude better than the first generation of PHOLEDs, such as those using PtOEP, Ir(ppy)3 , and iridium(III)bis(2-phenylquinolyl-N,C2 )acetylacetonate (PQ2 Ir(acac)) which have demonstrated lifetimes of several thousands of hours [152, 153]. Recent

Phosphorescent OLEDs for Power-Efficient Displays

31

PHOLEDs have now demonstrated lifetimes in excess of many hundreds of thousands of hours at display brightness and are used in the vast majority of OLED products around the world.

References 1. 2. 3. 4. 5. 6. 7. 8. 9.

10.

11. 12.

13. 14. 15.

16. 17. 18. 19.

20.

21.

https://www.lg.com/us/tvs/lg-OLED88Z9PUA-signature-oled-8k-tv https://www.samsung.com/us/mobile/galaxy-fold/ https://www.motorola.com/us/smartphones-razr https://www.lg.com/us/business/oled-displays https://www.lg.com/global/lg-signature/oled-tv-r9 C.W. Tang, S.A. VanSlyke, Organic electroluminescent diodes. Appl. Phys. Lett. 51, 913–915 (1987) S.A. VanSlyke, C.W. Tang, Organic electroluminescent devices having improved power conversion efficiencies. U.S. Patent 4,539,507 (1985). Granted 3 Sept 1985 M.A. Baldo, D.F. O’Brien, M.E. Thompson, S.F. Forrest, The excitonic singlet–triplet ratio in a semiconducting organic thin film. Phys. Rev. B 60, 14422–14428 (1999) M.A. Baldo, D.F. O’Brien, Y. You, A. Shoustikov, S. Sibley, M.E. Thompson, S.R. Forrest, Highly efficient phosphorescent emission from organic electroluminescent devices. Nature 395, 151–154 (1998) H. Yersin, A.F. Rausch, R. Czerwieniec, T. Hofbeck, T. Fischer, The triplet state of organotransition metal compounds: triplet harvesting and singlet harvesting for efficient OLEDs. Coord. Chem. Rev. 255, 2622–2652 (2011) C. Adachi, M.A. Baldo, M.E. Thompson, S.R. Forrest, Nearly 100% internal phosphorescence efficiency in an organic light emitting device. J. Appl. Phys. 90, 5048–5051 (2001) Z.B. Wand, M.G. Helander, J. Qiu, D.P. Puzzo, M.T. Greiner, Z.M. Hudson, S. Wang, Z.W. Liu, Z.H. Lu, Unlocking the full potential of organic light-emitting diodes on flexible plastic. Nat. Photon. 5, 753–757 (2011) R.J. Holmes, S.R. Forrest, Saturated deep blue organic electrophosphorescence using a fluorine-free emitter. Appl. Phys. Lett. 87, 243507 (2005) A. Zampetti, A. Minotto, F. Cacialli, Near-Infrared (NIR) Organic Light-Emitting Diodes (OLEDs): challenges and opportunities. Adv. Funct. Mater. 29, 1807623 (2019) H. Li, L. Duan, Y. Sun, D. Zhang, L. Wang, Y. Qiu, Study of the hole and electron transport in amorphous 9,10-Di-(2 -naphthyl)anthracene: the first-principles approach. J. Phys. Chem. C 117, 16336–16342 (2013) W. Brutting, S. Berleb, A.G. Muckl, Device physics of organic light-emitting diodes based on molecular materials. Org. Electron. 2, 1–36 (2001) C. Hosokawa, H. Tokailin, H. Higashi, T. Kusumoto, Transient behaviour of thin film electroluminescence. Appl. Phys. Lett. 60, 1220–1222 (1992) K-C Tang, K.L. Liu, I-C. Chen, Rapid intersystem crossing in highly phosphorescent iridium complexes. Chem. Phys. Lett. 386, 437–441 (2004) B.W. D’Andrade, J. Brooks, V. Adamovich, M.E. Thompson, S.R. Forrest, White emission using triplet excimers in electrophosphorescent organic light-emitting devices. Adv. Mater. 14, 1032–1036 (2002) K. Itano, H. Ogawa, Y. Shirota, Exciplex formation at the organic solid-state interface: yellow emission in organic light-emitting diodes using green fluorescent tris(8quinolinolato)aluminum and hole-transporting molecular materials with low ionization potentials. Appl. Phys. Lett. 72, 636–638 (1998) C.W. Tang, S.A. VanSlyke, C.H. Chen, Electroluminescence of doped organic thin films. J. Appl. Phys. 65(9), 3610–3616 (1989)

32

T. Fleetham and M. S. Weaver

22. D.Y. Kondakov, T.D. Pawlik, T.K. Hatwar, J.P. Spindler, Triplet annihilation exceeding spin statistical limit in highly efficient fluorescent organic light-emitting diodes. J. Appl. Phys. 106, 124510 (2009) 23. H. Uokyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Highly efficient organic lightemitting diodes from delayed fluorescence. Nature 492, 234–238 (2012) 24. M.H. Lu, J.C. Sturm, Optimization of external coupling and light emission in organic lightemitting devices: modelling and experiment. J. Appl. Phys. 91, 595–604 (2002) 25. K.-H. Kim, J.-L. Liao, S.W. Lee, B. Sim, C.-K. Moon, G.-H. Lee, H.J. Kim, Y. Chi, J.-J. Kim, Crystal organic light-emitting diodes with perfectly oriented non-doped pt-based emitting layer. Adv. Mater. 28, 2526–2532 (2016) 26. S.H. Shin, J.-H. Lee, C.-K. Moon, J.-S. Huh, B. Sim, J.-J. Kim, Sky-blue phosphorescent OLEDs with 34.1% external quantum efficiency using a low refractive index electron transporting layer. Adv. Mater. 28, 4920–4925 (2016) 27. A. Bernanose, Electroluminescence of organic compounds. Br. J. Appl. Phys. (Suppl. 4), S54–S55 (1955) 28. M. Pope, H.P. Kallmann, P.J. Magnante, Electroluminescence in organic crystals. J. Chem. Phys. 38, 2042–2043 (1963) 29. P.S. Vincett, W.A. Barlow, R.H. Hann, G.G. Roberts, Electrical conduction and low voltage blue electroluminescence in vacuum-deposited organic films. Thin Solid Films 94, 171–183 (1982) 30. M. Pfeiffer, S.R. Forrest, K. Leo, M.E. Thompson, Electrophosphorescent p-i-n organic lightemitting devices for very-high-efficiency flat-panel displays. Adv. Mater. 14, 1633–1636 (2002) 31. C. Adachi, S. Tokito, T. Tsutsui, S. Saito, Electroluminescence in organic films with three-layer structure. Jpn. J. Appl. Phys. 27(Part 2), L269–L271 (1988) 32. D.F. O’Brien, M.A. Baldo, M.E. Thompson, S.R. Forrest, Improved energy transfer in electrophosphorescent devices. Appl. Phys. Lett. 74, 442–444 (1999) 33. N. Chopra, J.S. Swensen, E. Polikarpov, L. Cosimbescu, F. So, A.B. Padmaperuma, High efficiency and low roll-off blue phosphorescent organic light-emitting devices using mixed host architecture. Appl. Phys. Lett. 97, 033304 (2010) 34. E.B. Grapper, in Handbook of Thin Film Process Technology, 1st ed., ed. by D.A. Glocker, S.I. Shah (IOP Publishing, Bristol, 1995), pp. A1.0:1–A1.1:7 35. A. Böhler, S. Dirr, H.-H. Johannes, D. Ammermann, W. Kowalsky, Influence of the process vacuum on the device performance of organic light-emitting diodes. Synth. Met. 91, 95–97 (1997) 36. H. Yamamoto, J. Brooks, M.S. Weaver, J.J. Brown, T. Murakami, H. Murata, Improved initial drop in operational lifetime of blue phosphorescent organic light emitting device fabricated under ultra high vacuum condition. Appl. Phys. Lett. 99, 033301 (2011) 37. H. Yamamoto, C. Adachi, M.S. Weaver, J.J. Brown, Identification of device degradation positions in multi-layered phosphorescent organic light emitting device using water probes. Appl. Phys. Lett. 100, 183306 (2012) 38. Y. Yang, A.J. Heeger, Polyaniline as a transparent electrode for polymer light-emitting diodes: lower operating voltage and higher efficiency. Appl. Phys. Lett. 64, 1245–1247 (1994) 39. H. Riel, S. Karg, T. Beierlein, B. Ruhstaller, W. Riess, Phosphorescent top-emitting organic light-emitting devices with improved light outcoupling. Appl. Phys. Lett. 82, 466–468 (2002) 40. S. Tokito, K. Nada, Y. Taga, Metal oxides as a hole-injecting layer for an organic electroluminescent device. J. Phys. D: Appl. Phys. 29, 2750–2753 (1996) 41. K. Furukawa, Y. Terasaka, H. Ueda, M. Matsumura, Effect of a plasma treatment of ITO on the performance of organic electroluminescent devices. Synth. Met. 91, 99–101 (1997) 42. M.G. Mason, L.S. Hung, C.W. Tang, S.T. Lee, K.W. Wong, M. Wang, Characterization of treated indium–tin-oxide surfaces used in electroluminescent devices. J. Appl. Phys. 86, 1688– 1692 (1999) 43. Z.B. Deng, X.M. Ding, S.T. Lee, W.A. Gambling, Enhanced brightness and efficiency in organic electroluminescent devices using SiO2 buffer layers. Appl. Phys. Lett. 74, 2227–2229 (1999)

Phosphorescent OLEDs for Power-Efficient Displays

33

44. I.-M. Chan, T.-Y. Hsu, F.C. Hong, Enhanced hole injections in organic light emitting devices by depositing nickel oxide on indium tin oxide anode. Appl. Phys. Lett. 81, 1899–1901 (2002) 45. C.C. Wu, C.I. Wu, J.C. Sturm, A. Kahn, Surface modification of indium tin oxide by plasma treatment: an effective method to improve the efficiency, brightness, and reliability of organic light emitting devices. Appl. Phys. Lett. 70, 1348–1350 (1997) 46. P.E. Burrows, G.L. Graff, M.E. Gross, P.M. Martin, M. Hall, E. Mast, C. Bonham, W. Bennet, L. Michalski, M.S. Weaver, J.J. Brown, D. Fogarty, L.S. Sapochak, Gas permeation and lifetime tests on polymer-based barrier coatings, in Proceedings of the International Society for Optical Engineering, vol. 4105 (San Diego, 2000), pp. 75–83 47. E.I. Haskal, A. Curioni, P.F. Seidler, W. Andreoni, Lithium–aluminum contacts for organic light-emitting devices. Appl. Phys. Lett. 71, 1151–1153 (1997) 48. D.G. Moon, R.B. Pode, C.J. Lee, J.I. Han, Failure of the top emission organic light-emitting device with a Ca/Ag semitransparent cathode. Synth. Met. 146, 63–68 (2004) 49. L.S. Hung, C.W. Tang, M.G. Mason, P. Raychaudhuri, J. Madathil, Application of an ultrathin LiF/Al bilayer in organic surface-emitting diodes. Appl. Phys. Lett. 78, 544–546 (2001) 50. V. Bulovic, G. Gu, P.E. Burrows, S.R. Forrest, M.E. Thompson, Transparent light-emitting devices. Nature 380, 29 (1996) 51. G. Parthasarathy, P.E. Burrows, V. Khalfin, V.G. Kozlov, S.R. Forrest, A metal-free cathode for organic semiconductor devices. Appl. Phys. Lett. 72, 2138–2140 (1998) 52. G. Gu, V. Bulovic, P.E. Burrows, S.R. Forrest, Transparent organic light emitting devices. Appl. Phys. Lett. 68, 2606–2608 (1996) 53. V. Bulovic, P. Tian, P.E. Burrows, M.R. Gokhale, S.R. Forrest, M.E. Thompson, A surfaceemitting vacuum-deposited organic light emitting device. Appl. Phys. Lett. 70, 2954–2956 (1997) 54. J. Kalinowski, in Organic Electroluminescent Materials and Devices, ed. by S. Miyata, H.S. Nalwa (Gordon & Breach, Amsterdam, 1997), pp. 1–72 55. N.C. Greenham, R.H. Friend, Semiconductor device physics of conjugated polymers. Solid State Phys. 49, 2–149 (1995) 56. K. Sugiyama, H. Ishii, Y. Ouchi, Dependence of indium–tin–oxide work function on surface cleaning method as studied by ultraviolet and x-ray photoemission spectroscopies. J. Appl. Phys. 87, 295 (2000) 57. B.W. D’Andrade, S. Data, S.R. Forrest, P. Djurovich, E. Polikarpov, M.E. Thompson, Relationship between the ionization and oxidation potentials of molecular organic semiconductors. Org., Electon. 6, 11–20 (2005) 58. S.A. VanSlyke, C.H. Chen, C.W. Tang, Organic electroluminescent devices with improved stability. Appl. Phys. Lett. 69, 2160–2162 (1996) 59. Y. Cao, G. Yu, C. Zhang, R. Menon, A.J. Heeger, Polymer light-emitting diodes with polyethylene dioxythiophene–polystyrene sulfonate as the transparent anode. Synth. Met. 87, 171 (1997) 60. A. Gyoutoku, S. Hara, T. Komatsu, M. Shirinashihama, H. Iwanaga, K. Sakanoue, An organic electroluminescent dot-matrix display using carbon underlayer. Synth. Met. 91, 73–75 (1997) 61. J.X. Tang, Y.Q. Li, L.R. Zheng, L.S. Hung, Anode/organic interface modification by plasma fluorocarbon films. J. Appl. Phys. 95, 4397–4403 (2004) 62. H. Sasabe, H. Nakanishi, Y. Watanabe, S. Yano, M. Hirasawa, Y.-J. Pu, J. Kido, Extremely low operating voltage green phosphorescent organic light-emitting devices. Adv. Funct. Mater. 23, 5550–5555 (2013) 63. K.S. Yook, S.O. Jeon, J.Y. Lee, Efficient hole injection by doping of hexaazatriphenylene hexacarbonitrile in hole transport layer. Thin Solid Films 517, 6109–6111 (2009) 64. P.M. Borsenberger, D.S. Weiss, Organic Photoreceptors for Imaging Systems (Marcel Dekker, New York, 1993). 65. P.E. Burrows, Z. Shen, V. Bulovic, D.M. McCarty, S.R. Forrest, J.A. Cronin, M.E. Thompson, Relation between electroluminescence and current transport in organic heterojunction lightemitting devices. J. Appl. Phys. 79, 7991–8006 (1996)

34

T. Fleetham and M. S. Weaver

66. M. Stolka, J.F. Yanus, D.M. Pai, Hole transport in solid solutions of a diamine in polycarbonate. J. Phys. Chem. 88, 4707–4714 (1984) 67. P.M. Borsenberger, L. Pautmeier, R. Richert, H. Bässler, Hole transport in 1,1-bis(di-4tolylaminophenyl)cyclohexane. J. Chem. Phys. 94, 8276 (1991) 68. J. Blochwitz, M. Pfeiffer, M. Hofman, K. Leo, Non-polymeric OLEDs with a doped amorphous hole transporting layer and operating voltages down to 3.2 V to achieve 100 cd/m2 . Synth. Met. 127, 169–173 (2002) 69. Y. Sun, S.R. Forrest, High-efficiency white organic light emitting devices with three separate phosphorescent emission layers. Appl. Phys. Lett. 91, 263503 (2007) 70. H. Nakanotani, K. Masui, J. Nishide, T. Shibata, C. Adachi, Promising operational stability of high-efficiency organic light-emitting diodes based on thermally activated delayed fluorescence. Sci. Rep. 3, 2127 (2013) 71. S.O. Jeon, S.E. Jang, H.S. Son, J.Y. Lee, External quantum efficiency above 20% in deep blue phosphorescent organic light-emitting diodes. Adv. Mater. 23, 1436–1441 (2011) 72. L.S. Hung, C.W. Tang, M.G. Mason, Enhanced electron injection in organic electroluminescent devices using an Al/LiF electrode. Appl. Phys. Lett. 70, 152–155 (1997) 73. C. Schmitz, H. Schmidt, M. Thelakkat, Lithium−quinolate complexes as emitter and interface materials in organic light-emitting diodes. Chem. Mater. 12, 3012–3019 (2000) 74. X. Cai, A.B. Padmaperuma, L.S. Sapochak, P.A. Vecchi, P.E. Burrows, Electron and hole transport in a wide bandgap organic phosphine oxide for blue electrophosphorescence. Appl. Phys. Lett. 92, 083308 (2008) 75. H.F. Chen, S.-J. Yang, Z.-H. Tsai, W.-Y. Hung, T.-C. Wang, K.-T. Wong, 1,3,5-Triazine derivatives as new electron transport–type host materials for highly efficient green phosphorescent OLEDs. J. Mater. Chem. 19, 8112–8118 (2009) 76. K. Togashi, S. Nomura, N. Yokoyama, T. Yasuda, C. Adachi, Low driving voltage characteristics of triphenylene derivatives as electron transport materials in organic light-emitting diode. J. Mater. Chem. 22, 20689–20695 (2012) 77. D. Yokoyama, H. Sasabe, Y. Furukawa, C. Adachi, J. Kido, Molecular Stacking Induced by Intermolecular C-H···N hydrogen bonds leading to high carrier mobility in vacuum-deposited organic films. Adv. Funct. Mater. 21, 1375–1382 (2011) 78. N. Chopra, J. Lee, Y. Zheng, S.-H. Eom, J. Xue, F. So, Effect of the charge balance on highefficiency blue-phosphorescent organic light-emitting diodes. ACS Appl. Mater. Interfaces. 1, 1169–1172 (2009) 79. L. Xiao, S.-J. Su, Y. Agata, H. Lan, J. Kido, Nearly 100% internal quantum efficiency in an organic blue-light electrophosphorescent device using a weak electron transporting material with a wide energy gap. Adv. Mater. 21, 1271–1274 (2009) 80. D. Tanaka, T. Takeda, T. Chiba S. Watanabe, J. Kido, Novel electron-transport material containing boron atom with a high triplet excited energy level. Chem. Lett. 36, 262–263 (2007) 81. Y. Kawamura, K. Goushi, J. Brooks, J.J. Brown, H. Sasabe, C. Adachi, 100% phosphorescence quantum efficiency of Ir(III) complexes in organic semiconductor films. Appl. Phys. Lett. 86, 071104 (2005) 82. N.C. Giebink, B.W. D’Andrade, M.S. Weaver, J.J. Brown, S.R. Forrest, Direct evidence for degradation of polaron excited states in organic light emitting diodes. J. Appl. Phys. 105, 124514 (2009) 83. N.C. Giebink, B.W. D’Andrade, M.S. Weaver, P.B. Mackenzie, J.J. Brown, M.E. Thompson, S.R. Forrest, Intrinsic luminance loss in phosphorescent small-molecule organic light emitting devices due to bimolecular annihilation reactions. J. Appl. Phys. 103, 044509 (2008) 84. N.C. Giebink, S.R. Forrest, Quantum efficiency roll-off at high brightness in fluorescent and phosphorescent organic light emitting diodes. Phys. Rev. B 77, 235215 (2008) 85. T. Sano, H. Fijii, Y. Nishio, Y. Hamada, H. Takahashi, K. Shibata, Organic electroluminescent devices doped with condensed polycyclic aromatic compounds. Synth. Met. 91, 27–30 (1997) 86. Y. Hamada, T. Sano, K. Shibata, K. Kuroki, Influence of the emission site on the running durability of organic electroluminescent devices. Jpn. J. Appl. Phys. 34, L824–L826 (1995)

Phosphorescent OLEDs for Power-Efficient Displays

35

87. Y.Y. Yimer, P.A. Bobbert, R. Coehoorn, Charge transport in disordered organic host–guest systems: effects of carrier density and electric field. Synth. Met. 159, 2399–2401 (2009) 88. D.-G. Ha, J.-J. Kim, M.A. Baldo, Link between hopping models and percolation scaling laws for charge transport in mixtures of small molecules. AIP Adv. 6, 045221 (2016) 89. S. Sanderson, B. Philippa, G. Vamvounis, P.L. Burn, R.D. White, Understanding charge transport in Ir(ppy)3:CBP OLED films. J. Chem. Phys. 150, 094110 (2019) 90. N.C. Erickson, R.J. Holmes, Investigating the role of emissive layer architecture on the exciton recombination zone in organic light-emitting devices. Adv. Funct. Mater. 23, 5190–5198 (2013) 91. C. Adachi, T. Tsutsui, S. Saito, Blue light-emitting organic electroluminescent devices. Appl. Phys. Lett. 56, 799–801 (1989) 92. B. Chen, X.H. Zhang, X.Q. Lin, H.L. Kwong, N.B. Wong, C.S. Lee, W.A. Gambling, S.T. Lee, Improved color purity and efficiency of blue organic light-emitting diodes via suppression of exciplex formation. Synth. Met. 118, 193–196 (2001) 93. S.W. Yin, Z. Shuai, Y. Wang, A quantitative structure–property relationship study of the glass transition temperature of OLED materials. J. Chem. Inf. Comput. Sci. 43, 970–977 (2003) 94. D.F. O’Brien, P.E. Burrows, S.R. Forrest, B.E. Koene, D.E. Loy, M.E. Thompson, Hole transporting materials with high glass transition temperatures for use in organic light-emitting devices. Adv. Mater. 10, 1108–1112 (1998) 95. H. Aziz, Z.D. Popovic, N.-X. Hu, A.-M. Hor, G. Xu, Degradation mechanism of small molecule based organic light emitting diodes. Science 283, 1900–1902 (1999) 96. Q. Wang, Q.-S. Tian, Y.-L. Zhang, X. Tang, L.-S. Liao, High-efficiency organic light-emitting diodes with exciplex hosts. J. Mater. Chem. C 7, 11329–11360 (2019) 97. Y. Tao, C. Yang, J. Qin, Organic host materials for phosphorescent organic light-emitting diodes. Chem. Soc. Rev. 40, 2943–2970 (2011) 98. R.A. Marcus, On the theory of chemiluminescent electron-transfer reactions. J. Chem. Phys. 43, 2654–2657 (1965) 99. M.A. Baldo, S. Lamansky, P.E. Burrows, M.E. Thompson, S.R. Forrest, Very high-efficiency green organic light-emitting devices based on electrophosphorescence. Appl. Phys. Lett. 75, 4–6 (1999) 100. V.E. Choong, Y. Park, N. Shivaparan, C.W. Tang, Y. Gao, Deposition-induced photoluminescence quenching of tris-(8-hydroxyquinoline) aluminum. Appl. Phys. Lett. 71, 1005–1007 (1997) 101. V. Adamovich, J. Brooks, A. Tamayo, A.M. Alexander, P.I. Djurovich, B.W. D’Andrade, C. Adachi, S.R. Forrest, M.E. Thompson, High efficiency single dopant white electrophosphorescent light emitting diodes. New J. Chem. 26, 1171–1178 (2002) 102. Y. Wang, N. Herron, V.V. Grushin, D. LeCloux, V. Petrov, Highly efficient electroluminescent materials based on fluorinated organometallic iridium compounds. Appl. Phys. Lett. 79, 449– 451 (2001) 103. W. Brutting, J. Frischeisen, T. Schmidt, B.J. Scholz, C. Mayr, Device efficiency of organic light-emitting diodes: Progress by improved outcoupling. Phys. Status Solidi, A 210(1), 44–65 (2013) 104. V. Bulovic, V.B. Khalfin, G. Gu, P.E. Burrows, D.Z. Garbuzov, S.R. Forrest, Weak microcavity effects in organic light-emitting devices. Phys. Rev. B 58, 3730–3740 (1998) 105. S.R. Forrest, P.E. Burrows, D.Z. Garbuzov, Displays having mesa pixel configuration. U.S. Patent 6,091,195. Granted 18 July 2000 106. B.J. Matterson, J.M. Lupton, A.F. Safonov, M.G. Salt, W.L. Barnes, I.D.W. Samuel, Increased efficiency and controlled light output from a microstructured light-emitting diode. Adv. Mater. 13, 123–127 (2001) 107. S.H. Fan, P.R. Villeneuve, J.D. Joannopoulos, E.F. Schubert, High extraction efficiency of spontaneous emission from slabs of photonic crystals. Phys. Rev. Lett. 78, 3294–3297 (1997) 108. M. Boroditsky, R. Vrijen, T.F. Krauss, R. Coccioli, R. Bhat, E. Yablonovitch, Spontaneous emission extraction and Purcell enhancement from thin-film 2-D photonic crystals. J. Lightwave Technol. 17, 2096–2112 (1999)

36

T. Fleetham and M. S. Weaver

109. I. Schnitzer, E. Yablonovitch, C. Caneau, T.J. Gmitter, A. Scherer, 30% external quantum efficiency from surface textured, thin-film light-emitting diodes. Appl. Phys. Lett. 63, 2174– 2176 (1993) 110. R.H. Haitz, Light-emitting diode with diagonal faces. U.S. Patent 5,087,949. Granted 11 Feb 1992 111. T. Yamasaki, K. Sumioka, T. Tsutsui, Organic light-emitting device with an ordered monolayer of silica microspheres as a scattering medium. Appl. Phys. Lett. 76, 1243–1245 (2000) 112. T. Tsutsui, M. Yahiro, H. Yokogawa, K. Kawano, M. Yokoyama, Doubling coupling-out efficiency in organic light-emitting devices using a thin silica aerogel layer. Adv. Mater. 13, 1149–1152 (2001) 113. S. Möller, S.R. Forrest, Improved light out-coupling in organic light emitting diodes employing ordered microlens arrays. J. Appl. Phys. 91, 3324–3327 (2001) 114. K.-H. Kim, J.-J. Kim, Origin and control of orientation of phosphorescent and TADF dyes for high-efficiency OLEDs. Adv. Mater. 30, 42 (2018) 115. Commission Internationale de l’Eclairage, in Proceedings of the 8th Session, Cambridge, 1931 (Cambridge University Press, London, 1932) 116. J. Sworakowski, J. Lipi´nski, K. Janus, On the reliability of determination of energies of HOMO and LUMO levels in organic semiconductors from electrochemical measurements: a simple picture based on the electrostatic model. Org. Electron. 33, 300–310 (2016) 117. P.L. Djurovich, E.L. Mayo, S.R. Forrest, M.E. Thompson, Measurement of the lowest unoccupied molecular orbital energies of molecular organic semiconductors. Org. Electron. 10, 515–520 (2009) 118. N.J. Turro, Modern Molecular Photochemistry (University Science Books, Mill Valley, CA, 1991). 119. D.B. Papkovsky, New oxygen sensors and their application to biosensing. Sensor. Actuat. B-Chem. 29, 213–218 (1995) 120. Natural transition orbitals for fac-Ir(ppy)3 from TDDFT. The molecule was optimized in the ground state (S0) with B3LYP functionals and 6–31G(d) basis set using Gaussian 16 Revision B.01 121. S. Lamansky, P. Djurovich, D. Murphy, F. Abdel-Razzaq, H.-E. Lee, C. Adachi, P.E. Burrows, S.R. Forrest, M.E. Thompson, Highly phosphorescent bis-cyclometalated iridium complexes: synthesis, photophysical characterization, and use in organic light emitting diodes. J. Am. Chem. Soc. 123, 4304–4312 (2001) 122. J. Brooks, Y. Babayan, S. Lamansky, P.I. Djurovich, I. Tsybe, R. Bau, M.E. Thompson, Synthesis and characterization of phosphorescent cyclometalated platinum complexes. Inorg. Chem. 41, 3055–3066 (2002) 123. O. Moudam, B.C. Rowan, M. Alamiry, P. Richardson, B.S. Richards, Europium complexes with high total photoluminescence quantum yields in solution and in PMMA. Chem. Commun. 43, 6649–6651 (2009) 124. S. Hoshino, H. Suzuki, Electroluminescence from triplet excited states of benzophenone. Appl. Phys. Lett. 69, 224–226 (1996) 125. M. Ikai, S. Tokito, Y. Sakamoto, T. Suzuki, Y. Taga, Highly efficient phosphorescence from organic light-emitting devices with an exciton-block layer. Appl. Phys. Lett. 79, 156–158 (2001) 126. C. Adachi, R.C. Kwong, P. Djurovich, V. Adamovich, M.A. Baldo, M.E. Thompson, S.R. Forrest, Endothermic energy transfer: a mechanism for generating high-energy phosphorescent emission in organic materials. Appl. Phys. Lett. 79, 2082–2084 (2001) 127. M-C. Tang, C-H. Lee, S-L. Lai, M. Ng, M-Y. Chan, V.W-W. Yam, Versatile design strategy for highly luminescent vacuum-evaporable and solution-processable tridentate gold(III) complexes with monoaryl auxiliary ligands and their applications for phosphorescent organic light emitting devices. J. Am. Chem. Soc. 139, 9341–9349 (2017) 128. T. Fleetham, Y. Ji, L. Huang, T.S. Fleetham, J. Li, Efficient and stable single-doped white OLEDs using a palladium-based phosphorescent excimer. Chem. Sci. 8, 7983–7990 (2017)

Phosphorescent OLEDs for Power-Efficient Displays

37

129. K.A. King, P.J. Spellane, R.J. Watts, Excited-state properties of a triply ortho-metalated iridium(III) complex. J. Am. Chem. Soc. 107, 1431–1432 (1985) 130. H. Sasabe, T. Chiba, S.-J. Su, Y.-J. Pu, K.-I. Nakayama, J. Kido, 2-Phenylpyrimidine skeletonbased electron-transport materials for extremely efficient green organic light-emitting devices. Chem. Comm. 44, 5821–5823 (2008) 131. A.J. Wilkinson, H. Puschmann, J.A.K. Howard, C.E. Foster, J.A.G. Williams, Luminescent complexes of Iridium(III) containing N∧C∧N-Coordinating terdentate ligands. Inorg. Chem. 45, 8685–8699 (2006) 132. Y.-J. Su, H.-L. Huang, C.-L. Li, C.-H. Chien, Y.-T. Tao, P.-T. Chou, S. Datta, R.-S. Liu, Highly efficient red electrophosphorescent devices based on iridium isoquinoline complexes: remarkable external quantum efficiency over a wide range of current. Adv. Mater. 15, 884–888 (2003) 133. J. Lee, H.-F. Chen, T. Batagoda, C. Cobern, P.I. Djurovich, M.E. Thompson, S.R. Forrest, Deep blue phosphorescent organic light-emitting diodes with very high brightness and efficiency. Nat. Mater. 15, 92–98 (2016) 134. H. Liu, G. Cheng, D. Hu, F. Shen, Y. Lu, G. sun, B. Yang, P. Lu, Y. Ma, A highly efficient, bluephosphorescent device based on a wide-bandgap host/firpic: rational sign of the carbazole and phosphine oxide moieties on tetraphenylsilane. Adv. Funct. Mater. 22, 2830–2836 (2012) 135. V. Sivasubramaniam, F. Brodkorb, S. Hanning, H.P. Loebl, V. Van Elsbergen, H. Boerner, U. Scherf, M. Kreyenschmidy, Fluorine cleavage of the light blue heteroleptic triplet emitter FIrpic. J. Fluorine Chem. 130, 640–649 (2009) 136. K. Udagawa, H. Sasabe, J. Kido, Low-driving-voltage blue phosphorescent organic lightemitting devices with external quantum efficiency of 30%. Adv. Mater. 26, 5062–5066 (2014) 137. J. Li, P.I. Djurovich, B.D. Alleyne, M. Yousufuddin, N.N. Ho, J.C. Thomas, J.C. Peters, R. Bau, M.E. Thompson, Synthetic control of excited-state properties in cyclometalated Ir(III) complexes using ancillary ligands. Inorg. Chem. 44, 1713–1727 (2005) 138. J. Kalinowski, V. Fattori, M. Cocchi, J.A.G. Williams, Light-emitting devices based on organometallic platinum complexes as emitters. Coord. Chem. Rev. 255, 2401–2425 (2011) 139. G. Li, T. Fleetham, E. Turner, X.-C. Hand, J. Li, Highly Efficient and Stable Narrow-Band Phosphorescent Emitters for OLED Applications. Adv. Opt. Mater. 3, 390–397 (2015) 140. Z.-Q. Zhu, K. Klimes, S. Holloway, J. Li, Efficient cyclometalated platinum(II) complex with superior operational stability. Adv. Mater. 29, 1605002 (2017) 141. S. Huo, J. Carroll, D.A.K. Vezzu, Design, synthesis, and applications of highly phosphorescent cyclometalated platinum complexes. Asian J. Org. Chem. 4, 1210–1245 (2015) 142. T. Fleetham, G. Li, J. Li, Phosphorescent Pt(II) and Pd(II) complexes for efficient, high-colorquality, and stable OLEDs. Adv. Mater. 29, 1601861 (2017) 143. V.W.-W. Yam, A.S.-Y. Law, Luminescent D8 metal complexes of platinum(II) and gold(III): from photophysics to photofunctional materials and probes. Coord. Chem. Rev. 414, 213298 (2020) 144. E. Turner, N. Bakken, J. Li, Cyclometalated platinum complexes with luminescent quantum yields approaching 100%. Inorg. Chem. 52, 7344–7351 (2013) 145. T.B. Fleetham, L. Huang, K. Klimes, J. Brooks, J. Li, Tetradentate Pt(II) complexes with 6-membered chelate rings: a new route for stable and efficient blue organic light emitting diodes. Chem. Mater. 28, 3276–3282 (2016) 146. T. Fleetham, J. Ecton, N. Bakken, Z. Wang, J. Li, Single-doped white organic light-emitting device with an external quantum efficiency over 20%. Adv. Mater. 25, 2573–2576 (2013) 147. T. Fleetham, L. Huang, J. Li, Tetradentate platinum complexes for efficient and stable excimerbased white OLEDs. Adv. Funct. Mater. 24, 6066–6073 (2014) 148. E.A. Margulies, P.-L. Boudreault, V.I. Adamovich, B.D. Alleyne, M.S. Weaver, J.J. Brown, “Narrow spectrum deep red emitter for OLED lighting and display” Society for Information Display Digest of Technical Papers, L, II, pp. 911–913 (2019) 149. S.R. Forrest, P.E. Burrows, M.E. Thompson, in Organic Electroluminescent Materials and Devices, ed. by S. Miyata, H.S. Nalwa. (Gordon & Breach, Amsterdam, 1997), pp. 447–453

38

T. Fleetham and M. S. Weaver

150. Z.D. Popovic, H. Aziz, Reliability and degradation of small molecule-based organic lightemitting devices (OLEDs). IEEE J. Sel. Top. Quant. Electron. 8, 362–371 (2002) 151. F. So, D. Kondakov, Degradation mechanisms in small-molecule and polymer organic lightemitting diodes. Adv. Mat. 22, 3762–3777 (2010) 152. P.E. Burrows, S.R. Forrest, T.X. Zhou, L. Michalski, Operating lifetime of phosphorescent organic light emitting devices. Appl. Phys. Lett. 76, 2493–2495 (2000) 153. R.C. Kwong, M.R. Nugent, L. Michalski, T. Ngo, K. Rajan, Y.-J. Tung, M.S. Weaver, T.X. Zhou, M. Hack, M.E. Thompson, S.R. Forrest, J.J. Brown, High operational stability of electrophosphorescent devices. Appl. Phys. Lett. 81, 162–164 (2002)

TADF and Hyperfluorescence Junji Adachi and Hisashi Okada

Abstract TADF (Thermally Activated Delayed Fluorescence) is recognized as the third generation of OLED-emitting technology, which provides highly efficient emission without using any rare metals, such as iridium. Fundamental molecule design strategies of TADF were based on the introduction of electron donor and acceptor units in which the π-conjugation was significantly distorted by steric hindrance introduced through bulky substituents. In this design, the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) orbitals can be located around donor and acceptor moieties, respectively, leading to a small Est while maintaining a reasonably high radiative decay rate (kr ). This concept enabled efficient spin up-conversion from triplet to singlet. As a result, highly efficient emission from singlet as high as 100% internal quantum efficiency was achieved. One disadvantage of TADF is low color purity if it is applied to displays because of its wide emission spectrum. The emission color purity is indicated by the index full width at half maximum (FWHM). FWHM of TADF, in general, is from 80 to 100 nm, wider than that of fluorescence. Hyperfluorescence (HF) combines TADF and fluorescence and is considered as the 4th-Gen OLED emitting technology. TADF acts as exciton generator and transfers excitons to fluorescence by Förster resonance energy transfer (FRET). Fluorescence molecules receive excitons and emit light as high as 100% internal quantum efficiency which is four times higher efficiency than a conventional fluorescence emitting technology. Color purity of HF, as indicated by FWHM, is similar to fluorescence—from 30 to 40 nm.

J. Adachi (B) · H. Okada Kyulux Inc., Suite 227, FiaS Bidg.2, 4-1 Kyudai-Sinmachi, Nishi, Fukuoka 819-0388, Japan e-mail: [email protected] H. Okada e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_2

39

40

J. Adachi and H. Okada

1 Fluorescence and Phosphorescence Four different emitting technologies are widely recognized in OLED. In OLEDs, electrically injected charge carriers recombine to form singlet and triplet excitons in a 1:3 ratio. Fluorescence, recognized as the first generation of emitting technology, was demonstrated by Tang et al. [1] in 1987. It emits light by utilizing the singlet excitons. Fluorescence-based OLEDs have been utilized for decades because of their long operational lifetime, high color purity in a variety of colors, including deep blue, and potentially low material cost. As shown in Fig. 1a, however, internal quantum efficiency (IQE) of fluorescence is limited to 25% because fluorescence only utilizes the singlet excitons. The second generation of emitting technology, phosphorescence, which was demonstrated using iridium phenyl–pyridine complexes by Baldo et al. [2] in 1999, emits light by utilizing three triplet excitons. This means that the IQE of phosphorescence is three times higher than that of fluorescence. In addition, phosphorescence utilizes a singlet exciton by Inter System Crossing (ISC) and achieves almost one hundred percent (100%) IQE as shown in Fig. 1b, which was demonstrated by Adachi et al. [3] in 2001. Phosphorescence has been widely adopted in OLED displays because of its high efficiency and long operational lifetime in red and green. The color purity of phosphorescence emitters is not as pure as that of fluorescence emitters because of their wide emission spectrum. One indicator of color purity is known as Full Width at Half Maximum (FWHM). The typical FWHM of fluorescence is from 30 to 40 nm. The FWHM of phosphorescence emitters, on the other hand, is from 60 to 80 nm. Another issue of phosphorescence is deep blue emission. Even after decades of intensive R&D efforts, no deep blue phosphorescence is commercially available.

Fig. 1 Energy diagrams of fluorescence (a), phosphorescence (b), and TADF (c)

TADF and Hyperfluorescence

41

2 TADF (Thermally Activated Delayed Fluorescence) Thermally Activated Delayed Fluorescence (TADF) is recognized as the third generation of OLED-emitting technology. TADF is a way of harvesting the triplet excitons to the singlet by controlling the energy gap between singlet and triplet (Est) small enough to upconvert from triplet to singlet by using Reverse InterSystem Crossing (RISC) (Fig. 1c). The concept of TADF was first realized by Endo et al. in 2009 by applying short electrical pulse excitation of an OLED based on an Sn(IV)–porphyrin complex [4]. The real breakthrough of pure organic TADF-based OLEDs to clarify the mechanism and molecular design rules of TADF was demonstrated by Endo et al. in 2011 [5]. In this work, strategies for molecule design were based on the introduction of electron donor and acceptor units in which the π-conjugation was significantly distorted by steric hindrance introduced through bulky substituents. In this design, HOMO and LUMO orbitals can be located around donor and acceptor moieties, respectively, leading to a small Est while maintaining a reasonably high radiative decay rate (kr ). A guest–host system was introduced to suppress nonradiative decay from the triplet excited states, because neat films usually resulted in intense concentration quenching, leading to complete disappearance of fluorescence, phosphorescence, and TADF. Based on these concepts, a TADF material possessing both a small Est and a high kr was designed. Figure 2 shows the molecular structure of 2-biphenyl-4,6bis(12-phenylindolo[2,3-a]carbazole-11-yl)-1,3,5-triazine (PIC-TRZ) containing an indolocarbazole donor unit and a triazine acceptor unit. PIC-TRZ exhibited Est as small as 0.11 eV along with a radiative decay rate of kr ~107 s−1 , providing both efficient up-conversion from T1 to S1 levels and intense fluorescence that led to high EL efficiency. In 2012, Uoyama et al. reported a series of highly efficient TADF emitters based on Carbazolyl Dicyanobenzene (CDCB) pure aromatic compound, with carbazole donor unit and dicyanobenzene electron acceptor unit (Fig. 3a) [6]. The highest occupied Natural Transition Orbital (NTO) of 4CzIPN, one of CDCBs, was delocalized over

Fig. 2 Molecular structure of PIC-TRZ and its HOMO and LUMO calculated by Gaussian 03. “Reproduced from [5], with the permission of AIP Publishing.”

42

J. Adachi and H. Okada

4CzPN: R = carbazolyl 2CzPN: R = H

4CzIPN

4CzTPN: R = H 4CzTPN-Me: R = Me 4CzTPN-Ph: R = Ph

Fig. 3 Molecular structures of CDCBs and natural transition orbits (NTO) of 4CzIPN. a Molecular structures of CDCBs. Me, methyl; Ph, phenyl. b Highest occupied NTO. c Lowest unoccupied NTO

the four carbazolyl moieties as shown in Fig. 3b, and the lowest unoccupied NTO was centered on the dicyanobenzene moiety as shown in Fig. 3c. Two NTOs of 4CzIPN were spatially well separated, leading to a small Est and enhanced RISC from T1 to S1 . Using 4CzIPN in a multilayer OLED device, a very high EQE of 19.3% was achieved. All results of CDCBs showed a significant breakthrough compared to the conventional fluorescence-based OLEDs, demonstrating that the utilization of organic TADF materials in harvesting triplet excitons by delayed fluorescence opened up a path to achieve very high-performance OLED devices without using any rare metals, such as iridium or platinum. Various studies on TADF have been conducted since this epoch-making discovery. The EQE of OLEDs based on pure organic TADF reached over 20% in various colors including red, yellow, green, and blue.

3 Hyperfluorescence Pure organic TADF provides highly efficient emission without using any rare metals, such as iridium. One disadvantage of TADF is low color purity if it is applied to displays because of its wide emission spectrum. The typical FWHM of TADF is from

TADF and Hyperfluorescence

43

80 to 100 nm. Pure color emission of red, green, and blue enables a wide color space which is required for the next generation of displays, e.g., Rec. 2020. Fluorescence has the advantage of higher color purity compared to phosphorescence and TADF because of its narrow emission spectrum. Nakanotani et al. proposed another way to harvest and utilize the triplet energy by using TADF as an assistant dopant in 2014 [7]. This emitting technology, called hyperfluorescence, which combines TADF as an assistant dopant and fluorescence as an emitter is recognized as the fourth generation of emitting technology. Once excitons are generated, 25% of the excitons go to singlet energy state and 75% of the excitons go to the triplet energy state. In TADF molecule, the triplet energy is upconverted to the singlet (T1t → S1t ) by RISC. Then the singlet energy is transferred from TADF to fluorescence (S1t → S1f ) by Förster Resonance Energy Transfer (FRET) [8]. Lastly, fluorescence emits light as high as one hundred percent (100%) IQE as shown in Fig. 4. The emitting technology is now recognized as TADF-Assisted Fluorescence (TAF) or Hyperfluorescence. In order to achieve high efficiency, the concentration ratio of host, TADF, and fluorescence dopant is optimized to enhance FRET and minimize Dexster energy transfer as shown in Fig. 5 [9]. A pair of hole and electron is recombined in TADF. Excitons are transferred from TADF to fluorescence by FRET. A fluorescence received excitons from TADFs within FRET radius region which showed by sky blue circles. The triplet energy of TADF may not be upconverted to the singlet because the triplet energy is transferred by Dexter energy transfer from the triplet of TADF to the triplet of fluorescence directly if fluorescence and TADF are located side by side. This Dexter energy transfer decreases the total efficiency of Hyperfluorescence. An additional benefit of Hyperfluorescence is utilizing fluorescence emitters which are composed of simple aromatic compounds, which have continued to attract interest because of their longer operational lifetimes in blue OLEDs, higher color purity EL and broad freedom of molecular design. Blue, green, yellow, and red Hyperfluorescence with various combinations of TADF and fluorescence were reported as shown in Fig. 6. Each Hyperfluorescence device showed significantly higher EQE than the

Fig. 4 Energy diagrams and energy transfer mechanisms of hyperfluorescence

44

J. Adachi and H. Okada

Fig. 5 Energy transfer diagram of hyperfluorescence. A pair of hole and electron is recombined in TADF molecules. Excitons are transferred from TADF to fluorescence by FRET. Dexter energy transfer occurs from the triplet of TADF to the triplet of fluorescence directly if fluorescence and TADF are located side by side

Fig. 6 a−d Fluorescence spectra of assistant dopant: host co-deposited film (upper), and absorption (dashed line) and fluorescence (solid line) spectra of emitter dopant in solution (10−5 mol l−1 in CH2 Cl2 ) (bottom) for blue, green, yellow and red. Device performance of OLEDs. e–h External EL quantum efficiency as a function of luminance for the blue, green, yellow, and red OLEDs. The external EL quantum efficiency for OLEDs without an assistant dopant is plotted as open symbols. Inset: chemical structures of emitter dopants used in this study

conventional fluorescence only device which uses the same fluorescence as an emitter [7]. A schematical diagram of spectrum comparison of emitting technologies, fluorescence, phosphorescence, TADF, and Hyperfluorescence are shown in Fig. 7 [10].

TADF and Hyperfluorescence

45

Fig. 7 Spectrum comparison

The color spectrum of Fluorescence is narrow but it suffers from low light intensity. Phosphorescence and TADF feature higher IQE and thus higher light intensity than fluorescence, however, the color purity of phosphorescence and TADF is lower than that of fluorescence because of their wider spectrum. Hyperfluorescence features a narrow spectrum, as narrow as fluorescence’s, and light intensity about four times higher than that of fluorescence because of its high efficiency. Although the IQEs of phosphorescence, TADF, and Hyperfluorescence reach almost one hundred percent (100%) in ideal conditions, light intensity of Hyperfluorescence at the peak top wavelength is higher than that of both phosphorescence and TADF. IQE is indicated by the area of spectrum in Fig. 7. Even if the IQEs of phosphorescence, TADF, and Hyperfluorescence are the same, this means the areas of spectrums are the same, narrow Hyperfluorescence spectrum shows higher light intensity than others. Comparison of characteristics from performance and commercialization standpoints are shown in Table 1 [10]. Fluorescence shows excellent features except efficiency. Phosphorescence shows high efficiency but low color purity and expected high cost because of the use of a rare metal, such as iridium. There is currently no commercially available blue phosphorescence emitter, even after decades of research. TADF shows high efficiency and low cost but low color purity. The commercialization of a blue TADF emitter is expected to be challenging due to its low color purity. Hyperfluorescence performs best in all categories: high efficiency, low cost, and high color purity. By considering yellow Hyperfluorescence being already on the market, Hyperfluorescence is expected to become an emitting technology which will satisfy all requirements for OLED display. A demonstration panel that uses a yellow Hyperfluorescence emitter is shown in Fig. 8 [10]. The right-hand side of the panel was fluorescence and the left-hand

46

J. Adachi and H. Okada

Table 1 Comparison of performances

Fig. 8 Mechanism of hyperfluorescence (left): excitons upconversion from T1 to S1 in TADF, FRET from TADF to fluorescence and light emission with 100% IQE from fluorescence. (middle): demonstration panel of hyperfluorescence (left) and conventional fluorescence (right), both sides of the panel emit light by the same fluorescence molecule. (Right): emitting diagram of fluorescences

side was Hyperfluorescence. The emitting fluorescence molecule was the same one on both sides but light intensity of Hyperfluorescence was more than four times brighter. Red, green, and blue Hyperfluorescence and TADF demonstration panels and EL characteristics are shown in Fig. 9 [10]. In Fig. 9a-1, b-1, c-1, the left-hand side of each panel was TADF as an emitter, and the right-hand side of each panel was Hyperfluorescence. The same TADF was used on both sides of each panel. Color coordinates of TADF and Hyperfluorescence are shown in (a-2, b-2, c-2). Color coordinates of Hyperfluorescence indicated a wider color space coverage compared with those of TADF. Energy of TADF which had a wide spectrum was successfully transferred to fluorescence with a narrow spectrum as shown in (a-3, b-3, c-3). In the red Hyperfluorescence device (b-1, b-2, b-3), yellow TADF transferred energy to red fluorescence. In the green Hyperfluorescence (a-1, a-2, a-3) and the blue Hyperfluorescence device (c-1, c-2, c-3), energy from TADF, which had a slightly longer peak top wavelength than fluorescence, successfully transferred. As the results

TADF and Hyperfluorescence

47

Fig. 9 Comparisons of TADF and hyperfluorescence, (a-1, b-1, c-1): demonstration panel, left-hand side; TADF as an emitter, right-hand side; hyperfluorescence, TADF as an assistant dopant. (a-2, b-2, c-2): CIE plots of hyperfluorescence and TADF. (a-3, b-3, c-3): emission spectrum comparison. Hyperfluorescence shows a higher light intensity and narrower spectrum than those of TADF

of an efficient energy transfer, the red, green, and blue Hyperfluorescence emitted a narrow spectrum with FWHM; 31 nm, 43 nm, and 31 nm, respectively, and higher light intensity than that of TADF. Color purity of Hyperfluorescence was enhanced further by applying a top emission (TE) device structure which was widely applied in smartphone OLED displays. Comparison of emission spectrums and CIE coordinates of TADF bottom emission (BE), Hyperfluorescence BE, and Hyperfluorescence TE devices are shown in Fig. 10a, b [10]. FWHM of TADF BE, Hyperfluorescence BE, and Hyperfluorescence TE were 88 nm, 31 nm, and 20 nm, respectively. The light intensity at the peak top wavelength of Hyperfluorescence TE was two times and four times higher than that of Hyperfluorescence BE and TADF BE, respectively. CIE coordinates of Hyperfluorescence TE achieved CIE requirement of Rec. 2020. Hyperfluorescence is recognized as the fourth generation of OLED emitting technology and has already proved its superiority from the viewpoints of efficiency, color

48

J. Adachi and H. Okada

Fig. 10 a EL spectrum of TADF BE, hyperfluorescence (HF) BE, and HF TE. b CIE coordinates of TADF BE, HF BE, and HF TE

purity, light intensity, and cost. It is expected to be adopted in all colors, red, yellow, green, and blue, for OLED displays.

4 TADF Molecules 4.1 Design Principles of TADF Molecules In this chapter, we describe the design principles of molecules that exhibit efficient TADF characteristics. The TADF process depends on a small Est to aid the RISC (Reverse InterSystem Crossing) from T1 to S1 . Considering the dominant factors of Est, Est is proportional to the exchange integral, meaning that the smaller the overlap between the ground state and excited state wave functions, the smaller the Est. That is, if the overlap between HOMO and LUMO is small enough, Est is small (Fig. 11). In π-electron aromatic compounds, Est usually has a large value of 1 eV or more due to large overlap of π-π* orbitals. In fact, although condensed polycyclic aromatic compounds such as anthracene and tetracene show fluorescence, TADF emission is not observed because Est exceeds 1 eV. In contrast, in the molecule that has an electron-donating site (donor) and an electron-accepting site (acceptor), HOMO and LUMO have the characteristic that they are largely separated into donor and acceptor sites, respectively. Therefore, by combining various donor groups and acceptor groups, HOMO and LUMO are separated, and molecules that exhibit efficient TADF design become possible. Particularly, in a molecule containing a donor and an acceptor in one molecule, small molecules can be obtained by introducing a

TADF and Hyperfluorescence

49

HOMO

Fig. 11 The strategy of realizing small Est in organic molecules. HOMO: the highest occupied molecular orbital, LUMO: the lowest occupied molecular orbital

LUMO Large Est

Small Est

sterically hindered group to spatially orthogonalize the donor group and the acceptor group. On the other hand, in order to obtain a TADF material exhibiting high PhotoLuminescense Quantum Yield (PLQY), an appropriate radiation rate constant (kr) is required. However, as large kr and small Est are conflicting, fine molecular design is required to realize them simultaneously for efficient TADF OLEDs. Tao et al. [11] have summarized some design principles for D–A type TADF molecules according to the relevant literature regarding TADF, especially for highperformance OLED characters (Fig. 12).

Donor

X

Acceptor

(1) X: Separation of HOMO and LUMO (Small Est) (a) Introduction of steric hindrance (b) Spiro linker, physical separation of donor and acceptor (c) X-shaped molecular structure (d) Multiple resonance effect (2) Increase the radiative decay rate (Large kr) Increasing the overlap density distribution between the S0 and S1 states; Large delocalization of molecular orbitals.

Fig. 12 The main design principles for D–A type TADF molecules

50

J. Adachi and H. Okada

4.2 Classification of Typical TADF Molecules The general classification of typical TADF molecules is as follows. (1) (1-1) (1-2) (2)

Donor–Acceptor molecular systems (D–A type) Intramolecular D–A type Bimolecular D–A type (Exciplex type) Multiple resonance type.

The following sections describe these representative types.

4.3 Intramolecular TADF Type A lot of various donor and acceptor groups have been used to construct D–A type TADF molecules. Representative donor groups and acceptor groups are shown in Fig. 13. Carbazole aromatic amine (Cz) and its condensed ring (such as indolocarbazole), diphenylamine, phenoxazine (PXZ), phenothiazine (PTZ), 9,9-dimethyl9,10-dihydroacridine (DMAC), 5-phenyl-5, 10-dihydrophenazine (PPZ), 5,10dihydrophenazine (DHPZ), and its derivatives are commonly used as donor groups (a) Representative examples of donor groups

(b) Representative examples of acceptor groups

Fig. 13 Representative donor and acceptor groups used for the construction of D–A type TADF molecules

TADF and Hyperfluorescence

51

Fig. 14 Molecular structures of carbazolyl dicyanobenzene (CDCB) type

HOMO

LUMO

Fig. 15 HOMO and LUMO of 2CzPN

for TADF molecules. Of these, aromatic amines such as carbazole are often used from the viewpoints of charge transport properties and redox stability. Various types of acceptors are used to tune their TADF characteristics, such as emission color, excited-state lifetime, and PLQY. The reported TADF molecules will be categorized by their acceptor units and discussed in detail below.

4.3.1

N-Containing UNIT as Acceptor Type

Cyano-Substituted Aromatic Type Owing to the strong electron-withdrawing ability, the cyano group has been widely used as a strong acceptor generally in the form of cyano-substituted aromatics for constructing TADF molecules with intramolecular D–A structure. Uoyama et al. [6] designed a series of highly efficient TADF emitters based on carbazolyl dicyanobenzene (CDCB), with carbazole as an electron donor and dicyanobenzene as an electron acceptor (Fig. 14). Because the carbazolyl unit is markedly distorted from the dicyanobenzene plane due to steric hindrance, the HOMOs and LUMOs of these emitters are localized on the donor and acceptor moieties, respectively, leading to a small Est (Fig. 15, 2CzPN as an example).

52

J. Adachi and H. Okada

Fig. 16 Photoluminescence of CDCB series. a Photoluminescence spectra measured in toluene. b Photograph under irradiation at 365 nm

Fig. 17 Dual-core TADF molecule (DDCzIPN)

Moreover, by varying the substituent number and the relative position of the cyano and carbazolyl groups, high PLQY and various emission colors can be achieved, i.e., 2CzPN (470 nm) with sky-blue emission, 4CzIPN with green emission (507 nm), 4CzTPN-Ph with orange emission (577 nm) (Fig. 16). Cho et al. [12] reported a dual-core TADF compound, DDCzIPN (Est = 0.13 eV), that has the coupling of two DCzIPN moieties (Fig. 17). Relative to a TADF emitter with a single emitting core, the coupling of two TADF emitter moieties into one molecule increased the EQE of the TADF device by enhancing the absorption coefficient (1.1 × 105 M−1 cm−1 for DCzIPN and 3.7 × 105 M−1 cm−1 for DDCzIPN) and PLQY (67% for DCzIPN and 91% for DDCzIPN) of the emitter. Due to the extension of conjugation, the PL emission peak of DDCzIPN red-shifted to 477 nm in the solid state. On the other hand, the lifetime of the delayed emission was shortened to ca. 2.8 μs, which may be because of the rigidity of the molecular structure caused by steric hindrance of the four CN units in the central biphenyl moiety. A high EQE of 18.9% was achieved using the DDCzIPN emitter in a green OLED (497 nm,) which is higher than that of the single-core TADF emitter DCzIPN (16.4%). Therefore, the dual-core TADF emitter design should be an efficient strategy to improve quantum efficiency. In contrast to directly connected donor and accepter units, a phenyl linker was introduced between the donor and the benzonitrile-type acceptor. Park et al. [13]

TADF and Hyperfluorescence

53

Fig. 18 TADF molecules with linker inserted between donor and acceptor

synthesized a series of full-color TADF emitters by combining a phthalonitrile acceptor core with various donor units (Fig. 18). These emitters possess small Ests that originate from the angular-linked D–A structures, with subsequent efficient TADF emission. The light-blue color OLED based on the emitter Ac-CNP exhibits a high maximum EQE of up to 18.9% and an extremely small efficiency roll-off.

N-Containing Heteroaromatic Type Triazine The highly electron-deficient triazine group is a suitable building block for the construction of organic semiconductor molecules. Aromatic amine–triazine compounds composed of electron-donating aromatic amine and electron-accepting triazine units are widely used as host materials for OLEDs. These types of materials show excellent thermal, morphological, optical, and electrical properties and many of this reported D–A type aromatic amine–triazine derivatives show effective TADF emission, due to the effective separation of their HOMO and LUMO with small Est values [5, 14]. Endo et al. [5] reported an effective TADF triazine derivative PIC-TRZ containing an indolocarbazole donor unit and a triazine acceptor unit. Because the indolocarbazole unit was bulky, it induced significant steric hindrance around the biphenyl triazole unit, leading mainly to the distribution of HOMO and LUMO electronic clouds on the indolocarbazole and biphenyl triazole units, respectively. This resulted in a limited overlap between its HOMO and LUMO levels and a small Est of 0.08 eV (Fig. 19). PIC-TRZ exhibits TADF emission with a lifetime of 2.30 μs and a high fluorescence radiative decay rate (kr = ~107 ). A PIC-TRZ doped OLED demonstrated an EQE of 5.3% at low current density with blue-green emission, approaching the theoretical limit of conventional fluorescence-based devices. Although a roll-off of the EQE was observed at a higher current density, which resulted from the long lifetime of PIC-TRZ, the maximum luminance reached nearly 10,000 cd m−2 . Sato et al. [14] reported a novel triazine derivative, PIC-TRZ2, that exhibits a very small Est of 0.003 eV, which is more than one order of magnitude smaller than that of PIC-TRZ. The HOMO and LUMO of PIC-TRZ and PIC-TRZ2 are shown in

54

J. Adachi and H. Okada

Fig. 19 Molecular structures of PIC-TRZ and PIC-TRZ2 with their HOMOs and LUMOs calculated at the PBE0/6-31G(d) level of theory. The calculated energy difference between the S1 and T1 excited states is Ecalc = 0.080 eV for PIC-TRZ and 0.003 eV for PIC-TRZ2. Reprinted figure with permission from [14]. Copyright (2013) by the American Physical Society

Fig. 19. In the case of PIC-TRZ2, the HOMO is localized over an indolocarbazole unit while the LUMO is located around the biphenyltriazine unit, indicating that they are well separated. PIC-TRZ2 clearly shows better HOMO–LUMO separation than PIC-TRZ, so it is most likely that the linear combination of one donor and one acceptor system in PIC-TRZ2 results in its predicted small Est. It is the almost ideal separation of the HOMO and LUMO in PIC-TRZ2 that enables a Est of almost zero. Its doped film shows a rather high PLQY of ~60% (~45% from a delayed component and ~14% from a prompt component) and a short TADF decay time of 2.7 μs. It was considered that PIC-TRZ2 exhibits appreciable oscillator strength (f = 0.0012) and the apparent radiative decay rate of TADF (kTADF ) in 6 wt% PIC-TRZ2:mCP is ~106 s−1 . A series of OLED devices with different structures containing this molecule as an emitting dopant were fabricated, and they exhibited blue-greenish emission (505 nm) with EQE = 14% ± 1%. Shizu et al. [15] used quantum mechanics to reveal the relationship between the molecular structures and the photophysical properties of TADF emitters and derived a direction for the molecular design of highly efficient TADF emitters. Theoretical analyses show that the luminous efficiency of TADF emitters largely depends on the overlap density (ρ10 ) between the electronic wave functions of the ground state and the S1 state, and increasing the ρ10 distribution promotes S1 → S0 radiative decay, and improves the PLQY of the TADF emitters. They developed an efficient sky-blue TADF emitter BCzT, by insertion of a phenyl group between the donor and acceptor units of compound CzT (Fig. 20). The HOMOs of BCzT

TADF and Hyperfluorescence

55

Fig. 20 HOMO and LUMO of BCzT and CzT: Reprinted with permission from [15]. Copyright (2015) American Chemical Society

and CzT show similar distribution patterns because BCzT and CzT contain identical electron-donating units. Their HOMOs are predominantly distributed over the electron-donating units. However, for BCzT, the LUMO is predominantly distributed on the electron-accepting unit but is also extended to the electron-donating unit, while the LUMO for CzT is strongly localized on the electron-accepting unit and does not extend to the electron-donating unit (Fig. 20), the orange circled region. Therefore, the HOMO–LUMO overlap for BCzT is larger than that for CzT, and ρ10 is thus more widely distributed for BCzT. Theoretical analyses based on quantum mechanics show that an increase in ρ10 leads to an increase in the transition dipole moment and the PLQY. When doped into a host material, BCzT produces a high PLQY of 95.6%, which is much higher than the corresponding value for a CzT-doped DPEPO (Bis[2(diphenylphosphino)phenyl]ether oxide) film (PLQY = 39.7%). From the transient PL decays of the doped film, the lifetime for the delayed component is 33 μs and the efficiency of excited triplet state conversion into light is estimated to be 76.2%. A sky-blue OLED using BCzT as an emitter produces an EQE of 21.7%, which is due to the high triplet-to-light conversion efficiency. It was suggested that the material design based on ρ10 distribution provides a rational approach for developing TADF emitters for high-efficiency OLEDs. Wada et al. [16] developed a TADF emitter with a very small Est of 0.0088 eV, 3ACR-TRZ. Its energy levels are shown in Fig. 21. When doped into the CBP host at 16 wt%, 3ACR-TRZ shows a high PLQY of 98%. Transient photoluminescence decay measurements of the 16 wt% 3ACR-TRZ:CBP film confirmed that 77 exhibits efficient TADF with a triplet-to-light conversion efficiency of 96%. This high conversion efficiency makes 3ACR-TRZ attractive as an emitting dopant in OLEDs. Using 3ACR-TRZ as the emitter, a solution-processed OLED was fabricated resulting in a maximum EQE of 18.6%.

56

J. Adachi and H. Okada

(a)

ca.90

(b)

Fig. 21 a Chemical structure of 3ACR-TRZ and its HOMO–2, HOMO–1, HOMO, LUMO, and LUMO+1 calculated at the PBE0/6-31G(d) level of theory. b Energy level diagram for low-lying singlet and triplet excited states of 3ACR-TRZ calculated with TD-PBE0/6-31G(d)

Diphenyl Sulfoxide Type To achieve efficient blue TADF emission, it is necessary to consider both the πconjugation length and the redox potential of the donor and acceptor moieties and interrupt the conjugation between them. Based on this principle, diphenylsulfoxide (DPS) derivatives were used to construct TADF molecules. This is because the oxygen of the sulfonyl group has a significant electronegativity and gives the sulfonyl group an electron-withdrawing property. In addition, the sulfonyl group of diphenylsulfoxide exhibits a tetrahedral shape, limiting the conjugation of the compound. A series of analogs based on DPS as acceptor groups, DPA-DPS, tDPA-DPS, and tDCz-DPS were prepared by Zhang et al. (Fig. 22) [17]. The Est values of compounds DPA-DPS, tDCz-DPS, and tDCz-DPS were calculated to be 0.54, 0.45, and 0.32 eV, respectively, in which it was considered that the

TADF and Hyperfluorescence

57

Fig. 22 Molecular structures of diphenyl sulfoxide type

Fig. 23 Molecular structures of Triarylboron type

introduction of tert-butyl groups on the diphenylamine unit enhances its electrondonating ability and consequently lowers the CT energy and the Est. On the other hand, replacing a diphenylamine unit with a carbazole unit slightly increases the 1 CT state and considerably increases the 3 ππ* state, resulting in a further decrease in its Est. The PLQY of DPA-DPS, tDPA-DPS, and tDCz-DPS in the DPEPO films were 0.60 (421 nm), 0.66 (430 nm), and 0.80 (423 nm), respectively. Using these doped DPEPO films as the emitting layers, multilayer OLEDs based on DPA-DPS were fabricated and their EQE was only 2.9%, while the values for the devices based on tDPA-DPS and tDCz-DPS increased dramatically to 5.6% (tDPA-DPS) and 9.9% (tDCz-DPS). The CIE coordinates for the EL of the tDCz-DPS-based device were (0.15, 0.07), which were very close to those of the National Television Standards Committee (NTSC) standard blue of (0.14, 0.08).

Triarylboron Type Triarylboron compounds have a vacant p-orbital at the central boron atom, resulting in attractive electron-accepting properties. D-A systems with triarylboron acceptor and amine-based donor groups are of considerable interest due to their strong ICT properties. Numata et al. [18] reported highly efficient blue TADF materials combining a boron-containing acceptor (10H-phenoxaborin) with various donors. 10HPhenoxaborin was chosen as the acceptor because of its wide bandgap and good electron-withdrawing characteristics of the boron atom with an expanded π-plane (Fig. 23) [19]. The Est ranges from 0.06–0.12 eV for all compounds. These compounds have high PLQY (56–100%) and emission spectra ranging from light blue (λPL = 475 nm) to deep blue (λPL = 443 nm). A doped OLED device produced the deepest blue emission (450 nm) with CIE coordinates (0.14, 0.16) to achieve an

58

J. Adachi and H. Okada

EQE of 20%. In addition, the materials exhibit concentration-independent PLQY in a doped film. This is desirable for simplified device fabrication.

Bimolecular D–A Type (Exciplex Type) TADF can also be realized from exciplex formation utilizing the intermolecular excited state between electron-donating and electron-accepting molecules. The exciplex is well known as a charge transfer state and its emission occurs as a result of electron transition from the LUMO of an acceptor to the HOMO of a donor. Thus, it is reasonable to consider that such HOMO and LUMO levels contribute to the exchange of energy. In the exciplexes, there are very close triplet and singlet levels as a result of spatially separated HOMO and LUMO levels, which are mainly located on the donor and acceptor molecules, respectively. Compared to intramolecular excited states in single molecular TADF material, the intermolecular excited state of an exciplex system can provide a smaller Est. After the pioneering work on exciplex emitters for TADF by Goushi et al. [20, 21], many highly efficient delayed fluorescence exciplex systems have been reported. High-performance OLEDs utilizing exciplexes have also been demonstrated. Two representative ways to form TADF exciplex systems have been developed. The first common way is to mix hole-transporting type donor materials with electron-transporting type acceptor materials. The second way is to blend with D–A type bipolar materials to form an exciplex, including TADF materials, which generates an intermolecular CT excited state. Goushi et al. [20] proposed a TADF exciplex system that was formed between 4,4 ,4 -tris[3-methylphenyl(phenyl)-amino]triphenylamine (m-MTDATA) as the donor and different acceptors of 2-(biphenyl-4-yl)-5-(4-tert-butylphenyl)1,3,4-oxadiazole (t-Bu-PBD) and tris-[3-(3-pyridyl) mesityl]-borane (3TPYMB) (Fig. 24). Using m-MTDATA: t-Bu-PBD exciplex system, they demonstrated EQE greater than 5%, even though the PLQY was low of 26%. It was also pointed out that further enhanced the performance of exciplex-based OLEDs requires the provision of a donor molecule with both a shallow HOMO level and high triplet excitation energy. Furthermore, Li et al. [22] designed a highly efficient exciplex system incorporating HAP-3MF as an electron acceptor and mCP as an electron donor with a very small Est, resulting in efficient exciton upconversion and a high PLQY of 66.1% (Fig. 24). Using this exciplex system as an emitting layer in an OLED, a high EQE of 11.3% was demonstrated. These findings are of fundamental interest for the development of highly efficient OLEDs based on exciplex-TADF systems.

Multiple Resonance Type Hatakeyama et al. [23] reported a new design concept for organic molecules that exhibit ultrapure blue TADF emission based on efficient HOMO–LUMO separation by the multiple resonance effect, as shown in Fig. 25. The molecules are constructed

TADF and Hyperfluorescence

Donor molecule

59

Acceptor molecule

Fig. 24 Donor and acceptor molecules forming the TADF exciplex system

from a triphenylboron moiety, possessing a rigid polycyclic aromatic framework with two nitrogen atoms (DABNA-1 and DABNA-2, Fig. 26). Since the nitrogen atom exhibits a resonance effect opposite to that of the boron atoms, the HOMO and LUMO in such framework can be significantly separated by the enhanced resonance effect without the need for donor or acceptor groups. Benefited from the rigid πconjugated framework and the large oscillator strength of the S0 –S1 transition (f = 0.205), OLEDs fabricated using DABNA-1 as the emitter show emission at 459 nm with a small FWHM of 28 nm, CIE coordinates (0.13, 0.09) and a high EQE of 13.5%. The introduction of substituents seems to improve the oscillator strength (f = 0.415) without affecting the localization of molecular orbitals, as demonstrated by the derivative DABNA-2. An OLED fabricated with DABNA-2 shows pure blue emission at 467 nm with a very narrow FWHM of 28 nm (Fig. 26), CIE coordinates (0.12, 0.13), and a higher EQE of 20.2%.

4.4 Soluble TADF Materials The development of TADF materials for solution processes is required from the viewpoint of simplifying the manufacturing process of OLED panels both for display and lighting.

60

J. Adachi and H. Okada

Fig. 25 a Conventional design and b new design for efficient HOMO–LUMO separation. The localized HOMO and LUMO are indicated by the blue and red colors, respectively. Typical absorption and emission spectra are indicated by the black and green lines, respectively. The blue emission spectrum of a commercial OLED display is indicated by the green dotted line

TADF and Hyperfluorescence

61

Fig. 26 Molecular structures and electroluminescence spectra of DABNA type

Soluble TADF materials are classified into the following three patterns. (a) (b) (c) 4.4.1

Solubility-improved small molecule type Dendrimer type Polymer type. Solubility-Improved Small Molecule Type

Cho et al. [24] reported soluble 4CzIPN derivatives. To develop highly efficient solution-processed TADF devices by modifying 4CzIPN with methyl or tert-butyl groups to improve the solubility of TADF materials. Two soluble TADF materials, m4CzIPN and t4CzIPN (Fig. 27) were compared to 4CzIPN, and solution-processed TADF OLED devices were created for this purpose. t4CzIPN was found to be effective in improving EQE, demonstrating a high EQE of 18.3% and green emission comparable to vacuum-deposited TADF OLEDs made using the same molecule. Furthermore, Cho et al. [25] also reported 5CzCN (Fig. 27), a benzonitrile core surrounded by five carbazole units, as a soluble blue TADF emitter. The emitter Fig. 27 Solubility-improved small molecule TADF

62

J. Adachi and H. Okada

5CzCN exhibits a small Est of 0.16 eV and a high film PLQY (81%). High efficiency can be achieved in the 5CzCN OLED by both vacuum evaporation and solution processing, resulting in maximum EQE of 19.7% and 18.7%, respectively.

4.4.2

Dendrimer Type

Dendrimers are fully branched polymers of the entire molecule [26]. Due to their unique structure with a high steric hindrance, they are generally soluble and amorphous, allowing the core to separate chromophores and prevent concentration quenching. Therefore, they have been extensively studied as solution-processable OLED materials [27]. Carbazole dendrimers are efficient hole-transporting materials and phosphorescence hosts, due to their excellent hole-transporting properties, high triplet energy level, and good thermal ability [28]. Carbazole dendrimers used in TADF OLED applications have a potential gradient with an electron-rich periphery (HOMO) and an electron-deficient core (LUMO). The bulky dendrimer structure is expected to reduce important intermolecular effects and minimize concentration quenching. Following the above idea, a carbazole dendrimer with a triphenyltriazine core was designed and synthesized by Albrecht et al. (Fig. 28) [29]. The solubility of the G1TAZ was extremely low and hence it could not be isolated. However, other generations of dendrimers were very well soluble in common organic solvents such as toluene, tetrahydrofuran, and chloroform, and were successfully separated. The neat film Est values were calculated to be 0.03 eV (G2TAZ), 0.06 eV (G3TAZ), and 0.06 eV (G4TAZ), which are low enough values

Fig. 28 Dendrimer TADF materials

TADF and Hyperfluorescence

63

to allow for RISC from the T1 to the S1 state at room temperature. The temperaturedependent emission lifetime confirmed that the resulting dendrimers exhibited TADF with a decay time of approximately 5-8 μs in either toluene solution or pure film. The dendrimer PLQY in toluene solution was almost 100% for all generations and decreased in a neat film under an inert atmosphere, showing a clear quenching effect. A device incorporating these dendrimers as a light-emitting layer by spin coating showed an EQE value of maximum 3.4%. These carbazole dendrimers were the first solution-processable, non-doped, high-molecular-weight TADF materials.

4.4.3

Polymer Type

There are few reports of TADF polymers. This is mainly due to the difficulty in observing TADF in oligomeric and polymeric materials. It is very difficult to achieve a small Est and suppress the internal conversion of molecules containing many atoms at the same time. Furthermore, polymer and oligomer triplet populations are often efficiently quenched by intra- and intermolecular TTA. Nikolaenko et al. [30] reported the synthesis of TADF polymer (Fig. 29) for OLEDs. They presented a strategy called “intermonomer TADF” based on the formation of a single polymeric material containing intermonomeric CT emitters separated by a high T1 spacer unit. The intermonomer CT emitter is formed from the donor and acceptor monomers during the polymerization process and also acts as a charge transport unit for EL applications (Fig. 29). This approach can provide a promising route to transfer the TADF concept to a fully solution-amenable polymer platform. This is the first example of an efficient TADF from polymeric materials that can achieve the highest device efficiency of 10% EQE. Despite the low ~3% EQE at higher current densities of 10 mA cm−2 , this work represents the first step towards TADF-based OLEDs that are highly efficient solution processable.

Fig. 29 Intermonomer TADF polymer

64

J. Adachi and H. Okada

Like other TADF polymers, conjugated copolymer-type [31–33] and side-chaintype polymers [34] have been reported.

References 1. C.W. Tang, S.A. VanSlyke, Appl. Phys. Lett. 51, 913–915 (1987) 2. M.A. Baldo, S. Lamansky, P.E. Burrows, M.E. Thompson, S.R. Forrest, Very high-efficiency green organic light-emitting devices based on electrophosphorescence. Appl. Phys. Lett. 75, 4–6 (1999) 3. C. Adachi, M.A. Baldo, M.E. Thompson, S.R. Forrest, Nearly 100% internal phosphorescence efficiency in an organic light emitting device. J. Appl. Phys. 90, 5048–5051 (2001) 4. A. Endo, M. Ogasawara, A. Takahashi, D. Yokoyama, Y. Kato, C. Adachi, Adv. Mater. 21, 4802–4806 (2009) 5. A. Endo, K. Sato, K. Yoshimura, T. Kai, A. Kawada, H. Miyazaki, C. Adachi, Appl. Phys. Lett. 98, 083302 (2011) 6. H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 492, 234–238 (2012) 7. H. Nakanotani, T. Higuchi, T. Furukawa, K. Masui, K. Morimoto, M. Numata, H. Tanaka, Y. Sagara, T. Yasuda, C. Adachi, Nat. Commun. 5, 4016 (2014) 8. C. Adachi, Private communication (2014) 9. Th. Förster, Delocalized excitation and excitation transfer, in Modern Quantum Chemistry. Istanbul Lectures. Part III: Action of Light and Organic Crystals, vol. 3, ed. by O. Sinanoglu (Academic Press, New York and London, 1965), pp. 93–137 10. J. Adachi, H. Kakizoe, A. Endo, T. Oyamada, Towards commercialization of hyperfluorescence, in 4th International TADF Workshop (2019) 11. Y. Tao, K. Yuan, T. Chen, P. Xu, H. Li, R. Chen, C. Zheng, L. Zhang, W. Huang, Adv. Mater. 26, 7931–7958 (2014) 12. Y.J. Cho, S.K. Jeon, B.D. Chin, E. Yu, J.Y. Lee, Angew. Chem. 54, 5201–5204 (2015) 13. I.S. Park, S.Y. Lee, C. Adachi, T. Yasuda, Adv. Funct. Mater. 26, 1813–1821 (2016) 14. K. Sato, K. Shizu, K. Yoshimura, A. Kawada, H. Miyazaki, C. Adachi, Phys. Rev. Lett. 110, 247401 (2013) 15. K. Shizu, H. Noda, H. Tanaka, M. Taneda, M. Uejima, T. Sato, K. Tanaka, H. Kaji, C. Adachi, J. Phys. Chem. C 119, 26283–26289 (2015) 16. Y. Wada, K. Shizu, S. Kubo, K. Suzuki, H. Tanaka, C. Adachi, H. Kaji, Appl. Phys. Lett. 107, 183303 (2015) 17. Q. Zhang, J. Li, K. Shizu, S. Huang, S. Hirata, H. Miyazaki, C. Adachi, J. Am. Chem. Soc. 134, 14706–14709 (2012) 18. M. Numata, T. Yasuda, C. Adachi, Chem. Commun. 51, 9443–9446 (2015) 19. Kobayashi, K. Kato, T. Agou, T. Kawashima, Chem. Asian J. 4, 42–49 (2009) 20. K. Goushi, K. Yoshida, K. Sato, C. Adachi, Nat. Photonics 6, 253–258 (2012) 21. K. Goushi, C. Adachi, Appl. Phys. Lett. 101, 023306 (2012) 22. J. Li, H. Nomura, H. Miyazaki, C. Adachi, Chem. Commun. 50, 6174–6176 (2014) 23. T. Hatakeyama, K. Shiren, K. Nakajima, S. Nomura, S. Nakatsuka, K. Ki noshita, J. Ni, Y. Ono, T. Ikuta, Adv. Mater. 28, 2777–2781 (2016) 24. Y.J. Cho, K.S. Yook, J.Y. Lee, Adv. Mater. 26, 6642–6646 (2014) 25. Y.J. Cho, S.K. Jeon, J.Y. Lee, Adv. Opt. Mater. 4, 688–693 (2016) 26. S.M. Grayson, J.M. Frechet, Chem. Rev. 101, 3819–3868 (2001) 27. S.H. Hwang, C.N. Moorefield, G.R. Newkome, Chem. Soc. Rev. 37, 2543–2557 (2008) 28. K. Albrecht, R. Pernites, M.J. Felipe, R.C. Advincula, K. Yamamoto, Macromolecules 45, 1288–1295 (2012) 29. K. Albrecht, K. Matsuoka, K. Fujita, K. Yamamoto, Angew. Chem. Int. Ed. 54, 5677–5682 (2015)

TADF and Hyperfluorescence

65

30. A.E. Nikolaenko, M. Cass, F. Bourcet, D. Mohamad, M. Roberts, Adv. Mater. 27, 7236–7240 (2015) 31. R.S. Nobuyasu, Z. Ren, G.C. Griffiths, A.S. Batsanov, P. Data, S. Yan, A.P. Monkman, M.R. Bryce, F.B. Dias, Adv. Opt. Mater. 4, 597–607 (2016) 32. S.Y. Lee, T. Yasuda, H. Komiyama, J. Lee, C. Adachi, Adv. Mater. 28, 4019–4024 (2016) 33. Y. Zhu, Y. Zhang, B. Yao, Y. Wang, Z. Zhang, H. Zhan, B. Zhang, Z. Xie, Y. Wang, Y. Cheng, Macromolecules 49, 4373–4377 (2016) 34. J. Luo, G. Xie, S. Gong, T. Chen, C. Yang, Chem. Commun. 52, 2292–2295 (2016)

Small Molecules in Ink Jet Printed OLEDs—History, Status, and Prospects Sebastian Meyer, Manuel Hamburger, Sebastian Stolz, Miriam Engel, Anna Hayer, Hsin-Rong Tseng, Rouven Linge, and Rémi Anémian

Abstract In this chapter, we will show that RGB side-by-side IJP is a very promising technology for large-scale, top-emission, and high-resolution applications.

1 OLED Deposition Technologies Initial market penetration of OLED technology started based on vapor thermal evaporation (VTE). Here, individual sub-pixels of r(ed), g(reen) and b(lue) are patterned in a side-by-side (SBS) architecture by a fine metal mask (FMM). This technology dominates nowadays the market for small and medium displays up to 15.6 . By this, not only smartphones and tablets but also the entry class of IT-displays in notebooks are realized. Larger display sizes including typical 55 TVs have been realized as well. However, FMM patterning is considered the main limitation for reasonable yields on large scale due to mask sticking of material and ductility issues for large mask dimensions [1]. For the realization of TV displays with 55 and beyond, an open mask evaporation process was successfully established [2]. Here, a white-emitting OLED (WOLED) is created by blue and yellow/red/(green) sub-cells, which are stacked on top of each other. The final emission color is tuned by color filters (CFs) on top. TVs based on this technology are amongst the best-commercialized TVs as shown in numerous consumer reports [3]. The use of two or more blue sub-cells allows to compensate the intrinsically much lower efficiency of fluorescent blue compared to phosphorescent red and green [4]. Downsides of the tandem architecture are relatively high driving voltages (i.e., power consumption) and a very complex multi-layer structure compared to RGB SBS. Common for both evaporation techniques are moderate to low material utilization rates. Depending on the type of evaporation source, values are typically in the range from 10 to 50% [5, 6]. S. Meyer · M. Hamburger · S. Stolz · M. Engel · A. Hayer · H.-R. Tseng · R. Linge · R. Anémian (B) Darmstadt, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_3

67

68

S. Meyer et al.

Fig. 1 Comparison of OLED-based display technologies: “RGB + FMM” and “WOLED + CF” are realized from VTE. In case of “B-OLED + QD-PCC” the pixel color converted is printed on top of evaporated blue OLED. In “RGB IJP” a large extent of the stack is realized by IJP

Next-generation display technologies are meant to decrease production complexity and extend the color gamut at the same time. Two candidates are seriously considered as alternatives to evaporation technologies, see Fig. 1: blue OLED with quantum-dot-based pixel color converter (QD-PCC) and RGB from ink jet printing (IJP). “blue OLED + QD-PCC” bares the charm of open mask evaporation with one or more blue cells stacked on top of each other. Red and green color is realized by the QD-PCC, which is deposited on top of the blue OLED by IJP. Likely challenges are the compromise on performance (for a low number of blue units) versus power consumption (for a higher number of blue units) in the OLED stack as well as the performance and stability control of the quantum dot. Additionally, these devices still show a high complexity, due to the additional quantum dot film, which could be a challenge in mass production lines. RGB IJP is the promise to enable SBS architecture in mass production and on large size applications by patterning of the substrate and precise deposition by the IJP process. In addition, IJP can allow top emission displays with high resolution as discussed below. Typically, hole injection layer (HIL), hole transport layer (HTL), as well as the emissive layer (EML), are printed, whereas subsequent layers of hole blocking, electron transport, and electron injection are evaporated by an open mask process, i.e., without using an FMM. The pixelated bank structure is realized by photolithography, while for the high-precision ink placement drop volumes down to the pico-liter regime are required. The combination guarantees excellent material usage rates up to 95% [7]. IJP can be done with small molecules and polymers as active materials; small molecules show a clear advantage over polymers in terms of purification process as all standard methods used in organic chemistry (e.g., column chromatography, recrystallization, sublimation, …) can also apply for the purification of soluble small molecules. In addition, small molecules are monodisperse species contrary to polymers, offering a better batch to batch reproducibility in the production process [8]. Market analysts by IHS have estimated cost savings for a 65 4 K TV on a Gen10 line to be 15–25% compared to WOLED + CF. Even for medium sizes (exemplified by a 13.3 panel) produced on a Gen6 line, cost savings were calculated up to 20%

Small Molecules in Ink Jet Printed OLEDs …

69

compared to FMM use [9]. In addition to high material utilization, the low complexity of device structure, and the relative freedom in terms of scalability add up to costsaving arguments compared to evaporation technologies. However, process-related requirements for applicable materials differ quite strongly when comparing IJP and VTE.

2 Material Requirements for Ink Jet Printing The basic requirement for any material to be adopted to an IJP process is solubility in the IJP solvent. It allows to solubilize a certain amount of solid at a given concentration (typically given in weight percent, wt%). The concentration is chosen among the parameters of target thickness, substrate resolution (i.e., pixel size), and number of drops to be placed. Thicker layers, smaller pixels, and lower drop numbers require higher concentrations up to 10 wt%. Such solubilities are usually hard to realize by using a single solvent only, especially as some of the layers in solution-processed OLEDs feature more than one material. Therefore, typical solvent systems consist of two or more solvents, i.e., a solvent blend. Any ink needs to provide physical long-term stability (e.g., no precipitation over time, gelation, or other phase separation) including inert behavior toward chemical reactions. These could potentially be between two solutes and reaction of solvent molecules with the OLED active materials as well as reaction of impurities (e.g., water or oxygen) before the ink is being applied in the IJP process. A big effort is put to avoid impurities by solvent purification, design of storage containers and related ink jet printer hardware. The latter is often referred to as compatibility. Especially the solvent blend needs to be carefully checked for two reasons: (1) potential damage to the printing equipment resulting in significantly shortened production campaigns and (2) potential leaching of impurities into the OLED ink and thus into the printed OLED pixels leading to unstable device performance. A special class of impurities is particle contamination. They can come along with the deposition of the ink causing non-emissive areas (so-called dark spots) within the pixel. Thus, quality criteria for inks include a strict particle removal, in the best case at the point-of-use. The number of particles should be reduced to levels such that in the total ink needed for the preparation of a panel, there is less than one particle. With regard to the printing process, solvents are chosen among their vapor pressure which must be low enough to prevent drying of the wet film during printing. The latter can cause high in-panel inhomogeneity. At the same time, the vapor pressure should not unnecessarily impede the drying step. This results in a relatively narrow range of tolerated vapor pressure ≤0.025 mbar, respectively boiling points of ≥240 °C. The combination with a melting point below room temperature limits the choice even further. The (potentially negative) impact of vapor pressure on drying is illustrated in Fig. 2. For “too low” vapor pressure, the material dries from the outside causing agglomeration at the pixel center with reduced layer thickness at the pixel edge. Thus, the emitted light appears brighter at the pixel edge. For “too high” vapor

70

S. Meyer et al.

Fig. 2 EL images of printed green pixels from three different solvent blends. Each blend represents either “too low”, “optimum” or “too high” vapor pressure during drying. Aside of “optimum”, the EML thickness is inhomogeneously distributed causing an uneven pixel illumination

pressure the pixel edge is not emitting. This indicates a lack of material, which is often referred to as de-wetting, i.e., the ink does not properly spread inside the pixel or is repelled to the inside during drying. The two most important parameters to predict the processability of an IJP ink are its viscosity and surface tension. Both parameters, although determined macroscopically on the bulk of the formulation, define the performance of the ink during the event of printing as well as the behavior in the pixel on the panel surface. Typical determination techniques of these two parameters are shear rheometry and pendant drop shape analysis, whose models are far off from the printing steps like drop ejection or inksubstrate-interaction. Nevertheless, they have empirically proven to be invaluable for the development of OLED inks. In short and oversimplified, IJP relies on the piezo-induced ejection of a drop of ink through a nozzle. Numerous parameters can be tuned to modify the droplet speed, size, and shape: e.g., the design of ink chamber, arrangement of orifice and piezo, and the pulse waveform to drive the piezo. For the ink to stay inside the nozzle, the surface tension of the ink should be high enough to produce enough capillary force to counterbalance the hydrostatic pressure. If the surface tension of the ink is too low, ink will leak out of the nozzles and nozzle plate wetting will occur. This dramatically reduces the accuracy of drop volume control and placement accuracy. The drop formation during the ejection of the drop is determined by both viscosity and surface tension. Within the typical target parameters for IJP-based display fabrication (e.g., a drop size 10 kHz), the upper limit for the viscosity is roughly at 10 mPa*s. Organic small molecules (typical molecular weight ≤1 kg/mol) have some advantage here because they show little dependency of viscosity on solid concentration, see light blue in Fig. 3. In fact, the solubility limit of quite a few molecules is reached before a viscosity increase compared to the pure solvent by more than one unit is registered. In contrast, linear conjugated polymers show a significant influence of their concentration on ink viscosity—especially at average molecular weights of ≥100 kg/mol. As illustrated in Fig. 3, this dependency easily limits the applicable concentration to moderate 1

Small Molecules in Ink Jet Printed OLEDs …

71

Fig. 3 Viscosity plot versus solid concentration for conjugated polymers (dark blue) and small molecules (light blue). Upper viscosity limit for IJP applicable inks is 10 mPa s−1 [12]

wt% to stay below the viscosity limit of 10 mPa*s. The predictability of polymer inks is further reduced by their tendency toward non-Newtonian behavior. Too viscous inks exit the nozzle at too low speed, which delays the firing of the next drop and reduces the drop frequency. In addition, nozzle clogging can appear when small amounts of ink evaporate thereby further increasing the viscosity locally. Ideal drops have short-lived ligaments after detachment of the printhead nozzle before forming a spherical shape during transit to the substrate without any satellite (secondary drop) formation (Fig. 4). The drop shape is defined by minimization of the free energy and the surface tension is the governing parameter. A high surface tension will assist in forming a spherical droplet shortly after leaving the nozzle. Too high surface tension results in poor droplet ejection, too low surface tension results in secondary droplet formation (satellites) or spraying. For most relevant classes of small molecule semiconductors used for printed OLEDs, there is no impact of dissolved material on the surface

Fig. 4 Snapshots of a drop ejection from an inkjet nozzle taken with a dropwatcher camera. Small molecule-based ink top row, polymer-based ink bottom row: a nozzle plate, b ink ejection, c drop detachment with visible ligament, and d catch-up and spherical drop formation during flight. A longer ligament is formed for the polymer ink and consequentially, the catch-up time is longer

72

S. Meyer et al.

tension of the ink—the surface tension of the solvent blend stays constant when dissolving the active components. High viscosity slows down any movement within the droplet and the resulting residual ligament can result in poor drop placement accuracy. Especially shear rate-dependent viscosity (e.g., shear-thinning) poses a big challenge when trying to control both the behavior during drop ejection (shear rates 105 1/s) and in flight (shear rate 0). Fortunately, small molecule-based OLED inks show only minimal shear thinning property.

2.1 Ink-substrate interaction When the ink gets in contact with the bank structure in the pixelated substrate, the attraction or repulsion of the liquid by the bank material and the substrate surface mainly determine the homogeneity of the pixel coverage. This phenomenon is usually referred to as “wetting”. To ensure a reproducible printing result, the pixel base should be fully covered with one to two drops. Wetting can be quantified by the contact angle, i.e., the angle at the liquid/vapor/solid interface of a sessile ink droplet on a surface. The two main contributing material factors are the surface energy of the substrate and the surface tension of the fluid. The surface energy heavily depends on and can be influenced by the substrate treatment (like baking, ozone, plasma or corona treatment, etc.) and the substrate roughness. While it may be possible to measure the surface energy of, e.g., the semi-transparent electrode or a hole transport layer, it is very challenging to measure contact angles in a pixelated substrate, especially at bank structure walls with taper angles of up to 80°. Thus, usually, the surface tension of the ink is kept below a certain value (e.g., 40 mN/m) and adequate wetting is ensured experimentally. For evenly distributed ink in pixels, the final drying step by applying vacuum and subsequent thermal annealing needs to result in homogenously flat films. During vacuum drying, most of the solvent leaves the pixel starting from the outside to the inside, in accordance with the observation of a sessile droplet on an open substrate. The evacuation speed will thereby influence the homogeneity of the drying process. Jumps in solubility can appear for solvent blends with one solvent predominantly leaving, while the remaining ones may cause sudden precipitation in case of insufficient solubility. This can be understood based on the order of vapor pressures of the individual solvents, although physicochemical interaction often occurs and increases the complexity of predicting the change in solvent composition during drying. Adjustment to the drying process is possible by tuning the pumping speed as well as modifying the solvent blend composition (compounds & ratio). Thermal annealing of the thin film as a final step removes residual solvent and allows for close packing of the molecules. For materials which need to be thermally crosslinked to allow further printing steps on top, the annealing temperature usually has a lower limit given by the reactivity of the crosslinking species, and in general, a higher annealing temperature is beneficial for the device lifetime. For small moleculecontaining layers, the glass transition temperature (Tg) forms an upper limit to the

Small Molecules in Ink Jet Printed OLEDs …

73

annealing temperature. Above Tg, (poly-)crystalline phases form. This results in charge transport along grain boundaries and very different optical properties within crystalline or amorphous domains, significantly affecting excited states’ energies and lifetimes.

2.2 Fluorescent Blue In contrast to red and green, blue emissive layers are nowadays still realized by fluorescent materials instead of phosphorescent ones. The reason is mainly the challenge in phosphorescent blue to achieve reasonable lifetimes. Fluorescent blue, however, suffers from intrinsic efficiency limitations, where not 100% excitons like in phosphorescence can be used, but only the 25% with singlet character. A moderate increase to about 40% can be achieved by maximizing triplet-triplet annihilation (TTA), where a non-radiative triplet exciton is converted with the help of a second triplet to one singlet state. The TTA process is very sensitive to charge balance, requiring a strong local density of triplets. Early-phase “soluble blue” OLEDs relied on transferring vapor-based materials to solution processing. Due to mismatch in charge balance of the whole device stack, as well as low performing generations of material, device performance for blue was very poor with external quantum efficiency (EQE) around 5% and lifetime values for 5% luminance decrease (LT95) at 1000 cd/m2 in the single or low dual-digit range. A temporary workaround was established by the “blue common layer”, see Fig. 5a. Here, the blue subpixel was left empty during the EML print while a subsequently evaporated state-of-the-art blue EML was finally covering all

Fig. 5 Comparison of stack architecture of RG with blue common layer from 2016 versus truly solution-processed RGB from 2017. Significant efficiency improvements are observed for all three colors by changing from evaporated to truly printed blue. The benefit on color is evident from the change in CIE x/y coordinates for Red and Green [13]

74

S. Meyer et al.

Fig. 6 a Energy diagram for interface of HTL and EML with respective jumps in LUMO and HOMO. b Preferred energy level alignment with respect to smooth injection of holes and confinement of electrons inside the EML. Energic situation can be considered same for EML/HBL by exchanging HOMO and LUMO

three subpixels. Obvious drawbacks have been the additional evaporation step, the color purity of the red/blue and green/blue sub-unit as well as increased driving voltage in red/green sub-unit due to the additional blue layer on top. Meanwhile, the performance of solution-processed blue has significantly improved as can be seen in Fig. 5b and the table below. Key driving factors have been the introduction of suitable blocking layers adjacent to the EML, i.e., HTL and HBL. Both layers were optimized to prevent charge leakage out of the EML. Hole confinement inside the EML is realized by applying an HBL with sufficiently low HOMO not accessible from the occupied HOMO of the EML. The electron is confined in the EML by the sufficiently high step in LUMO from the EML to the HTL, see Fig. 6. While charge confinement is the first essential requirement to allow for efficient charge-recombination, confinement of excitons within the emissive layer is as important. This is achieved by applying adjacent materials to the EML (i.e., HTL and HBL) with sufficiently high levels of the excited states, which cannot be accessed by excitons out of the EML. In a final step, charge balance is adjusted by fine-tuning both hole- and electron-current, which leads to maximized contribution from triplettriplet-annihilation. Measured TTA rates of the highest efficient blue devices from solution meet the values for state-of-the-art vapor-based devices [10]. Apart from the charge-related contributions, outcoupling is the remaining parameter to influence the device efficiency. When the generated light passes the interface of two different layers (i.e., typically made of different materials), its direction is changed due to the law of refraction (Snell’s law). (Total) internal reflection is the main limitation, that light is captured inside the device stack. However, for an anisotropic spatial distribution of the emissive species, all (more realistic: a major fraction of) transition dipole moments are within the horizontal plane of the emissive layer. Light is emitted orthogonal to this plane, which decreases the losses due to

Small Molecules in Ink Jet Printed OLEDs …

75

refraction and thereby enhances the outcoupled portion. A small drawback appears for “off-normal”, i.e., higher viewing angles. In this case, the change to a non-Lambertian light distribution decreases the intensity of perceived light, which is accompanied by subtle, but still recognizable change in color, especially when using a cavity-type device. Anisotropic molecular shapes with high aspect ratio are favored over sphere-like ones. The potential increase in external quantum efficiency by maximizing the outcoupled light fraction can add up to 50% to efficiency.

2.3 Phosphorescence Red and Green State-of-the-art red and green emissive layers are based on phosphorescence. Typically, these are guest: host systems with organometallic complexes as emitters. The host transports the charges to the organometallic complex, on which the excited states are formed. Due to strong spin-orbit coupling caused by the heavy metal atom, excited singlet states are rapidly converted to triplets. This process is referred to as intersystem crossing and happens on a femtosecond time scale. The radiative decay of the triplet state, however, is on a much lower time scale of micro- to milliseconds. The relatively long-lived T1 state causes intrinsic broadening of the emission spectra (i.e., increase in FWHM) and increased probability toward non-radiative decay (lowering the lifetime of the material) compared to fluorescent emitters. Depending on the choice of emitter core, FWHM can go down to 40 nm with a decent PLQE (photoluminescent quantum efficiency, experimental analog to internal quantum efficiency) of 90%, see Fig. 7 shown for phosphorescent red emitters.

Fig. 7 Impact of emitter generation for phosphorescent red on full width half maximum of the emission spectra as well as photoluminescent quantum efficiency both in Toluene solution [14]

76

S. Meyer et al.

Fig. 8 a Spectral overlap of photopic curve (dotted red line) and various red emission spectra of same peak emission but varying spectral width (FWHM, black to grey lines). b Corresponding luminous efficiency and c CIEx coordinate: while CIEx stays almost constant, luminous efficiency drops significantly for stronger intensities at higher wavelength [14]

However, due to harvesting of both singlet and triplet excitons, the exciton yield is 100%, which significantly boosts the efficiency over fluorescent type emitting systems (max. 40%). Efficiency is typically quantified by luminous efficiency (cd/A), which is strongly color affected due to the photopic curve resembling the perception by the human eye. Since the photopic curve has its maximum in the green wavelength regime, minor changes in wavelength have a bigger impact on cd/A in red and blue than in green. Figure 8 illustrates the effect for a series of red emission spectra of same peak emission but different vibronic broadening. The higher the broadening (i.e., more light contribution from higher wavelengths), the lower the contribution to the luminous efficiency. Best practice values for red state EQE of 23.5% (17.1 cd/A @CIEx = 0.69) in bottom emission devices with LT95 of ~4900 h at 1000 cd/m2 . Best performance in green is 22.5% EQE at a CIEy = 0.64 and LT95 at 1000 cd/m2 ~6800 h. Flat film-forming OLED inks: toward top emission Given the trend of increased resolution from full high definition (HD) to 4 K or even 8 K for current TV models, bottom emission devices are becoming rapidly replaced by top emission architecture devices. Since cavity effects play a much stronger role for the latter, thickness uniformity over the whole emissive area of the pixel becomes very crucial. Essential criteria to be met are (1) flat film profiles along both pixel axes and (2) limited pile-up, i.e., material agglomeration at the edge formed by the substrate and the bank material. Control of drying behavior is key to obtain uniform and flat layers. Main driving parameters are boiling point, surface tension, solubility, and especially viscosity. During drying of a multi-solvent OLED ink, surface tension and viscosity will change as the drying proceeds. As the HIL is the first layer to form on the substrate, its flatness is most crucial providing the template for the subsequently deposited layer of HTL and EML, but it also faces the most diverse surface materials like substrate, bank wall, and bank surface. As described for sessile droplets on an open surface, two drying regimes can be envisaged, constant contact angle and constant contact line. As the ink volume that is printed in a subpixel in one swath is usually more than the volume defined by the bank, the initial phase

Small Molecules in Ink Jet Printed OLEDs …

77

Fig. 9 a U-shaped HIL profile with target thickness of 150 nm (lower thickness in the pixel center compared to the edges) resulting from a printed single-solvent formulation. The profile of the empty pixel is depicted in grey. b Profile of same HIL solid composition but printed from multi-component solvent blend. The profile shape perfectly resembles the shape of the empty pixel and results in a flat and homogenous film of 150 nm. c Determination of pixel uniformity, which is the ratio of A /A. A represents the total pixel area and A represents the pixel area with thickness range of 10% difference of bottom height

can be described with a drying regime of constant contact line—which is the top rim of the bank in this case. Once the ink volume is reduced to roughly the subpixel volume, the further drying regime is a result of the inks’ surface tension, the bank wall properties, and the taper angle. A high SFT in combination with steep taper angles will result in a receding contact line (constant contact angle). Consequently, very low material deposition will happen on the bank wall, which will affect the deposition of following layers. Especially at the late stage of drying, material movement needs to be controlled, as Marangoni-type flow usually leads to excessive material deposition at the outer part of the wetted surface. This can be achieved by a formulation leading to a high viscosity at the late stage of drying or inverting the surface tension gradient on the surface of the drying ink. The effect of the drying from different solvent systems is illustrated in Fig. 9: for the single solvent-based HIL formulation, a U-shaped film profile is observed. This is characteristic for excessive material agglomeration on the bank of the pixel well, which area in an inhomogeneous lateral thickness distribution with a minimum thickness in the pixel center, see left profile of Fig. 9a. A multi-solvent system allows for optimized drying along the whole process of solvent leaving the pixel and thereby constantly changing the solubility of the remaining solvent-solid mix. In an optimized case, as depicted in Fig. 9b, the remaining film profile is flat and follows the shape of the empty pixel. This effect can be understood from the effective boiling point as well as surface tension and viscosity during vacuum drying. A quantitative judgement can be done by calculating the so-called “pixel uniformity”: Here the total pixel area of A is divided by the area of A , which covers the area with 10% thickness difference from the bottom height. The pixel uniformity increases from 60 to 85% from the U-shape profile of Fig. 9a to the flat film profile of Fig. 9b.

78

S. Meyer et al.

2.4 Top Emission Performance The change from a reflective to a semi-transparent cathode (typically Mg:Ag alloy) in top emission adds the cavity aspect to the optical and electrical characteristics of the device. For optimized optical path lengths, the Purcell effect can enhance the radiative conversion rate (i.e., the internal quantum efficiency). Roughly speaking, constructive inference is achieved by matching the optical path length inside the device to the emission wavelength. In case of layer non-uniformity however, a mismatch in the optical path length occurs in certain areas of the pixel. The effect is illustrated in Fig. 10, where an evaporated HIL film (flat by nature) is compared to a (non-)flat printed HIL film according to their profiles. On top of these three different HIL layers, the very same stack from HTL to cathode was evaporated. The comparison of their corresponding EL images shows a clear inhomogeneous appearance for the non-flat type printed HIL, while the evaporated and flat-printed HIL are equally uniform. The same trend is observed in OLED parameters: for the non-flat HIL luminous efficiency is reduced and driving voltage increased, while the flat-printed HIL perfectly matches the evaporated HIL performance. Besides the sensitivity toward the film profile’s shape, another critical parameter is the layer thickness precision. Figure 11 illustrates a red-emitting TE device with varying HIL thickness ranging from ~115 nm (28 drops) to ~150 nm (34 drops) [drop size ~10 pl; 80 ppi substrate resolution]. The optimum of external quantum efficiency is achieved for 31 drops (~130 nm HIL thickness) at an emission wavelength of

Fig. 10 a Evaporated blue stack of HTL/EML/HBL/ETL/cathode/capping layer on varying HIL: vapor, non-flat and flat HIL from IJP. b Film profiles of non-flat (u-shape) and flat HIL from IJP. c EL images. d Driving voltage and blue index (cd/A/CIEy)

Small Molecules in Ink Jet Printed OLEDs …

79

Fig. 11 Luminous efficiency and electroluminescent spectra of a red-emitting device for varying HIL layer thickness [~115 to ~150 nm] achieved by printing 28–34 drops of HIL ink. Incremental difference to the maximum in drop number has huge negative impact on emission maximum and peak efficiency [15]

Fig. 12 a EL spectra for RGB TE data set. b CIE 1931 color space chromaticity diagramm with respective entries. EL maxima, FWHM, and CIE x/y coordinates are given in the table below. Color space DCI-P3 and REC 2020 serve for comparison [15]

631 nm. However, already one drop in excess causes a drop in EQE by 1.2% and a shift in EL maximum by 6 nm. In case all layer thicknesses are well adjusted, the cavity effect in TE leads to very narrow emission spectra and thus, to a higher color purity compared to BE devices. Full width half maxima (FWHM) values are 31.2 nm for red, 37 nm for green, and 25 nm for blue, see Fig. 12. The corresponding CIE x/y coordinates match quite well the DCI-P3 color space.

80

S. Meyer et al.

Table 1 Comparison of best practise BE versus TE performance from Q1/2020 [15]

Best practice examples for TE performance of RGB are depicted in Table 1. As can be seen from the comparison with the set of Bottom emission data, luminous efficiency is increased substantially in top emission for red and green. For both colors, the higher wavelength intensities of the BE spectra are cut, while the main emission at lower wavelength (but closer to the maximum of human eye sensitivity) is increased by the cavity effect. In case of blue, the effect is inverted. Since higher wavelength contributions offside from the main peak are closer to the maximum of human eye sensitivity, their cut in top emission leads to a reduced luminous efficiency. Same trend holds true in EQE, which is color-independent. Significant improvements are observed for red and green, while EQE drops in case of blue. The latter can be understood from a subtle change in charge balance for the transition from BE to TE which is not favorable for the blue performance.

3 Outlook In this chapter, we have shown that RGB side-by-side IJP is a very promising technology for large-scale, top-emission, and high-resolution applications. According to the best of our knowledge, evaporated OLED displays using an RGB SBS architecture are currently only commercially available until display sizes of up to 15 . The IJP technology remains consequently the only technology enabling a side-by-side architecture for large-size displays. We have also demonstrated the strong performance improvement of the IJP OLED technology, which has caught up significantly both in terms of material development and process optimization against the VTE technology in the recent years. Nevertheless, VTE-based devices are still ahead in terms of performance as OLED vapor materials and device understanding have also continuously improved. To further close the gap, material development combined with process improvement should be accelerated. For example, future development should focus on further improving efficiency by optimizing light outcoupling via optical anisotropy. Lifetime increase is also a key topic and will remain a joint effort to be tackled not only by the development of new materials showing better stability but also by the optimization of the full soluble device. Increased color gamut by higher color purity has become another very important requirement recently.

Small Molecules in Ink Jet Printed OLEDs …

81

The concept of solution-processable ETMs, if competitive in terms of performance compared to evaporated ones, is in general very appealing also for OLED application [11]. It would allow to even further reduce the complexity of production equipment and manufacturing steps with potential benefits on lower costs.

References 1. Y. Jeong, J. Yu, K. Shin et al., Microstructure-based analysis of fine metal mask cleaning in organic light emitting diode display manufacturing. Micro Nano Syst. Lett. 7, 2 (2019). https:// doi.org/10.1186/s40486-019-0081-x 2. C.-W. Han et al., 53.2: invited paper: advanced technologies for large-sized OLED TV, in SID Symposium Digest of Technical Papers (SID ‘14) (2014), pp. 770–773 3. N. Di Giovanni, The 7 Best TVs—Summer 2020–29 May 2020 [cited 25 June 2020]. Available from: https://www.rtings.com/tv/reviews/best/tvs-on-the-market#conclusion 4. C.-W. Han, et al., 3−1: Invited paper: 3 stack–3 color white OLEDs for 4 K premium OLED TV, in SID Symposium Digest of Technical Papers, vol 48. https://doi.org/10.1002/sdtp.11555 5. H.W. Kim et al., Improvement of material utilization of organic evaporation source for manufacturing large-sized AMOLED devices, in Proceedings of the International Symposium, Seminar, and Exhibition (SID ‘08), pp. 1450–1453, May 2008, Technical Papers 39 6. C.C. Hwang, Plane source and in-line deposition system for OLED manufacturing, in Proceedings of the 44th International Symposium, Seminar, and Exhibition (SID ‘06), pp. 1542–1545, Technical Papers 37 7. T. Hayashi, A. Shimamura, Y. Konta, K. Oshima, OLED3-3: important technologies of inkjet printer and VF unit for OLED display fabrication, in Proceedings of the International Display Workshops (2018) 8. K.S. Yook, J.Y. Lee, Small molecule host materials for solution processed phosphorescent organic light-emitting diodes. Adv. Mater. 26, 4218–4233 (2014). https://doi.org/10.1002/ adma.201306266 9. C. Li, Inkjet printing for AMOLED technology & market report Li C.—05 August 2019 [cited 25 June 2020]. Available from: https://technology.informa.com/615284/inkjet-oledtechn ology-on-the-verge-of-mass-production-withcapacity-set-to-rise-12-fold-from-2020-to-2024 10. S. Meyer et al., OLED 4-4L: Recent advances in solution processed fluorescent blue OLEDs— closing the gap to vapor processed devices, in Proceedings of the International Display Workshops (2016) 11. N. Aizawa, Y. Pu, M. Watanabe et al., Solution-processed multilayer small-molecule lightemitting devices with high-efficiency white-light emission. Nat. Commun. 5, 5756 (2014). https://doi.org/10.1038/ncomms6756 12. G. Bernatz et al. 80−2: invited paper: ink jet printed film formation and its impact on OLED device performance, in SID Symposium Digest of Technical Papers, vol. 50 (2019), 1154–1156. https://onlinelibrary.wiley.com/doi/abs/10.1002/sdtp.13134 13. E. Böhm et al., 57-1: ink-jet-printed OLED displays, in SID Symposium Digest of Technical Papers, vol. 48 (2017), pp. 842–844 14. A. Hayer et al., OLED 4-2: novel materials for highly efficient long-lived solution-processed phosphorescent red OLED devices, in Proceedings of the International Display Workshops (2015) 15. S. Meyer et al., 27-2: Invited Paper: Soluble small molecules in top emission OLED devices from ink jet printing: requirements and performance status. SID Symposium Digest of Technical Papers, 51(2020), pp. 391–393. https://doi.org/10.1002/sdtp.13886

Solution-Processible OLED Material: Based on Conjugated Polymer Technology Takeshi Yamada

Abstract The basic guidelines of solution-processible OLED material design for higher efficiency and longer lifetime based on the conjugated polymer technology, and the latest progress and status are discussed. The materials are highly suitable for printing OLED panel fabrication, especially for mid- to large-sized displays. We also show the comparison of the performance between inkjet (IJ) and spin devices, and its theoretical fundamentals to secure high performance in IJ printing. Finally we discuss the expectation for IJP-OLED panel production from a viewpoint of its process, resolution and panel size, and performance as well as the comparison with WOLED.

1 Introduction The development of organic printed electronics has been expanding to a variety of applications such as organic light-emitting diodes (OLEDs), organic photovoltaic cells (OPVs), electronic papers, wearable electronics, and various sensors. It is expected that flexible, thin, and light-weight devices fabricated by cost-efficient process should bring innovation to our lives, which encourages the research and development in this area so competitive and global. As self-emitting devices, OLEDs have features of fine image-quality, are ultrathin, and are lightweight. They have become familiar to general consumers as they have been equipped broadly in smartphones such as Samsung’s Galaxy series. Such small or middle size displays have been the main products for OLEDs so far, and an evaporation process is adopted for the mass production of OLEDs. Although this process is practical when used for the mass production of small size OLED panels, it is still considered that critical difficulties exist in the fabrication of large-size displays from technical and cost points of view. The recent trends in OLED applications are shifting to large-sized TVs, such as 55-inch OLED-TV launched by LG in 2012. As this trend develops, soluble light-emitting materials have received lots of attention T. Yamada (B) Sumitomo Chemical Co. Ltd., 6, 27-1, Shinkawa 2, Chuo-ku, Tokyo, Japan e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_4

83

84

T. Yamada

from a viewpoint of low-cost mass production for larger sized mother glass over 6th generation. We have been developing organic materials applicable to printed electronics based on conjugated polymer technology, because we strongly expect the printing process to be essential for cost-competitive mass production of large-sized displays [1, 2]. In this chapter, material technologies in polymer OLEDs are reviewed, from their development history to recent progress, including our findings and experiences. A lot of materials used for organic printed electronics applications have been reported, and systematic reviews are frequently updated [3]. Such reports are very useful for grasping the characteristics of each conjugated polymer structure. On the other hand, organic electronics materials exhibit their performances as electric devices, hence their performance depends strongly on the combinations of materials, layer structures, process conditions in their fabrication, etc. These circumstances make analysis and speculation complicated. Furthermore, device durability is also an important factor from the practical application perspective. The durability is governed not only by the chemical or physical stability of materials, but also by the factors other than material properties. Therefore, parallel activities of basic analytical research and experimental screening are necessary to promote material development. Such intrinsic circumstances in this field may lead to a situation where a few years (or more) are necessary to find new materials and/or working mechanisms to bring them to practical usage.

2 Polymer OLED 2.1 Development History In 1989, a group from Cambridge University observed electroluminescence (EL) from conjugated polymer [4]. Almost at the same time, Sumitomo Chemical observed a similar phenomenon [5]. This was only a weak emission of 0.1% external quantum efficiency (EQE), and the device had a very short lifetime of a few minutes. After this discovery, Sumitomo Chemical, Covion in Germany, Dow Chemical in the US, and Cambridge Display Technology (CDT) in the UK started the energetic research and development of polymer OLED materials. As the results of over 20 years of R&D, high EQE of over 10% and long operating lifetime of over several tens of thousands hours have been achieved.

2.2 Characteristics of Polymer OLED Materials OLED emissive materials are categorized into a small-molecule group and a polymer group as shown in Fig. 1. The polymer group is further classified into non-conjugated

Solution-Processible OLED Material …

85

Fig. 1 Typical classification of emissive materials

or conjugated subgroups. Dendritic compounds can be used as an intermediate material between small-molecule and polymer groups [6]. Currently, vacuum thermal evaporation (VTE) is practically a main process for OLED production, therefore small molecule are widely used for the OLED panels. But, the shift of fabrication process from VTE to printing surely exists. Along the trend, the development of solubilized small molecules was started. The introduction or replacement of substituent groups by small molecules is the main modification to achieve solubility in ink solvents. Figure 2 shows an example of how dithienyl-benzothiadiazole derivatives were modified to become soluble materials [7]. As explained in the next section, small-molecule OLEDs should employ multilayer device architecture. Because of the requirement to form stacked layers, further modifications in chemical structures are necessary in addition to solubility. Although a lot of patents regarding this technology have been filed, solubilized small-molecule OLEDs seem to still be under development for overcoming such requirements. On the contrary, polymer OLED materials are readily soluble by ink solvents, and this results in appropriate materials for wet processes (printing processes) such as inkjet-printing and die-coating, which are considered to be appropriate for the fabrication of large-sized OLED devices. Conjugated polymers, having sp2 carbons in the main chain, exhibit high electron and hole mobility originating from the delocalized π-electrons. Many conjugated polymers with this characteristic have been reported as emissive materials. Polyphenylenvinylene (PPV) [4, 8], polyfluorene (PF) [9–12], and poly-p-phenylene (PPP) [13] are typical examples of conjugated polymer OLED materials. Two types of non-conjugated polymer OLEDs, the pendant type and the host–guest blend type, have been developed to maximize device performance. In the former case, functional units are attached to the non-conjugated backbone (polymer) and, in the latter case, functional guests are doped into the host polymer. Of these non-conjugated types, polyvinylcarbazole (PVCz) is one of the typical polymers, and its optical and electrical properties were well-analyzed in the early-stage of OLED developments [14,

86

T. Yamada

Fig. 2 Solubilized emitters of dithienyl-benzothiadiazole derivatives

15] where PVCz was used extensively as a host material. A fluorescent [16] or phosphorescent [17] host–guest polymer OLED containing PVCz in an emitting layer has also been reported. Furthermore, PVCz has been developed as a pendant type phosphorescent polymer where an iridium complex is incorporated in its side chain [18, 19]. As the fluorescent pendant version, triarylamine pendant type polymers have been studied, which offer the potential benefit of ease of synthesis compared to the conjugated type polymers [20]. Their OLED performance was evaluated as part of a mixture with the conjugated polymer in order to complement mutually insufficient functions. Owing to its ease of synthesis, non-conjugated type polymers have been used for basic studies of OLEDs rather than for practical development because the independent controllability of emissive and carrier transporting functions can be relatively easily achieved by selecting suitable functional units. Dendritic compounds are also useful for the printing process, particularly in combination with polymers. Dendrimers containing an iridium complex as the phosphorescent emitting core are known in red [21], green [22], and blue [23]. Polymer OLEDs utilizing conjugated polymers and dendrimers already achieve a high enough performance for their practical usage [24]. The conjugated polymer system works well without a complex multilayer device structure, because the required functions are introduced into polymer chains by copolymerization technology. A conjugated polymer consists of functional backbone units, electron affinity units, hole affinity units, or emissive units, as shown in Fig. 3.

Solution-Processible OLED Material …

87

Fig. 3 Co-polymerization structure of functional units of a typical light-emitting polymer

Emission color, which is one of the most important properties, is usually controlled by introducing an emissive unit to the polymer chains. Therefore, backbone units are generally selected from wide band gap structures. From this standpoint, polyfluorene [25–29] or polycarbazole [30], whose band gap is wide enough even for blue emission, are commonly used for a bacbone structure. As a typical design of a polymer, fluorene-acene co-polymers are shown in Table 1 [25]. Each acene unit was expected to emit RGB color; anthracene for blue, naphthalene for green, and pentacene for red. Devices with PDOFA (fluorene: anthracene = 9:1) with CIE color point of (0.21, 0.19), PDOFN (fluorene: naphthalene = 9:1) with (0.37, 0.45) and PDOFP (fluorene: pentacene = 9:1) with (0.64, 0.30) showed blue, green, and red EL emissions, respectively, as designed. In a similar scheme, a variety of unit structures were co-polymerized with fluorene units to obtain polymers with the desired colors [31, 32].

2.3 Features of Polymer OLED Polymer OLEDs, as well as small-molecule OLEDs, have characteristics of (a) high contrast ratio coming from a self-emissive device (luminance-on/-off), (b) wide viewing angle, (c) vivid colors, (d) thin devices, (e) high-speed image switching, and (f) low power consumption. The noteworthy feature of polymer OLEDs is the applicability of a cost-efficient fabrication process in mass production. As shown in Table 2, conventional small-molecule OLEDs, consisting of complex device struc-

88 Table 1 EL emission color of fluorine-acene co-polymers

Table 2 Comparison of characteristics between small-molecule and polymer OLED

T. Yamada

Solution-Processible OLED Material …

89

tures (multilayer structure), are fabricated by a vacuum evaporation process. On the other hand, polymer OLEDs have a simple device architecture, which can be fabricated with a printing process such as inkjetting. A typical polymer OLED consists of an anode, a hole injection layer, an emissive layer, and a cathode. The introduction of an interlayer between the hole injection layer and the emissive layer has brought a big improvement in device performance [33]. The cross-linking properties of the interlayer make it possible to stack three layers of hole injection, interlayer, and emissive layers by a printing process.

2.4 Polymer OLED Performance: Emission Efficiency The OLED efficiency can be expressed as [34] (1) EQE(external quantum efficiency) consists of (i) charge balance and exciton formation efficiency (ii) recombination and exciton generation rate (iii) material PLQY and (iv) photon out-coupling efficiency. According to Eq. (1), well-controlled injection of electron and hole, higher recombination/exciton generation rate, higher photoluminescence efficiency, and improvement of out-coupling lead to the improvement of the OLED efficiency. The basic understanding and primitive design policy for high efficiency were reviewed in a previous article [35] in detail. We would like to introduce our typical approach to polymer design with amine units to obtain blue polymers with high efficiency [36]. By screening triarylamine units, promising candidates are tested as co-polymers with fluorene backbone units for various co-polymerization ratios. In these polymers, amine units work as hole transporter as well as blue emitter. It was found that the amine content can control the hole mobility in the range of 10–7 to 10–3 cm2 /Vs, and efficiency reaches its maximum at the lowest mobility, indicated in Fig. 4. It is considered that a small amount of amines work as hole traps in the polymer host, resulting in the increase of electron and hole recombination. In the history of smallmolecule OLED development, the introduction of emissive dopants into double-layer devices was one of the breakthroughs for increasing emission efficiency [37]. As already pointed out, an interlayer is one of the remarkable technologies for a dramatic increase in the emission efficiency of polymer OLEDs [38]. The interlayer is a layer introduced between the hole injection layer and the emission layer and is considered to work as an electron and exciton blocking layer in addition to its hole-transporting function. It was demonstrated that a thin interlayer (∼10 nm) with a conjugated polymer, poly[9,9-dioctyl fluorine-co-N-(4butylphenyl) diphenylamine] (TFB), between poly(3,4-ethylenedioxythiophene) and poly(styrene sulfonic acid) (PEDT: PSS, hole injection layer) and an emissive polymer poly(9,9-dioctyl fluorine-co-benzothiadiazole) (F8BT) prevented exciton

90

T. Yamada

Fig. 4 The dependence of a hole mobility and b current efficiency of EL devices on the content of hole-transporting units in light-emitting polymers [36]

quenching by the PEDT:PSS, resulting in an improvement in the device efficiency. After this finding, a lot of investigations and material screenings were performed so that lower hole mobility by trapping could make carrier recombination confined to near the interface of the interlayer and the emission layer. In summary, the introduction of the interlayer is considered to play two roles: (1) separation of the emission zone and the hole injection layer; and (2) accumulation of electrons at the interface of the interlayer/emission layer due to its electron blocking property. For the maximum use of the interlayer function, it is preferable for the electron mobility of the interlayer to be lower than that of emission layer. One of the fundamental limits to the efficiency of a fluorescent OLED is the ratio of emissive singlet excitons to non-emissive triplet excitons formed in the electroluminescent process. Although, from simple spin statistics, only 25% of the excitons formed are singlets, there are a number of suggested mechanisms for exhibiting greater efficiency than the statistically expected value. In small-molecule OLEDs, it was reported that triplet–triplet annihilation (TTA) increased the emission efficiency by converting triplet excitons into singlets [39, 40]. It was also shown that approximately 20% of the device efficiency originated from the production of singlet excitons by TTA in the fluorescent conjugated polymer OLED using F8-PFB and an interlayer of F8-TFB [41]. A similar phenomenon was observed in a polymer OLED with a spirofluorene derivative [42]. This mechanism yielded a very high emission efficiency of 10% EQE with y = 0.13 in a blue-emitting device [43]. These technologies brought about remarkable progress in the performance of polymer OLEDs and became close to that of small-molecule type OLEDs with a multilayer device structure. To understand further the TTA mechanism in polymer OLED system, we measured and analyzed triplet behavior of fluorescent blue polymer material through time-resolved absorption and electroluminescent (TREL) spectroscopy. We observed from time-resolved absorption measurement a clear decay of triplet signal on polymer and the decay at initial stage can be fitted by 2-body collision model. This strongly

Solution-Processible OLED Material …

91

Fig. 5 Energy diagram for polymer/TCP

suggests that even in photoluminescence when the excited power is large TTA is occurred in the polymer system. We also measured direct delayed fluorescent signal originated from TTA by TREL, and it amounts ~30% of total emission from singlet. Furthermore, by addition of a polymer which has a function of TTA-sensitizing (triplet control polymer: TCP), very effective TTA was occurred on the TCP polymer (Fig. 5) and the enhancement of TTA yield was observed. It also accompanied with improvement of device lifetime. We are using this TTA mechanism in the blue-emitting polymer (Fig. 6) and realizes better efficiency and lifetime. Improvement of out-coupling efficiency has been proved through the emitting dipole alignment, which is obtained by the careful design of emitter molecule alignment in polymer system. We have evaluated the alignment of polymers in a thin film through ellipsometry and angular distribution of PL emission. From these two methodologies, same results were derived and they indicate that polymer material has a high tendency of orientation and emitting dipole alignment. Recently we have been investigating the polymer system in which emitter monomer is embedded in polymer chain, and compare it with polymer/emitter blend system. It has been clearly shown that the emitter-embedded system has a higher tendency of dipole alignment

Fig. 6 (Left) Contribution of TTA enhanced by TCP blend, and (Right) Efficiency gain by TCP blend

92

T. Yamada

Table 3 Observed emitting dipole alignment in polymer/emitter system Blend or in-chain

Emitter

Attachment

Polymer

αabs

αemit

1

Blend

Isotropic

–

Aligned

0.53

1.07

2

Blend

Isotropic

–

Aligned

0.53

0.70

3

In-chain

Anisotropic

Linear

Aligned

0.39

0.31

4

In-chain

Anisotropic

Linear

Highly aligned

0.26

0.17

than the emitter-blended system. We consider that in addition to the emitter structure and dipole design, embedding of the emitter in the polymer chain strongly forces the emitter dipole to be aligned in plane. In summary, for the higher emitting dipole alignment, we need the material conditions of (1) anisotropic emitter (2) attached linearly in (3) aligned polymer. Case 4 in Table 3 shows the highest emitting dipole alignment (αemit = 0.17) and this indicates the expected EQE from this alignment would be X1.5 higher than non-aligned case. From these findings we designed stateof-the-art green material an observed over 100 cd/A efficiency for Green (EQE > 24%) with non-cavity device.

2.5 Polymer OLED Performance: Emission Color The recent OLED display development is oriented to wider color gamut such as BT.2020 color standard. We are focusing on improvement in emission spectrum of emitter. We have found new emitters which show better peak position and narrow FWHM as shown in Fig. 7. Peak positions of RGB emitters are tunable for topemission micro-cavity and target color. Very narrow blue spectrum with FWHM of 19 nm has been achieved. A new red emitter with deeper red spectrum has been developed for wider color gamut.

2.6 Polymer OLED Performance: Device Operating Lifetime Device operating lifetime means the decay of luminance by constant-current driving. We measured lifetime and the change of PL intensity in a device to understand the degradation phenomena. We observed a decrease in PL as EL intensity decreased in the polymer OLED (see Fig. 8). From this observation, PL decay is considered as the main cause of EL decay. However, at the same time we should also consider possibilities of the change in carrier balance and/or some quenching sites due to diffusion of electrode materials as other degradation pathways. In the investigation of the PL decay mechanism, we have found that PL decay was not observed in the driving of monopolar devices (hole-major or electron-major device) as shown in Fig. 9. The device structures of the hole-major and electron-

Solution-Processible OLED Material …

93

Fig. 7 Improved spectrum from newly developed emitters

Fig. 8 Degradation curves of EL and PL intensities of EL devices containing a blue polymer during constant-current operation. Device structure: ITO/HIL/IL(20 nm)/LEP(60 nm)/cathode

major devices are ITO/HIL/IL/LEP/Au and Al/LEP/Alkali metal compound l/Al, respectively. We also performed degradation analysis of the host–guest polymer OLED device by reverse engineering techniques. The polymer OLED device was fabricated with a host polymer and guest emitter compounds. After degradation, the device was decomposed for chemical analysis of the host and guest materials, and the following results have been obtained:

94

T. Yamada

Fig. 9 PL spectra of OLEDs before and after they were operated at constant-current density with same duration. (Left) bipolar-device, (middle) hole-only device, and (right) electron-only device

• No change in the photoluminescence efficiency of the guest emitter; • Formation of an insoluble layer of polymer. These observations indicate that (1) the PL decay of emissive materials is related to the excited state by the recombination of carriers, (2) PL decay does not originate from the degradation of the emissive unit, but from the creation of quenching sites in the host polymer. Now we consider PL-stability, similarly to photo-stability, should be one of good measures to estimate device stability, given the situation that PL is major factor for lifetime. The experimental condition and results of PL-stability are summarized in Table 4. We use several excitation wavelength to excite host/blended emitters independently, and observe PL emission from host, G-em, or R-em also independently. We monitor PL emission intensity during excitation and compare them for several material classes. In Table 4, case 1 and 4 show almost identical PL-stability for Host 1, 2, and 3, whereas case 2 and 3 show significant differences in PL-stability for these hosts. This difference is supposed to be related to a difference of population of exciton on host. When the energy transfer from host to emitter is effective and the back energy transfer is prohibited, the exciton population on host will be very Table 4 Comparison of PL photo-stabilities among several excited conditions and materials

Solution-Processible OLED Material …

95

small and negligible. As a result, relatively reduced quenching site formation would be occurred on host, then better stability is realized. As reported for small-molecule OLED degradation mechanism, there is at least 2 steps in degradation (=loss of efficiency) profile for polymer system. One is an early-stage degradation and second is a longer term degradation. It is very important to understand the mechanism of early-stage degradation, as the initial loss of luminance (several % loss) results in shorter image-sticking lifetime of display. We try to understand whether the mechanism that the loss of PL is main cause of degradation is still plausible at very early-stage of device operation or not. From the PL/EL simultaneous measurement, at the very early-stage we still observe very linear relationship between PL and EL (Fig. 10). EL at T95 is totally proportional to PL, this suggests that origin of initial decay is loss of PL. And TTA contribution at T95 is also proportional to EL, this suggests that no special decay occurred for TTA. These findings strongly indicate that PL loss is a dominant factor of device degradation even at early-stage. From this observation we are trying to design more robust and durable material against environmental stress and device operation through intensive study of chemical structure screening and methodology development to create super-purity material. Generation of traps, which was observed when polymer OLED (anode/holeinjection layer/inter-layer/emitting polymer layer/cathode) device was operated, was investigated using thermally stimulated current (TSC) method. Generally for the operated p-OLED device, deep-trap was generated, whose energy level was very

Fig. 10 (left) PL/EL relationship at very early-stage degradation, and (right) Delayed-fluorescence contribution (TTA contribution) at very early-stage degradation

96

T. Yamada

Fig. 11 Observation of trap-formation from El-degraded device (left) and UV-degraded device (right)

deep and was not de-trapped under thermal treatment until 130 C temperature. Importantly the linear relationship was observed between the amount of generated deep-trap and the loss of PL intensity not only for electrically operated device but also UVdegraded device (Fig. 11). From these observations we currently understand that this trap-generation is one of the cause of PL loss in operated device, i.e., this trap acts as PL quenching site, and also one of the governing factor of device lifetime. Our understanding is that the degradation is mainly dominated by the loss of PLQY of emitting material, and this loss comes along the pathway of exciton-induced degradation. Considering this mechanism, first approach to reduce degradation should be to shorten lifetime of singlet and triplet exciton, in order to avoid further interaction. Thus, we need higher PLQY materials and an efficient TTA system to recycle triplet energy and up-convert to singlet. Figure 12 shows the relationship of triplet lifetime, photo-stability, and device lifetime. The shorter triplet lifetime, which means TTA is effective, gives a better photo-stability and a longer device lifetime. In addition to the exciton-sole pathway, we consider that SSA (singlet–singlet annihilation) process greatly affects the decay of PLQY, as the higher energy state generated by SSA has a great possibility of deteriorating materials. Thus, second approach should be to prevent SSA by confining the exciton in one-polymer or in the emitter. Based on these findings, we designed more robust polymer materials by optimizing substituents to backbone and/or using more rigid units to reach double digit improvements of device lifetime.

Solution-Processible OLED Material …

97

Fig. 12 Triplet lifetime, photo-stability, and device lifetime

2.7 Features of OLED Printing We have been also investigating the similarities and dissimilarities among OLEDfabrication processes, i.e., small-molecule evaporation, small-molecule printing, and polymer printing. The most significant differences are (i) material purities, (ii) interface formation between layers, and (iii) deposition condition of organic layers. Interface between interlayer (IL) and emissive layer (EML) should be a key to the ideal OLED-stack, which is similar to evaporative device, by soluble system. We observed three kinds of phenomena at the interface, (1) penetration of EML material into IL, (2) inter-mixing of EML material with IL polymer, and (3) dissolution of noncross-linked IL polymer into EML. To prevent these phenomena and form a clear and distinguishable interface, our system consists of (1) polymer EML (no penetration into IL) and (2) highly cross-linked polymer IL. For the effective cross-linking, we design and control an activity of cross-linker, polymer Tg, and Mw. Inkjet parameters We identified “IJP parameters” to realize the best material performance on the inkjet printed test-cell or panel [7]. Those are categorized into the areas of (1) (2) (3) (4) (5) (6)

Jettability of ink Film morphology Solvent impact (residue and impurity) Film uniformity (in pixel and across panel) IL intractability Interaction with bank.

98

T. Yamada

By precisely controlling these parameters, now we achieve a quite identical OLED performance with IJP and Spin devices. We formulate ink by employing the suitable solvents to dissolve the materials and to secure process robustness. The formulation consists of the mixture of low-bp solvent and high-bp solvent both have good solubility for the materials. The mixture solvent system gives very flat layer profile inside pixel as well as very high-uniformity across panel, when combined with best optimized drying condition. Flood-spin is a methodology to check the performance of devices, which is considered as a “transition” device between Spin and IJP. For flood-spin, we use IJP-ink even for spin-coating. Through this methodology we can check the film morphology from IJP-ink and the solvent impact (residue and impurity) by simple spin-coating. For example, when we change the drying condition of flood-spun device, we see significant variety of OLED lifetime performances (Fig. 13), this comes from the different level of residual solvent in emitting layer. Residual solvent is governed by the bp of solvent, Tg of material, and drying condition (such as temperature and vacuum pressure), and we can derive the best appropriate conditions from this very simple flood-spin method. To secure the required and uniform layer thicknesses inside pixel and across panel, we should optimize ink-formulation and drying condition. In addition to these, the printing methodology is also a major factor to secure best printing performance. We propose (1) Drive per nozzle and (2) 1-pass printing to minimize nozzle MURA and swathe-join MURA. These days, major IJ printer vendors are proposing their own methodologies for IJ printing, so we think the material and ink should be optimized independently to designated printing apparatus which is set at customer. We can optimize the material and ink, keeping the OLED performance, to match customer’s need based on our deep understanding of IJP parameters.

Fig. 13 Flood-spin and its performance

Solution-Processible OLED Material …

99

As discussed in many papers, “coffee-ring” may appear in liquid droplet drying. A side-flow occurs in any solute and solvent generally, which may cause the “coffeering” of the solute. To prevent this side-flow and to obtain a flat-film inside a pixel, increasing viscosity during drying is one of the powerful measures besides drying condition control, as increasing viscosity has a resistant effect against the side-flow. We consider that the flat-film formation in a pixel is realized through the balance of increasing viscosity and controlling the pinning position at the bank-wall, as schematically shown in Fig. 14. One example of how the layer profile inside pixel affects the device performance is shown by a model experiment. The parallel diodes with different EML thicknesses, which is a model of the deviated EML layer thickness inside one pixel, show the decreased luminescent lifetime compared to the diodes with same thicknesses. The origin of decreased lifetime is thought to be an enhanced aging at thinner area in pixel. This shows the importance of the layer thickness profile in IJP printed device. There are much higher possibilities of contamination in IJP than spin (flat) coating. The major factors of contamination are (1) dissolution of non-X-linked part of interlayer to EML (cross-contamination), (2) residual solvent in EML film, and (3) contamination from bank chemicals. To prevent the cross-contamination between interlayer/EML, we need over 98% of X-linking efficiency of interlayer. For this purpose, we developed a new X-linking system and unique polymer engineering technology to reduce small-molecular weight part (less than Mw 50 k) and to realize narrow Mw distribution. By controlling all aspects of the IJP parameters discussed above, now we have reached the status where the performance of Spin-device and IJP-device are quite identical to each other.

Fig. 14 Layer formation inside pixel

100

T. Yamada

The resolution of 55 inch-4 K and 20 inch-4 K panels are 80 ppi and 200 ppi, respectively. Also higher resolutions up to 300 ppi are desired to widen the application of IJP-OLED devices. Material/ink development for higher resolution printing is underway, by reducing Mw and improving solubility and viscoelasticity of ink for jetting properties.

2.8 Current Design of Polymer OLED Materials Red and Green emitting materials Emitter-embedded conjugated polymers. Backbone-monomers and the sequences of polymers have adequate triplet (T1) levels with controlled conjugation length through monomer design and “twisting” between adjacent monomers. Emitters are metal-centered dendritic complexes for phosphorescent emission. Blue-emitting materials Fully conjugated polymers consist of monomers of backbone, electron-transporting unit, hole-transporting unit, and emitting unit. These monomers are integrated into one-polymer chain by Suzuki-polymerization. Recently a very unique polymerization technology to make a “blocky” copolymer was established, and this brings us much better OLED performance than the materials of conventional “random” copolymers. Especially, an efficient TTA (triplet–triplet annihilation) process for better efficiency and lifetime is realized by inserting TCP (triplet control polymer) block parts into the main chain with controlled block sizes and sequences, using this blocky polymer technology. Interlayer (hole-transporting layer) materials (IL) Similar design with the blue-emitting polymers, but the interlayer polymers do not contain electron-transporting and emitting units. The polymers are designed to show much higher hole mobility than RGB emitting polymers. In addition, the polymers contain thermally cross-linking units, so the deposited layer turns to be insoluble after thermal annealing and this enables sequential layer formation by solution-process on top of IL without any inter-mixing at the interface. The soluble polymer system has several advantages over small-molecule soluble system: (1) Very soluble and ink-stable: materials are very soluble up to 3–5 wt% and very stable as ink without any precipitation or crystallization, (2) Good film formation: smooth, uniform, and flat layer formation are possible inside any size of pixels, (3) Good morphology of deposited film: film is very amorphous and without any phase separation, aggregation, or crystallization, and (4) Distinctive layer formation can be realized by thermally cross-linkable polymeric HTL layer. Given these advantages, uniform, high-performance, and low-cost OLED panel can be fabricated through printing technology.

Solution-Processible OLED Material …

101

Table 5 Current performance of polymer OLED

2.9 Current Performance of Polymer OLEDs In recent years we made a big progress in efficiency, color, and lifetime of our RGBOLED materials. In 2012, RGB T50 lifetime reached to commercially viable level; 300 khrs for Red and Green, 20 khrs for Blue. After this achievement, we have been focusing on T95 lifetime and achieved significant improvement recently. In parallel with lifetime, we have also been focusing on improving the efficiency and color, based on theoretical analysis of efficiency and discovery of emitter utilizing QC-modeling. Best achieved performance is summarized in Table 5. Based on the comparison of soluble and evaporative materials, there is still a gap to be filled for soluble Blue materials regarding efficiency and lifetime. Therefore, our main challenge is the improvement of Blue efficiency and lifetime.

2.10 Next Direction of R&D As already mentioned, polymer emissive materials have the unique feature that various required functions can be introduced into a single polymer by designing functional units and co-polymerization techniques. This feature takes advantage of the applicability of printing processes for large area device fabrication or color patterning. In terms of the aforementioned points, many kinds of emissive polymers have been

102

T. Yamada

investigated [44, 45]. For higher efficiency, phosphorescent materials with a combination of small-molecular phosphorescent compounds and carrier transporting polymers have been proposed as polymer phosphorescent OLED materials. In that study, non-conjugated polymers with carrier transporting units attached to their side chains were commonly used. This type of polymer was considered to be easier for controlling singlet or triplet energy levels and other physical parameters. Basic research for understanding the effect of carrier transporting units on phosphorescent units using non-conjugated polymers other than PVCz has continued [46, 47]. The combination of polymers and phosphorescent emitters is very important to obtain high efficiency in further investigations. Recently new conjugated polymers which can be used as blue phosphorescent materials were reported [48]. They have polyphenylene chains connected by meta-linkage in the main chains, and exhibit 4.69 cd/A efficiency originating from a typical light-blue iridium complex, Firpic. As conjugated polymers with delocalized electrons, they possess excellent carrier transport properties, and the design policy of soft-controlled conjugation is studied for better and improved materials. As another approach, research on the phase separation in polymer blends is ongoing to uncover clues about high efficiency. In some trials, adding polystyrene to conjugated polymers has shown an increase in efficiency [49], and the investigation of a relation between EL properties and phase separation in conjugated polymer mixtures was reported [50]. We would like to introduce new topical high-efficiency materials. There have been two approaches of materials systems to obtain high efficiency; TTA fluorescent materials using an up-conversion mechanism, and phosphorescent. materials utilizing emissive triplet levels. As an up-conversion mechanism other than TTA, a thermally activated delayed fluorescence (TADF) process was proposed for converting non-emissive triplet excited states into emissive singlet excited states [51]. The energetic development of TADF materials was started to give us other options to avoid potential issues such as the use of the rare metal iridium, difficulties in developing pure blue phosphorescent materials, and so on. As in the energy diagram shown in Fig. 15, the feature of TADF materials is considered as a small energy difference (EST) between the singlet (S1) and the triplet (T1) energy levels [52, 53]. TADF materials, which thermally activate the reverse intersystem-crossing (RISC) from the triplet excited state to the singlet excited state, lead to an increase of fluorescence intensity, therefore high EL efficiency can be achieved in OLEDs with TADF materials. Furthermore, the roll-off characteristics of the EL efficiency can be reduced if the rate constant of RISC is significantly larger than the phosphorescent decay rate. Adachi’s group started a search for efficient TADF materials from tin (IV) fluoride porphyrin complexes. TADF materials with various colors were reported [53, 54], and remarkable materials of pure blue with 10% EQE [55], deep blue with 14.5% EQE [56], and very high EQE of 16.5% [57] were obtained. This material design is expected to be introduced into polymer OLED materials. Small-molecule OLEDs have a longer development history, so it takes a lead in practical usage compared with polymers. A vacuum evaporation method is used to form layers of OLED and many technical improvements have been developed in this

Solution-Processible OLED Material …

103

Fig. 15 Process of formation of singlet and triplet excitons in the electroluminescence process [53]

area. Recently, soluble small-molecule materials have received much attention as many material companies have reported or press released their progress. Introducing soluble groups to existing emissive structures are common molecular designs [50]. As an OLED performance of soluble phosphorescent small molecules, LT50 of about 130 000 h in red, 150 000 h in green, and 8000 h in blue [58] have been observed, and are estimated to be close to evaporated type OLEDs. Soluble small molecules are becoming competitive technologies versus polymers. It is considered that further progress in these soluble materials should accelerate the expansion and growth of the organic printed electronics market, including OLEDs.

2.11 Future Perspectives Up to now, WOLED (white OLED + color filter) is the only one technology to produce large-sized OLED panels, and the technology has been adopted to produce OLED-TV. IJP-OLED (Inkjet printed RGB-OLED) is expected to be one of the next technology to follow WOLED, as IJP is essentially scalable and limitless on panel-size. In addition, IJP-OLED is also considered to have several merits over WOLED, which include (1) lower depreciation cost, (2) lower power consumption, (3) possibility for higher resolution up to 8 K with top-emission configuration, (4) better sub-pixel lifetime, and (5) better color-reproducibility by effective microcavity device architecture. Currently, in addition to IJP-OLED, QD-OLED (Blue stack + QD color filter) technology is also proposed and intensively developed. IJP-OLED and QD-OLED are thought to be competitive technologies for large-sized panel production.

104

T. Yamada

3 Conclusion In this chapter, the material technology in polymer OLEDs was reviewed from the development histories to recent progress, including our experiences. The features of OLEDs with polymers (PLEDs) were explained in comparison with OLEDs with small molecules. The main strengths of PLEDs are suitability to the printing process for device fabrication, and their simple device structure. Conjugated polymers can be well designed to meet the required performances by co-polymerization technology. A variety of units have been developed, and polyfluorene is widely and practically used from standpoints of wide band gap and high conductivity of charge carriers. Regarding polymer OLED performances, higher emission efficiency and longer operating device lifetime were picked up as the main subjects for development. We introduced an example of polymer design whereby controlling the hole mobility with amine unit contents in the polymer chain optimizes the emission efficiency. The impact of introducing an interlayer on the efficiency is an epoch-making technology in PLED, and TTA that is commonly used in fluorescent materials as an up-conversion technology was also introduced. By using these techniques, over 10% of EQE in a blue fluorescent PLED was obtained. Concerning device operating lifetimes, forming excited states was considered to be a key cause of the degradation process from the comparison of photoluminescence changes by operation of bipolar and monopolar devices. In the latter section, future trends of polymer OLED materials were indicated. Improvements of sticking image lifetime by the development of new materials is expected, and such investigation is predicted to become more active. Combinations of polymer and phosphorescent emitters, especially blue, was listed as one of these new materials. TADF, that is another up-conversion type material, was also introduced as a topical high-efficiency material. As very high efficiency has already been obtained with small-molecule type materials, it is expected that TADF will be applied to polymer materials as well. For the technologies of polymer OLED, it can be said that fundamental material performances have become close to practical levels. However, we should understand that further efforts are necessary to finish the material development to be used for real device products. Soluble-OLED materials are strongly tied to device architectures and fabrication processes to demonstrate their performances, thus parallel developments in materials and device/process are expected to be effectively promoted continuously.

References 1. G. Gustafsson, Y. Cao, G.M. Treacy, F. Klavetter, N. Colaner, A.J. Heeger, Nature 357, 477 (1992) 2. S. Tekoglu, G. Hernandez-Sosa, E. Kluge, U. Lemmer, N. Mechau, Org. Electron. 14, 3493 (2013)

Solution-Processible OLED Material …

105

3. X. Guo, M. Baumgarten, K. Müllen, Prog. Polym. Sci. 38, 1832 (2013) 4. J.H. Burroughes, D.D.C. Bradley, A.R. Brown, R.N. Marks, K. Mackay, R.H. Friend, P.L. Burn, A.B. Holmes, Nature 345, 539 (1990) 5. T. Ohnishi, T. Nakano, D. Doi, T.N. Tokukaihei 3-244630 6. P.L. Burn, C. Victor, S.-C. Lo, J. Pillow, N. Gerard, L.J. Mark, I.D.W. Samuel, WO02/66552 7. J. Huang et al., Adv. Funct. Mater. 19, 2978 (2009) 8. N.C. Greenham, S.C. Moratti, D.D.C. Bradley, R.H. Friend, A.B. Holmes, Nature 365, 628 (1993) 9. M. Bernius, M. Inbasekaran, J. O’Brien, W. Wu, Adv. Mater. 12, 1737 (2000) 10. M. Bernius, M. Inbasekaran, E. Woo, W. Wu, L. Wujkowski, J. Mater. Sci. Mater. Electron. 11, 111 (2000) 11. M. Bernius, M. Inbasekaran, E. Woo, W. Wu, L. Wujkowski, Thin Solid Films 363, 55 (2000) 12. M. Inbasekaran, E. Woo, M. Bernius, L. Wujkowski, Synth. Met. 111–112, 397 (2000) 13. G. Grem, G. Leditzky, B. Ullrich, B. Leising, Adv. Mater. 4, 36 (1992) 14. R.H. Partrige, Polymer 24, 733 (1983) 15. R.H. Partrige, Polymer 24, 755 (1983) 16. Y. Mori, H. Endo, Y. Hayashi, Oyo Butsiri 61, 1044 (1992) 17. S. Takizawa, J. Nishida, T. Tsuzuki, S. Tokito, Y. Yamashita, Inorg. Chem. 46, 4308 (2007) 18. X.-Y. Wang, A. Kimyonok, M. Weck, Chem. Comm. 37, 3933 (2006) 19. S. Tokito, M. Suzuki, F. Sato, M. Kmamachi, K. Shirane, Org. Electron. 4, 105 (2003) 20. A. Gupta, S.E. Watkins, A.D. Scully, T.B. Singh, G.J. Wilson, L.J. Rozanski, R.A. Evans, Synth. Met. 161, 856 (2011) 21. M.J. Frampton, E.B. Namdas, S.C. Lo, P.L. Burn, I.D.W. Samuel, J. Mater. Chem. 14, 2881 (2004) 22. S.C. Lo, T.D. Anthopoulos, E.B. Namdas, P.L. Burn, I.D.W. Samuel, Adv. Mater. 17, 1945 (2005) 23. S.C. Lo, R.E. Harding, C.P. Shipley, S.G. Stevenson, P.L. Burn, I.D.W. Samuel, J. Am. Chem. Soc. 131, 16681 (2009) 24. J. Pillow, Z. Liu, C. Sekine, S. Mikami, M. Mayumi, SID’05 Digest 1071 (2005) 25. S. Tokito, K.H. Weinfurtner, H. Fujikawa, Y. Taga, Gekkan Display 6, 26 (2000) 26. C. Ego, A.C. Grimsdale, F. Uckert, G. Yu, G. Srdanov, M. Inbasekaran, E. Woo, Adv. Mater. 14, 809 (2002) 27. T. Miteva, A. Meisel, W. Knoll, H.G. Northofer, U. Scherf, D.C. Müller, K. Meertholz, A. Yasuda, D. Neher, Adv. Mater. 13, 565 (2001) 28. J.H. Lee, D.H. Hwang, Chem. Comm. 2836 203 29. D. Vak, C. Chun, C.L. Lee, J.J. Kim, D.Y. Kim, Mater. Chem. 14, 1342 (2004) 30. Y. Ohmiri, Oyo Butsuri 76, 522 (2007) 31. W. Wu et al., Microelectron. J. 35, 343 (2004) 32. Q. Hou, Q. Zhou, Y. Zhang, W. Yang, R. Yang, Y. Cao, Macromolecules 37, 6299 (2004) 33. A.C. Morteani, A.S. Dhoot, J.-S. Kim, C. Silva, N.C. Greenham, C. Murphy, E. Moons, S. Ciná, J.H. Burroughes, R.H. Friend, Adv. Mater. 15, 1708 (2003) 34. S. Tokito, C. Adachi, H. Murata, Organic EL displays, 21, Ohm-sha (2004) 35. T. Tsutsui, N. Takata, J. Appl. Phys. 110001 (2013) 36. T. Yamada, Y. Tsubata, C. Sekine, T. Ohnishi, SID Sym. Digest Tech. Papers 39, 404 (2008) 37. C.W. Tang, S.A. VanSlyke, C.H. Chen, J. Appl. Phys. 65, 3610 (1989) 38. J.S. Kim, R.H. Friend, I. Grizzi, J.H. Burroughes, Appl. Phys. Lett. 87, 023506 (2005) 39. D.Y. Kondakov, T.D. Pawlik, T.K. Hatwar, J.P. Spindler, J. Appl. Phys. 106, 124510 (2009) 40. H. Kuma, The 7th Yuki EL Toronkai S9-1 (2008) 41. S.M. King, M. Cass, M. Pintani, C. Coward, F.B. Dias, A.P. Monkman, M. Roberts, J. Appl. Phys. 109, 074502 (2011) 42. A.P. Monkman, ISRN Mater. Sci. 19, 670130 (2013) 43. M. Roberts, S. King, M. Cass, M. Pintani, C. Coward, N. Akino, H. Nakajima, M. Anryu, SID Sym. Digest Tech. Papers 42, 1820 (2011) 44. J. Liu, Q. Pei, Curr. Org. Chem. 14, 2133 (2010)

106

T. Yamada

45. 46. 47. 48. 49. 50. 51.

C. Lee, K. Lee, J. Kim, Appl. Phys. Lett. 77, 2280 (2000) Motomura, The 9th Yuki EL Toronkai S4-2 (2009) Motomura, The 10th Yuki EL Toronkai S2-2 (2010) J. Liu, Q. Pei, Macromolecules 43, 9608 (2010) G. He, Y. Li, J. Liu, Y. Yang, Appl. Phys. Lett. 80, 4247 (2002) K. Kim, W. Doherty, W. Salaneck, C. Murphy, R. Friend, J. Kim, Nano Lett. 10, 385 (2010) A. Endo, M. Ogasawara, A. Takahashi, D. Yokoyama, Y. Kato, C. Adachi, Adv. Mater. 21, 4802 (2009) H. Uoyama, K. Goushi, K. Shizu, H. Nomura, C. Adachi, Nature 492, 235 (2012) C. Adachi, Oyo Butsuri 82, 458 (2013) H. Tanaka, K. Shizu, H. Nakanotani, C. Adachi, Chem. Mater. 25, 3766 (2013) Q. Zhang, J. Li, K. Shizu, S. Huang, S. Harata, H. Miyazaki, C. Adachi, J. Am. Chem. Soc. 134, 14706 (2012) S. Wu, M. Aonuma, Q. Zhang, S. Huang, T. Nakagawa, K. Kuwabara, C. Adachi, J. Mater. Chem. C 2, 421 (2014) K. Nasu, T. Nakagawa, H. Nomura, C.J. Lin, C.H. Cheng, M.R. Tseng, T. Yasuda, C. Adachi, Chem. Commun. 49, 10385 (2013) E. Bohm et al., SID-ME Spring Meeting (2011)

52. 53. 54. 55. 56. 57. 58.

Chemical Mechanisms of Intrinsic Degradation of Emitting Layers in Organic Light-Emitting Devices Youngmin You

Abstract The short operation lifetime of organic light-emitting devices (OLEDs) remains as a major hurdle. The poor device longevity results from accumulations of defects during normal operation. Strategies to improve device stability require understanding of the chemical processes underlying the generation and annihilation of defects. This chapter summarizes the current knowledge about the chemical degradation of organic materials in OLEDs. Focus is placed on the chemical mechanisms of the defect generation. Unimolecular degradation from excitons or polarons is discussed with key examples. Chemical degradation by bimolecular processes, including exciton-exciton annihilation and exciton-polaron annihilation, are summarized. Strategies toward minimizing bimolecular degradation are also introduced. Finally, a recently identified bimolecular degradation pathway involving exciton-mediated electron transfer is overviewed.

1 Introduction Organic light-emitting devices (OLEDs) have emerged as important components in display applications. Flat-panel displays incorporating OLEDs are commercially available, and they are rapidly replacing conventional displays based on rigid liquidcrystal panels. The potential utility of OLEDs is indeed broad, as the devices can be fabricated in nonplanar form factors. One challenge in the exploitation of OLEDs is their moderate operation lifetime. Although substantial advances have been made, poor device longevity impedes the full realization of the benefits of OLEDs. The very short lifetime of blue-emitting OLEDs, in particular, is the largest remaining hurdle. OLEDs based on recently developed dopants that exhibit thermally activated delayed fluorescence (TADF) also experience lifetime problems. Although extensive research has been devoted Y. You (B) Division of Chemical Engineering and Materials Science, and Graduate Program in System Health Science and Engineering, Ewha Womans University, Seoul 03760, Republic of Korea e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_5

107

108

Y. You

Fig. 1 Deterioration of device performances by defects

to solving this problem, the efforts commonly suffer from a dearth of useful strategies for creating robust devices. This lack of strategies is due mostly to insufficient understanding of the processes responsible for the poor device stability. There is broad agreement that the instability of OLEDs results from an accumulation of defects. Device degradation during normal operation is unavoidable because the device cycle, which includes injection, transport, and recombination of charge carriers, involves vulnerable species such as polarons and excitons. The defects adversely affect the performance of OLEDs (Fig. 1): (1) defects trap charge carriers, reducing power efficiency; (2) defects quench emissive excitons through energy transfer, decreasing quantum efficiency; (3) defects serve as nonemissive centers for charge-carrier recombination, further lowering quantum efficiency; (4) defects are chemically reactive, and therefore propagate. Thus, minimizing the generation of defects during device operation is critical. In this chapter, I classify the chemical mechanisms of the defect generation, focusing on defects produced by the intrinsic degradation of emitting layers. I hope that this chapter will serve as a beginner’s guide to the study of intrinsic degradation of organic materials in OLEDs. Interested readers are strongly encouraged to read the recent excellent reviews [1–5].

2 Types of Defects Defects can be classified into physical and chemical defects, depending on whether the molecular structure changes (chemical defect) or not (physical defect) (Fig. 2). Routinely observed physical defects include a phase transition during device operation. Excitons relax to the ground state, accompanied by the emission of light and heat. “Thermal relaxation” refers to a nonradiative transition or phonon scattering from a microscopic perspective. “Joule heating” is a macroscopic description of this process. Because the internal quantum yield of an OLED is usually less than 100%, Joule heating is unavoidable. The heating promotes the phase transition of organic films from an amorphous state to a crystalline state. The crystalline state usually consists of multigrain boundaries; thus, charge-carrier transport

Chemical Mechanisms of Intrinsic Degradation …

109

Fig. 2 Types of intrinsic physical defects

behaviors are strongly affected by the thermal phase transision. The amorphousto-crystalline phase transition also leads to changes in the emission properties. For example, Reineke and co-workers recently found that the deposition temperature during device fabrication was intimately linked to the operational lifetime of OLEDs [6]. Multilayer OLEDs with a phosphorescent [Ir(ppy)2 (acac)] dopant (ppy = 2-phenylpyridinate; acac = acetylacetonate) and a 2,2 ,2 -(1,3,5-benzinetriyl)tris(1-phenyl-1-H-benzimidazole) (TPBi) host were fabricated at different deposition temperatures (32, 69, and 90 °C). Device stability was found to increase with increasing deposition temperature. This improvement was attributed to denser packing of the organic molecules. Nonradiative relaxation and the other dissociative processes could be kinetically avoided at such densely packed states. Another physical defect is adhesion failure between two successive layers. This failure is due to different thermal expansion behaviors of the two layers. Accumulations of physical defects can be avoided by using organic materials with high glass transition temperatures or with similar thermal expansion coefficients. Chemical defects are major sources of degradation. Chemical defects involve a bond cleavage in a material. The generation of excitons or polarons frequently leads to the irreversible bond breakage because the former is a high-energy species and the latter usually has a lower activation barrier for the cleavage reaction. Figure 3 summarizes the routes to the formation of chemical defects. Holes (or positive polarons) injected from an indium tin oxide (ITO) anode can accumulate at the interface between a hole-transporting layer (HTL) and an emitting layer (EML). Although the majority of HTLs exhibit high stability against oxidative stress, the interfacial hole-charge build-up can initiate degradation of the radical cations of the HTL. The accumulated holes can also undergo interfacial exciton-polaron annihilation (EPA) if an exciton-formation zone in the EML is located in the vicinity of the HTL. EPA can occur within the EML at large hole concentrations. The EPA between exciton and negative polarons cannot be excluded, although its probability is usually lower than that of EPA involving a positive polaron. Charge-carrier recombination generates excitons. The exciton is also subjected to degradation when the excited-state potential energy surfaces are repulsive. Examples of this exciton-localized degradation will be outlined later. Exciton-exciton annihilation (EEA) can occur at large current densities. The occurrence of EEA becomes

110

Y. You

Fig. 3 Various pathways to the generation of chemical defects in an OLED. Symbols: excitonD , dopant exciton; excitonH , host exciton; polaronH + , positive polaron of the host; polaronH − , negative polaron of the host; polaronHTL + , positive polaron of the HTL

more likely if excitons form within triplet dopants with a long lifetime. EEA yields high-energy species that are highly susceptible to degradation. The penetration of molecular oxygen (O2 ) and water into the EML can lead to degradation. The mechanism of this degradation reaction has not been thoroughly investigated but may involve sensitization of highly oxygenating singlet oxygen (1 O2 ) by triplet excitons. In many cases, hole transport is faster than electron transport within an EML. This different charge-carrier transport behavior can lead to the accumulation of holes at the interface between an EML and a hole-blocking layer (HBL). The hole-charge buildup leads to increased probability of EPA. Leakage of holes through the HBL into an electron-transporting layer (ETL) generates a radical cation in the ETL. Because ETLs are designed to be more stable in the negative polaron state than in the positive polaron state, the hole penetration results in prompt degradation of the ETL. Finally, the hole can further reach the interface between the ETL and the metal cathode. In this case, the cathode can suffer exciton-induced degradation. Notably, all of the destructive routes outlined in Fig. 3 result from normal device operation, implying that the accumulation of chemical defects is inevitable. Compared with the approaches to minimizing the physical defects, the approach to minimizing chemical defects are more difficult to establish because the formation of chemical defects depends sensitively on the molecular structure. For example, phosphorescent dopants based on cyclometalated Ir(III) complexes experience excitonic degradation resulting from hexacoordinate-to-pentacoordinate geometric distortion. On the contrary, such a change in coordination does not occur in organic molecules,

Chemical Mechanisms of Intrinsic Degradation …

111

including TADF compounds. Therefore, thorough chemical backgrounds are essential for establishing strategies toward suppressing chemical defects. Examples of chemical degradation will be discussed in the following sections.

3 Chemical Mechanisms of Defect Formation Chemical processes related to the intrinsic degradation of organic materials are under active research. This section classifies chemical degradation into unimolecular and bimolecular mechanisms, depending on the chemical nature of the reactive intermediate. The unimolecular mechanism usually involves either an exciton or a polaron as a degradation intermediate. Bimolecular mechanisms include degradation resulting from homo- or heterobimolecular interactions such as EEA and EPA, respectively. A recently discovered bimolecular mechanism based on exciton-mediated intermolecular electron transfer is also introduced.

3.1 Unimolecular Degradation Mechanism 3.1.1

Excitonic Degradation

The majority of organic materials are inert in the ground state. This ground-state stability is due to huge activation barriers for destructive reactions. However, the materials can degrade in their excitonic state if the dissociative (repulsive) potential energy surface crosses the excitonic potential energy surface. As a result, the exciton can bypass the ground-state activation barrier, producing degradation byproducts. Figure 4 presents a simplified mechanism of the excitonic degradation. Extensive efforts have been devoted to elucidating the excitonic degradation of phosphorescent dopants. Cyclometalated Ir(III) complexes have been widely studied. Leo and co-workers investigated the chemical stability of [Ir(MDQ)2 (acac)] (MDQ = 2-methyldibenzo[f ,h]quinoxaline) [7]. Their investigation revealed decoordination of the acac ligand in the excitonic state. The resultant positively charged [Ir(MDQ)2 ]+ fragment further dissociated into [Ir(MDQ)]2+ and an MDQ radical (Fig. 5). These fragments could react with organic materials in the vicinity of the EML, leading to spectral deterioration. Similar excitonic decoordination of ligands from Ir(III) complexes has also been proposed, and its plausibility is supported by quantum chemical calculations [8–10] In particular, the calculations suggest that partial decoordination as well as full decoordination of a bidentate ligand can adversely affect device performance (Fig. 5) [11]. Excitonic degradation can be avoided by preventing access to dissociative states. Yam and co-workers investigated the photophysical properties of a series of Au(III) pincer complexes (Fig. 6) [12]. The Au(III) complexes displayed unique prompt and thermally stimulated delayed dual phosphorescence emission. The complexes

112

Y. You

Fig. 4 Mechanism of excitonic degradation * N

+

2+

N N Ir

N N

O

+

Ir

O

O

N

N Ir

+

N

O

2

2

* N

N N

N

N Ir

N

N

hexa-coordiate

N N

N Ir N

N

penta-coordiate

Fig. 5 Excitonic degradation of phosphorescent Ir(III) complexes

possessed monodentate ligands with different electron densities. The electronic control perturbed the energy of the intraligand charge-transfer (ILCT) transition state within the tridentate ligand. A higher lying intraligand (IL) state was present and was presumed to be responsible for degradation of the complexes. Therefore, access to the destructive IL state was avoided by the use of an appropriate monodentate ligand. In addition to the inner-sphere changes, excitons of phosphorescent dopants are also susceptible to degradation by 1 O2 . Highly reactive 1 O2 can be generated by triplet–triplet energy transfer from a phosphorescent exciton. König and co-workers characterized the 1 O2 -mediated oxygenative degradation of Ir(III) complex dopants

Chemical Mechanisms of Intrinsic Degradation … t-Bu

N

t-Bu

N

N

N t-Bu

t-Bu

N

N

N t-Bu

Au

F

113

t-Bu

N N

t-Bu

Au

t-Bu

Au

t-Bu

F t-Bu

CF3

3

IL

3

ILCT

prompt phosphorescence

degradation

delayed phosphorescence

Fig. 6 Phosphorescent Au(III) complexes and their photophysical processes

[13]. As depicted in Fig. 7, a benzylic carbon in the cyclometalating ligand can be oxidized into a formyl group by 1 O2 . This result demonstrates the importance of protecting the EML against air. Kondakov and co-workers reported an interesting degradation in which dehydrogenative cyclization was observed for 9,10-dinaphthylanthracene, a well-known fluorescence host (Fig. 8) [14]. The cyclization produced an annulated benzofluoFig. 7 Singlet oxygen-mediated degradation of an Ir(III) complex

Fig. 8 Degradation of a fluorescence host by dehydrogenative cyclization

114

Y. You

rancene compound. Although the chemical mechanisms of the cyclization remain elusive, the observations demonstrate that even highly stable aromatic hydrocarbons are susceptible to excitonic degradation. Excitonic degradation occurs in both hosts and dopants; however, operational stability strongly depends on where the exciton is generated. Duan and co-workers have reported that exciton formation on a host (Langevin recombination) led to worse device stability than exciton formation on a dopant (Shockley–Read–Hall recombination) [15]. They attributed this contrasting stability to the slow reverse intersystem crossing (rISC) in the host exciton. Contrary to a TADF dopant with kinetically promoted rISC, pure organic hosts are incapable of converting a triplet exciton into a singlet exciton. This inability leads to a very long triplet exciton lifetime of the host, which results in degradation. The results of Duan and co-workers demonstrate the importance of charge-carrier recombination within dopants for improved stability. Kondakov et al. conducted chemical analyses of the excitonic degradation of 4,4-bis(N-carbazolyl)-1,1-biphenyl (CBP) [16]. Using various chemical techniques, they identified a multitude of degradation byproducts of CBP. They subsequently proposed that the CBP exciton promoted homolysis of the C-N bonds in CBP, generating radical species. The reactive radical species initiated subsequent reactions to produce multiple degradation products (Fig. 9). Han and co-workers reported a similar degradation of 4,4-bis(N-carbazolyl)-2,2-dimethylbiphenyl (CDBP) [17]. In the case of CDBP, excitonic degradation involves the cleavage of the C–C bond between the two phenyl rings. Adachi and co-workers compared the stabilities of a 3,3 -di(9H-carbazol-9-yl)-1,1 -biphenyl (mCBP) host and a TADF dopant [18]. Specifically, they compared the photolytic degradation behaviors for neat films of a TADF dopant, a mCBP host, and their composite films. Photoluminescence measurements during the photolytic degradation revealed that the TADF dopant was more stable than the host. Mass spectrometric analyses of the degradation byproducts indicated C-N bond cleavage in both the host and the TADF dopant.

Fig. 9 Excitonic degradation of CBP

Chemical Mechanisms of Intrinsic Degradation …

3.1.2

115

Polaronic Degradation

Compared with the neutral state, polaron states (i.e., radical ions) are usually unstable even in the absence of exciton formation. This instability originates from a lowered activation barrier for degradation (Fig. 10). The plausibility of polaronic degradation has been predicted by Brédas and co-workers. They performed quantum chemical calculations to simulate degradation of the radical anion of a bipolar host consisting of carbazole and dibenzothiophene (Fig. 11) [19]. They predicted that the added electron in the radical anion would be localized within the dibenzothiophene moiety. Spontaneous cleavage of the C-N bond connecting the carbazole and dibenzothiophene moieties was found to occur through intramolecular electron transfer from the dibenzothiophene radical anion to carbazole. Quantum chemical/molecular dynamics calculations by Van Voorhis and coworkers also suggested that polaronic degradation is likely [20]. They simulated a disordered cell comprising host molecules and [Ir(ppy)3 ] dopants. The electron

Fig. 10 Unimolecular mechanism for the intrinsic degradation of a polaron

Fig. 11 Polaronic degradation of a bipolar host

116

Y. You

affinity and ionization potential values were calculated for the simulated ensembles of the host and dopant. The calculations indicated that hole trapping by the dopant was faster than electron trapping. The different charge-carrier trapping rates suggest non-negligible concentrations of polarons of the dopant even at a perfect charge-carrier balance within an EML. There are few experimental reports on the polaronic degradation of materials. Tang and co-workers investigated the operational stability of a multilayer OLED with a configuration of ITO/MoO3 /NPB/mCBP/FIrpic/Alq3/LiF/Al (NPB = N,N  -di(1-naphthyl)-N,N  -diphenyl-(1,1 -biphenyl)-4,4 -diamine; FIrpic =  iridium(III) bis[2-(4,6-difluorophenyl)-pyridinato-N,C 2 ]picolinate; Alq3 = tris(8hydroxyquinolinate)aluminum) [21]. FIrpic was used as a blue-phosphorescent dopant. Note that the authors employed an unusual mCBP/FIrpic double-layer structure to compare the intrinsic stability of the dopant (i.e., FIrpic) and the host (i.e., mCBP) in their polaronic states. They concluded that FIrpic was unstable in its positive polaronic state. Leo and co-workers had previously proposed that this instability originates from oxidative decarboxylation (Fig. 12) [22]. Degradation of FIrpic was also found to be mediated from a negative polaron [23]. Li and coworkers reported that suppression of the interfacial charge build-up can minimize electrochemical degradation of a phosphorescent Pt(II) complex [24]. Qiu and co-workers investigated the excitonic stability of a series of ambipolar hosts (Fig. 13) [25, 26]. Chemical analyses of the degradation byproducts revealed cleavage of carbon-heteroatom bonds. Quantum chemical calculations of bonddissociation energies (BDEs) in the neutral and radical ion states identified vulnerable bonds that are cleaved in various states. In the case of CzSF in Fig. 13, BDE of the CS bond in the negative polaron state was calculated to be smaller (29.7 kcal mol−1 ) than those of the charge-neutral (66.7 kcal mol−1 ) and the positive polaron states (91.6 kcal mol−1 ). Similarly, the C-P bonds in CzPO2 become significantly weaker in the negative polaron state, as inferred from a BDE smaller (42.9 kcal mol−1 )

Fig. 12 Polaronic degradation of FIrpic

Chemical Mechanisms of Intrinsic Degradation …

117

Fig. 13 Bond dissociation energies of the C-X (X = S, C, and P) bonds in ambipolar hosts in their charge-neutral, positive polaron, and negative polaron states

than those of the charge-neutral (81.2 kcal mol−1 ) and the positive polaron states (70.5 kcal mol−1 ). On the contrary, the bond cleavage of SBFK was predicted to occur in its excitonic state, as its polaron states have BDEs greater than that of the neutral form. The results demonstrate that the identity of a dissociative intermediate depends on the chemical structure of a material.

3.2 Bimolecular Degradation Mechanism 3.2.1

Degradation by Exciton-Exciton Annihilation

EEA is a bimolecular upconversion process because it generates a species higher in energy than the original states. Triplet–triplet annihilation (TTA), which is responsible for roll-offs in device efficiencies, is one type of EEA. In this process, two triplet excitons interact to generate one singlet exciton and one ground-state species. Although TTA has served as a viable approach to achieving exciton-harvested electrofluorescence, it frequently adversely affects the stability of devices. The product of TTA (i.e., a singlet exciton) is in a high vibrational energy state and is, therefore, susceptible to bond dissociation (Fig. 14). In addition, the high-energy electronic state produced by EEA can cross the repulsive states, leading to degradation. The occurrence of EEA requires high exciton densities and is therefore usually observed at large current densities. Avoiding degradation by EEA remains a challenge because bright electroluminescence inevitably requires high densities of excitons. To minimize EEA, the exciton lifetime should be as short as possible. One viable approach is to use dopants that exhibit fast radiative decay. An alternative approach is to avoid triplet exciton formation within hosts.

118

Y. You

Fig. 14 Bimolecular mechanism for intrinsic degradation initiated by exciton-exciton annihilation

Auger processes, including EEA, have been proposed to provide access to dissociative “hot” excited states. Long-lived triplet excitons of Ir(III) complexes are prone to EEA. To minimize degradation by EEA, Forrest and co-workers incorporated a manager Ir(III) complex into an EML [27]. The added manager compound had a triplet state located between the “hot” excited state and the lowest triplet state of the emissive dopant (Fig. 15). This energetic alignment promoted energy transfer from the hot emissive dopant generated from EEA to the triplet state of the manager compound, followed by subsequent back energy transfer to the lowest triplet state of the emissive dopant. This cyclic energy transfer enabled the dissociative state accessible from the hot states to be bypassed. EEA is not limited to identical excitons. Annihilation between host excitons and dopant excitons can also lead to degradation. Adachi and co-workers recently proposed that a polaron can be generated by energy transfer from host excitons to dopant excitons [28]. TADF dopants are long-lived because of an equilibrium between the singlet and triplet excited states. The longer exciton lifetime might enable energy transfer from the host exciton. If a host exciton has an energy greater than the ionization energy of a dopant exciton, one electron in the LUMO can be ejected (Fig. 16). The resulting positive polaron of a dopant leads to polaronic degradation. The Adachi group investigated the plausibility of this process using donor–acceptor type TADF dopants and high-bandgap-energy hosts, including bis(2-(diphenylphosphino)phenyl)ether oxide (DPEPO), 9-(4-tert-butylphenyl)-3,6bis(triphenylsilyl)-9H-carbazole (CzSi), mCBP, and SF3K [28]. The extent of degradation of dopants was found to be proportional to the bandgap energy of the hosts.

Chemical Mechanisms of Intrinsic Degradation …

119

Fig. 15 Cyclic bimolecular photophysical processes that bypass the Auger-induced dissociation hosts Si

Si

O P O

dopants N

N

O P

N

N

N

N

N

N

N N

N

N

DPEPO E(host) = 4.0 eV

t-Bu CzSi E(host) = 3.5 eV N

O N

N N mCBP

N N

N

N

O S O

SF3K

E(host) = 3.5 eV

E(host) = 3.4 eV

vacuum level

vacuum level excited-state ionization potential of dopant (Ip*(dopant)) LUMO

LUMO

bandgap energy of host ( E(host))

energy transfer

excited-state absorption when E(host) > Ip*(dopant) degradation HOMO

HOMO host exciton

dopant exciton

host

dopant

Fig. 16 Heteromolecular exciton-exciton annihilation to produce a vulnerable dopant polaron

120

3.2.2

Y. You

Degradation by Exciton-Polaron Annihilation

EPA is characterized as transfer of energy from a neutral exciton to a doublet polaron (Fig. 17). This energy transfer produces an excited-state polaron that is highly unstable. Pioneering studies by Forrest and co-workers indicated the occurrence of EPA, as well as unimolecular degradation processes, in OLEDs [29, 30]. The occurrence of EPA and its contribution to intrinsic degradation could be monitored, using various spectroscopic and electric techniques. Operating devices under photoillumination led to high concentrations of polarons and excitons. Increases in the capacitance or operating voltage served as indicators of the generation and accumulation of degradation byproducts. Control experiments performed in the absence of either photoillumination or charge-carrier injection revealed a large contribution of EPA to the overall degradation. Similar observations have been reported by So and co-workers [31]. A subsequent study by Forrest and co-workers suggested that EPA involves a dopant exciton and the negative polaron of a host [30]. EPA was observed for various EMLs. Phosphorescent EMLs incorporating Ir(III) complexes have been suggested to degrade following EPA [32, 33]. Adachi and co-workers also found that degradation of an EML by EPA facilitated subsequent degradation of a vicinal layer [34]. This secondary degradation was due to a loss in electron trapping capability in the bulk of the EML. Holes that did not recombine with electrons penetrated into the HBL, leading to the formation of electromers of HBL molecules. In addition to chemical degradation, EPA has been proposed to facilitate the aggregation of materials. Aziz and co-workers investigated changes in the photophysical properties of an OLED with a configuration of ITO/MoO3 /CBP/TPBi/LiF/Al [35]. Operation of this device resulted in the emergence of bathochromically shifted bands in both the corresponding electroluminescence and UV–vis absorption spectra. The spectral changes were ascribed to the generation of CBP aggregates produced through

Fig. 17 Bimolecular mechanism for intrinsic degradation initiated by exciton-polaron annihilation

Chemical Mechanisms of Intrinsic Degradation …

121

the annihilation of a CBP exciton and a CBP polaron. Similar aggregation behaviors have been observed from devices comprising an EML doped with phosphorescent Ir(III) complexes [36]. The occurrence of EPA requires an overlap between an exciton-formation zone and a charge-carrier accumulation zone. Aziz and a co-worker reported that the extent of EPA increased in proportion with hole accumulation in an EML [37]. The stability of an OLED with a configuration of ITO/NPB/CBP:Ir dopant/HBL/Alq3/Mg:Ag/Ag depended on the identity of the HBL. The device lifetime decreased with increasing hole-blocking ability of the HBL. The use of a HBL was, however, unavoidable because devices without a HBL suffered from hole-leakage-induced degradation of Alq3 [38]. EPA can be avoided if the exciton-formation zone is delocalized over the entire region of an EML. Localization of an exciton-formation zone in the vicinity of an ETL or HTL promotes EPA [33, 39, 40]. Therefore, balancing the chargecarrier mobility within an EML is critical. The location of an exciton-formation zone depends on a charge-carrier injection barrier. In recent studies, Kim and a co-worker revealed that direct hole injection from a HTL to dopants led to chargecarrier balance within an EML [41]. On the contrary, hole injection from a HTL to hosts resulted in an accumulation of holes at the EML-HBL interface, thereby facilitating EPA. The latter device exhibited poor stability relative to the former device. Interfacial EPA was proposed to be responsible for the degradation of the host in the EML. Noguchi et al. investigated the degradation behavior of an OLED with a configuration of ITO/NPD/CBP:TADF dopant/TPBi/LiF/Al [42]. The device displayed poor stability because of the degradation of the CBP host. They proposed that the degradation was initiated by interfacial EPA between the EML and the TPBi HBL. A strong propensity for the accumulation of interfacial negative charge due to spontaneous orientation polarization of TPBi was observed. Their results demonstrate that, in addition to charge-carrier mobility, the choice of HTL or ETL provides a useful strategy to control EPA. One viable approach to suppressing EPA is reducing the exciton lifetime. Recent advances in the understanding of rISC provide guidance for achieving a short exciton lifetime. The key process in rISC is the second-order vibronic coupling between the triplet local excited (3 LE) state and the singlet intramolecular charge-transfer (1 ICT) state. Therefore, locating the 1 ICT state in the vicinity of the 3 LE state facilitates rISC, thereby resulting in a short exciton lifetime. Adachi and co-workers demonstrated this strategy using a series of TADF molecules shown in Fig. 18 [43]. As expected, the rate constant for rISC (k rISC ) increased inversely with the energy difference between the 3 LE and ICT states. A substantial improvement in the operational lifetime was achieved using TADF dopants with greater k rISC values. An energy donor-energy acceptor approach was demonstrated to be effective for decreasing exciton lifetime. Lovrincic and co-workers incorporated a pyrenebased fluorescent dopant into an EML that already contained a blue-phosphorescent dopant [44]. The emission spectra of the fluorescent and phosphorescent dopants overlapped, although the former exhibited a smaller Stokes shift than the latter (Fig. 19). This spectral alignment promoted energy transfer from the phosphorescent

122

Y. You

Fig. 18 TADF dopants with different k rISC values

dopant to the fluorescent dopant. This energy transfer shortened the lifetime of the vulnerable triplet exciton. An alternative approach is to broaden the exciton-formation zone. Aziz and co-workers demonstrated that thick EMLs provide broad exciton-formation zones, thereby reducing the probability of EPA [45]. In a recent study, Kim and co-workers demonstrated that exciplex hosts could also suppress EPA [46]. A mixed host capable of forming exciplexes accelerates electron-carrier transport and, also, broadens an exciton-formation zone. A combined result is lowered propensity of EPA.

Chemical Mechanisms of Intrinsic Degradation …

123

Fig. 19 Phosphor-sensitized fluorescence

Conventional devices usually have higher hole densities than electron densities in their EMLs. This disparity in charge-carrier densities was responsible for EPA. To suppress annihilation between excitons and positive polarons, Zhang and a co-worker fabricated a device with an inverted structure with a configuration of ITO/Mg/TPBi/CBP:[Ir(ppy)3 ]/CBP/MoO3 /Al [47]. This structure led to balanced densities of oppositely charged carriers within the EML, which in turn led to a longer device lifetime. Adachi and co-workers used a similar approach to balancing chargecarrier transport by combining electron-transporting hosts with a hole-transporting TADF dopant, which substantially improved device stability [48]. Designing molecules with robust resistance to EPA and EEA remains a formidable hurdle. Molecules should include stable chemical bonds such that (i) the BDE of polarons is greater than the exciton energy and (ii) the BDE of excitons is greater than twice the exciton energy. These requirements are, however, difficult to achieve because of the limited information about the BDEs of excitons and polarons. A potential approach addressing this challenge could rely on quantum chemical tools. Kwon and co-workers recently introduced a quantum chemical strategy to create stable host materials [49].

3.2.3

Degradation by Exciton-Mediated Generation of Polaron Pair

Another bimolecular degradation process is initiated by electron transfer between two different molecules (Fig. 20). In many EMLs, electron transfer from a dopant to a host exciton is thermodynamically allowed. The exergonicity is attributable to the excited-state reduction potential (E*red ) of the host being more positive than the ground-state oxidation potential (E ox ) of the dopant. The heterobimolecular electron transfer is usually ultrafast. The resultant electron-transfer product is a pair composed

124

Y. You

Fig. 20 Bimolecular mechanism for intrinsic degradation initiated by exciton-mediated formation of a polaron pair

of the negative polaron (or radical anion) of the host and the positive polaron (or radical cation) of the dopant. The individual component within the polaron pair is susceptible to polaronic degradation. However, this degradation differs from the usual polaronic degradation, as it does not require charge-carrier trapping. Notably, a radical ion pair is generated in the absence of extra holes or electrons. The exciton-mediated electron transfer provides an additional pathway to polaronic degradation. One interesting feature of this degradation process is the presence of a self-annihilation pathway. Within the polaron pair, back electron transfer from the negative polaron of the host to the positive polaron of the dopant can occur. This charge recombination competes with polaronic degradation of the radical ion pair; thus, it serves as a protective step. Kim et al. investigated electron-transfer interactions between a dopant and a host exciton [50]. Spectroscopic measurements revealed the formation of a radical ion pair. The kinetic parameters of the charge recombination were determined and were found to show good linear correlation with the operation lifetime of devices. Subsequent studies by Moon et al. established the central effect of the electron-transfer degradation of TADF OLEDs [51].

4 Conclusions This chapter overviews the chemical mechanisms for the intrinsic degradation of emitting layers in OLEDs. Even a small amount of degradation byproducts can dramatically reduce the performance of OLEDs. Therefore, understandings of the chemistry underlying the initiation and progression of degradation are profoundly important. This chapter focuses on the chemical degradation of EMLs. Degradation

Chemical Mechanisms of Intrinsic Degradation …

125

mechanisms are classified into unimolecular and bimolecular processes, depending on the stoichiometry of the reactive intermediate in the key degradation step. Unimolecular degradation involves an exciton or a polaron as reactive intermediates. Accumulated reports have indicated that both the dopant and the host suffer from unimolecular degradation. Excitonic degradation competes with radiative transition of an exciton. Therefore, dopants with high photoluminescence quantum yields are less prone to degradation. Potential approaches to minimizing the unimolecular degradation could involve designing molecules that avoid repulsive states in their excited state and strengthening weak chemical bonds in their polaron state. Chemical strategies to improve the stability of polarons remain insufficiently established, which calls for future research. Bimolecular degradation becomes more prominent at high current densities. The excessive creation of excitons leads to EEA, which is frequently observed for longlived excitons, such as those generated at phosphorescent complexes and TADF molecules. EEA produces high-energy species that readily dissociate. To suppress degradation by EEA, the exciton lifetime should be as short as possible. Shockley– Read–Hall recombination is favored over Langevin recombination because the latter can produce a very long-lived triplet exciton of a host vulnerable to degradation. In addition to EEA, energy transfer from excitons to polarons occurs. The EPA product is a highly unstable excited polaron. The probability of EPA can be lowered by reducing the collisional interaction between excitons and polarons. Delocalizing an exciton-formation zone within an EML is a viable strategy. Finally, electron transfer from a dopant to a host exciton provides an additional polaronic degradation pathway. This recently identified mechanism has been demonstrated to be relevant to the operational stability of devices. Kinetic control over charge recombination of the electron-transfer product (i.e., polaron pair) is critically important for minimizing this degradation path. Notably, the degradation mechanisms discussed in this chapter can occur simultaneously. Quantification of the dominant mechanism is, thus, critical for improving device longevity.

References 1. H. Aziz, Z.D. Popovic, Degradation phenomena in small-molecule organic light-emitting devices. Chem. Mater. 16, 4522–4532 (2004) 2. S. Schmidbauer, A. Hohenleutner, B. König, Chemical degradation in organic light-emitting devices: mechanisms and implications for the design of new materials. Adv. Mater. 25, 2114– 2129 (2013) 3. F. So, D. Kondakov, Degradation mechanisms in small-molecule and polymer organic lightemitting diodes. Adv. Mater. 22, 3762–3777 (2010) 4. S. Scholz, D. Kondakov, B. Luessem, K. Leo, Degradation mechanisms and reactions in organic light-emitting devices. Chem. Rev. 115, 8449–8503 (2015) 5. S.K. Jeon, H.L. Lee, K.S. Yook, J.Y. Lee, Recent progress of the lifetime of organic lightemitting diodes based on thermally activated delayed fluorescent material. Adv. Mater. 31, 1803524 (2019)

126

Y. You

6. J. Ràfols-Ribé, P.-A. Will, C. Hänisch, M. Gonzalez-Silveira, S. Lenk, J. Rodríguez-Viejo, S. Reineke, High-performance organic light-emitting diodes comprising ultrastable glass layers. Sci. Adv. 4, eaar8332 (2018) 7. R. Meerheim, S. Scholz, S. Olthof, G. Schwartz, S. Reineke, K. Walzer, K. Leo, Influence of charge balance and exciton distribution on efficiency and lifetime of phosphorescent organic light-emitting devices. J. Appl. Phys. 104, 014510 (2008) 8. G. Treboux, J. Mizukami, M. Yabe, S. Nakamura, Blue phosphorescent iridium(III) complex. A reaction path on the triplet potential energy surface. Chem. Lett. 36, 1344–1345 (2007) 9. M.J. Jurow, A. Bossi, P.I. Djurovich, M.E. Thompson, In situ observation of degradation by ligand substitution in small-molecule phosphorescent organic light-emitting diodes. Chem. Mater. 26, 6578–6584 (2014) 10. I. Rabelo de Moraes, S. Scholz, B. Luessem, K. Leo, Role of oxygen-bonds in the degradation process of phosphorescent organic light emitting diodes. Appl. Phys. Lett. 99, 053302 (2011) 11. D. Jacquemin, D. Escudero, The short device lifetimes of blue PhOLEDs: insights into the photostability of blue Ir(III) complexes. Chem. Sci. 8, 7844–7850 (2017) 12. M.-C. Tang, M.-Y. Leung, S.-L. Lai, M. Ng, M.-Y. Chan, V.W.-W. Yam, Realization of thermally stimulated delay phosphorescence in arylgold(III) complexes and efficient gold(III) based blueemitting organic light-emitting devices. J. Am. Chem. Soc. 140, 13115–13124 (2018) 13. S. Schmidbauer, A. Hohenleutner, B. König, Studies on the photodegradation of red, green and blue phosphorescent OLED emitte. Beilstein J. Org. Chem. 9, 2088–2096 (2013) 14. D.Y. Kondakov, C.T. Brown, T.D. Pawlik, V.V. Jarikov, Chemical reactivity of aromatic hydrocarbons and operational degradation of organic light-emitting diodes. J. Appl. Phys. 107, 024507 (2010) 15. M. Cai, D. Zhang, J. Xu, X. Hong, C. Zhao, X. Song, Y. Qiu, H. Kaji, L. Duan, Unveiling the role of Langevin and trap-assisted recombination in long lifespan OLEDs employing thermally activated delayed fluorophores. ACS Appl. Mater. Interfaces 11, 1096–1108 (2019) 16. D.Y. Kondakov, T.D. Pawlik, W.F. Nichols, W.C. Lenhart, Free-radical pathways in operational degradation of OLEDs. J. SID. 16, 37–46 (2008) 17. A.Y. Choi, T. Yamaguchi, C.-H. Han, A photochemical investigation into operational degradation of arylamines in organic light-emitting diodes. Res. Chem. Intermed. 39, 1571–1579 (2013) 18. A.S.D. Sandanayaka, T. Matsushima, C. Adachi, Degradation mechanisms of organic lightemitting diodes based on thermally activated delayed fluorescence molecules. J. Phys. Chem. C. 119, 23845–23851 (2015) 19. M. Hong, M.K. Ravva, P. Winget, J.-L. Bredas, Effect of substituents on the electronic structure and degradation process in carbazole derivatives for blue OLED host materials. Chem. Mater. 28, 5791–5798 (2016) 20. S.R. Yost, L.-P. Wang, T. Van Voorhis, Molecular insight into the energy levels at the organic donor/acceptor interface: a quantum mechanics/molecular mechanics study. J. Phys. Chem. C. 115, 14431–14436 (2011) 21. K.P. Klubek, C.W. Tang, L.J. Rothberg, Investigation of blue phosphorescent organic lightemitting diode host and dopant stability. Org. Electron. 15, 1312–1316 (2014) 22. I. Rabelo de Moraes, S. Scholz, B. Luessem, K. Leo, Analysis of chemical degradation mechanism within sky blue phosphorescent organic light emitting diodes by laserdesorption/ionization time-of-flight mass spectrometry. Org. Electron. 12, 341–347 (2011) 23. V. Sivasubramaniam, F. Brodkorb, S. Hanning, H.P. Loebl, V. van Elsbergen, H. Boerner, U. Scherf, M. Kreyenschmidt, Fluorine cleavage of the light blue heteroleptic triplet emitter FIrpic. J. Fluorine Chem. 130, 640–649 (2009) 24. Z.-Q. Zhu, K. Klimes, S. Holloway, J. Li, Efficient cyclometalated platinum(II) complex with superior operational stability. Adv. Mater. 29, 1605002 (2017) 25. N. Lin, J. Qiao, L. Duan, H. Li, L. Wang, Y. Qiu, Achilles heels of phosphine oxide materials for OLEDs: chemical stability and degradation mechanism of a bipolar phosphine oxide/carbazole hybrid host material. J. Phys. Chem. C. 116, 19451–19457 (2012)

Chemical Mechanisms of Intrinsic Degradation …

127

26. N. Lin, J. Qiao, L. Duan, L. Wang, Y. Qiu, Molecular understanding of the chemical stability of organic materials for OLEDs: a comparative study on sulfonyl, phosphine-oxide, and carbonylcontaining host materials. J. Phys. Chem. C. 118, 7569–7578 (2014) 27. J. Lee, C. Jeong, T. Batagoda, C. Coburn, M.E. Thompson, S.R. Forrest, Hot excited state management for long-lived blue phosphorescent organic light-emitting diode. Nat. Commun. 8, 15566 (2017) 28. L.-S. Cui, Y.-L. Deng, D.P.-K. Tsang, Z.-Q. Jiang, Q. Zhang, L.-S. Liao, C. Adachi, Controlling synergistic oxidation processes for efficient and stable blue thermally activated delayed fluorescence devices. Adv. Mater. 28, 7620–7625 (2016) 29. N.C. Giebink, B.W. D’Andrade, M.S. Weaver, P.B. Mackenzie, J.J. Brown, M.E. Thompson, S.R. Forrest, Intrinsic luminance loss in phosphorescent small-molecule organic light emitting devices due to bimolecular annihilation reactions. J. Appl. Phys. 103, 044509 (2008) 30. N.C. Giebink, B.W. D’Andrade, M.S. Weaver, J.J. Brown, S.R. Forrest, Direct evidence for degradation of polaron excited states in organic light emitting diodes. J. Appl. Phys. 105, 124514 (2009) 31. S. Liu, C. Peng, A. Cruz, Y. Chen, F. So, Degradation study of organic light-emitting diodes with solution-processed small molecule phosphorescent emitting layers. J. Mater. Chem. C. 4, 8696–8703 (2016) 32. Y. Zhang, H. Aziz, Degradation mechanisms in blue phosphorescent organic light-eEmitting devices by exciton-polaron interactions: loss in quantum yield versus loss in charge balance. ACS Appl. Mater. Interfaces 9, 636–643 (2017) 33. H.Z. Siboni, H. Aziz, Causes of driving voltage rise in phosphorescent organic light emitting devices during prolonged electrical driving. Appl. Phys. Lett. 101, 173502 (2012) 34. M. Tanaka, H. Noda, H. Nakanotani, C. Adachi, Effect of carrier balance on device degradation of organic light-emitting diodes based on thermally activated delayed fluorescence emitters. Adv. Electron. Mater. 5, 1800708 (2019) 35. Q. Wang, B. Sun, H. Aziz, Exciton-polaron-induced aggregation of wide-bandgap materials and its implication on the electroluminescence stability of phosphorescent organic light-emitting devices. Adv. Funct. Mater. 24, 2975–2985 (2014) 36. Y. Zhang, H. Aziz, Influence of the guest on aggregation of the host by exciton-polaron interactions and its effects on the stability of phosphorescent organic light-emitting devices. ACS Appl. Mater. Interfaces 8, 14088–14095 (2016) 37. H.Z. Siboni, H. Aziz, The influence of the hole blocking layers on the electroluminescence stability of phosphorescent organic light emitting devices. Org. Electron. 12, 2056–2060 (2011) 38. H.Z. Siboni, Y. Luo, H. Aziz, Luminescence degradation in phosphorescent organic lightemitting devices by hole space charges. J. Appl. Phys. 109, 044501 (2011) 39. R.Y. Yang, X.M. Li, X.A. Cao, Role of wide bandgap host in the degradation of blue phosphorescent organic light-emitting diodes. J. Appl. Phys. 122, 075501 (2017) 40. K. Klimes, Z.Q. Zhu, J. Li, Efficient blue phosphorescent OLEDs with improved stability and color purity through judicious triplet exciton management. Adv. Funct. Mater. 29, 1903068 (2019) 41. J.-M. Kim, J.-J. Kim, Charge transport layers manage mobility and carrier density balance in light-emitting layers influencing the operational stability of organic light emitting diodes. Org. Electron. 67, 43–49 (2019) 42. Y. Noguchi, H.-J. Kim, R. Ishino, K. Goushi, C. Adachi, Y. Nakayama, H. Ishii, Charge carrier dynamics and degradation phenomena in organic light-emitting diodes doped by a thermally activated delayed fluorescence emitter. Org. Electron. 17, 184–191 (2015) 43. H. Noda, H. Nakanotani, C. Adachi, Excited state engineering for efficient reverse intersystem crossing. Sci. Adv. 4, eaao6910 (2018) 44. P. Heimel, A. Mondal, F. May, W. Kowalsky, C. Lennartz, D. Andrienko, R. Lovrincic, Unicolored phosphor-sensitized fluorescence for efficient and stable blue OLEDs. Nat. Commun. 9, 4990 (2018) 45. Y. Zhang, Q. Wang, H. Aziz, Simplified organic light-emitting devices utilizing ultrathin electron transport layers and new insights on their roles. ACS Appl. Mater. Interfaces 6, 1697–1701 (2014)

128

Y. You

46. J.-M. Kim, C.-H. Lee, J.-J. Kim, Mobility balance in the light-emitting layer governs the polaron accumulation and operational stability of organic light-emitting diodes. Appl. Phys. Lett. 111, 203301 (2017) 47. Y. Zhang, H. Aziz, Enhanced stability in inverted simplified phosphorescent organic lightemitting devices and its origins. Org. Electron. 22, 69–73 (2015) 48. L.-S. Cui, S.-B. Ruan, F. Bencheikh, R. Nagata, L. Zhang, K. Inada, H. Nakanotani, L.-S. Liao, C. Adachi, Long-lived efficient delayed fluorescence organic light-emitting diodes using n-type hosts. Nat. Commun. 8, 2250 (2017) 49. A.Y. Freidzon, A.A. Safonov, A.A. Bagaturyants, D.N. Krasikov, B.V. Potapkin, A.A. Osipov, A.V. Yakubovich, O. Kwon, Predicting the operational stability of phosphorescent OLED host molecules from first principles: a case study. J. Phys. Chem. C. 232, 22422–22433 (2017) 50. S. Kim, H.J. Bae, S. Park, W. Kim, J. Kim, J.S. Kim, Y. Jung, S. Sul, S.-G. Ihn, C. Noh, S. Kim, Y. You, Degradation of blue-phosphorescent organic light-emitting devices involves exciton-induced generation of polaron pair within emitting layers. Nat. Commun. 9, 1211 (2018) 51. Y.K. Moon, H.J. Jang, S. Hwang, S. Kang, S. Kim, J. Oh, S. Lee, D. Kim, J.Y. Lee, Y. You, Modeling electron-transfer degradation of organic light-emitting devices. Adv. Mater. 2003832 (2021)

Encapsulation Technology for Flexible OLEDs Eun Gyo Jeong and Kyung Cheol Choi

Abstract With the ongoing advances in nanomaterials and fabrication technologies, the form factor of electronic devices is also evolving. Displays have evolved from rigid flat panels to flexible, rollable, and foldable formats. Such changes in form factor can provide improvements in consumer utility and convenience, including portability and ease of use. Developing a stable flexible display has attracted considerable attention for these reasons. Comprehensive studies have investigated various approaches, including flexible liquid crystal displays (LCD), displays employing light-emitting diodes (LED), and organic light-emitting diodes (OLED). Because they provide superior flexibility, OLEDs are highly desirable in the flexible display industry. But to realize the full potential of flexible OLEDs, they not only require enhanced characteristics to withstand rolling and folding, but a highly effective thin-film encapsulation barrier is also essential. To date, most existing encapsulation studies have focused on the low water vapor transmission rate (WVTR) characteristic, which is related to gas barrier properties. This book chapter covers developments in encapsulation technologies; their structure designs; and materials for realizing flexible, rollable, and foldable OLEDs. Special focus is given to the existing hurdles to flexibility and how to overcome these limitations. Finally, further insights on the evolution of encapsulation technologies are discussed.

1 Introduction Form factor refers to the physical shape or appearance of an object, and the display serves to optimally implement various functions of electronic devices in the form of visual information. Electronic devices are traditionally heavy and rigid, but modern devices are breaking away from this stereotype, and are being reborn in lighter and E. G. Jeong · K. C. Choi (B) School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea e-mail: [email protected] E. G. Jeong Department of Clothing and Textiles, Chonnam National University, Gwangju, Republic of Korea © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_6

129

130

E. G. Jeong and K. C. Choi

more flexible forms. Various functions of electronic devices that have typically been performed by hardware are now being served by application services. As internal components become increasingly simplified, the display will occupy a relatively larger proportion of the electronic devices’ form factor. The liquid crystal display (LCD), which has recently been the leader in development and market growth, has entered technical and commercial maturity, and new growth will depend on further innovation. Display form factor is particularly important in this regard. A free and convenient form factor that is robust under various environmental conditions is increasingly desirable, to address the increasing mobility of consumers, convergence between devices, and the use of smart devices. This evolution in display form factor will be from the current “rigid square shape” to a “flexible form” that can be freely transformed. To realize the full potential of the flexible display, various types of form factors are being explored, from the fixed type “Curved and Bendable,” to the single-axis changeable “Rollable and Foldable,” and the multi-axis variable “Stretchable” type. Since glass has some flexibility when it is thin, it can be used for curved and bendable displays which are deformed, and can be fixed in a single direction. For example, curved televisions (TV) usually employ a thin glass substrate with a bending radius of 4000–5000 mm, which is a highly curved shape. On the other hand, a curved smartphone, which provides advantages of graspability and stability, requires a bending radius between 400 and 700 mm, and a bendable smartphone requires a bending radius of less than 10 mm for edge bending. Since glass cannot satisfy these ranges of flexibility, a flexible plastic material is typically used for the substrate. While curved and bendable displays have a fixed bending radius, rollable and foldable displays require more dramatic innovation in bending radius. The rollable display is usually stored in a rolled up state, like a scroll, and the bending radius determines how small they can be rolled. Since a rolled display requires less storage space, the rollable format is especially suited for large displays of 65–75 in. or more. In such applications, it is desirable to minimize the bending radius to less than 50 mm. Mobile devices that are mainly used with one hand are likely to have a foldableform display rather than a rollable one because of its convenience. Although the display is not fully folded like paper, the bending radius of a foldable display should be reduced to around 1 mm. Two types of foldable displays, an out-folding and in-folding design, have been proposed, as shown in Fig. 1. The out-folding type can have a higher display bending radius because it folds outward. However, in this configuration the display is exposed to the outside, a disadvantage which leaves it

Fig. 1 Types of foldable display: a out-folding, and b in-folding

Encapsulation Technology for Flexible OLEDs

131

vulnerable to breaking. Since the in-folding type folds inward, its structural design helps to protect the display, but its thickness remains a limitation. Considering the frequent repetition of deformation (rolling or folding) during use, the flexible display needs to meet the following technical requirements. First, the display must be able to restore its own shape while enduring repeated external pressure. To this end, it is important to ensure its flexibility and elasticity as well as make the display thinner. Since the physical pressure can affect not only the performance of the display but also life span, it is essential that it be durable enough to avoid degradation of inherent display performance and lifetime. Efforts to determine light sources that can meet these requirements have been underway since the beginning of flexible display research. The LCD, which is currently the most widely used type, has several drawbacks. It is difficult to make them flexible and to uniformly arrange the liquid crystals, which can cause leakage. In contrast, organic light-emitting diodes (OLEDs) are attractive because of their wide viewing angles and clear image quality, due to the self-emitting property of the organic materials. Above all, the OLED structure is simple and thin, which makes it possible to reproduce images without distortion even when the panel is deformed. As a result, most flexible display research is now based on OLED technology [1–4]. However, OLEDs need an encapsulation barrier to protect them from the external environment, because the organic materials are very vulnerable to moisture and oxygen. As flexible rollable and foldable displays are being developed, various kinds of encapsulation technologies are being proposed that can prevent moisture and oxygen damage even when the shape changes. For example, the barrier properties and flexibility of an encapsulation barrier can be improved by alternately stacking multiple layers of inorganic and organic films, with nano-thickness inorganic and inorganic films. While thick film is ideal to effectively block moisture and oxygen, it requires a complicated fabrication process and reduces flexibility. To realize a flexible display and meet its major technical requirements, further technological innovations in the encapsulation barrier must be achieved. To solve these challenges, several approaches have been developed, including nano-scale structural design, and placing the encapsulation barrier in neutral axis to avoid external stress. At the same time, a new progressive approach to materials is still required. In this chapter, the authors will review the comprehensive development trends in encapsulation barriers and the technical issues that remain to be addressed.

2 Encapsulation Barrier Technologies 2.1 Dark Spots and Defects in OLEDs Two types of OLED deterioration are common, intrinsic degradation, and extrinsic degradation over time. Typical OLED lifetimes are approximately 100,000 h at

132

E. G. Jeong and K. C. Choi

500 cd/m2 for red and green and 10,000 h at 200 cd/m2 for blue [5]. Although the mechanism of chemical degradation in OLEDs is not known in detail, intrinsic degradation caused by the dissociation of phosphors and the product of phosphors near the blocking layer has been studied using laser desorption/ionization time-offlight mass spectrometry (LDI-TOF-MS) [6–8]. Accordingly, while understanding of intrinsic degradation is relatively mature, extrinsic degradation still remains a challenge. Dark spots on OLEDs are believed to be caused by extrinsic degradation and factors, such as the roughness of the substrate and electrode, particle contamination, oxygen, and water. All of these should be avoided to ensure reliability, as shown in Fig. 2. In the first case, when the surface of substrate or electrode is very rough, with sharp spikes, for example, local discharges can occur during operation. The high local current from the discharge process can lead to excessive heat, which results in the crystallization of the organic materials. With crystallization, an electrochemical change can occur which can cause the organic material to peel off of the electrode. As a result, a dark spot will be generated on that space [9, 10]. Contamination due to particles is another important factor that can cause dark spots. If the particle has low conductivity, it can cause local insulation that interrupts current flow, generating dark spots. On the other hand, high-conductivity particles will generate bright spots, resulting from locally high current. In addition, large particles may cause discontinuous film and cracks in the organic layer, leading to dark

Fig. 2 Dark spots generated in OLEDs. Reproduced with permission [40, 45]. Copyright 2016, Elsevier and Copyright 2016, Royal Society of Chemistry

Encapsulation Technology for Flexible OLEDs

133

spots. Therefore, it is important to maintain cleanliness to strictly control particles such as falling dust during the loading process or existing particles in equipment. Above all, the main cause of OLED degradation is the effect of water moisture and oxygen. OLED electrodes need to have a low work function, and as a consequence are relatively active metal or metal alloy, which can be easily oxidized or corroded by water moisture and oxygen. This produces an insulating layer at the interface of the layers, resulting in a change in charge injection. When the electrode deterioration becomes severe, dark spots eventually occur. Not only the electrodes but also some organic materials are very sensitive to oxygen and water. Electrochemical reactions between organic layers induced by water molecules will deteriorate the lifetime of OLEDs.

2.2 Thin-Film Encapsulation In the past, glass-lid encapsulation with a desiccant of barium oxide (BaO) or calcium oxide (CaO) has been used for rigid OLEDs, and it provides an acceptable level of water vapor transmission rate (WVTR) [11–13]. However, the brittle nature of glass makes it unsuitable for flexible OLEDs. In addition, the conventional epoxy sealing is insufficient to block penetration, and desiccants are costly, making the glass-lid approach inappropriate for commercialization. The glass lid also increases the display thickness, making it rigid and heavy. To solve these problems, many kinds of encapsulation technologies have been studied and proposed as alternatives. Among these technologies, thin-film encapsulation (TFE) is considered particularly promising since it can eliminate edgepermeation while providing some flexibility. Many types of TFE have been developed for flexible OLEDs, including a multi-barrier structure consisting of alternating inorganic/organic layers, and an inorganic-based nanolaminate system. Multi-barriers consisting of inorganic/organic layers have been widely used for OLED encapsulation [13, 15–19]. As shown in Fig. 3, the inorganic layer is an excellent gas barrier and has superior chemical and thermal stabilities. However, defects (pinholes) which work as pathways for water moisture and oxygen exist in the layer [20]. Although the organic layer is a poor gas barrier, it can prevent the formation of pinholes and has low residual stress and provides planarization. The strategy of alternating inorganic/organic layers is used to effectively pack the pinholes with organic materials and distributing the pinholes. By stacking the inorganic/organic layers in pairs, known as a dyad, the multi-barrier creates an effectively complicated diffusion path, low WVTR, flexibility, and chemical and thermal stabilities. Since the WVTR mainly depends on the quality and thickness of the inorganic layer and the possibility for using it as a superstrate depends on the organic layer, optimizing each layer and the number of the dyads is necessary.

134

E. G. Jeong and K. C. Choi

Fig. 3 Schematic diagram of a multi-barrier system. Reproduced with permission [20]. Copyright 2017, IEEE

2.3 Atomic Layer Deposition (ALD) Many deposition techniques can be used to fabricate thin films for the TFE, including physical vapor deposition (PVD), chemical vapor deposition (CVD), and atomic layer deposition (ALD). The PVD method, which deposits a vaporized material in a vacuum chamber, has the advantage of high throughput at a low price. In spite of its convenience, the WVTR of PVD-TFE is too high for OLEDs because it is difficult to form a dense film, and contains a lot of impurities. In contrast, CVD which uses gas reactions and ions can form a film that is dense enough to obtain a low WVTR for OLEDs, but involves a high-temperature process. Although a low-temperature process is possible, with plasma-enhanced CVD (PECVD), direct deposition on OLEDs is still difficult because of its plasma damage. On the other hand, ALD, which has been used to form conformal dielectric layers since the 1970s, has attracted attention as a method of depositing TFEs because it forms an almost pinhole-free thin film. However, forming a pinhole-free film requires a high-temperature process of 350 °C and a perfectly clean environment. Since flexible OLEDs often have a plastic substrate and some organic materials, they cannot endure high temperature, and as a result the barrier characteristics of ALD-TFE are not good as expected. Low-temperature ALD-TFE has been found to generate pinholes. Considerable study has been devoted to solving this limitation by controlling pulse time and pressure. In spite of its slow deposition rate, ALD is still widely used for TFE because it has the best barrier characteristics. ALD is a method of depositing a film on a substrate using a chemical adsorption reaction between alternate pulses of precursor gas and reactant vapor. There is a purge process between each pulse using an inert gas such as nitrogen to remove residual

Encapsulation Technology for Flexible OLEDs

135

Fig. 4 The ALD process

precursor or reactant. The basic ALD deposition principle is shown in Fig. 4. To form a thin film, which is called AB, gaseous AX and BY are used as a precursor and reactant, respectively, and the by-product gaseous XY is discharged through the purge process. Since the reaction only happens between the reactant and surface during deposition, atomic-scale deposition is possible. This kind of reaction is called a self-limiting reaction, and has the advantage of precisely controlling the thickness of the thin film through the cycle. In addition to having fewer defects or pin holes, ALD thin film has excellent step coverage because of its chemisorption reaction on the substrate.

2.4 Measurement Methods To design an appropriate encapsulation barrier for OLEDs, it is important to measure the barrier property of the barrier film. For this purpose, encapsulation barrier researchers continuously evaluate its WVTR. A gravimetric cup filled with desiccants or water covered by the barrier can be used to measure WVTR by mass change [21]. More accurate WVTR can be measured using a mass spectrometer, which has two chambers (a high-pressure gas chamber and an ultra-high vacuum chamber) separated by the test barrier [22, 23]. It is also possible using tritiated water instead of the mass spectrometer, know as a tritium test [24, 25]. The MOCON test, which has been commercialized by MOCON, Inc., uses a desiccant carrier gas to transport

136

E. G. Jeong and K. C. Choi

water through the barrier, and then its WVTR is calculated by sensor. Deltaperm by Technolox Ltd. can measure the amount of water moisture concentration inside the barrier film by direct pressure measurement. Since organic devices are very sensitive to water moisture, it is also possible to use them as a detector. Among these measurement methods, the calcium (Ca) corrosion test, also known as the Ca test, has been the most widely used because of its convenience and accuracy [26]. Calcium becomes transparent with oxidation, and the light intensity passing through the calcium also changes. Estimating WVTR based on the remaining calcium, which blocks light, was studied by Nisato et al. There is another method which calculates WVTR using the growth rate of corroded calcium; however, it is much more appropriate for defect imaging. In addition, the electrical characteristics of calcium films also vary, because calcium oxide is an insulator, while calcium film is a conductor. The WVTR can be calculated by the increase in calcium film resistance with oxidation.

2.5 ALD-Based Multi-barrier The inorganic layer is the main barrier to water and oxygen permeation, and numerous studies have been conducted to develop an excellent impermeable inorganic layer. Various types of inorganic layer have been developed using deposition methods. To compare different deposition methods, the WVTR of PVD deposited magnesium (MgO) and ALD deposited MgO was measured by electrical Ca test [27, 28]. In this study, electron beam (E-beam) evaporation was used to deposit 100-nm-thick PVD-MgO while bis(ethyl cyclopentadienyl) magnesium (Mg(CpEt)2 ) and water (H2 O) were used to fabricate a 40-nm-thick ALD-MgO layer. In spite of its thinness, the ALD-MgO layer boasted a better WVTR (7.13 × 10−2 g/m2 /day) than PVDMgO (0.46 g/m2 /day) due to the defect-minimizing ALD method. To compensate its poor flexibility and barrier properties, 6 dyads of PVD-MgO and 4.5 dyads of ALDMgO were utilized in a multi-barrier approach to prepare flexible TFE, as shown in Fig. 5a, b. Although MgO has a getter property, which is suited to TFE, aluminum oxide (Al2 O3 ) is considered the most suitable material for TFE because it can be deposited with an amorphous structure in an almost defect-free layer at relatively low temperature. Therefore, in recent years, studies on TFE using ALD-Al2 O3 have been actively conducted to prepare highly impermeable TFEs [29, 30]. However, although the amorphous Al2 O3 ALD film can be formed at a low temperature of less than 100 °C, which do not cause thermal damage to OLEDs, a low-temperature Al2 O3 single layer did not provide enough barrier performance (7.94 × 10−4 g/m2 /day for 30 nm and 8.06 × 10−5 g/m2 /day for 60 nm) or sufficient mechanical properties to use as a flexible TFE. More than 3 dyads of multi-barriers were fabricated to achieve a WVTR of 10−5 g/m2 /day, as shown in Fig. 5c. Figure 5d and Table 1 summarize the properties of ALD-based conventional multi-barriers.

Encapsulation Technology for Flexible OLEDs

137

Fig. 5 Various kinds of multi-barrier: a cross-section image of a PVD-MgO multi-barrier, b crosssection image of a ALD-MgO multi-barrier, c results of a Ca test and cross-section image of ALD-Al2 O3 multi-barrier, and d WVTR estimation based on the number of dyads. a Reproduced with permission [27]. Copyright 2011, Elsevier. b Reproduced with permission [28]. Copyright 2013, Elsevier. c Reproduced with permission [29]. Copyright 2013, Elsevier

Table 1 Estimations of WVTR multi-barriers according to various kinds of inorganic materials Inorganic materials

WVTR (g/m2 /day) 0.5 dyad 10−1

1.5 dyads 2.8 ×

10−3

2.5 dyads 6×

10−4

3.5 dyads 4.5 ×

10−4

4.5 dyads 1.8 ×

10−4

5.5 dyads

PVD-MgO

4.6 ×

4.7 × 10−5

ALD-MgO

7.13 × 10−2

7.80 × 10−2

5.28 × 10−2

7.88 × 10−2

1.49 × 10−5

N/A

ALD-Al2 O3

7.94 × 10−4

1.14 × 10−4

5.43 × 10−5

1.14 × 10−5

N/A

N/A

Heat dissipation is another issue for flexible OLEDs, because flexible OLEDs are placed on plastic substrates with very low thermal conductivity. Heat generated by electrical device operation can cause performance degradation, which seriously affects OLED efficiency and lifetime, and is a hurdle to realizing flexible displays. In particular, the thermally insulating TFE makes heat dissipation in OLEDs difficult. Therefore, attempts have been made to dissipate heat in OLEDs using metal substrates and metal heat sinks. Although the metal heat sink is effective for dissipating heat in

138

E. G. Jeong and K. C. Choi

Fig. 6 Metal incorporated multi-barrier: a schematic diagram of encapsulation barrier, b results of Ca test, and c optical transmittance of the metal incorporated multi-barrier. a–c Reproduced with permission [14]. Copyright 2017, American Chemical Society

OLEDs, the addition of a thick and opaque heat sink is a disadvantage in a transparent flexible OLED. A thin-film-based thermally conductive TFE, shown in Fig. 6, was developed using a dielectric/metal/dielectric (DMD) structure, with a 15-nm-thick Ag thin film inserted between Al2 O3 films [14]. It provided high thermal conductivity and maintained the good flexibility of the Ag thin film. Although the DMD-TFE encapsulated FOLED did not show a significant improvement in OLED lifetime, thermally conductive TFEs and methodologies are still considered to have high potential to improve the lifetime and efficiency of flexible OLEDs.

3 Flexible and Reliable Thin Film Encapsulation 3.1 TFE Mechanical Failure Although the multi-barrier approach can address the lack of flexibility of inorganic layers, it is still difficult to overcome the higher mechanical stress, as shown in Fig. 7. For example, ALD-Al2 O3 on polyethylene naphthalate (PEN) substrate was found to endure tensile stresses of up to 5.08% for a 5 nm layer, and 0.88% for a 125-nmthick layer [31] or 2.4% for a 5 nm and 0.52% for 80-nm-thick layer [32]. When mechanical stress greater than a critical stress was applied, cracks which initiated from small defects expanded, following a direction perpendicular to the strain, or new cracks were formed in the direction of strain. This means that external mechanical

Fig. 7 Failure of an encapsulated OLED with mechanical stress. Reproduced with permission [40, 50]. Copyright 2016, Royal Society of Chemistry

Encapsulation Technology for Flexible OLEDs

139

stress or loads can cause delamination or fracture, such as brittle fracture, ductile fracture, and fatigue fracture of the TFE barrier. Delamination is the separation of material at the interface between two adjacent layers. Not only the direct effects of external mechanical strain and stress, but also stress due to the difference in coefficient of hygroscopic expansion (CHE) can contribute to delamination. A mismatch in CHE induces degradation of interfacial adhesion and hygroscopic swelling in TFE [33–36]. As the degradation accelerates, the moisture-induced crack propagates along the interface, leading to delamination. When delamination happens, the WVTR will increase since the total thickness of the TFE decreases, providing another diffusion path for oxygen and water moisture. Since the resulting interface is rough and broken, it can lead to the initiation and propagation of other voids and cracks. Materials that exhibit low levels of yielding and inelasticity usually suffer brittle fracture which rapidly leads to cracking. This kind of fracture can occur in TFE depending on materials such as silica fillers, when exposed to external mechanical stress. When excessive mechanical stress is applied to the TFE, rapid cracking due to existing defects such as voids, inclusions, or discontinuities can be initiated. Ductile fractures can also occur with external mechanical stress, because some materials are sensitive to environmental factors such as temperature and viscoelastic properties. To predict such limitations, methods of measuring the critical stress intensity factor have been developed [37]. Even when the applied stress is lower than the critical stress, a fracture, called a fatigue fracture, can occur in a TFE when subjected to repetitive cyclic stress. Cyclic stresses include various kinds of stress, such as hygroscopic, thermal, or combined loads, as well as mechanical stress. Typically, a fatigue fracture does not initiate from a new spot, but at an existing point of discontinuity or defect. Fatigue fractures are usually categorized in three steps: initiation, stable propagation, and catastrophic failure. Stress-life (S-N) curves, also known as Wohler curves, can be used to predict fatigue behavior with a coupon testing machine. A small metal coupon is periodically subjected to cyclic stress below the ultimate strength of the material until failure. With this method, the S-N curve can be constructed and one can determine the accumulated fatigue damage to materials.

3.2 Neutral Axis Engineering for TFE To avoid these kinds of fractures and to mitigate the limitations of mechanical properties, it is important to design the TFE considering the distribution of bending stress over the entire structure. Recently, the composite beam theory has been used to analyze the distribution of mechanical stress in nano-scale structures, to determine its potential as a flexible device [38, 39]. The nonlinear two-dimensional finite-element analysis (FEA), in particular, based on finite-element method (FEM) solving by ANSYS 16.0, has been used to interpret the theoretical behavior of a neutral axis (NA) position. When the location of a device is utilized in a NA position, the device

140

E. G. Jeong and K. C. Choi

will have no more bending stress, since the NA position is the location of zero stress. Therefore, it is possible to improve its bending characteristics by optimizing its structure. Using this approach, a method which improves the mechanical flexibility of the Al2 O3 -based TFE was proposed, to control the NA position by inserting a thick buffer layer onto encapsulated OLEDs as shown in Fig. 8a, b [40]. To control the position of the NA, both the thickness and the corresponding Young’s modulus of the buffer material must be considered. For optimization, only elastic deformation was considered, and the following mechanical simulation model was used. The mechanical stress from bending strain assumed a force of 25 mN, which was focused at the center of the bottom surface. Considering the bending test, the degrees of freedoms (DOFs) were fixed to eliminate rigid body motion while constraints in the x-direction DOF were applied on other nodes. Young’s moduli of the materials were obtained via nano-indentation based on Oliver and Pharr analysis. Poisson’s ratio was not considered important since Poisson’s ratio is not a dominant factor [41]. Layers under 5 nm thickness were not considered in the FEA model either. Based on these conditions, the FEA model showed the relation between Young’s modulus and NA position, or thickness and NA position. Both Young’s modulus and the thickness of the buffer layer had a similar tendency, in that the position of the NA moved toward the TFE side when they were increased. Although the change could be observed with either Young’s modulus or thickness, the total amount of change was not that effective. To address this limitation, the correlation between Young’s modulus and thickness needs to be determined. As the stress along the cross section, summing the force in the x-direction, must be in equilibrium, an equation considering geometrical symmetry can be derived when the NA is located at the center of the

Fig. 8 Neutral axis engineering of an Al2 O3 -based multi-barrier: a schematic diagram of the simulation model, b the position of the NA according to the thickness of the buffer layer, and c images of activated flexible OLEDs over time. a–c Reproduced with permission [40]. Copyright 2016, Royal Society of Chemistry

Encapsulation Technology for Flexible OLEDs

141

TFE. The correlation between the thickness and Young’s modulus of buffer layer can be understood using this method. The increase in Young’s modulus results in a decrease in thickness, while the other values are determined through geometrical factors and physical properties. With this NA engineering method, the mechanical properties of an Al2 O3 -based TFE can be improved. The NA-engineered TFE maintained its WVTR after 1000 iterations of 1.18% bending strain, while the Al2 O3 -based TFE lost its property. As shown in Fig. 8c, the encapsulated FOLEDs did not show dark spots after 30 days. However, the fabrication process is more complex, since the buffer layer in the NA engineering method is thick and requires an additional deposition step. In addition, it is more difficult to adjust the NA engineering for real fabrication because it needs to consider all other subsequent panel processes, such as the polarizer, touch screen panel, module, and so forth, which makes it hard to calculate. Inorganic layers need to be developed that can replace existing homogeneous ALD inorganic films to simplify the barrier structure, and to improve moisture-resistance and mechanical properties with respect to WVTR, as well as mechanical flexibility, for use as a flexible TFE.

3.3 Nano-stratified Structure TFE Since thickness is an important factor for WVTR, to design a more impermeable barrier the inorganic layer is becoming thicker. Although the WVTR of the TFE becomes lower with thickness at first, with increasing thickness the WVTR becomes saturated and then increases after fracture thickness. In the first step, the substrate becomes covered with the target layer material via an island growth mechanism. Since the substrate is not yet covered completely with such a thin layer, the WVTR decreases until a critical thickness is reached. After the critical thickness, the layer maintains its WVTR with increasing thickness or slowly decreases based on the Fickian diffusion mechanism. The growth of existing defects due to residual stress neutralizes the barrier property during this stage, and the WVTR does not decrease any more. The critical thickness mainly depends on the residual stress of the existing film. Above the fracture thickness, the WVTR starts to increase because of the relaxation of residual stress. In recent years, as an alternative to increasing thickness, a nanolaminate structure with different alternating inorganic layers has been proposed, to fabricate a more impermeable TFE. The ALD method can be used to fabricate uniform ultrathin (a few nanometers thick) layers through cycle controlling. The ALD nanolaminate prevents the formation of microscopic voids and nano-crystals. In addition, the diffusion path in the nanolaminate structure is complicated, because each layer is independent and not continuous. The superior barrier property of an ALD nanolaminate compared to a single inorganic layer was proved experimentally. ALD nanolaminate TFEs combined with Al2 O3 such as Al2 O3 /SiO2 , Al2 O3 /ZnO, Al2 O3 /ZrO2 , Al2 O3 /TiO2 , and Al2 O3 /HfO2 have begun to be developed [25, 42–44]. Despite its structural

142

E. G. Jeong and K. C. Choi

innovation, the nanolaminate structure still has limitations because it is composed of rigid and brittle inorganic materials. To improve its flexibility, a new type of TFE with the nano-stratified moisture barrier, utilizing a nanolaminate structure instead of a homogeneous inorganic layer in a multi-barrier, has been proposed, as shown in Fig. 9. Usually, the nano-stratified layers contain Al2 O3 and ZnO to enhance their mechanical properties via a defect suppression mechanism [45, 46]. The defect suppression mechanism is based on the microcrack toughening model and the Griffith crack model, which describe the critical mechanical stresses that cause the failure of brittle inorganic materials [47, 48]. There are six main assumptions in this model: all materials have their own cracks, some cracks are oriented in the direction of applied loads, the crack grows with mechanical stress, the energy of crack propagation is released with crack extension, the surface energy increases with crack growth, and the crack grows continuously until the energy needed to form a new surface exceeds the release energy. Based on these assumptions, it is obvious that the concentration of mechanical stress will happen at the edge of an existing crack. This means that the bigger crack edge will release the stress dramatically, compared with a small crack edge. This is the main concept for the microcrack toughening model. The internal microcrack, crack arrester, can enlarge the tip of a propagating crack and postpone its growth, as shown in Fig. 10a, b. In general, a microcrack occurs due to internal stress, which is critical for the TFE. However, the nano-stratified TFE creates a crack arrester which acts like a microcrack, maintaining its optimized internal stress via the defect suppression mechanism. The crack arrester is created by a chemical etching process, between ZnO and trimethylaluminium (TMA), which is a reactant for Al2 O3 , at the interface between the ALD-ZnO layer and the ALD-Al2 O3 layer, as shown in Fig. 10c. With this strategy, the Al2 O3 /ZnO-based nano-stratified TFE showed enhanced mechanical properties compared to a single Al2 O3 -based multi-barrier. When both TFEs were compared under the same strain and stress, the WVTR of the multi-barrier increased sharply, while the nano-stratified barrier increased slowly. The WVTR of the 3.5 dyads multi-barrier and the nano-stratified barrier changed from 1.77 × 10−5 g/m2 /day to 1.35 × 10−2 g/m2 /day and 7.87 × 10−6 g/m2 /day to 7.78 × 10−5 g/m2 /day

Fig. 9 a Schematic diagram of a nano-stratified encapsulation barrier and b cross-section images of a nano-stratified encapsulation barrier. a Reproduced with permission [52]. Copyright 2019, Royal Society of Chemistry. b Reproduced with permission [46]. Copyright 2017, Royal Society of Chemistry

Encapsulation Technology for Flexible OLEDs

143

Fig. 10 Defect suppression mechanism of nano-stratified TFE: a schematic diagram of microcrack toughening model, b the propagation of a crack in an Al2 O3 -based multi-barrier and a nano-stratified barrier, c schematic diagram of Zn etching, and d cross-section images of Al2 O3 -based multi-barrier and nano-stratified barrier after bending. a–d Reproduced with permission [46]. Copyright 2017, Royal Society of Chemistry

with 0.63% strain, respectively. The acceleration of defect growth was suppressed in the nano-stratified barrier because of its defect suppression mechanism, while a bending crack, which is the main diffusion path for oxygen and water moisture, was observed in the multi-barrier, as shown in Fig. 10d. The previous nano-stratified structures were composed of two different sublayers, and considering material design, a more complicated structure was developed and studied. For this purpose, nanolaminate structures combining Al2 O3 , ZnO, and MgO were proposed, using structural and material design to fabricate highly impermeable/flexible and environmentally stable TFEs, as shown in Fig. 11a [49]. The ALDZnO, ALD-Al2 O3 , and ALD-MgO sublayers were used as the main gas permeation barrier, residual stress reliever, and water absorber, respectively. In addition, the Al2 O3 sublayers formed chemical bonds at the respective interfaces with the ZnO and MgO sublayers. To obtain better TFE properties, the thickness of the ultrathin sublayers, which is the main factor for the multi-interfacial structure, needs to be optimized. A 1.5 dyads TFE using 50-nm-thick ALD-ZAM (ZnO, Al2 O3 , MgO) layers composed of a 1-nm-thick sublayer was able to achieve a WVTR of 10−6 g/m2 /day and a mechanical reliability up to a tensile strain of 1%. Figure 11b and Table 2 show the WVTR properties of the nano-stratified structure TFE. Another form of flexible display, a wearable display, can be fabricated on a textile substrate instead of a plastic substrate, and then applied in various e-textile industries, such as in the fashion, healthcare, and automobiles. This chapter should contribute

144

E. G. Jeong and K. C. Choi

Fig. 11 a Schematic diagram of the ZAM-based nano-stratified encapsulation barrier and b WVTR estimation according to bending strain. a Reproduced with permission [49]. Copyright 2017, American Chemical Society

Table 2 The summary of WVTR estimation based on different inorganic layer structures at various strains Structure of inorganic layer

WVTR under strain (g/m2 /day) 0%

0.21%

0.32%

0.63%

0.89%

1.04%

1.25%

Al2 O3

1.77 × 10−5

3.60 × 10−5

4.13 × 10−4

1.35 × 10−2

N/A

N/A

N/A

Al2 O3 /ZnO

7.87 × 10−6

1.96 × 10−5

4.05 × 10−5

8.57 × 10−5

N/A

N/A

N/A

Al2 O3 /ZnO/MgO

2.06 × 10−6

N/A

3.59 × 10−6

8.19 × 10−6

2.22 × 10−5

7.34 × 10−5

3.67 × 10−4

to the evolution of the wearable display, as well as provide insights on fabric-based devices. OLEDs are the most promising candidates for wearable displays because of their low operating voltages, flexibility, lightweight, and ultrathinness. These features are compatible with the physical properties of fabric, such as drape behavior. However, OLED thickness (a few hundreds of nm) reduces the stability of textile-based OLEDs, since the textiles have a micro-scale rough and curved surface. To date, several methods of fabricating fabric substrates which maintain their intrinsic properties have been reported [50, 51]. However, further advances in performance, environmental stability, mechanical stability, interconnection among components, and washability need to be accomplished for practical wearable applications. Recently, a textile-based washable OLEDs module was developed which met the requirements for the successful commercialization of wearable displays. A new type of washable encapsulation barrier, a modified form of the nanostratified TFE barrier, was used to protect a wearable display module composed of OLEDs and polymer solar cells (PSCs) from the washing process [52]. Al2 O3 , which is mainly used for its barrier property in TFE, usually undergoes deterioration during the wash process. When the Al2 O3 is exposed to an aqueous environment, a phase transition into boehmite occurs, which has a crystalline structure. The crystalline

Encapsulation Technology for Flexible OLEDs

145

structure includes a lot of grains which can be used as a diffusion path for water moisture and oxygen, and the WVTR of the TFE increases rapidly. To prevent its degradation, a protonation–deprotonation reaction between the Al2 O3 layer and the polymer capping layer was used. For the capping layer, a SiO2 polymer composite was used because of its high density of Si–O bonds. The textilebased wearable display module maintained its properties even after 20 times of washing and bending at a 3 mm bending radius. As interest in flexible, rollable, and foldable displays continues to grow, more advanced flexible display technologies have been required. Among the various required technologies, the TFE is the most critical core technology for commercialization because it is directly related to display reliability and flexibility. As shown in Fig. 12a, to be suitable for flexible display OLEDs require a WVTR of 10−5 g/m2 /day and a bending radius of 30R to 1R, from flexible to foldable. A number of TFE technologies have been developed to meet both constraints. Encapsulation technologies, in particular, from conventional multi-barrier to nano-stratified barrier are needed to improve flexibility, as highlighted in this chapter. Figure 12b and Table 3 summarize

Fig. 12 a Required WVTR for different devices and bending radius for flexible, rollable, and foldable devices and b estimated WVTR for various kinds of TFE. a Reproduced with permission [4]. Copyright 2020, Taylor & Francis

Table 3 Summary of WVTR estimates based on different TFE structures Structure

WVTR (g/m2 /day)

Remark

Before bending

After bending

Al2 O3 /S-H nanocomposite

1.26 × 10−5

1.76 × 10−5 (0.26%)

Conventional

Al2 O3 /S-H nanocomposite/Ag

8.7 × 10−6

4.46 × 10−5 (0.41%)

Metal included

Al2 O3 /S-H nanocomposite/Buffer

4.4 × 10−5

8.2 × 10−5 (1.18%)

NA engineering

Al2 O3 /ZnO/S-H nanocomposite

7.87 × 10−6

7.78 × 10−5 (0.63%)

Crack arrester

Al2 O3 /ZnO/MgO/S-H nanocomposite

2.06 × 10−6

3.67 × 10−4 (1.25%)

Crack deflection

146

E. G. Jeong and K. C. Choi

the representative TFEs of each method introduced in this chapter. Since each technology has its own disadvantages, combining various technologies to compensate their limitations will likely help to optimize TFEs.

4 Conclusion Current flexible encapsulation technology is typically developed in specific experimental environments with stable conditions, but for commercialization, more intensive verification in various environments is necessary. Accordingly, it is not easy to predict with exact timing when the commercialization of a highly reliable flexible display will occur. However, if R&D is conducted with a focus on the following technical issues, it will not only accelerate the time to market, but also exert a ripple effect on flexible display technology. First, it is necessary to note that further innovation in materials must be accomplished in parallel, to dramatically improve flexibility. Although the strategies mentioned in this chapter can be used for a transitional approach, they have fundamental constraints, such as complexity in fabrication and limitations in mechanical properties. A substrate composed of a new material with sufficient modulus of elasticity to overcome plastic deformation and materials for TFE that do not break with stretching need to be developed for the form factor innovations of next-generation flexible technology. Second, it will be necessary to develop a technology that substitutes for naturally rigid and brittle components. In addition to enhancing the flexibility of the TFE, which has been extensively discussed in this chapter, the flexibility of other components with presently rigid characteristics, such as semiconductors, modules, and batteries, will also emerge as major issues. Since these components are themselves closely tied to advances in the performance of the flexible display, they will emerge as key challenges in next-generation flexible display technology. Third, the further development of flexible technology will require simplifying the internal structure of the flexible display as much as possible. This will include, for example, the development of a composite film incorporating a touch panel, a polarizing plate, a cover window, and the like. As evolving into a thinner and simpler display, a case, such as the backlight is no more required with the transition from LCD to OLED, may occur in which certain materials and components are integrated or disappear. In the long-term evolution of form factor innovation, as opposed to the short term, it will be necessary to integrate materials and components. In order to commercialize highly reliable flexible displays and provide new momentum for the form factor, the innovative development of components such as materials, processes, substrate, and TFE needs to continue. However, it is important that the innovations in technology could not immediately open the flexible display product market. Although some technologies secured enough mature development, they are failed to create new markets in the process of commercialization while others created innovative products.

Encapsulation Technology for Flexible OLEDs

147

To be successful, new markets should not only be based on the maturity of the technology, but also the surrounding conditions, such as the needs of consumers and the composition of suppliers and partners. No matter how innovative a technology is, if a business group that commercializes it does not care about strategic choices, it will not lead to changes in everyday life. To connect the technology innovation and the display form factor to an actual market, it is also necessary to perform strategic activities such as creating a product concept that can reach customers, discover new application markets, and achieve attractive prices.

References 1. J. Song, H. Lee, E.G. Jeong, K.C. Choi, S. Yoo, Organic light-emitting diodes: pushing toward the limits and beyond. Adv. Mater. 1907539 (2020). https://doi.org/10.1002/adma.201907539 2. S. Kwon, Y.H. Hwang, M. Nam, H. Chae, H.S. Lee, Y. Jeon, S. Lee, C.Y. Kim, S. Choi, E.G. Jeong, K.C. Choi, Recent progress of fiber shaped lighting devices for smart display applications—a fibertronic perspective. Adv. Mater. 32, 1903488 (2020). https://doi.org/10. 1002/adma.201903488 3. S. Kwon, H. Kim, S. Choi, E.G. Jeong, D. Kim, S. Lee, H.S. Lee, Y.C. Seo, K.C. Choi, Weavable and highly efficient organic light-emitting fibers for wearable electronics: a scalable, low-temperature process. Nano Lett. 18, 347–356 (2018). https://doi.org/10.1021/acs.nanolett. 7b04204 4. E.G. Jeong, J.H. Kwon, K.S. Kang, S.Y. Jeong, K.C. Choi, A review of highly reliable flexible encapsulation technologies towards rollable and foldable OLEDs. J. Inf. Disp. 21, 19–32 (2020). https://doi.org/10.1080/15980316.2019.1688694 5. R. Meerheim, B. Lüssem, K. Leo, Efficiency and stability of p-i-n type organic light emitting diodes for display and lighting applications. Proc. IEEE 97, 1606–1626 (2009). https://doi.org/ 10.1109/JPROC.2009.2022418 6. S. Scholz, R. Meerheim, K. Walzer, K. Leo, Chemical degradation mechanisms of organic semiconductor devices, in Organic Optoelectronics and Photonics III, ed. by P.L. Heremans, M. Muccini, E.A. Meulenkamp (SPIE, 2008), p. 69991B 7. S. Scholz, R. Meerheim, B. Lüssem, K. Leo, Laser desorption/ionization time-of-flight mass spectrometry: a predictive tool for the lifetime of organic light emitting devices. Appl. Phys. Lett. 94, 043314 (2009). https://doi.org/10.1063/1.3075607 8. I.R.De Moraes, S. Scholz, B. Lüssem, K. Leo, Analysis of chemical degradation mechanism within sky blue phosphorescent organic light emitting diodes by laser-desorption/ionization time-of-flight mass spectrometry. Org. Electron. 12, 341–347 (2011). https://doi.org/10.1016/ j.orgel.2010.11.004 9. S.T. Lee, Z.Q. Gao, L.S. Hung, Metal diffusion from electrodes in organic light-emitting diodes. Appl. Phys. Lett. 75, 1404–1406 (1999). https://doi.org/10.1063/1.124708 10. Z.D. Popovic, H. Aziz, N.X. Hu, A.M. Hor, G. Xu, Long-term degradation mechanism of tris(8hydroxyquinoline) aluminum-based organic light-emitting devices. Synth. Met. 111, 229–232 (2000). https://doi.org/10.1016/S0379-6779(99)00353-7 11. D.J. Williams, G. Rajeswaran, Direct view AM OLED technology for camera and other portable electronic applications, in Cockpit Displays X, ed. by D.G. Hopper (SPIE, 2003), p. 166 12. Y. Tsuruoka, S. Hieda, S. Tanaka, H. Takahashi, 21.2: Transparent thin film desiccant for OLEDs. SID Symp. Dig. Tech. Pap. 34, 860 (2003). https://doi.org/10.1889/1.1832406 13. J.S. Lewis, M.S. Weaver, Thin-film permeation-barrier technology for flexible organic lightemitting devices. IEEE J. Sel. Top. Quantum Electron. 10, 45–57 (2004). https://doi.org/10. 1109/JSTQE.2004.824072

148

E. G. Jeong and K. C. Choi

14. J.H. Kwon, S. Choi, Y. Jeon, H. Kim, K.S. Chang, K.C. Choi, Functional design of dielectric– metal–dielectric-based thin-film encapsulation with heat transfer and flexibility for flexible displays. ACS Appl. Mater. Interfaces 9, 27062–27072 (2017). https://doi.org/10.1021/acsami. 7b06076 15. M.S. Weaver, L.A. Michalski, K. Rajan, M.A. Rothman, J.A. Silvernail, J.J. Brown, P.E. Burrows, G.L. Graff, M.E. Gross, P.M. Martin, M. Hall, E. Mast, C. Bonham, W. Bennett, M. Zumhoff, Organic light-emitting devices with extended operating lifetimes on plastic substrates. Appl. Phys. Lett. 81, 2929–2931 (2002). https://doi.org/10.1063/1.1514831 16. C.C. Chiang, D.S. Wuu, H.B. Lin, Y.P. Chen, T.N. Chen, Y.C. Lin, C.C. Wu, W.C. Chen, T.H. Jaw, R.H. Horng, Deposition and permeation properties of SiNX/parylene multilayers on polymeric substrates. Surf. Coatings Technol. 200, 5843–5848 (2006). https://doi.org/10.1016/ j.surfcoat.2005.08.133 17. T.-N. Chen, D.-S. Wuu, C.-C. Wu, C.-C. Chiang, Y.-P. Chen, R.-H. Horng, Improvements of permeation barrier coatings using encapsulated parylene interlayers for flexible electronic applications. Plasma Process. Polym. 4, 180–185 (2007). https://doi.org/10.1002/ppap.200 600158 18. M.S. Weaver, A.B. Chwang, M.A. Rothman, J.A. Silvernail, M.G. Hack, J.J. Brown, P.E. Burrows, G.L. Graff, M.E. Gross, P.M. Martin, M. Hall, E. Mast, C.C. Bonham, W.D. Bennett, M. Zumhoff, Recent progress in flexible displays, in Cockpit Displays IX: Displays for Defense Applications, ed. by D.G. Hopper (SPIE, 2002), pp. 237–249 19. A.B. Chwang, M.A. Rothman, S.Y. Mao, R.H. Hewitt, M.S. Weaver, J.A. Silvernail, K. Rajan, M. Hack, J.J. Brown, X. Chu, L. Moro, T. Krajewski, N. Rutherford, Thin film encapsulated flexible organic electroluminescent displays. Appl. Phys. Lett. 83, 413–415 (2003). https://doi. org/10.1063/1.1594284 20. S.-M. Lee, J.H. Kwon, S. Kwon, K.C. Choi, A review of flexible OLEDs toward highly durable unusual displays. IEEE Trans. Electron. Dev. 64, 1922–1931 (2017). https://doi.org/10.1109/ TED.2017.2647964 21. A. Dameron, M.O. Reese, T.J. Moriconie, M. Kempe, Understanding moisture ingress and packaging requirements for photovoltaic modules. Photovoltaics Int. 9 (2010) 22. P. Hülsmann, D. Philipp, M. Köhl, Measuring temperature-dependent water vapor and gas permeation through high barrier films. Rev. Sci. Instrum. 80, 113901 (2009). https://doi.org/ 10.1063/1.3250866 23. X.D. Zhang, J.S. Lewis, C.B. Parker, J.T. Glass, S.D. Wolter, Measurement of reactive and condensable gas permeation using a mass spectrometer. J. Vac. Sci. Technol. A Vac. Surf. Film. 26, 1128–1137 (2008). https://doi.org/10.1116/1.2952453 24. R. Dunkel, R. Bujas, A. Klein, V. Horndt, Method of measuring ultralow water vapor permeation for OLED displays. Proc. IEEE 93, 1478–1482 (2005). https://doi.org/10.1109/JPROC.2005. 851494 25. A.A. Dameron, S.D. Davidson, B.B. Burton, P.F. Carcia, R.S. McLean, S.M. George, Gas diffusion barriers on polymers using multilayers fabricated by Al2 O3 and rapid SiO2 atomic layer deposition. J. Phys. Chem. C 112, 4573–4580 (2008). https://doi.org/10.1021/jp076866+ 26. S. Cros, M. Firon, S. Lenfant, P. Trouslard, L. Beck, Study of thin calcium electrode degradation by ion beam analysis. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. Atoms. 251, 257–260 (2006). https://doi.org/10.1016/J.NIMB.2006.06.014 27. Y.C. Han, C. Jang, K.J. Kim, K.C. Choi, K. Jung, B.-S. Bae, The encapsulation of an organic light-emitting diode using organic–inorganic hybrid materials and MgO. Org. Electron. 12, 609–613 (2011). https://doi.org/10.1016/J.ORGEL.2011.01.007 28. E. Kim, Y. Han, W. Kim, K.C. Choi, H.-G. Im, B.-S. Bae, Thin film encapsulation for organic light emitting diodes using a multi-barrier composed of MgO prepared by atomic layer deposition and hybrid materials. Org. Electron. 14, 1737–1743 (2013). https://doi.org/10.1016/J. ORGEL.2013.04.011 29. Y.C. Han, E. Kim, W. Kim, H.-G. Im, B.-S. Bae, K.C. Choi, A flexible moisture barrier comprised of a SiO2 -embedded organic–inorganic hybrid nanocomposite and Al2 O3 for thinfilm encapsulation of OLEDs. Org. Electron. 14, 1435–1440 (2013). https://doi.org/10.1016/ J.ORGEL.2013.03.008

Encapsulation Technology for Flexible OLEDs

149

30. J.H. Kwon, E.G. Jeong, Y. Jeon, D.-G. Kim, S. Lee, K.C. Choi, Design of highly water resistant, impermeable, and flexible thin-film encapsulation based on inorganic/organic hybrid layers. ACS Appl. Mater. Interfaces 11, 3251–3261 (2019). https://doi.org/10.1021/acsami.8b11930 31. D.C. Miller, R.R. Foster, Y. Zhang, S.H. Jen, J.A. Bertrand, Z. Lu, D. Seghete, J.L. O’Patchen, R. Yang, Y.C. Lee, S.M. George, M.L. Dunn, The mechanical robustness of atomic-layer- and molecular-layer-deposited coatings on polymer substrates. J. Appl. Phys. 105, 093527 (2009). https://doi.org/10.1063/1.3124642 32. S.H. Jen, J.A. Bertrand, S.M. George, Critical tensile and compressive strains for cracking of Al2 O3 films grown by atomic layer deposition. J. Appl. Phys. 109, 084305 (2011). https://doi. org/10.1063/1.3567912 33. J. Zhou, T.Y. Tee, X. Zhang, J.E. Luan, Characterization and modeling of hygroscopic swelling and its impact on failures of a flip chip package with no-flow underfill, in Proceedings of 7th Electronics Packaging Technology Conference, EPTC 2005 (2005), pp. 561–568 34. E. Stellrecht, B. Han, M.G. Pecht, Characterization of hygroscopic swelling behavior of mold compounds and plastic packages. IEEE Trans. Compon. Packag. Technol. 27, 499–506 (2004). https://doi.org/10.1109/TCAPT.2004.831777 35. H. Ardebili, E.H. Wong, M. Pecht, Hygroscopic swelling and sorption characteristics of epoxy molding compounds used in electronic packaging. IEEE Trans. Compon. Packag. Technol. 26, 206–214 (2003). https://doi.org/10.1109/TCAPT.2002.806172 36. E.H. Wong, R. Rajoo, S.W. Koh, T.B. Lim, The mechanics and impact of hygroscopic swelling of polymeric materials in electronic packaging. J. Electron. Packag. Trans. ASME. 124, 122– 126 (2002). https://doi.org/10.1115/1.1461367 37. M. Kitano, S. Kawai, A. Nishimura, K. Nishi, A study of package cracking during the reflow soldering process. Trans. Jpn. Soc. Mech. Eng. Ser. A 55, 356–363 (1989). https://doi.org/10. 1299/kikaia.55.356 38. C.J. Chiang, C. Winscom, S. Bull, A. Monkman, Mechanical modeling of flexible OLED devices. Org. Electron. 10, 1268–1274 (2009). https://doi.org/10.1016/j.orgel.2009.07.003 39. S. Lee, J.Y. Kwon, D. Yoon, H. Cho, J. You, Y.T. Kang, D. Choi, W. Hwang, Bendability optimization of flexible optical nanoelectronics via neutral axis engineering. Nanoscale Res. Lett. 7, 256 (2012). https://doi.org/10.1186/1556-276X-7-256 40. Y.C. Han, E.G. Jeong, H. Kim, S. Kwon, H.-G. Im, B.-S. Bae, K.C. Choi, Reliable thinfilm encapsulation of flexible OLEDs and enhancing their bending characteristics through mechanical analysis. RSC Adv. 6, 40835–40843 (2016). https://doi.org/10.1039/C6RA06571F 41. G.M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J. Mater. Res. 7, 1564–1583 (1992). https:// doi.org/10.1557/JMR.1992.1564 42. J. Meyer, H. Schmidt, W. Kowalsky, T. Riedl, A. Kahn, The origin of low water vapor transmission rates through Al2 O3 /ZrO2 nanolaminate gas-diffusion barriers grown by atomic layer deposition. Appl. Phys. Lett. 96, 243308 (2010). https://doi.org/10.1063/1.3455324 43. C.-Y. Wang, C. Fuentes-Hernandez, M. Yun, A. Singh, A. Dindar, S. Choi, S. Graham, B. Kippelen, Organic field-effect transistors with a bilayer gate dielectric comprising an oxide nanolaminate grown by atomic layer deposition. ACS Appl. Mater. Interfaces 8, 29872–29876 (2016). https://doi.org/10.1021/acsami.6b10603 44. L.H. Kim, K. Kim, S. Park, Y.J. Jeong, H. Kim, D.S. Chung, S.H. Kim, C.E. Park, Al2 O3 /TiO2 nanolaminate thin film encapsulation for organic thin film transistors via plasma-enhanced atomic layer deposition. ACS Appl. Mater. Interfaces 6, 6731–6738 (2014). https://doi.org/10. 1021/am500458d 45. E.G. Jeong, Y.C. Han, H.-G. Im, B.-S. Bae, K.C. Choi, Highly reliable hybrid nano-stratified moisture barrier for encapsulating flexible OLEDs. Org. Electron. 33, 150–155 (2016). https:// doi.org/10.1016/J.ORGEL.2016.03.015 46. E.G. Jeong, S. Kwon, J.H. Han, H.-G. Im, B.-S. Bae, K.C. Choi, A mechanically enhanced hybrid nano-stratified barrier with a defect suppression mechanism for highly reliable flexible OLEDs. Nanoscale 9, 6370–6379 (2017). https://doi.org/10.1039/C7NR01166K

150

E. G. Jeong and K. C. Choi

47. A.A. Griffith, The phenomena of rupture and flow in solids. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 221, 163–198 (1921). https://doi.org/10.1098/rsta.1921.0006 48. G.R. Irwin, Analysis of stresses and strains near the end of a crack traversing a plate. J. Appl. Mech. 24, 361–364 (1957) 49. J.H. Kwon, Y. Jeon, S. Choi, J.W. Park, H. Kim, K.C. Choi, Functional design of highly robust and flexible thin-film encapsulation composed of quasi-perfect sublayers for transparent, flexible displays. ACS Appl. Mater. Interfaces 9, 43983–43992 (2017). https://doi.org/10.1021/ acsami.7b14040 50. S. Choi, S. Kwon, H. Kim, W. Kim, J.H. Kwon, M.S. Lim, H.S. Lee, K.C. Choi, Highly flexible and efficient fabric-based organic light-emitting devices for clothing-shaped wearable displays. Sci. Rep. 7, 6424 (2017). https://doi.org/10.1038/s41598-017-06733-8 51. W. Kim, S. Kwon, Y.C. Han, E. Kim, K.C. Choi, S.-H. Kang, B.-C. Park, Reliable actual fabric-based organic light-emitting diodes: toward a wearable display. Adv. Electron. Mater. 2, 1600220 (2016). https://doi.org/10.1002/aelm.201600220 52. E.G. Jeong, Y. Jeon, S.H. Cho, K.C. Choi, Textile-based washable polymer solar cells for optoelectronic modules: toward self-powered smart clothing. Energy Environ. Sci. 12, 1878– 1889 (2019). https://doi.org/10.1039/C8EE03271H

Oxide Thin-Film Transistors for OLED Displays Hyeon Joo Seul, Min Jae Kim, and Jae Kyeong Jeong

Abstract Since the invention of IGZO by Prof. Hosono in 2004, the development of IGZO thin-film transistors has been accelerated by material optimization including cation composition, processing conditions, and careful architecture design such as a self-aligned structure. In this chapter, the baseline of the current IGZO backplane technology for AMOELD TVs will be addressed including the architecture and process optimization of IGZO TFTs. In particular, the next-generation backplane technology beyond IGZO and SiO2 will be briefly presented from the viewpoint of high mobility and low-voltage operation. Finally, the device instability of IGZO TFTs will be discussed, which is critical for their implementation in AMOLED products. The degradation mechanisms for the bias thermal stress and light illumination will be summarized including the carrier trapping/injection, defect creations such as oxygen vacancies, oxygen interstitials, and a hydrogen complex model.

1 High-Performance Metal-Oxide Thin-Film Transistors Since the discovery of amorphous indium gallium zinc oxide (a-IGZO) by Prof. Hosono and coworkers in 2004, intensive research and development have been carried out to improve the performance of their thin-film transistors (TFTs) for the backplane in active-matrix (AM) liquid-crystal display (LCD) and organic light-emitting diodes display (OLED) applications. Their intriguing properties such as high carrier mobility, extremely low leakage current, good gate swing, and low-temperature processing capability led to the commercialization of an IGZO TFTs in high-end AM displays. The cost and visual quality of OLED panels strongly depends on the architecture of TFTs, which can be classified as a bottom gate and coplanar structure. Thus, the configuration and critical fabrication process of IGZO TFTs are firstly briefly described.

H. J. Seul · M. J. Kim · J. K. Jeong (B) Department of Electronic Engineering, Hanyang University, Seoul 133-791, South Korea e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_7

151

152

H. J. Seul et al.

1.1 TFT Configuration and Fabrication Process There are several types of device architectures used for metal oxide TFTs. They can be generally classified as a bottom or top gate structure on the basis of the gate electrode position with respect to the channel layer. The conventional structure adopted by most LCD manufactures is the bottom gate with a back-channel etch configuration because it has the merits of being a simple and low-cost fabrication process (Fig. 1a). Also, an opaque gate electrode can shield the channel region from the back light in LCDs. However, the IGZO channel layer is easily degraded by either energetic bombardment or chemical attack during the back-channel etch process. To solve this problem, the etch-stopper (ES) structure has been preferred where an additional protecting layer is deposited on the channel layer prior to depositing the S/D electrode and data line. Although an ES-type bottom gate structure exhibits high performance and acceptable uniformity, it requires an additional deposition and lithography step compared to the BCE-type bottom gate structure. In addition, it has a large parasitic capacitance due to the inevitable overlap between the gate (G) and S/D electrode, which increases the RC delay time constant, and thus reduces its switching speed. Nowadays, OLED TVs are driven by the IGZO transistor with a selfaligned structure because a self-aligned architecture enables high-speed operation due to negligible parasitic capacitance between the G and S/D electrode, as shown in Fig. 1c [1]. The critical process in a self-aligned structure is the metallization of the access IGZO region for good ohmic contact and low parasitic resistance. Ar plasma treatment is usually used to convert the semiconducting IGZO to metallic IGZO during the dry etching of a SiO2 film as a gate insulator and/or interlayer. Plasma exposure on the IGZO top region during the over-etching of SiO2 causes preferential oxygen loss (oxygen deficiency) via the physical collision of Ar radicals, resulting in a rapid increase (1019 /cm3 ) of the free electron density (N e ). This method is used in the mass production of IGZO backplanes for OLED TVs because of its simple process and acceptable metallization capability. Sony reported the metal oxidation-induced metallization process to obtain a low ohmic contact resistance and sheet resistance in a self-aligned architecture [2]. After gate patterning, the ultrathin (~5 nm) reactive Al or Ti film is conformally deposited on the IGZO access region. The subsequent thermal annealing allows the oxidation of an Al or Ti film onto AlOx

Fig. 1 Schematic device configurations of the a BCE bottom gate, b ES bottom gate, and c selfaligned coplanar top gate structure

Oxide Thin-Film Transistors for OLED Displays

153

Fig. 2 Process flowchart of metal oxidation-induced metallization to form the low ohmic contact [2]

or TiOx , where the underneath IGZO top region loses oxygen anions to this capping layer due to the lower oxidation power. The oxygen-deficient top IGZO layer serves as an excellent metallic region, which results in a low ohmic contact resistance and sheet resistance (Fig. 2).

1.2 TFT Optimization Cation Composition in the IGZO Channel Layer Prof. Hosono’s group designed the IGZO quaternary system in which the role of each binary oxide was identified on the basis of the compositional-dependent Hall effect results. Indium (In) cations play an essential role in boosting the carrier mobility, which is one of the most important metrics for transistors. The intercalation of In 5 s orbitals provides a percolation pathway in the conduction band due to its relatively large ionic radius and effective low electron mass of In2 O3 . The gallium (Ga) cation has a high ionic field strength due to its small radius and 3 + valence state, preventing the formation of oxygen vacancy (V o ) defects due to its strong oxygen binding capability. Thus, it diminishes the N e value (so-called carrier suppressor). The role of In and Ga as a carrier enhancer and suppressor, respectively, can be clearly seen in Fig. 3. A zinc (Zn) cation stabilizes the formation of an amorphous phase. ZnO has a wurtzite crystal structure based on the corner sharing of tetrahedral ZnO4 , whereas In2 O3 and Ga2 O3 consist of edge-sharing octahedral. Thus, the coexisting nature of In, Ga, and Zn makes it difficult to cause crystallization due to dissimilarity of the crystallographic structure. Iwasaki et al. reported a systematic combinatorial study on the performance of IGZO transistors rather than relying on indirect Hall measurements using three targets of In2 O3 , Ga2 O3 , and ZnO in an RF magnetron co-sputtering apparatus. A clear tradeoff relationship between the field-effect mobility (FE ) and I ON/OFF modulation ratio

154

H. J. Seul et al.

Fig. 3 Variations of the carrier mobility and concentration from Hall effect measurements (left) and the amorphous formation region (right). The numbers in parenthesis denote N e (×1018 cm−3 ) from Hosono et al. [11]

(or V TH ) was observed, as shown in Fig. 4 [3]. It can be seen that the μFE value was enhanced by increasing the In fraction in the IGZO channel layer. However, the high fraction of In cations causes negative displacement of V TH and a loss in terms of the I ON/OFF ratio because a high In composition favors the formation of shallow V O donors due to its lower binding energy to oxygen anions, leading to the creation of free electron carriers. Therefore, the optimal composition of In:Ga:Zn was determined to be 1:1:1 or 2:2:1 for acceptable performance and reliability of the resulting transistors [4, 5]. Optimization of the Deposition Conditions of the Oxide Semiconducting Film The performance of oxide TFTs is greatly affected by not only the aforementioned channel composition but also the deposition conditions in the sputtering system. Thus, the effect of major factors such as the chamber pressure, RF power, and oxygen flow ratio during the sputtering process on the device performance has been intensively investigated. Fine-tuning of the chamber pressure turns out to be crucial in Fig. 4 Variations of μFE , V TH , and I ON/OFF ratio as a function of In/(In + Ga) ratio in the IGZO channel layer where the Zn fraction was fixed at 31–34%. Reproduced from Iwasaki et al. [3]

Oxide Thin-Film Transistors for OLED Displays

155

obtaining high μFE and low subthreshold swing (SS) characteristics in IGZO TFTs [6]. The μFE and SS values for the device with the IGZO channel prepared at a chamber pressure of 1 mTorr were improved to 21.8 cm2 /Vs and 0.17 V/decade, respectively, whereas the device fabricated at a chamber pressure of 5 mTorr exhibited modest μFE (11.4 cm2 /Vs) and large SS (0.87 V/decade) values. A lower chamber pressure facilitates more effective densification and a smooth morphology as a result of the enhanced diffusion length upon the arrival of adatoms on the growing surface. A similar result was also reported by Yasuno et al. [7]. There have been many reports on the effect of the target power and oxygen flow ratio on the performance of amorphous metal oxide TFTs [8–10]. For example, increasing the RF power frequently improves the device performance because it leads to fast deposition through the enhanced kinetic energies of the adatoms and densification of the film [8, 9]. In some cases, an increased RF power can cause the adverse lattice damage in the channel layer via an energetic bombardment effect, degrading the carrier mobility in the resulting TFTs [10]. Lee et al. emphasized the importance of the densification of a semiconducting oxide channel layer for high-performance TFTs with excellent photo-bias stability [11]. The effects of the chamber pressure, RF power, and oxygen flow ratio during sputtering of the ZTO channel layer were examined in detail. A major finding was that securing a high mass density of the channel layer is a key factor in achieving high mobility and good stability irrespective of specific processing conditions, as shown in Figs. 5 and 6. Post-deposition Annealing of the Oxide Semiconducting Film Defects such as dangling bonds, non-stoichiometry, and impurities can be removed by an appropriate annealing process. Generally, TFTs with an as-deposited oxide channel layer suffer from inferior performance and instability including a low μFE , large SS, and significant hysteresis. A temperature in the range of 300–400 °C is accepted to be the optimal annealing temperature for high-performance oxide TFTs [12]. It is noted that a high temperature (≥500 °C) can create a new undesirable trap state resulting from the dissociation and subsequent elimination of hydrogen species.

Fig. 5 Representative transfer characteristics of the oxide TFTs fabricated under different deposition conditions: chamber pressure (left), RF power (middle), and oxygen flow ratio (right). Reproduced from Lee et al. [11]

156

H. J. Seul et al.

Fig. 6 Correlation between the mass density of the channel layers and a μFET , b SS, c bulk-trap density, and d VTH . Reproduced by Lee et al. [11]

The thermal annealing atmosphere is also important in determining the performance and reliability of IGZO TFTs. Nomura et al. examined the effect of wet and dry O2 annealing at 400 °C on the performance of IGZO TFTs. Wet-annealed IGZO TFTs have the strongest immunity to external stress conditions. This behavior was explained by the higher diffusivity of OH groups due to its smaller size compared to O2 molecules in the amorphous IGZO network where the diffusing OH group eliminates local defect sites [13]. Similarly, other annealing methods such as plasma treatment [14–16] or high-pressure annealing [17, 18] have been investigated, which indicates that the purification and densification of the oxide channel layer are crucial to the high performance of oxide TFTs. Optimal Channel Layer Thickness It is necessary to control the V TH value of a switching transistor for the design of OLED panels. The enhancement mode (EM) is usually preferred in terms of the low standby power consumption because the channel region is completely turned off at a zero gate voltage. However, the depletion mode (DM) is frequently observed for oxide TFTs when the level of In cations (or oxygen anions) in the oxide channel layer

Oxide Thin-Film Transistors for OLED Displays

157

(t ch ) increases (or decreases) leading to high carrier mobility. In fully depleted thinfilm transistors, the V TH value can be represented by the following Eq. (2), assuming that the effect of gap states is negligible: [19] VT H = Vo −

2 q Ne tch q Ne tch − Ci 2εo εr

(1)

where V O is a non-ideality related constant, and εo and εr are the vacuum permittivity and relative dielectric constant of oxide semiconductor, respectively. It can be shown that the V TH sensitivity with respect to the channel thickness (V TH /t ch ) is mainly determined by the N e value. As the thickness of the channel layer decreases, the V TH values of the resulting TFTs will be positively displaced. This behavior can be useful in controlling the V TH value of oxide TFTs. Typically, the t ch value of oxide TFTs fabricated in a production line is approximately 30 nm. The geometrical downscaling of a self-aligned TFT is required to meet the ever-increasing demand for an ultra-high pixel density and low-voltage driving. Shao et al. examined the dependence of t ch from 108 , which can be ascribed to

Oxide Thin-Film Transistors for OLED Displays

159

effective mass densification resulting from the simultaneous packing of cations with different sizes and reduction of tail states [33]. The gate-bias stability of IGZTO TFTs was also comparable to that of IGZO TFTs, suggesting that this composition can be considered as an alternative to the IGZO system in next-generation backplane technology [34]. High-k Dielectric for Low-Voltage Driving The drain current drivability in oxide TFTs is proportional to the capacitance per unit area. The PECVD-derived SiO2 , which is a standard gate dielectric of oxide TFTs, has a low dielectric constant (κ) of ~ 3.9, which requires a high operation voltage (approximately 20 V) due to inefficient capacitive coupling. Therefore, the usage of a high-κ dielectric such as ZrO2 [35, 36], HfO2 [37, 38], Y2 O3 [39, 40], or their ternary alloy [41–44] has been investigated to reduce the operation voltage of oxide TFTs. The variations of the relative κ and bandgap (E g ) values for various insulating materials are summarized in Fig. 8 [45]. ZnO TFTs with a polycrystalline monoclinic HfO2 dielectric layer showed a high μFE of ~42 cm2 /Vs and low operating voltage of ~6 V [45]. The same group reported a remarkably enhanced μFE of ~85 cm2 /Vs and similar low operating voltage (~6 V) for Li-doped ZnO TFTs with a ZrO2 gate dielectric [46]. However, the rather high leakage current (I OFF ) and low I ON/OFF ratio due to their lower E g values and facile formation of trap centers are penalties for high-κ dielectric-based oxide TFTs [47–49]. In order to overcome these issues, ternary alloy dielectrics such as HfLaOx , Yx Sc1-x O3 , or Al2-x Yx O3 as well as a nano-laminated structure have been intensively researched for low-voltage operating oxide TFTs. The novel gate dielectric stack consisting of Al2 O3 /HfO2 /Al2 O3 (AHA) exhibited low leakage characteristics as well as a high capacitance value due to the synergic effect of Al2 O3 with a wide E g (~8 eV) and HfO2 with a high permittivity (κ ≈ 20). TFTs with a bilayer IGZO channel and triple AHA dielectric stack exhibited a remarkably high μFE of ~112.8 cm2 /Vs, small SS of 0.08 V/decade, and an I ON/OFF ratio of 2 × 109 [50]. The non-negligible hysteresis of bilayer IGZO TFTs with a HfO2 dielectric layer can be eliminated by inserting a 2-nm-thick thin Al2 O3 layer between the IGZO channel and HfO2 dielectric layer, suggesting excellent interface matching between the IGZO and Al2 O3 layers [51, 52].

2 Reliability of Amorphous Oxide TFTs An AMOLED panel consists of TFT arrays as a backplane and a light-emitting diodes display mode as a frontplane. From the viewpoint of the backplane, the simplest unit pixel comprises two transistors (switching and driving transistors) and one capacitor. The switching transistor selects an image signal for the given pixel, which loads it to the storage capacitor as a form of a charge carrier. The charges stored in the storage capacitor retain the gate voltage applied to the driving transistor, which supplies the target current to the light-emitting diodes during a frame time. In a conventional active-matrix driving scheme, the select time is determined by the frame rate and

160

H. J. Seul et al.

Fig. 8 Bandgap versus static dielectric constant for gate dielectric oxides grown by different deposition methods

pixel density. The typical frame rate of 120 Hz means that the pixel is addressed 120 times per second. Thus, the duty time of a single frame is 8.33 ms. Assuming a UHD TV, the ratio of the turn-on (~2.1 μs) and turn-off times (~8.33 ms) during a single frame time is approximately 0.00025. It is evident that most of the gate-bias duty is given to the off-state condition, which corresponds to a negative gate-bias thermal stress (NBTS). In contrast, the driving transistors are always subjected to the on-state stress (PBTS) because it must supply a constant current to the OLED. Therefore, the PBTS and NBTS stabilities must be satisfied at the same time for the AMOLED. Unfortunately, the gate-bias stabilities of oxide TFTs are a big concern. Thus, it is very important to understand the degradation mechanism of oxide TFTs depending on the magnitude and duration of the applied PBTS and NBTS. Carrier Trapping/Injection Model Generally, the V TH value of oxide TFTs is positively and negatively shifted upon the application of PBTS and NBTS, respectively. When a positive voltage is applied to the gate electrode, the majority electron carriers will be accumulated at the gate/channel interface region shown in Fig. 9a. The positive V TH displacement means a reduction of the N e value, which occurs via either the trapping of electron carriers near the channel/dielectric interface or injection into the bulk-trap sites of the gate insulator [53–55]. Here, the interfacial states or channel gap states provide the potential trapping sites for any type of carrier. Convincing experimental evidence supporting the carrier trapping/injection model was provided by Moon et al. [56]. The oxygen (O2 ) plasma treatment on the SiNx gate insulator allows the resulting IGZO TFTs to possess superior PBTS stability compared to the device with an untreated SiNx gate insulator. An ultrathin SiON layer created as a result of O2 plasma treatment between IGZO and a SiNx layer acts as a good electron blocking barrier layer due to its wider bandgap nature compared to the SiNx layer. This reduces the interfacial trap density (Dit ) due to the better interface matching with respect to the IGZO channel film, which

Oxide Thin-Film Transistors for OLED Displays

161

Fig. 9 Energy band diagrams of the MIS structure under a PBS, b PBIS, and c NBIS. In the PBIS condition, the photo-generated electron–hole pairs quickly recombine after PBIS application because of the absence of an electric field, which is responsible for the better immunity against external PBIS application

results in the superior PBTS stability. The application of NBTS into the gate electrode will create complete depletion of the thin IGZO channel layer due to the quasi-Fermi energy level (E F ), decreasing toward the valence band (VB) edge. The NBTS-induced negative V TH shift turns out to be negligible or small compared to the PBTS-induced positive V TH shift, which is strongly related to the rarity of hole carriers. Exposure of visible or ultraviolet (UV) light greatly accelerates the amount of the negative V TH shift upon the NBTS condition, which is known as negative bias illumination stress (NBIS) instability, as shown in Fig. 9c. The absorption of light creates electron– hole pairs in the IGZO channel layer, which are separated by the local electric field. Photo-generated hole carriers will move toward the channel/dielectric interfaces and can be trapped (or injected) at the channel/dielectric interface (into dielectric bulk region), leading to the negative V TH shift. Obviously, the blocking energy barrier for the hole carrier is a critical factor to mitigate the NBIS-induced negative V TH shift. The SiNx dielectric film, which is a standard gate insulator for amorphous Si TFTs, suffers from a huge negative V TH shift upon NBIS application. This is attributed to the lack of a hole energy barrier for the SiNx /IGZO interface due to the smaller bandgap of SiNx (~5.1 eV). Conversely, a PECVD-derived SiO2 film has a wider E g of 8.8 eV, which results in a significant hole barrier of ~2.4 eV at the interface for the SiO2 /IGZO stack. Thus, the injection of photo-created hole carriers into the SiO2 dielectric layer can be effectively prevented, leading to better NBIS stability [57]. This is why the SiO2 becomes a standard gate dielectric layer in the IGZO backplane for AMOLED displays. Adsorption Model of Ambient Gas ZnO or SnO2 materials are good candidate systems for sensing different gases such as O2 , NO2 , and H2 O due to their extreme sensitivity. The application of PBTS on TFTs with an unpassivated IGZO channel layer causes a positive V TH shift, as mentioned earlier, which can be attributed to either carrier trapping/injection or the dynamical adsorption of ambient gases. In particular, the oxygen molecules in air absorb on the defective sites of the IGZO back surface, becoming a negatively charge

162

H. J. Seul et al.

Fig. 10 a Comparison of the IGZO PBTS instability with several passivation layers. Schematics showing the PBTS mechanism of ambient effects: b electric field-induced adsorption of oxygen molecules and c electric field-induced desorption of water molecules. Reproduced from Jeong et al. [58]

adsorbed oxygen ion or molecule due to its high electron negativity. The chemisorption of oxygen depletes the underneath IGZO channel layer, making the V TH value positively displaced. The application of PBTS induces a huge accumulation of free electron carriers in the IGZO channel layer, which calls for more absorption of oxygen molecules, leading to the positive V TH shift. Therefore, the passivation of an IGZO layer using high-quality SiO2 or Al2 O3 results in a significant improvement in terms of the PBTS and NBIS stability of the resulting TFTs due to the prevention of gate-bias and photo-induced adsorption or desorption of oxygen or hydroxyl groups [58–60] (Fig. 10). Oxygen Vacancy Model In an ionic IGZO semiconductor, the V O defect is identified as a meta-stable hole-trap center. In crystalline ZnO, the formation energy of a neutral VO state is calculated to be smaller than those of single (V +O ) and double ionized (V 2+ O ) defects. In the thermodynamical equilibrium state, a neutral V O state is the most stable, making a V O defect electrically inactive under normal transistor operation. However, the transition of the neutral V O state to the V 2+ O excited state is allowed upon external light exposure. Outward relaxation during this photoionization renders the V 2+ O state to be a meta-stable shallow donor, donating the delocalized free electrons in the conduction band [61–63]. This model is accepted as an origin of the persistent photoconductivity (PPC) effect in crystalline ZnO materials. In HfInZnO TFTs, a significant negative V TH shift was observed under NBIS conditions, which was attributed to the photoionization of the V O state to the meta-stable V 2+ O state [64, 65]. This PPC effect is also consistent with the slow recovery of the V TH value to the positive direction after NBIS application, which is on the time scale of days. Subsequently, the phototransition model of V O has been used to explain the negative V TH shift upon NBIS application for IGZO TFTs. The question has been raised whether the deep-to-shallow transition of V O in crystal ZnO can be extended to an amorphous multicomponent system. The electronic structure of various V O defects in amorphous IGZO was calculated using the first-principles density functional theory [66]. It was

Oxide Thin-Film Transistors for OLED Displays

163

found that most V O defects are deep states whereas some of the defects are shallow donors due to local coordination, which is different from the crystalline state. Because the V 2+ O shallow donors can capture free electrons, these defects are responsible for the positive V TH shift under PBS application. Under NBIS conditions, the deep V O state can be ionized, becoming a V 2+ O defect due to the negative-U behavior. The theoretical V O model is supported by experimental evidence where an intentional supply of oxygen species in an amorphous IGZO channel using high-pressure O2 annealing (10 atm) improved the NBIS stability of the resulting TFTs significantly compared to the device treated under air ambient at 1 atm [67]. The relative V O related peak area for the IGZO sample under high-pressure O2 annealing at 10 atm was diminished by 7% compared to that with air annealing at 1 atm. The terminology used for V O in a crystalline solid can be unclear in an amorphous solid due to the lack of a strict neighbor configuration. Nahm et al. pointed out that the under-coordinated In* is an intrinsic acceptor-like defect, which can be converted to a (In*–M)2− center by capturing two free electrons under PBS conditions [68]. Here, the In*-M configuration can be a more accurate description of V O in amorphous IGZO. The following equation explains the positive and negative V TH shifts by application of PBTS and NBTS, respectively (Fig. 11).    2− VO2+ + 2e− ↔ VO or I n ∗ − M + 2e− ↔ I n ∗ − M

(2)

Interstitial Oxygen Model In a highly disordered IGZO, the photo-induced hole carriers are suggested to facilitate the formation of an O22− peroxide state (PS) from the normal disordered state

Fig. 11 Energy change when structural relaxation occurs in under-coordinated indium. The normal state (NS) has two M–O bonds (blue square) and when In* exists, it makes In*–M and receives electrons as (In*–M)2− , becoming a meta-stable state (red circle). Reproduced from Nahm et al. [68]

164

H. J. Seul et al.

Fig. 12 Local atomic structure of amorphous InGaZnO4 in the a disordered (DS and DS*), b (2+) charge TS*, c peroxide (PS and PS*), and d neutral TS states. The small red balls are O atoms and the large ones are cations. e Calculated total energies as a function of the O–O distance in (0) and (2+) charge states, with respect to the PS energy. Adapted from [69]

(DS). A local atomic structure of amorphous InGaZnO4 is depicted in Fig. 12 and the local structure is denoted as a DS. When the excited hole carriers during NBIS are introduced into the supercell, the DS* in the (2+) hole injected case becomes metastable and is converted to O22− PS*. The relevant transformation can be represented by the following reaction: O 2− + O 2− + 2h+ ↔ O22−

(3)

The valence band tail (VBT ) is characterized by the anti-bonding pp* because the VB edge is mostly comprised of oxygen 2p anti-boding states. With the peroxide defect in the neutral states, the two electrons are donated to the CB, leading to the negative V TH shift of IGZO TFTs under the NBIS condition because the pp* antiboding state is empty. The recovery after NBIS or the PBTS-induced positive V TH shift can be explained by the reverse reaction where the O22− PS captures two free electrons and goes back to the DS state with an activation energy barrier (E A ) of ~0.9 eV, which is supported by the experimental observation. This model predicts that the reduction of localized VBT states can improve the gate-bias and photo-bias stability of the amorphous IGZO TFTs because the peroxide defect formation is mediated by the VBT holes. Indeed, the stability of IGZO TFTs is improved by appropriate post-deposition annealing or thermal annealing [69]. The peroxide model clearly explains the negative impact of a disordered oxygen network on NBIS instability. However, the effect of the non-stoichiometry and hydrogen incorporation in IGZO was not explicitly considered. Robertson et al. reported the interstitial oxygen model as a NBIS instability mechanism on the basis of excess oxygen and hydrogen incorporation [70]. They calculated the addition

Oxide Thin-Film Transistors for OLED Displays

165

Fig. 13 Defect formation energy for the interstitial oxygen in a-InGaZnO4 , crystalline InGaZnO4 , and ZnO in the O-rich limit (left). Schematic bonding diagrams for the a formation of the O–O bond at a neutral Oi site, b transfer of electrons from two hydrogens to O0i creating the O−2 i state, 0 . Adapted from [70] and c photo-excitation from the O−2 state in the CB edge, leaving a neutral O i i

effect of one oxygen and/or two hydrogens on the electron structure for the amorphous stoichiometric IGZO network with ~ 84 atoms. In the random network of IGZO with one neutral oxygen interstitial (O0i ), the extra oxygen bonds to a bulk oxygen ion with a form of a dimer peroxide unit. The corresponding localized O0i defect is located above the VB edge. When two additional hydrogens are introduced in this network, these hydrogens exist as O–H bonds, donating their electrons to the network due to the shallow donor nature. Interstitial oxygen captures two electrons and becomes an O2i − site separated from the other O2− sites, which is similar to a network O2− site. In moderately doped n-type a-IGZO, the formation energy of O2i − is very small or even negative, indicating that its formation is spontaneous. The relevant reaction can be described by the following equation: Oi0 + 2e− ↔ Oi2−

(4)

Light exposure during NIBS excites two electrons to the CB edge whereas the resulting holes localized on a single O site lead to the formation of an O–O bond (O0i ), which is the reverse of the reaction shown above. This modified interstitial oxygen model well explains the role of hydrogen and importance of the stoichiometric composition in the a-IGZO film. The theoretical Oi model is supported by the experimental result that the existence of excess oxygen near the channel/dielectric interface region in self-aligned coplanar a-IGZO TFTs negatively impacts the PBTS stability [71]. The V TH value for the a-IGZO TFTs after PBTS application is proportional to the peak Dit value for the given device, which is strongly correlated with the amount of excessive oxygen concentration with respect to the stoichiometric value. Therefore, it was concluded that a highly stable IGZO transistor can be fabricated

166

H. J. Seul et al.

by minimizing excess oxygen interstitial defects near the channel/dielectric region through an appropriate process and thermal annealing (Fig. 13). Hydrogen Bi-stability Model Hydrogen is a ubiquitous impurity in semiconductor devices. A dangling bond in a silicon semiconductor is a microscopic origin for the gap state trap. Hydrogen annealing (known as a forming gas annealing) is usually used to reduce the Dit or bulk-trap density (N T ) value because hydrogen incorporation effectively passivates the Si dangling bond. In crystal ZnO, hydrogen is identified as a shallow donor irrespective of the substitutional H (H O ) at the O site or interstitial H (H i ). However, the maximum N e value via hydrogen doping in ZnO is limited to ~1018 cm−3 even though the concentration of hydrogen in ZnO increases up to ~1020 cm−3 [72–74]. Bi-stability of H O in ZnO was suggested to explain this discrepancy between the experimental and theoretical results. According to DFT calculations, the normally positively charged H +O defect can capture electrons under the PBS condition (E F is closed to E C ), dissociating into V O and H +i through a large lattice relaxation where this Frenkel defect is referred to as a H-DX − center [75]. The shallow-to-deep transition indicates that the H i in the Frenkel defect acts as an acceptor rather than a donor, which explains the experimental results of the self-doping limit well. This hydrogen bi-stability model provides an alternative explanation for the PPC effect (NBIS instability) in ZnO (Fig. 14). The concept of hydrogen bi-stability was extended to account for the gate bias and photo-instability of the multicomponent a-IZTO system [76]. Two types of TFTs were fabricated with a-IZTO having different hydrogen contents. The device with a hydrogen-rich channel exhibited a high μFE of 48.0 cm2 /Vs, low SS of 0.14 V/decade, and an I ON/OFF ratio of 1010 whereas the μFE for the hydrogen-poor device was reduced to 20.5 cm2 /Vs. The higher μFE for the hydrogen-rich device results from its large N e value by virtue of the percolation conduction mechanism, which should be attributed to the donor behavior of incorporated hydrogen impurities in the aIZTO system. However, the hydrogen-rich devices suffer from inferior PBTS and NBTS stabilities, which were characterized by lower activation barrier energies (E A ) (0.10–0.41 eV) compared to those (0.31–0.91 eV) of the hydrogen-poor devices. The formation energies for the four possible hydrogen-related defects were calculated to investigate the responsible microscopic mechanism (Fig. 15). H +i and H +O are more − stable than H − i and H-DX in most regions of E F , as shown in Fig. 15c, d, which is consistent with the fact that the hydrogen impurity is a shallow donor in a-IZTO. + + – Conversely, H − i and H-DX were energetically comparable to H i and H O when E F is close to the CB edge. When PBTS is applied to a-IGZO TFTs, the rising E F near the CB edge favors the following reaction by capturing two free electrons: HO + 2e− ↔ (H − D X ) or Hi + 2e− ↔ Hi−

(5)

The E A values (1.05–2.35 eV) for the transition from H +i to H − i are higher than those (0.04–0.81 eV) for the transition from H +O to H-DX – due to the invoking of a strong O–H bond (Fig. 15g). The lower E A value for the hydrogen-rich device

Oxide Thin-Film Transistors for OLED Displays

167

Fig. 14 Electronic structure of H-DX − : Calculated a total and b partial densities of states for H-DX − in ZnO. c A schematic diagram for the electronic structures of H +O and H-DX − . The charge densities of d α and e α* states are shown with the iso-surface value of 3.50 × 10−5 Å−3 . Adapted from [75]

can be attributed to the activation of the transition from H +O to H-DX – due to its higher N e value, which was corroborated by the smaller experimental E A value. This model explains the following double-side effect of hydrogen doping well. Hydrogen incorporation in an amorphous oxide semiconductor enhances the carrier mobility at lower concentration ranges of [H] due to its shallow donor center role. However, above the critical concentration of [H], the shallow-to-deep transition (formation of H-DX − center) occurs through large lattice relaxation, which aggravates the gatebias and photo-bias stability of the resulting TFTs. The delicate controllability of hydrogen and V O is a key factor to achieve high-performance oxide TFTs with excellent stability.

168

H. J. Seul et al.

− Fig. 15 Partial charge densities of a H − i and b H-DX electronic states in a-IZTO. c The schematic + diagram of the energy distributions of the H levels. The formation energies of e H +i , H − i and f H O , HDX − with respect to the Fermi level assuming H-rich and O-rich conditions. The averaged thermal transition levels (ε(±)av ) for H i and H O are denoted by the vertical (green) lines. The solid and dashed lines indicate the average value and the deviation of the formation energies, respectively. An EF value of 0 corresponds to VBM. g Schematic configuration-energy diagrams for the transition + + − of the charge states between H − i (H-DX ) and H i (H O ). The solid red and blue lines denote the − + + − transitions from H i to H i and H-DX to H O , respectively, in the negative charge state. The dashed red and blue lines indicate the energy curves for the structural transitions of Hi and HO , respectively, when the electrons are excited to the conduction band in the positive charge state. Adapted from reference [76]

Oxide Thin-Film Transistors for OLED Displays

169

3 Summary and Outlook This chapter reviewed the current status of a-IGZO TFTs in the mass production of AMOLED TV from the viewpoint of architecture and the core unit process. The responsible mechanisms of PBTS and NBTS instabilities of a-IGZO TFTs were also addressed, which has been considered as a critical issue for their implementation in commercial AMOLED products. However, technical challenges should be overcome to exploit the advantage of oxide TFTs such as the extremely low leakage current and low-cost fabrication. The ever-increasing demand on backplane electronics in terms of high pixel density and low power consumption strongly requires the downscaling of transistors to less than 1 μm. A high pixel density (>2,000 ppi) for next-generation smartphones as well as AR and VR applications will be the next potential market for oxide TFTs. New routes such as ALD or CVD to deposit the oxide channel, highpermittivity dielectric, and electrode film on the non-planar structure as well as the pattering process should be developed in the near future. In addition, the fundamental device research on the hot carrier effect or short channel effect such as DIBL should be a major topic.

References 1. J.Y. Kwon, J.K. Jeong, Recent progress in high performance and reliable n-type transition metal oxide-based thin film transistors. Semicond. Sci. Technol. 30(2), 024002 (2015) 2. N. Morosawa, Y. Ohshima, M. Morooka, T. Arai, T. Sasaoka, Novel self-aligned top-gate oxide TFT for AMOLED displays. J. Soc. Inf. Disp. 20(1), 47–52 (2012) 3. T. Iwasaki, N. Itagaki, T. Den, H. Kumomi, K. Nomura, T. Kamiya, H. Hosono, Combinatorial approach to thin-film transistors using multicomponent semiconductor channels: An application to amorphous oxide semiconductors in In–Ga–Zn–O system. Appl. Phy. Lett. 90(24), 242114 (2007) 4. T. Kamiya, H. Hosono, Material characteristics and applications of transparent amorphous oxide semiconductors. NPG Asia Mater. 2(1), 15–22 (2010) 5. P. Barquinha, L. Pereira, G. Goncalves, R. Martins, E. Fortunato, Toward high-performance amorphous GIZO TFTs. J. Electrochem. Soc. 156(3), H161–H168 (2009) 6. J.H. Jeong, H.W. Yang, J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, J.W. Song, C.S. Hwang, Origin of subthreshold swing improvement in amorphous indium gallium zinc oxide transistors. Electrochem. Solid State Lett. 11(6), H157–H159 (2009) 7. S. Yasuno, T. Kita, A. Hino, S. Morita, K. Hayashi, T. Kugimiya, Physical properties of amorphous In–Ga–Zn–O films deposited at different sputtering pressures. Jpn. J. Appl. Phys. 52(3S), 03BA01 (2013) 8. B.S. Yang, S. Oh, Y.J. Kim, S.J. Han, H.W. Lee, H.J. Kim, Effect of sputter power on the photobias stability of zinc-tin-oxide field-effect transistors. J. Vac. Sci. Technol. B(32), 011202 (2014) 9. S. Junfei, D. Chengyuan, D. Wenjun, W. Jie, C. Yuting, Z. Runze, The influence of RF power on the electrical properties of sputtered amorphous In–Ga–Zn–O thin films and devices. J. Semicond. 34(8), 084003 (2013) 10. J.K. Jeong, J.H. Jeong, H.W. Yang, J.-S. Park, Y.-G. Mo, H.D. Kim, High performance thin film transistors with cosputtered amorphous indium gallium zinc oxide channel. Appl. Phys. Lett. 91(11), 113505 (2007)

170

H. J. Seul et al.

11. H.W. Lee, B.S. Yang, Y.J. Kim, A.Y. Hwang, S. Oh, J.H. Lee, J.K. Jeong, H.J. Kim, Comprehensive studies on the carrier transporting property and photo-bias instability of sputtered zinc tin oxide thin film transistors. IEEE Trans. Electron. Dev. 61(9), 3191–3198 (2014) 12. Y. Hanyu, K. Abe, K. Domen, K. Nomura, H. Hiramatsu, H. Kumomi, H. Hosono, T. Kamiya, Effects of high-temperature annealing on operation characteristics of a In–Ga–Zn–O TFTs. J. Display Technol. 10(11), 979–983 (2014) 13. K. Nomura, T. Kamiya, Y. Kikuchi, M. Hirano, H. Hosono, Comprehensive studies on the stabilities of A–In–Ga–Zn–O based thin film transistor by constant current stress. Thin Solid Films 518(11), 3012–3016 (2010) 14. K. Takenaka, M. Endo, G. Uchida, Y. Setsuhara, Fabrication of high-performance InGaZnOx thin film transistors based on control of oxidation using a low-temperature plasma. Appl. Phys. Lett. 112(15), 152103 (2018) 15. A. Abliz, Q. Gao, D. Wan, X. Liu, L. Xu, C. Liu, C. Jiang, X. Li, H. Chen, T. Guo, J. Li, L. Liao, Effects of nitrogen and hydrogen codoping on the electrical performance and reliability of InGaZnO thin-film transistors. ACS Appl. Mater. Interfaces. 9(12), 10798–10804 (2017) 16. P. Liu, T.P. Chen, Z. Liu, C.S. Tan, K.C. Leong, Effect of O2 plasma immersion on electrical properties and transistor performance of indium gallium zinc oxide thin films. Thin Solid Films 545, 533–536 (2013) 17. W.-G. Kim, Y.J. Tak, B.D. Ahn, T.S. Jung, K.-B. Chung, H.J. Kim, High-pressure gas activation for amorphous indium-gallium-zinc-oxide thin-film transistors at 100 °C. Sci. Rep. 6, 23039 (2016) 18. S. Yoon, Y.J. Tak, D.H. Yoon, U.H. Choi, J.S. Park, B.D. Ahn, H.J. Kim, Study of nitrogen high-pressure annealing on InGaZnO thin-film transistors. ACS Appl. Mater. Interfaces. 6(16), 13496 (2014) 19. J.S. Park, J.K. Jeong, Y.G. Mo, H.D. Kim, C.J. Kim, Control of threshold voltage in ZnO-based oxide thin film transistors. Appl. Phys. Lett. 93(3), 033513 (2008) 20. B. Zhang, H. Li, X. Zhang, Y. Luo, Q. Wang, A. Song, Performance regeneration of InGaZnO transistors with ultra-thin channels. Appl. Phys. Lett. 106(9), 093506 (2015) 21. T.H. Chiang, B.S. Yeh, J.F. Wager, Amorphous IGZO thin-film transistors with ultrathin channel layers. IEEE Trans. Electron. Dev. 62(11), 3692–3696 (2015) 22. L. Shao, K. Nomura, T. Kamiya, H. Hosono, Operation characteristics of thin-film transistors using very thin amorphous In–Ga–Zn–O channels. Electrochem. Solid-State Lett. 14(5), H197 (2011) 23. K. Nomura, T. Kamiya, H. Hosono, Stability and high-frequency operation of amorphous In–Ga–Zn–O thin-film transistors with various passivation layers. Thin Solid Films 520(10), 3778–3782 (2012) 24. H. Ohara, T. Sasaki, K. Noda, S. Ito, M. Sasaki, Y. Toyosumi, Y. Endo, S. Yoshitomi, J. Sakata, T. Serikawa, S. Yamazaki, 4.0 In. QVGA AMOLED display using In–Ga–ZnOxide TFTs with a novel passivation layer. SID Symp. Dig. Tech. Papers 40(1), 284–287 (2009) 25. K.H. Ji, J. Noh, P.S. Yun, J.U. Bae, K.S. Park, I. Kang, P-2: Novel high mobility oxide TFT with self-aligned s/d regions formed by wet-etch process. SID Symp. Dig. Tech. Papers 47(1), 1129–1131 (2016) 26. M. Nakata, M. Ochi, H. Tsuji, T. Takei, M. Miyakawa, Y. Fujisaki, H. Goto, T. Kugimiya, T. Yamamoto, Fabrication of a short-channel oxide TFT utilizing the resistance-reduction phenomenon in In–Ga–Sn–O. SID Symp. Dig. Tech. Papers 48(1), 1227–1230 (2017) 27. J.-Y. Noh, H. Kim, H.-H. Nahm, Y.-S. Kim, D.H. Kim, B.-D. Ahn, J.-H. Lim, G.H. Kim, J.-H. Lee, J. Song, Cation composition effects on electronic structures of In–Sn–Zn–O amorphous semiconductors. J. Appl. Phys. 113(18), 183706 (2013) 28. J.H. Song, K.S. Kim, Y.G. Mo, R. Choi, J.K. Jeong, Achieving high field-effect mobility exceeding 50 cm2 /Vs in In–Zn–Sn–O thin-film transistors. IEEE Electron. Dev. Lett. 35(8), 853–855 (2014) 29. W. Zhong, G. Li, L. Lan, B. Li, R. Chen, Effects of annealing temperature on properties of InSnZnO thin film transistors prepared by co-sputtering. RSC Adv. 8(61), 34817–34822 (2018)

Oxide Thin-Film Transistors for OLED Displays

171

30. H.A. Kim, J.O. Kim, J.S. Hur, K.S. Son, J.H. Lim, J. Cho, J.K. Jeong, Achieving high mobility in IGTO thin-film transistors at a low temperature via film densification. IEEE Trans. Electron. Dev. 65(11), 4854–4860 (2018) 31. J. Lee, D. Kim, S. Lee, J. Cho, H. Park, J. Jang, High field effect mobility, amorphous In–Ga– Sn–O thin-film transistor with no effect of negative bias illumination stress. IEEE Electron. Dev. Lett. 40(9), 1443–1446 (2019) 32. M. Nakata, M. Ochi, T. Takei, H. Tsuji, M. Miyakawa, G. Motomura, H. Goto, Y. Fujisaki, High-mobility back-channel-etched IGZTO-TFT and application to dual-gate structure. SID Symp. Dig. Tech. Papers 50(1), 1226–1229 (2019) 33. I.M. Choi, M.J. Kim, N. On, A. Song, K.B. Chung, H. Jeong, J.K. Park, J.K. Jeong, Achieving high mobility and excellent stability in amorphous In–Ga–Zn–Sn–O thin-film transistors. IEEE Trans. Electron. Dev. 67(3), 1014–1020 (2020) 34. T. Sun, L.Q. Shi, C.Y. Su, W.H. Li, X.W. Lv, H.J. Zhang, Y.H. Meng, W. Shi, S.M. Ge, C.Y. Tseng, Y.F. Wang, C.C. Lo, A. Lien, Amorphous Indium–Gallium–Zinc–Tin–Oxide TFTs with high mobility and reliability. SID Symp. Dig. Tech. Papers 46(1), 766–768 (2015) 35. W.H. Choi, J. Sheng, H.J. Jeong, J.S. Park, M. Kim, W. Jeon, Improved performance and stability of In–Sn–Zn–O thin film transistor by introducing a meso-crystalline ZrO2 high-k gate insulator. J. Vac. Sci. Technol. A 37(2), 020924 (2019) 36. J.S. Lee, S. Chang, S.M. Koo, S.Y. Lee, High-performance a-IGZO TFT with ZrO2 gate dielectric fabricated at room temperature. IEEE Electron. Dev. Lett. 31(3), 225–227 (2010) 37. S.Y. Na, Y.M. Kim, D.J. Yoon, S.M. Yoon, Improvement in negative bias illumination stress stability of In–Ga–Zn–O thin film transistors using HfO2 gate insulators by controlling atomiclayer-deposition conditions. J. Phys. D: Appl. Phys. 50(49), 495109 (2017) 38. J. Yang, J.K. Park, S. Kim, W. Choi, S. Lee, H. Kim, Atomic-layer-deposited ZnO thin-film transistors with various gate dielectrics. Phys. Status Solid. A 209(10), 2087–2090 (2012) 39. Y.J. Cho, J.H. Shin, S.M. Bobade, Y.B. Kim, D.K. Choi, Evaluation of Y2 O3 gate insulators for a-IGZO thin film transistors. Thin Solid Films 517(14), 4115–4118 (2009) 40. K. Song, W. Yang, Y. Jung, S. Jeong, J.A. Moon, Solution-processed yttrium oxide gate insulator for high-performance all-solution-processed fully transparent thin film transistors. J. Mater. Chem. 22(39), 21265–21271 (2012) 41. J.Q. Song, L.X. Qian, P.T. Lai, Effects of Ta incorporation in Y2 O3 gate dielectric of InGaZnO thin-film transistor. Appl. Phys. Lett. 109(16), 163504 (2016) 42. Y. Gao, X. Li, L. Chen, J. Shi, X.W. Sun, J. Zhang, High mobility solution-processed hafnium indium zinc oxide TFT with an Al-doped ZrO2 gate dielectric. IEEE Electron. Dev. Lett. 35(5), 554–556 (2014) 43. S.Y. Je, B.G. Son, H.G. Kim, M.Y. Park, L.M. Do, R. Choi, J.K. Jeong, Solution-processable LaZrOx /SiO2 gate dielectric at low temperature of 180 °C for high-performance metal oxide field-effect transistors. ACS Appl. Mater. Interfaces. 6(21), 18693–18703 (2014) 44. J.B. Kim, C. Fuentes-Hernandez, B. Kippelen, High-performance InGaZnO thin-film transistors with high-k amorphous Ba0.5 Sr0.5 TiO3 gate insulator. Appl. Phys. Lett. 93(24), 242111 (2008) 45. M. Esro, G. Vourlias, C. Somerton, W.I. Milne, G. Adamopoulos, High-mobility ZnO thin film transistors based on solution-processed hafnium oxide gate dielectrics. Adv. Funct. Mater. 25(1), 134–141 (2014) 46. G. Adamopoulos, S. Thomas, P.H. Wobkenberg, D.D.C. Bradley, M.A. McLachlan, T.D. Anthopoulos, High-mobility low-voltage ZnO and Li-doped ZnO transistors based on ZrO2 high-k dielectric grown by spray pyrolysis in ambient air. Adv. Mater. 23(16), 1894–1898 (2011) 47. A.S. Foster, F.L. Gejo, A.L. Shluger, R.N. Nieminen, Vacancy and interstitial defects in hafnia. Phys. Rev. B 65(17), 174117 (2002) 48. G.D. Wilk, R.M. Wallace, J. Anthony, High-κ gate dielectrics: current status and materials properties considerations. J. Appl. Phys. 89(10), 5243–5275 (2001) 49. X. Zou, G.J. Fang, L.Y. Yuan, X.S. Tong, X.Z. Zhao, A comparative study of amorphous InGaZnO thin-film transistors with HfOx Ny and HfO2 gate dielectrics. Semicond. Sci. Technol. 25(5), 055006 (2010)

172

H. J. Seul et al.

50. Z. Wu, Z. Yao, S. Liu, B. Yuan, Y. Zhang, Y. Liang, Z. Wang, X. Tang, G. Shao, Tuning the electrical performance of metal oxide thin-film transistors via dielectric interface trap passivation and graded channel modulation doping. J. Mater. Chem. C. 5(5), 1206–1215 (2017) 51. S.Y. Lee, S. Chang, J.S. Lee, Role of high-k gate insulators for oxide thin film transistors. Thin Solid Films 518(11), 3030–3032 (2010) 52. S. Chang, Y.W. Song, S. Lee, S.Y. Lee, B.K. Ju, Efficient suppression of charge trapping in ZnO-based transparent thin film transistors with novel Al2 O3 /HfO2 /Al2 O3 structure. Appl. Phys. Lett. 92(19), 192104 (2008) 53. J.M. Lee, I.T. Cho, J.H. Lee, H.I. Kwon, Bias-stress-induced stretched-exponential time dependence of threshold voltage shift in InGaZnO thin film transistors. Appl. Phys. Lett. 93(9), 093504 (2008) 54. A. Suresh, J.F. Muth, Bias stress stability of indium gallium zinc oxide channel based transparent thin film transistors. Appl. Phys. Lett. 92(3), 033502 (2008) 55. R.B.M. Crossa, M.M. De Souza, Investigating the stability of zinc oxide thin film transistors. Appl. Phys. Lett. 89(26), 263513 (2006) 56. Y.K. Moon, S. Lee, W.S. Kim, B.W. Kang, C.O. Jeong, D.H. Lee, J.W. Park, Improvement in the bias stability of amorphous indium gallium zinc oxide thin-film transistors using an O2 plasma-treated insulator. Appl. Phys. Lett. 95(1), 013507 (2009) 57. K.H. Ji, J.I. Kim, Y.G. Mo, J.H. Jeong, S. Yang, C.S. Hwang, S.H.K. Park, M.K. Ryu, S.Y. Lee, J.K. Jeong, Comparative study on light-induced bias stress instability of IGZO transistors with SiNX and SiO2 gate dielectrics. IEEE Electron. Dev. Lett. 31(12), 1404–1406 (2010) 58. J.K. Jeong, H.W. Yang, J.H. Jeong, Y.G. Mo, H.D. Kim, Origin of threshold voltage instability in indium-gallium-zinc oxide thin film transistors. Appl. Phys. Lett. 93(12), 123508 (2008) 59. J.K. Jeong, Photo-bias instability of metal oxide thin film transistors for advanced active matrix displays. J. Mater. Res. 28(16), 2071–2084 (2013) 60. S. Yang, D.H. Cho, M.K. Ryu, S.H.K. Park, C.S. Hwang, J. Jang, J.K. Jeong, Improvement in the photon-induced bias stability of Al–Sn–Zn–In–O thin film transistors by adopting AlOx passivation layer. Appl. Phys. Lett. 96(21), 213511 (2010) 61. H.W. Yeon, S.M. Lim, J.K. Jung, H. Yoo, Y.J. Lee, H.Y. Kang, Y.J. Park, M. Kim, Y.C. Joo, Structural-relaxation-driven electron doping of amorphous oxide semiconductors by increasing the concentration of oxygen vacancies in shallow-donor states. NPG Asia Mater. 8, e250 (2016) 62. A. Janotti, C.G.V. Walle, Oxygen vacancies in ZnO. Appl. Phys. Lett. 87(12), 122102 (2005) 63. T. Kamiya, K. Nomura, H. Hosono, Present status of amorphous In–Ga–Zn–O thin-film transistors. Sci. Technol. Adv. Mater. 11(4), 044305 (2010) 64. K. Ghaffarzadeh, A. Nathan, J. Robertson, S. Kim, S. Jeon, C. Kim, U.I. Chung, J.H. Lee, Instability in threshold voltage and subthreshold behavior in Hf–In–Zn–O thin film transistors induced by bias-and light-stress. Appl. Phys. Lett. 97(11), 113504 (2010) 65. K. Ghaffarzadeh, A. Nathan, J. Robertson, S. Kim, S. Jeon, C. Kim, U.I. Chung, J.H. Lee, Persistent photoconductivity in Hf–In–Zn–O thin film transistors. Appl. Phys. Lett. 97(14), 143510 (2010) 66. H.K. Noh, K.J. Chang, Electronic structure of oxygen-vacancy defects in amorphous In–Ga– Zn–O semiconductors. Phys. Rev. B 84(11), 115205 (2011) 67. K.H. Ji, J.I. Kim, H.Y. Jung, S.Y. Park, R. Choi, U.K. Kim, C.S. Hwang, D. Lee, H. Hwang, J.K. Jeong, Effect of high-pressure oxygen annealing on negative bias illumination stress-induced instability of InGaZnO thin film transistors. Appl. Phys. Lett. 98(10), 103509 (2011) 68. H.H. Nahm, Y.S. Kim, Undercoordinated indium as an intrinsic electron-trap center in amorphous InGaZnO4 . NPG Asia Mater. 6, e143 (2014) 69. H.H. Nahm, Y.S. Kim, D.H. Kim, Instability of amorphous oxide semiconductors via carriermediated structural transition between disorder and peroxide state. Phys. Status Solid. B 249(6), 1277–1281 (2012) 70. J. Robertson, Y. Guo, Light induced instability mechanism in amorphous InGaZn oxide semiconductors. Appl. Phys. Lett. 104(16), 162102 (2014) 71. S. Oh, J.H. Baeck, J.U. Bae, K.S. Park, I.B. Kang, Effect of interfacial excess oxygen on positive-bias temperature stress instability of self-aligned coplanar InGaZnO thin-film transistors. Appl. Phys. Lett. 108(14), 141604 (2016)

Oxide Thin-Film Transistors for OLED Displays

173

72. Y.M. Strzhemechny, H.L. Mosbacker, D.C. Look, D.C. Reynolds, C.W. Litton, N.Y. Garces, N.C. Giles, L.E. Halliburton, S. Niki, L.J. Brillson, Remote hydrogen plasma doping of single crystal ZnO. Appl. Phys. Lett. 84(14), 2545–2547 (2004) 73. N. Ohashi, Y.G. Wang, T. Ishigaki, Y. Wada, H. Taguchi, I. Sakaguchi, T. Ohgaki, Y. Adachi, H. Haneda, Lowered stimulated emission threshold of zinc oxide by hydrogen doping with pulsed argon–hydrogen plasma. J. Cryst. Growth 306(2), 316–320 (2007) ˇ 74. G. Brauer, W. Anwand, D. Grambole, J. Grenzer, W. Skorupa, J. Cížek, J. Kuriplach, I. Procházka, C.C. Ling, C.K. So, D. Schulz, D. Klimm, Identification of Zn-vacancy–hydrogen complexes in ZnO single crystals: a challenge to positron annihilation spectroscopy. Phys. Rev. B. 79(11), 115212 (2009) 75. H.H. Nahm, C.H. Park, Y.S. Ki, Bistability of hydrogen in ZnO: Origin of doping limit and persistent photoconductivity. Sci. Rep. 4, 4124 (2015) 76. Y. Kang, B.D. Ahn, J.H. Song, Y.G. Mo, H.H. Nahm, S. Han, J.K. Jeong, Hydrogen bistability as the origin of photo-bias-thermal instabilities in amorphous oxide semiconductors. Adv. Electron. Mater. 1(7), 1400006 (2015)

Pixel Circuits for OLED Displays Kee Chan Park

Abstract The structures, operating principles and technical issues of OLED pixel circuits are explained in this chapter. First, the necessity and role of the OLED pixel circuit is introduced. Then, three operating principles: diode-connection, sourcefollower and current programming schemes of in-pixel compensation techniques are explained. Other compensation techniques utilizing a complicated system outside the display panel are also introduced. Moreover, three examples of OLED pixel circuits employed in mobile display products are analysed in detail. Finally, the effect of the VT extraction time and dimming method on low-grey-level mura, is discussed.

1 The Need for OLED Pixel Circuit Unlike liquid crystal displays (LCD), organic light-emitting diode (OLED) displays require a circuit in every pixel. A single transistor as an on/off switch in each pixel is usually sufficient to operate an LCD, but at least two transistors are required for each pixel of an OLED display as shown in Fig. 1.

1.1 OLED Driving in a Display OLEDs exhibit non-uniform luminance-to-voltage (L-V) characteristics even on a single substrate due to the non-uniformity of the fabrication process. In addition, non-uniform degradation of L-V characteristics occurs during operation because the cumulative light-emitting time differs from pixel to pixel. However, the luminanceto-current (L-I) characteristics are more uniform and stable against the process nonuniformity and long-term operation. Therefore, it is difficult to express precise grey levels with an OLED by adjusting the voltage across it, as shown in Fig. 2a, or by modulating the duty ratio with a constant supply voltage. On the other hand, the K. C. Park (B) Konkuk University, Seoul, Korea e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_8

175

176

K. C. Park

Fig. 1 Comparison of the pixel structures of a an LCD and b an OLED display

Fig. 2 a Voltage driving of OLED. b Current driving of OLED

current driving scheme shown in Fig. 2b is more likely to be utilized for uniform and stable grey-level expression. The current source in the current driving scheme gives rise to additional power consumption for the same OLED current compared with the voltage driving scheme.

1.2 Role of the OLED Pixel Circuit The I-V characteristics of an OLED tend to degrade with operation time as illustrated in Fig. 3a. Therefore, the voltage across the OLED, VOLED increases to keep the current, IOLED constant as illustrated in Fig. 3b [1]. In this case, the voltage across the current driver, VDRV , in Fig. 2b, decreases when the supply voltage, ELVDD, is constant. This function—the maintenance of a constant IOLED against the inevitable VOLED change—is the first role of the OLED pixel circuit. If this is not fulfilled

Pixel Circuits for OLED Displays

177

Fig. 3 Degradation of OLED performance: a in constant-voltage driving and b in constant-current driving

properly, OLED degradation will result in the appearance of severe image sticking on the screen. The current driver is made of a thin-film transistor (TFT) in the saturation regime. Therefore, the drain-to-source voltage, VDS , of the TFT has little effect on the drain current, which is the same as the IOLED . The drain current of the TFT is predominantly determined by the gate-to-source voltage, VGS , which is stored in a capacitor, CST , and refreshed every frame as in the case of active-matrix LCD driving. However, the I-V characteristics of many TFTs exhibit large deviations so the current of the TFT is not uniform from pixel to pixel although precise data voltage is stored in CST . Usually, 256 different voltage levels are stored in CST and 256 corresponding current levels should be precisely generated by the TFT, as shown in Fig. 4. The voltage difference between adjacent grey levels is as small as several millivolts in some cases. However, for example, the deviation of the threshold voltage, VT , of the TFT is sometimes hundreds of millivolts, even on a single substrate. Therefore, without any special technique, the luminance of each pixel is not uniform. The second role of the OLED pixel circuit is to compensate for the non-uniform characteristics of the TFT and to generate uniform grey-level current according to the various data voltage levels. Figure 5 shows the display images of an OLED panel without and with the compensation function, respectively, in the pixel circuit [2]. Fig. 4 Analog voltage-to-current conversion in a pixel of an OLED display

178

K. C. Park

Fig. 5 Display images of OLED panel: a without compensation function, b with compensation function

The supply voltage ELVDD at a certain pixel drops compared with that at the input pad of the display panel because some current for OLED driving flows through the thin metal line of the ELVDD that has non-zero resistance. For a large-sized OLED display, this drop in supply voltage, the so-called IR drop, should not affect the IOLED value. This requirement should also be satisfied by the OLED pixel circuit.

1.3 TFT-to-OLED Connection Structure The OLED and the current driver TFT can be connected in two ways, as shown in Fig. 6. The first option is to connect the TFT to the anode of the OLED, as shown in Fig. 6a. This is called the common-cathode structure because the cathode is shared by all the pixels on a display panel and the anode is separated pixel by pixel. The second way is the common-anode structure where a TFT is connected to the cathode of the OLED and the anode is shared by all the pixels, as shown in Fig. 6b. All the commercially produced OLED displayed so far have had the commoncathode structure because it can be easily fabricated with high resolution. The reason for this is as follows. The TFT fabrication process normally precedes the OLED deposition process, since the water, oxygen, plasma, etc. utilized in the TFT process severely degrade the performance of the OLED. In the OLED process, an anode made of ITO is normally deposited before the organic light-emitting layer, which is apt to be damaged by the plasma during the anode deposition process. As a result, the anode is placed between the TFT and the OLED layers, so it is easy to connect the TFT to the anode rather than the cathode of the OLED. In addition, without the delicate OLED layers, the via hole connecting the TFT and the OLED anode can be

Pixel Circuits for OLED Displays

179

Fig. 6 Two ways of TFT and OLED connection: a common-cathode structure and b common-anode structure

formed using a photolithography process, resulting in a more efficient pixel layout than the common-anode structure, in which much larger connecting hole is required to connect the uppermost cathode and the TFT. Figure 7 shows examples of CST connection in the pixel circuit. When an nchannel TFT is used as the current driver, CST should be connected between the gate and source terminals of the current driver TFT, as shown in Fig. 7b, because only this structure preserves the VGS of the TFT against the variation in OLED characteristics. If CST is connected between the gate and a DC voltage, as shown in Fig. 7a, the VGS of the current driver TFT is affected by VOLED . So, this structure is vulnerable to severe image sticking. When a p-channel TFT is used as the current driver, CST can be connected between the gate of the TFT and any DC voltage, as shown in Fig. 7c. The variation of OLED I-V characteristics has little effect on IOLED because the OLED only affects the VDS of the TFT.

Fig. 7 a CST connection between the gate of an n-channel TFT and a DC voltage, b CST connection between the gate and source of an n-channel TFT, c CST connection between the gate of a p-channel TFT and a DC voltage

180

K. C. Park

2 Operating Principles of OLED Pixel Circuit There are many device parameters that determine the I-V characteristics of a TFT, e.g. threshold voltage VT , field-effect mobility μFE and subthreshold swing S.S. Among these parameters, VT is the most important because the current driver TFT operates with the VGS value around VT , where the drain current is exponentially dependent upon VGS . So, the operation of an OLED pixel circuit is primarily focused on compensating for the VT non-uniformity of the current driver TFT. However, VT is not everything. TFTs may have different μFE and S.S. values even if they have the same VT . We need to compensate for the non-uniformity at every current level over the whole grey scale in use to completely remove the mura, which is practically impossible. In fact, the pixel circuit designed to compensate for the non-uniform VT also compensates for the non-uniformity of μFE and S.S. Precisely speaking, such a circuit reduces the current deviation for the current level reached at the end of the compensation period. This is discussed in detail in Sect. 4.

2.1 In-Pixel Compensation The current of the driver TFTs over the entire panel becomes uniform if the VT of the driver TFT itself in each pixel is stored in CST , in addition to the data voltage VDATA , as shown in Fig. 8. For this purpose, we need to extract the VT of the current driver TFT in every pixel. Thus far, two methods have been developed to accomplish this. One method is to reduce the VGS of the TFT by its own drain current until it turns off after initially applying a sufficiently large VGS to turn on the TFT. In this method, here, VT is extracted at the instant of TFT turn-off and stored in CST . The other method is to make the current with a specific value flow through the driver TFT. The VGS of the driver TFT changes to make the drain current identical to the specific current value and is then stored in CST . The first method is called the voltage

Fig. 8 Bias state of the current driver TFT during the light-emission period of OLED as a result of VT compensation: a n-channel TFT case, b p-channel TFT case

Pixel Circuits for OLED Displays

181

programming scheme, since a data voltage is delivered to each pixel to express a certain grey level. In the second method, however, a data current is supplied to each pixel through the data line to set up an appropriate IOLED in the driver TFT. This method is called the current programming scheme. Both methods can be carried out within each pixel circuit or by a system outside the display panel. In this subsection, the mechanisms for how it is performed in the pixel circuit are explained. The first method mentioned above can be implemented in two ways. One is to change the gate voltage with the source voltage fixed and the other is to change the source voltage with the gate voltage fixed. In the former case, the drain and gate nodes of the TFT should be connected to each other—the so-called diode connection—for the drain current to charge or discharge the gate node. In the latter case, the source node is charged or discharged by the drain current and approaches the gate voltage until the TFT is turned off. Therefore, this is called the source-follower scheme.

2.1.1

Diode-Connection Scheme

The process of VT compensation using the diode connection is shown in Fig. 9. Two switch TFTs, S1 and S2, are open to preserve the charge in CST during the light-emission period, as shown in Fig. 9a. A large VGS , such as −7 V in Fig. 9b, is initially applied to the driver TFT to turn it on. This moment is called the reset or initialization period. The current of the driver TFT should not flow through the OLED during this period as such current flow precludes a high-quality expression of black. Then, S1 opens and S2 closes—diode connection—for the drain current to charge the gate node. The gate node is charged until the driver TFT is turned off—strictly speaking, it goes into the subthreshold regime rather than turn-off. At this instant, the drain current becomes negligibly small and the gate node voltage is almost saturated. The VGS at this instant is simply regarded as the VT of the driver TFT. In Fig. 9c, for example, the gate node voltage rises up to 9 V since the VT is −

Fig. 9 The process of VT compensation using the diode connection of driver TFT: a light-emission period, b reset or initialization period, c VT extraction or compensation period

182

K. C. Park

Fig. 10 The process of VT compensation using the source-follower configuration: a light-emission period, b reset or initialization period, c VT extraction or compensation period

1 V and ELVDD is 10 V. Finally, VDATA is added to the extracted VT in CST , which could happen in a variety of ways. Some examples are introduced in Sect. 3.

2.1.2

Source-Follower Scheme

The process of VT compensation using the source-follower configuration is shown in Fig. 10. Two switch TFTs, S1 and S2, are open to preserve the charge in CST during the light-emission period as shown in Fig. 10a. A large VGS , such as 5 V in Fig. 10b is applied to the driver TFT to turn it on during the initialization period. The current of the driver TFT should not flow through the OLED during this period as such current flow precludes a high-quality expression of black. Then, S2 opens with S1 still closed for the drain current to charge the source node. The source node voltage changes as it approaches the gate node voltage—source follower—until the driver TFT goes into the subthreshold regime. At this instant, the drain current becomes negligibly small and the source node voltage is almost saturated. The VGS at this instant is simply regarded as the VT of the driver TFT. In Fig. 10c, for example, the source node voltage rises up to 4 V since VT is 1 V and the gate node voltage is 5 V. Finally, VDATA is added to the extracted VT in CST , which could happen in a variety of ways. Some examples are introduced in Sect. 3.

2.1.3

Current Programming Scheme

The process of current programming is shown in Fig. 11. During the light-emission period, the switch TFT S1 is open to preserve the charge in CST , as shown in Fig. 11a. S2 is also open while S3 is closed to make the drain current of the driver TFT the same as IOLED . Before the next light-emission period, the data voltage in CST is refreshed, as shown in Fig. 11b. S1 and S2 are closed while S3 is open. The drain current of the

Pixel Circuits for OLED Displays

183

Fig. 11 Operation of current programming scheme: a light-emission period, b programming period

driver TFT, IPXL , is determined by ELVDD–VDL alone because the driver TFT is in the saturation regime by diode connection, where VDL is the voltage of the data line. The driver IC connected to the data line draws a data current, IDATA . The difference between IPXL and IDATA is ICDL , which charges or discharges the capacitance of the data line CDL . When IPXL is larger than IDATA , CDL is charged and VDL rises to decrease IPXL . When, on the contrary, IDATA is larger than IPXL , CDL is discharged and VDL falls to increase IPXL . As a result, IPXL eventually becomes the same as IDATA . Then, S1 opens again and the voltage set up in CST —VGS .DATA is preserved. S3 is closed again as shown in Fig. 11a, and IOLED becomes the same as IDATA if the driver TFT operates in the saturation regime. The current programming scheme, in principle, brings the best compensation results over a wide grey-scale range. IOLED is exactly same as IDATA regardless of the non-uniformity of VT , μFE and S.S. if the drain current of the driver TFT is constant in the saturation regime. However, it is not used in the commercial product at all because it takes too long time for IPXL to match IDATA . It sometimes takes hundreds or thousands of times longer programming time than what is available in practical display driving. In addition, a driver IC which generates hundreds of uniform analog current outputs is very difficult to fabricate.

2.2 Ext-Panel Compensation The characteristics of the current driver TFT and the OLED can be evaluated by a system outside the OLED display panel. For this, analog-to-digital converters (ADC) should be incorporated in the driver IC. The device parameters of the TFT and OLED

184

K. C. Park

Fig. 12 Schematic diagram of an OLED display module for ext-panel compensation by measuring the voltage of the sense line

in every pixel are converted into digital data by the ADCs and saved in an external nonvolatile memory, as shown in Fig. 12. When the data voltage in each pixel is refreshed, the input grey-level data for each pixel are modified based on the device characteristics data stored in the memory to display an image without mura and then transferred to the digital-to-analog converters (DAC). The ext-panel compensation scheme is usually employed in large display products such as TVs due to its bulky form factor and the high cost of additional parts.

2.2.1

Voltage Sensing Scheme

The simplest pixel circuit for ext-panel compensation is composed of three n-channel TFTs and a capacitor, as shown in Fig. 12 [3]. This circuit is used in OLED TV products. Figure 13 shows the timing diagrams for the (a) VT extraction and (b) μFE evaluation of TDR . For VT extraction, TDR is initially turned on by a large VGS . Then it becomes turned off as the sense line voltage, VSL , rises by its own drain current while the data line voltage, VDL , is kept constant with the scan switch TFT, TSC turned on. VSL is measured by ADC when it is almost saturated. The difference between VDL and VSL at this instant is regarded as the VT of TDR . Special care should be taken for VSL not to exceed ELVSS during this process, otherwise TDR never turns off and some current will flow through TDR and the OLED. As a result, we cannot obtain the correct VT value. In addition, the pixels under evaluation do not display an appropriate image during this process and it takes a long time because the capacitance of the sense line, CSL , is too large with respect to the drain current of TDR approaching the threshold state. For this reason, this process cannot be carried out while the display is in use, but it is usually performed when the display becomes powered off.

Pixel Circuits for OLED Displays

185

Fig. 13 Timing diagrams of an ext-panel compensation system employing a voltage sensing scheme for the a VT extraction and b μFE evaluation of TDR

For μFE evaluation, TDR is initially turned on by a moderate VGS , which is determined by the previously extracted VT value. Then the TSC in Fig. 12 is turned off. As a result, the VGS of TDR is kept constant by CST and a constant current flows through TDR . CSL begins to be charged by the constant current when the switch, SPRE , opens. After a designated time, the sense switch TFT, TSS turns off and VSL is measured by ADC. The μFE of TDR for a specific VGS -VT bias is calculated from the measured VSL . During this process, VSL should not exceed ELVSS for correct evaluation results. The pixels under evaluation do not display an appropriate image either but this process does not take long because the constant drain current of TDR is rather large. So, this process can be carried out while the display is in use and it is usually performed during a blank moment between each frame. The degradation of L-I characteristics of OLED results in burn-in on a display, even if TDR supplies the same current. The I-V characteristics of OLED are used to estimate the degradation of L-I characteristics and a larger current is supplied to the degraded OLED to hide the burn-in. For this purpose, the I-V characteristics of OLED are evaluated as follows. First, both TDR and OLED are turned on by applying an appropriate VG —corrected value based on the characteristics of TDR . Then, VS is determined depending on the I-V characteristics of the OLED. Next, TSC is turned off to keep the VGS of TDR constant and VSL is lowered sufficiently by VPRE to prevent the OLED from turning on hereafter. Finally, SPRE opens and VSL begins to rise. The speed of the VSL rise is determined by the VGS of TDR stored in CST , which is dependent on the I-V characteristics of the OLED.

186

K. C. Park

Fig. 14 Schematic diagram of an OLED display module for ext-panel compensation by measuring the current through the sense line with a current integrator

2.2.2

Current Sensing Scheme

The evaluation time for the device parameters in each pixel increases as the resolution and size of an OLED display panel increase because the CSL increases. However, it can be reduced and becomes independent of display resolution and size if VSL does not change during the evaluation process. This is implemented by employing a current integrator, as shown in Fig. 14 [4]. VSL is kept the same as VPRE due to the negative feedback configuration of the Op-Amp via CFB . The current integrator composed of an Op-Amp and a capacitor CFB outputs a voltage that is proportional to the current integration in CFB . The evaluation time of this current sensing scheme is much shorter than that of the voltage sensing scheme because a small CFB is charged more quickly than the large CSL by the same TDR current.

3 Examples of OLED Pixel Circuit In this section, three OLED pixel circuits used in the mobile display products are introduced. All of them adopt an in-pixel compensation scheme and the lowtemperature polycrystalline-silicon (LTPS) TFT as a current driver. The switch devices may be either LTPS TFTs or metal oxide (MO), e.g. IGZO TFTs. Compared with the IGZO TFT, the LTPS TFT has better stability and relatively larger S.S., which is beneficial for decreasing the mura because the driver TFTs in the mobile OLED displays operate mostly in the subthreshold regime.

Pixel Circuits for OLED Displays

187

3.1 7T-1C-6L PMOS Diode-Connection Pixel Circuit The OLED pixel circuit in Fig. 15 is composed of seven p-channel TFTs, one capacitor and six lines—ELVDD, Em[n], Scan[n-1], Scan[n], Data and Init (7T-1C-6L). It was originally developed as a 6T-1C circuit [5]. The switch TFT T7 was added later for better black expression. It has nothing to do with the compensation effect. T1 is the current driver and T2–T7 are simple on/off switches. The VGS of T1 is stored in CST and determines the IOLED according to the grey-level data. Independent of the driver TFT T1, T5 and T6 can be used to block IOLED under the control of the Em[n] signal. So, the Em[n] signal is used for the dimming control of the display panel. The operation of data refresh with VT compensation of T1 is carried out as follows. Initialization period 1 comes first with all the switch TFTs turned off, except T4, as shown in Fig. 16a. So, the gate voltage of the driver T1 is pulled down to VINIT and T1 is initially turned on. Then, T4 is turned off. Next, a data voltage VDAT is applied to the source of T1 and VT extraction is performed as T2, T3 and T7 are turned on at the beginning of period 2, as shown in Fig. 16b. During period 2, the gate node voltage of T1 changes from VINIT to VDAT + VTP , where VTP is the VT of T1. At the same time, the anode of the OLED is discharged through T7 to prevent charge accumulation in the intrinsic OLED capacitance, COLED , by a leakage current through T1. Otherwise, some light will be emitted even when black expression is wanted. Finally, the gate voltage of T1 is preserved by CST and the source of T1 is connected to ELVDD, as shown in Fig. 16c, during light-emission period 0. The VT extraction time of this circuit is limited to one horizontal line selection time—1H time—since the nth data voltage should be maintained throughout VT extraction period 2.

Fig. 15 a Circuit diagram and b timing diagram of the 7T-1C-6L PMOS diode-connection pixel circuit. 0: light-emission period, 1: initialization period, 2: data writing and VT extraction period

188

K. C. Park

Fig. 16 Operation of the 7T-1C-6L PMOS diode-connection pixel circuit: a initialization, b data writing and VT extraction, c light emission

3.2 4T-2C-6L NMOS Source-Follower Pixel Circuit The OLED pixel circuit in Fig. 17 is composed of four n-channel TFTs, two capacitors and six lines—ELVDD, Em[n], Scan1[n], Scan2[n], Data and Init (4 T-2C-6L). T1 is the current driver and T2–T4 are simple on/off switches. The VGS of T1 is stored in C1 and determines IOLED according to the grey-level data. Independent of the driver TFT T1, T4 can be used to block IOLED under the control of the Em[n] signal. So, the Em[n] signal is used for the dimming control of the display panel. The operation of VT compensation and data refresh is carried out as follows. Initialization period 1 comes first with only switch TFTs T2 and T3 turned, as shown in Fig. 18a. A predefined voltage VREF is applied through the data line during period 1 and 2. So, the VGS of the driver T1 is set as VREF –VINIT and T1 is initially turned on during period 1. Next, at the beginning of period 2, T2 is turned off and T4 is

Fig. 17 a Circuit diagram and b timing diagram of the 4 T-2C-6L NMOS source-follower pixel circuit. 0: light-emission period, 1: initialization period, 2: VT extraction period, 3: data writing period

Pixel Circuits for OLED Displays

189

Fig. 18 Operation of the 4 T-2C-6L NMOS source-follower pixel circuit: a initialization, b VT extraction, c data writing, d light emission

turned on to extract the VT of T1, as shown in Fig. 18b. At the end of period 2, the gate voltage of T1 is VREF and the source voltage of T1 becomes VREF – VTN , where VTN is the VT of T1. Then during period 3, T4 is turned off again and a data voltage VDAT is applied to the gate of T1. As a result, the source node voltage of T1 changes as much as (VDAT – VREF ) C1/(C1 + C2) from VREF – VTN , as shown in Fig. 18c. Finally, T3 is turned off and the VGS of T1 is maintained as (VDAT – VREF ) C2/(C1 + C2) + VTN by C1 with the drain of T1 connected to ELVDD, as shown in Fig. 18d during light-emission period 0. Special care should be taken for the source voltage of T1 not to exceed ELVSS during VT extraction period 2, otherwise T1 will never be turned off and the VT of T1 cannot be extracted. The anode of the OLED is discharged during the data refresh process, which is necessary for high-quality expression of black. The VT extraction time of this circuit is limited to less than 1H time, since neither (n − 1)th nor (n + 1)th data voltage is allowed in the data line during periods 1–3.

190

K. C. Park

Fig. 19 a Circuit diagram and b timing diagram of the 6T-1C-7L LTPO diode-connection pixel circuit. 0: light-emission period, 1: initialization period, 2: data writing and VT extraction period

3.3 6T-1C-7L LTPO Diode-Connection Pixel Circuit The OLED pixel circuit in Fig. 19 is composed of six n-channel TFTs, one capacitor and seven lines—ELVDD, Em1[n], Em2[n], Scan1[n], Scan2[n], Data and Init (6T1C-7L) [6, 7]. T1 is the current driver and T2–T6 are simple on/off switches. The VGS of T1 is stored in CST and determines IOLED according to the grey-level data. Independent of the driver TFT T1, T4 or T5 can be used to block IOLED . So, they can be used for the dimming control of the display panel. The voltage across CST is kept constant if T3 is completely turned off during lightemission period 0 in Fig. 20c. So, an IGZO TFT with extremely low off-state current, e.g. less than 10 fA, is used for T3 and the display is free from flicker problem even for very low frame rates such as 1 Hz. The combination of LTPS and MO TFTs, so-called LTPO, can be used to reduce the power consumption arising from data refresh by lowering the display frame rate [8]. The operation of data refresh with VT compensation of T1 is carried out as follows. Initialization period 1 comes first as T5 is turned off but T3 and T6 are turned on, as shown in Fig. 20a. So, the gate voltage of the driver T1 is pulled up to ELVDD and T1 is initially turned on. At the same time, the anode of the OLED is discharged through T6 to discharge COLED for a high-quality expression of black. Then, T4 is turned off. Next, as T2 is turned on at the beginning of period 2, a data voltage, VDAT , is applied to the source of T1 and VT extraction is performed, as shown in Fig. 20b. During period 2, the gate node voltage of T1 changes from ELVDD to VDAT + VTN , where VTN is the VT of T1. At the end of period 2, T3 and T6 are turned off and VDAT + VTN – VINIT is stored in CST as the VGS of T1. Finally, T4 and T5 are turned

Pixel Circuits for OLED Displays

191

Fig. 20 Operation of the 6T-1C-7L LTPO diode-connection pixel circuit: a initialization, b data writing and VT extraction, c light emission

on and a uniform IOLED flows through T4, T1 and T5, as shown in Fig. 20c during light-emission period 0. The VT extraction time of this circuit is limited to less than 1H time, since the nth data voltage should be maintained throughout VT extraction period 2.

4 Limitations of In-Pixel Compensation Circuits The pixel circuits of an OLED display do not eliminate mura, but only reduce it. For example, the LTPS TFT employed as the current driver in the mobile OLED display exhibits especially high short-range non-uniformity in the I-V characteristics, such as VT , μFE and S.S. Therefore, the VT compensation alone is not enough and some mura appears, even though the complicated pixel circuit is used.

4.1 Current Uniformity Versus VT Extraction Time The current driver TFT in the mobile OLED display operates mostly in the subthreshold regime since the current required in each pixel is below tens of nanoamperes. In principle, the pixel circuits introduced in the previous section do not compensate for the current non-uniformity of the driver TFT in the subthreshold regime if the S.S. is different among the TFTs. In addition, the extracted VT value is in reality not a single value, but may change depending on the extraction time, as the subthreshold current flows continuously yet decreasingly during the VT extraction period. Figure 21a shows how the current at the end of the VT extraction period decreases with the extraction time, which means that the VGS of the driver TFT does not saturate and that the extracted VT is determined by the extraction time [9].

192

K. C. Park

Figure 21 is the SPICE simulation results of ID —the current of the driver TFT— in the 7T-1C-6L PMOS diode-connection pixel circuit of Fig. 15 versus the VT extraction time, assuming 27 combinations of device parameter variations—VT = ±0.2 V, μFE = ±20%, S.S. =±10%. Figure 21a shows the ID values at the end of the VT extraction period but Figs. 21b–d show the relative errors of ID values for three current levels during the light-emission period. It should be noted that the relative standard deviation in Fig. 21a is very uniform—6–8%—though the absolute standard deviation σ certainly decreases as the VT extraction time increases, meaning that the compensation effect is improved with longer VT extraction time. The relative standard deviation of IOLED decreases monotonically along with the VT extraction time when the target IOLED is small, as shown in Fig. 21b. However, it does not only

Fig. 21 Current deviations of the 7T-1C-6L PMOS diode-connection pixel circuit versus VT extraction time for various TFT characteristics (VT = ±0.2 V, μFE = ±20%, S.S. = ± 10%). a ID values of driving TFTs at the end of the VT extraction period. IOLED values during the light-emission period for the target current of a 0.1, b 1 and c 10 nA

Pixel Circuits for OLED Displays

193

decrease but also increases along with the extraction time when the target IOLED is large, as shown in Fig. 21c, d. The reason why the relative error of IOLED decreases with the extraction time is illustrated in Fig. 22. The IVT s—the ID s at the end of the VT extraction time—of TFTs, T1 and T2, becomes similar, as shown in Fig. 21a. However, the IVT is rather large for 1 μs or 5 μs—33.9 nA and 4.9 nA, respectively, in Fig. 21a—compared with the low grey-level IOLED for mobile OLED displays—less than 1 nA. So, the VGS for low grey-level is more positive than VT , as shown in Fig. 22a. Therefore, ID s at the low grey-level exhibits a large difference if the S.S.s of the TFTs are different. However, they become similar as the IVT s approaches the low grey-level current value with increasing VT extraction time, as shown in Fig. 22b. On the other hand, the difference between the ID s for the high grey-level may increase as the extraction

Fig. 22 Effects of different VT extraction time on the current deviation after compensating for high and low grey levels. a short VT extraction time, b long VT extraction time

194

K. C. Park

time increases too much, because the VGS for the high grey-level once again differs from the extracted VT value. Figures 21c, d show that an optimum VT extraction time exists to minimize the deviation of IOLED and that it is different according to the target IOLED value. The optimum time is such that it makes IVT several times smaller than the target IOLED value. This is because the absolute deviation among IVT s decreases as the VT extraction time increases, though the extracted VT moves away from the VGS of the driver TFT during the light-emission period.

4.2 Low Grey-Level Mura Versus Dimming Method The grey scale of the OLED display is expressed by modulating IOLED independently in each pixel. However, dimming of the panel can be controlled in two ways. The first option is to modulate the IOLED in every pixel by a common rate over the entire panel. This method is referred to as pulse amplitude modulation (PAM). The other method is to modulate the light-emission time while keeping the IOLED constant according to the grey-level data. This method is referred to as pulse width modulation (PWM). Figure 23 shows the high-speed photographs of the OLED displays driven by PAM or PWM dimming scheme. The photographs are taken with a shutter speed of 0.5 ms. Every pixel over the entire screen displays a certain grey level in PAM dimming, regardless of changes in brightness, as shown in Fig. 23a–c. On the contrary, four black horizontal bands are shown in PWM dimming and the width of the black bands decreases as brightness increases, as shown in Fig. 23d–f. These bands are not detected by the human eye because they move vertically very quickly, i.e. at the same speed as data refresh. The luminance of light-emitting regions is constant regardless of changes in brightness. PWM dimming is implemented by turning on and off the switch TFT in the path of IOLED from ELVDD to the OLED in each pixel, rather than the current driver TFT. PAM has an advantage over PWM in terms of OLED lifetime because the IOLED range is shifted to lower levels for PAM dimming, whereas a constant IOLED range is maintained for PWM dimming [10]. However, mura originating from non-uniform TFT characteristics is severer for PAM dimming than PWM dimming, as shown in Fig. 24. This is because the IOLED range used for PAM at 10% brightness is ten times lower than for PWM at 10% brightness. The pixel circuit of the 7T-1C-6L PMOS diode-connection pixel circuit in Fig. 15 is used for both OLED displays and the VT extraction time is approximately 5 μs for both. Based on the results in Fig. 21, it is expected that a much longer VT extraction time is required to compensate for the non-uniform characteristics of the driver TFT for such a low current level. The mura of the PAM dimming OLED display is reduced as the brightness increases to 50%, as shown in Fig. 24b. Accordingly, most OLED displays for mobile devices employ PWM dimming to suppress mura. Recently, the vertical resolution of mobile OLED display is increasing to over 3 k and the frame rate is also on the rise from 60 to 120 Hz. These trends lead to shorter

Pixel Circuits for OLED Displays

195

Fig. 23 High-speed photographs of OLED displays driven by the PAM or PWM dimming scheme. Shutter speed is 0.5 ms. a PAM 10% brightness, b PAM 50% brightness, c PAM 90% brightness, d PWM 10% brightness, e PWM 50% brightness, f PWM 90% brightness

VT extraction time and higher power consumption. Adaptive frame rate driving, i.e. high frame rate for moving picture and low frame rate for still image, may be a good solution for suppressing the power consumption increase. However, a serious flicker problem occurs if the conventional LTPS backplane is used for the low frame rate display, e.g. below 10 Hz due to high off-state current. LTPO backplane can be used as a solution to the flicker problem of low frame rate driving owing to extremely low off-state current of the MO TFT. Accordingly, LTPO backplane with adaptive frame rate is expected to be widely used for the mobile OLED display [11]. The other

196

K. C. Park

Fig. 24 Normal-speed photographs of OLED displays with a 7T-1C-6L PMOS diode-connection pixel circuit driven by a PAM or PWM dimming scheme. Shutter speed is 100 ms. Full-screen 65th grey level is displayed. a PAM at 10% brightness, b PAM at 50% brightness, c PWM at 10% brightness, d PWM at 50% brightness

Pixel Circuits for OLED Displays

197

problem, shorter VT extraction time along with the improvement of OLED efficiency, makes the mura problem more and more serious because the difference between the current levels at the instant of VT extraction and during the light emission increases. New pixel circuits with the VT extraction time not limited to 1H time and extendable to hundreds of microseconds need to be developed [12]. The structure of the new OLED pixel circuit also needs to be as simple as the conventional circuit in order to be integrated within a small pixel area.

References 1. H. Kim, S. Kim, S.W. Chang, D. Lee, D.S. Jeong, H.K. Chung, Y. Hong, Frequency dependence of OLED voltage shift degradation, in Society for Information Display International Symposium, vol. 38 (Long Beach, 2007), pp. 1108–1111 2. K.C. Park, J.H. Jeon, Y.I. Kim, J.B. Choi, Y.J. Chang, Z.F. Zhan, C.W. Kim, A poly-Si AMOLED display with high uniformity. Solid-State Elec 52, 1691–1693 (2008) 3. I.H. Han, S.H. Choi, B.S. Kim, B.W. Kang, G.H. Oh, H.K. Shin, Organic light emitting diode display device. Korea Patent Appl. No. 10-2013-0169323 (2013) 4. O.S. Do, K.D. Woo, H.J. Kim, J.Y. Bae, Electroluminescent display device and driving method of the same. Korea Patent Appl No. 1020170095415 (2017) 5. S.M. Choi, O.K. Kwon, N. Komiya, H.K. Chung, A self-compensated voltage programming pixel structure for active-matrix organic light emitting diodes, in 10th International Display Workshop (Fukuoka, 2003), pp. 535–538 6. T.T. Tsai, V. Gupta, C.W. Lin, S.C. Chang, Organic light-emitting diode display with reduced capacitive sensitivity. US Patent Appl. No. 14/469,513 (2014) 7. T.K. Chang, C.W. Lin, S. Chang, LTPO TFT technology for AMOLEDs, in Society for Information Display International Symposium, vol. 50 (San Jose, 2019), 545–548 8. H. Watakabe, T. Jinnai, I. Suzumura, A. Hanada, R. Onodera et al., Development of advanced LTPS TFT technology for low power consumption and narrow border LCDs, in Society for Information Display International Symposium, vol. 5 (San Jose, 2019), pp. 541–544 9. J.H. Lee, D.H. Kim, J.W. Yang, K.C. Moon, S.Y. Lee et al., Correlation between the compensation time and the current deviation of OLED pixel circuit. J. Soc. Inf. Display 28, 882–891 (2020) 10. J.P. Spindler, T.K. Hatwar, M.E. Miller, A.D. Arnold, M.J. Murdoch et al., System considerations for RGBW OLED displays. J. Soc. Inf. Disp. 14, 37–48 (2006) 11. R. Yonebayashi, K. Tanaka, K. Okada, K. Yamamoto, K. Yamamoto et al., High refresh rate and low power consumption AMOLED panel using top-gate n-oxide and p-LTPS TFTs. J. Soc. Inf. Display 28, 350–359 (2020) 12. N.H. Keum, S.K. Hong, O.K. Kwon, An AMOLED pixel circuit with a compensating scheme for variations in subthreshold slope and threshold voltage of driving TFTs. J. Solid-State Circ. 55, 3087–3096

Large-Size OLED TVs with White OLED Chang Wook Han, Yoon Deok Han, Hyun Chul Choi, and In Byeong Kang

Abstract Currently, commercially available large-sized OLED displays use white OLED devices. This chapter intends to cover the core technologies of tandem white OLED devices and their applications through white OLED devices have been developed in order to improve performance. The first half explains two-stack and threestack tandem WOLED structures, which were developed to improve the efficiency of bottom-emitting WOLEDs, and the configuration of a light-emitting layer to increase color reproduction. In addition, it introduces features of products with this technology such as wallpaper OLED, cinematic sound OLED, rollable OLED, and 8K OLED displays. The second half deals with issues such as the viewing angles and power consumption of microcavity top-emitting WOLEDs and introduces the sputtered transparent electrode technology and OLED device characteristics that make up the non-microcavity top-emitting WOLED devices. Additionally, we introduce transparent and flexible transparent OLED displays.

1 Introduction Organic light-emitting diodes (OLEDs) are rising as the next generation display device, surpassing liquid crystal displays (LCDs) in image quality. Although quantum dot film and LED backlights have been applied in order to improve the image quality of LCDs, it still cannot overcome the superior image quality of OLEDs. OLEDs have

C. W. Han (B) · Y. D. Han · H. C. Choi · I. B. Kang LG Display, LG Science Park, Seoul, Korea e-mail: [email protected] Y. D. Han e-mail: [email protected] H. C. Choi e-mail: [email protected] I. B. Kang e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_9

199

200

C. W. Han et al.

Fig. 1 Photograph of LG’s a wallpaper OLED, b cinema sound OLED, c rollable OLED, and d OLED tunnel signage display

superior characteristics that cannot be implemented with conventional display technologies, and the following OLED features are introduced by a specialized company that produces large-sized OLED displays [1]. First, OLED technology has an authentic emissive color capable of realizing deep and natural colors. Second, it delivers identical picture quality from any angle. Third, it enables thinner and more innovative designs. As shown in Fig. 1, thin and innovative design technology has lead to innovative products such as wallpaper OLED, cinema sound OLED with sound embedded in the display, rollable OLED, and OLED tunnel signage displays. Lastly, it is an “ocular guard” display that protects the eyes of viewers through the reduction of blue light. In order to manufacture TV products with these OLED characteristics, there needs to be productivity and price competitiveness. This requires large-scale process technology that enables six 55-in. panels with 8th generation glass substrate (2500 × 2200 mm2 ) board. In the case of the TFT backplane, much effort has been made to utilize the a-Si TFT used in LCDs, but it was concluded that the a-Si:H TFT is not suitable for use in OLED TVs due to the lack of driving reliability [2]. As an alternative, in order to make the best use of the a-Si:H TFT infrastructure, oxide TFT, which is relatively more reliable than a-Si:H TFT, has been applied in a combination with compensation circuit technology to compensate for the shifted threshold voltage and mobility of oxide TFT [3]. The next important issue to be solved is the need for OLED patterning technology that can utilize large-area substrates. RGB patterning technology using a fine metal mask (FMM), which is applied to mobile OLED products, is difficult to apply to 8th generation substrates. As the substrate size increases, problems such as color mixing occur.

Large-Size OLED TVs with White OLED

201

Therefore, although several companies have made attempts, RGB technology has not been adopted for the production of TV products using 8th generation substrates. WOLEDs have been studied as an alternative technology to RGB side-by-side (SBS), and WOLED panels from 14 to 15-in. grade have been studied and introduced [4, 5]. Finally, in 2013, 55-in. FHD products using 1/2 8th generation substrates were released, and in 2015, 55- and 65-in. UHD products were introduced to the market with WOLED technology which used 8th generation substrates [6, 7]. Also, in 2019, an 88-in. 8K ultra-high-resolution OLED TV with 4 times the resolution of 4K was introduced [8]. Currently released OLED TVs can only be produced using WOLED technology. This chapter introduces the evolution of bottom-emitting WOLED devices and the features of large-sized OLED displays that use this technology. We also cover how efforts have been made to overcome issues in top-emitting WOLEDs and apply them to transparent OLED displays.

2 White Organic Light-Emitting Devices OLEDs generally achieve full color using metal masks, but due to the limitations of metal masks, they are difficult to apply to large area and high-resolution displays. Therefore, there has been much research to develop an alternative method for creating color with a color filter using white OLEDs. One-stack white OLED was announced by co-evaporating red dopants in host material that emits light from blue to green [9]. At the time, phosphorescent materials had not been developed enough to be applied to products, and all applied fluorescent materials. Being a one-stack device, it was difficult to apply to displays because a high voltage of 15 V was required to produce a brightness of 5000 cd/m2 . In order to improve the current efficiency of WOLED, a multi-photon emission structure was introduced by stacking several light-emitting units using V2 O5 and 4FTCNQ as charge generation layers (CGLs) and the graph of the current efficiency increased according to the number of units that was introduced in Fig. 2 [10]. As a result, multi-stack tandem white OLEDs have been actively studied and developed in various institutions.

2.1 Two-Stack Tandem WOLEDs Multi-stack white OLED devices first introduced tandem devices in which white units were stacked in stacks of three as shown in Fig. 3 [11]. The characteristics of one-, two-, and three-stacked devices are summarized in Table 1. This structure has the advantage of increasing current efficiency and extending lifetime by adding more white units, but the number of stacks increases, the color shift value increases

202

C. W. Han et al.

Fig. 2 Left: structure of multiphoton emission. Right: current density versus efficiency of MPE device. Reproduced with permission [10]. Copyright 2003, John Wiley & Sons, Inc.

Fig. 3 Structure of three-stack tandem connected by P-N contact. Modified from Hatwar et al. Reproduced with permission [11]. Copyright 2006, KIDS

according to the viewing angle, and the blue lifetime is low, so that the white color largely changes as it ages. Stabilized blue OLED is introduced to improve the blue lifetime. The first white stack consists of two layers, blue, and yellow. The energy transfer from blue to yellow reduces the stress from the excited state in the blue dopant, and consequently, it is the stabilized blue OLED structure that increases the lifetime of blue. As shown in Fig. 4, the first stack forms white with blue and yellow, and the second stack uses red, yellow, and green to form two-stack white OLED. The current

Large-Size OLED TVs with White OLED

203

Table 1 Tandem white OLED performance with increasing white units

Modified from Hatwar et al. Reproduced with permission [11]. Copyright 2006, KIDS

Fig. 4 Two-stack tandem white OLED consisting of a first stack with blue and yellow, and second stack with red, yellow, and green. Reproduced with permission [12]. Copyright 2008, John Wiley & Sons, Inc.

efficiency was 23.6 cd/A and the voltage was 9.3 V, and the color shift with aging showed a stable value of less than 0.02 during the 50% luminance reduction [12]. Next, phosphorescent materials were applied instead of fluorescence to further increase current efficiency. As shown in Fig. 5, a hybrid tandem white OLED composed of one unit of fluorescent blue and one unit of phosphorescent red and phosphorescent green was constructed. A white OLED device with a current efficiency of 30 cd/A or more was proposed [13]. In general, phosphorescent yellow exhibits a longer lifetime than phosphorescent green, so a hybrid tandem white OLED with fluorescent blue and phosphorescent yellow (λmax = 570 nm) instead of red and green was presented, as shown in Fig. 6 [14]. The color coordinates were (0.32, 0.31), the current efficiency was 49.6 cd/A, and the voltage was 6.6 V. In this hybrid tandem white OLED, the stabilized blue OLED was not applied because the lifetime of phosphorescent yellow is shorter

204

C. W. Han et al.

Fig. 5 Hybrid tandem white OLED composed of one unit of fluorescent blue and one unit of phosphorescent red and phosphorescent green. Reproduced with permission [13]. Copyright 2007, John Wiley & Sons, Inc.

Fig. 6 Hybrid tandem white OLED with fluorescent blue and phosphorescent yellow (λmax = 570 nm). Reproduced with permission [14]. Copyright 2010, John Wiley & Sons, Inc.

than fluorescent yellow. The lifetime of fluorescent yellow was reported as 800 h at 80 mA/cm2 [12], whereas phosphorescent yellow lasted about 208 h at 80 mA/cm2 (calculated based on a phosphorescent yellow current efficiency of 57.5 cd/A, current acceleration factor of 1.7). This value is similar to the lifetime of fluorescent blue with 200 h at 80 mA/cm2 as presented in a previous paper [12]. Therefore, a white OLED device with stable color shift characteristics for aging can be manufactured. Next, a hybrid tandem white OLED with λmax = 560 nm phosphorescent yellowgreen (YG) and an exciton blocking layer (EBL) was applied to the blue unit to increase white efficiency as shown in Fig. 7 [5]. The performance was 56.7 cd/A, 7.5 V, and the color coordinate (0.33, 0.33). By applying EBL to the first blue unit, the lifetime could be improved while increasing current efficiency. The second unit coevaporated the YG dopant and the red dopant in the light-emitting layer. The energy transfer from the YG dopant to the red dopant reduces the excited-state lifetime of

Large-Size OLED TVs with White OLED

205

Fig. 7 Hybrid tandem white OLED with λmax = 560 nm phosphorescent yellow-green (YG) + red and exciton blocking layer. Reproduced with permission [5]. Copyright 2011, John Wiley & Sons, Inc.

the YG dopant, thus improving the lifetime of the green and red sub-pixels in an OLED panel. The lifetime curve of each unit and WOLED device is shown in Fig. 8. The lifetime of the WOLED device initially follows the lifetime curve of the second unit, but as time passes, the blue lifetime of the first unit is more affected. However, when the YG dopant and the red dopant are co-evaporated in the same light-emitting layer, the red dopant is doped by about 0.2%, so the color shifts according to the slight change in the doping amount. For example, when the red dopant doping ratio is 0.2%, the color coordinate of the second unit is (0.462, 0.526), but when the doping ratio is 0.3%, the color changes to (0.486, 0.504). Therefore, it is difficult to secure color uniformity with the co-evaporation of YG dopant and red dopant in large-area 8th generation substrates. Therefore, an efficiency of 78.7 cd/A, a voltage of 7.1 V, and (0.32, 0.33) color coordinates were obtained by applying the device structure without red dopant and Fig. 8 The lifetime curve of each unit and WOLED device. The lifetime of the WOLED device initially follows the lifetime curve of the YG + R unit, but as time passes, the blue lifetime of the blue unit is more affected. Reproduced with permission [5]. Copyright 2011, John Wiley & Sons, Inc.

206

C. W. Han et al.

Fig. 9 Hybrid tandem white OLED with fluorescent blue and phosphorescent yellowgreen. This device was applied to the world’s first 55-in. FHD OLED TV. Reproduced with permission [15]. Copyright 2014, John Wiley & Sons, Inc

Fig. 10 Photograph of the world’s first 55-in. FHD OLED TV with a full white luminance of 100 cd/m2 and a peak luminance (25% area on) of 400 cd/m2

applying a mixed host to YG, as shown in Fig. 9 [15]. Finally, the world’s first 55-in. FHD OLED TV was launched with this new WOLED device as shown in Fig. 10. The luminance specification of the product is a full white luminance of 100 cd/m2 and peak luminance (25% area on) of 400 cd/m2 .

2.2 Three-Stack Tandem WOLEDs After the launch of the first OLED TV, there was a positive market response due to its vivid image quality and fancy design, but there was also feedback that a higher

Large-Size OLED TVs with White OLED

207

brightness was needed for a brighter screen as the full white brightness is only around 100 cd/m2 . However, the low efficiency of blue has become a problem for higher luminance. Phosphorescent blue or TADF blue, which is more efficient than fluorescent blue, has not yet obtained sufficient characteristics for commercialization. Therefore, in order to increase the efficiency of blue, a three-stack tandem white OLED structure that considers the blue unit is needed in addition to the two-stack tandem white OLED structure mentioned above [16]. There are three cases where two blue units are applied: B/YG/B, B/B/YG, and YG/B/B. Since TV products are different from lighting products, there are many situations to consider when designing white OLED devices. For example, a cavity structure should be selected that not only maximizes efficiency for high brightness but also maintains image quality over all viewing angles. It is necessary to design the device structure with minimal change in the white color coordinate from aging and minimal change in the white color coordinate from grey scale change. The optimal structure to meet these requirements was the B/YG/B three-stack tandem white OLED structure shown in Fig. 11. The current efficiency for this structure is 85 cd/A, and the color coordinates are (0.287, 0.300). By applying the three-stack tandem WOLEDs, the brightness of the full white of the 65-inch UHD OLED TV is 150 cd/m2 , and the peak luminance (25% area on) is 450 cd/m2 . The brightness increased by 1.5 times compared to the two-stack tandem, but the panel’s power consumption was reduced by 20%, not only because the blue efficiency was enhanced, but also because the color temperature of WOLED increased, so that it used fewer blue pixels. Fig. 11 Three-stack tandem white OLED with two-stack fluorescent blue and one-stack phosphorescent yellow-green. This device was applied to the world’s first 65-in. UHD OLED TV. Reproduced with permission [16]. Copyright 2016, John Wiley & Sons, Inc.

208

C. W. Han et al.

Fig. 12 Color spectrum for white OLED. A green color filter with narrower spectrum was developed for a deeper green

Since the release of the product, market feedback has required products that meet the digital cinema initiative (DCI) color gamut. The DCI color gamut is the RGB color space required for digital movies in the US film industry, requiring 25% more area than the BT 709 color space, which requires deeper green and red colors [17]. To meet the DCI color space requirements, a new WOLED structure and color filter was suggested [18]. To meet this requirement, the green color should be realized using phosphorescent greens with shorter wavelengths than phosphorescent YG, but phosphorescent greens would still not have a long lifetime for use in TVs. Therefore, a deeper green color was developed using a color filter with a narrow green spectrum as shown in Fig. 12. The red case came up with a new white OLED structure that adds a red-emitting layer. As shown in Fig. 13, we have devised two structures in which phosphorescent red is applied and fluorescent red is applied. For phosphorescent red, a phosphorescent red layer was applied to the phosphorescent YG in the second unit. The characteristics of the white OLED device were a current density of 83 cd/A, a voltage of 12 V, and the color coordinates (0.308, 0.315). When applying fluorescent red, a fluorescent red layer was applied to fluorescent blue in the third unit. In this case, a buffer layer is inserted between the fluorescent blue layer and the fluorescent red layer. In general, without a buffer layer, fluorescent blue’s host material has excellent electron-transport ability, and electrons are transferred to an adjacent red layer so that blue does not emit light. In general, the buffer layer is made of a material having electron blocking properties. The white OLED characteristics at this time obtain a current density of 90 cd/A, a voltage of 12 V, and the color coordinates (0.297, 0.280). Through these studies, the 65-in. UHD OLED TV widened the color gamut of green and red as shown in Fig. 14 to get 99% of the DCI color gamut and allowed a full white luminance of 150 cd/m2 and a peak luminance (25% area on) of 500 cd/m2 .

Large-Size OLED TVs with White OLED

209

Fig. 13 Left: B/YG/B + R three-stack tandem WOLED structure. Right: B/YG + R/B three-stack tandem WOLED structure

Fig. 14 Color coordinates of displays with B/YG + R/B and B/YG/B three-stack tandem WOLED

3 Display Transformation with OLEDs As discussed in the previous session, OLEDs not only provide excellent picture quality, but also self-illuminating displays that do not require backlight, enabling innovative designs that are thin and light. Composed of organic materials, OLEDs

210

C. W. Han et al.

are vulnerable to substances in the air such as moisture. For example, moisture easily hydrolyzes electrodes or electron-injection layers, resulting in dark spots, which are non-light-emitting areas [19]. This causes defects such as pixel shrinkage. Therefore, OLED needs encapsulation technology to prevent external air and moisture from entering. For large-area OLED products using white OLEDs, an adhesive film with excellent moisture resistance was used, and a metal substrate similar to the thermal expansion coefficient of the glass substrate and having a thickness of 0.1 mm was applied [15]. Thus, wallpaper OLED composed of a TFT substrate and metal encapsulation is capable of achieving a 3.9 mm panel thickness and a lightweight panel weighing 7.6 kg as shown in Fig. 15, so the distance between the wall and the wallpaper OLED is 0 mm. It is possible to design the display to stick to the wall like wallpaper with paper-slim thickness [20]. However, in the case of LCDs, backlight is required and LED housing and a cover-shield should be attached, so the thickness of the thinnest panel is 6.9 mm and the thickness of the thickest part is 13.7 mm, making it difficult to attach to the wall like wallpaper. Therefore, the wallpaper all-in-one with the wall increases viewer immersion and the unobtrusiveness of the TV means it harmonizes with indoor interior without damaging it [21]. Fig. 15 Upper side: wallpaper OLED with a 3.9 mm panel thickness and 0 mm distance between the OLED panel and wall. Down side: LCD with a 6.9 mm panel thickness and a 13.7 mm distance between the LCD panel and wall [21]

Large-Size OLED TVs with White OLED

211

Next is cinema sound OLEDs (CSO). CSO technology attaches a vibrationgenerating drive exciter to the back of the OLED panel. The exciter converts electrical sound signals into physical vibration signals and transmits the vibrations to the front of the panel. OLED panels are thin and OLEDs don’t distort sound like liquid crystals, so the panel itself can act as a diaphragm. In order to improve the sound performance, a sounding space is created between the OLED panel and the exciter, and an enclosure structure is applied to control the reflected waves. Firstly, CSO technology has excellent acoustic characteristics. In ordinary TVs, the speaker is placed behind the panel and the sound is generated downward so that the acoustic characteristics are uneven and the environment in which the TV is placed is greatly affected. On the other hand, OLED TVs with this technology deliver clear, balanced sound from low to high as shown in Fig. 16, because the vibration from the exciter delivers sound directly to the viewer through the OLED panel. Second, this technology creates a wide sound field. Since the OLED panel itself is the diaphragm of a large speaker, it creates a wide sound field that fills the entire screen and increases the realism of the sound. Third, this technology is characterized by the matching of sound and video. In ordinary TVs, the screen and sound are projected from separate spots, which is unnatural and heterogeneous, but this technology converges the sound and video. In other words, the position of the speaker’s face in the video and the location of the sound are exactly the same, which Fig. 16 Sound pressure level comparison between Cinematic Sound OLED and ordinary TV speakers [21]

212

C. W. Han et al.

maximizes the viewer’s immersion by maximizing the realistic feeling of the scene. Applying LG’s CSO technology, Sony released the Bravia A1 OLED TV in 2017 and received favorable reviews from the market [21]. Next, through innovations in form factor, various types of displays can be realized. Form factor refers to the structured form of a product. In other words, most of the display form factors we use today are square. Since LCDs must use backlights, it is impossible to achieve a free form factor by overcoming thickness and rigidity. A very gently curved LCD is possible, but the freely foldable or bendable form factor is limited to OLEDs, self-luminous displays based on plastics and organic materials. Form factor innovations include step 1 with fixed types, such as curved and bendable, step 2 with single-axis variable types, such as foldable and rollable, and step 3 will evolve into a free form variable, such as stretchable [21]. The step 1 form factor innovation is already available on smartphones and TVs. For the evolution of the step 2 rollable form, the radius of the curve rule must be significantly lowered and the display thickness must be dramatically thinner. Figure 17 shows a rollable OLED TV prototype that can be rolled up or rolled down. The screen is hidden like a roll screen, revealing the interior as it is, and unfolding only when watching TV. In other words, the curvature radius determines how small the screen can be rolled up when it is rolled up. To reduce this radius of curvature, a 0.01 mm-thick plastic substrate is applied and the final OLED display panel thickness is only 0.8 mm. This is how rollable OLED TV can be realized [22]. The 33 million-pixel 8 K 88-in. OLED display, four times sharper (7680 × 4320) than UHD (3840 × 2160), was introduced for the first time in the world as shown in

Fig. 17 Left: rolled down rollable OLED TV, Right: rolled up rollable OLED TV. Both are applied 0.01 mm-thick plastic substrate and 0.8 mm-thick OLED display panel

Large-Size OLED TVs with White OLED

213

Fig. 18 Photograph of the 33 million-pixel 8 K 88-in. OLED displays with bottom emission white OLEDs

Fig. 18. As the number of pixels increases, it is difficult to secure the lifetime of the bottom emission WOLEDs. A technology developed to solve this problem has been introduced. Through 3.8 × 3.8 μm2 contact hole define technology, 120 Hz drive and RC delay minimization technology, an aperture ratio comparable to that of 4 K UHD products was secured, despite it being a bottom emission [8].

4 Top-Emitting White OLEDs and Display Application Top-emitting OLEDs have several advantages over bottom-emitting OLEDs. Topemitting in OLED displays can achieve a high aperture ratio regardless of how the circuit design of the pixel is constructed because the emitted light comes out to the opposite side of the TFT substrate. As a result, it has the advantage of lowering the current density injected into the pixel to increase the lifetime of the OLED device. Top-emitting generally emits light to the outside through the cathode, so the cathode transparency must be high. However, ITO deposited by sputtering is highly transparent, but it is difficult to use because it causes damage to the organic material, so metal with a semi-transparent cathode is used alongside metal with high reflectivity

214

C. W. Han et al.

for the anode. In general, Ag:Mg is used for the cathode, and an Ag reflector is mainly used for the anode [23]. The spectral shape and intensity of the light emitted are controlled by the microcavity effect, which is an interference effect between the cathode and anode reflectors. The full width at half maximum (FWHM) of the spectrum formed by the microcavity is expressed as follows by the Fabry–Perot model [24]. √ 1 − R1 R2 λ2max × 4√ FW H M = 2n L e f f π R1 R2

(1)

Here, L eff is the effective cavity length, n is the refractive index, and R1 , R2 is the reflectivity of the anode and cathode, reflectively. According to Eq. (1), as the reflectance of the anode and the cathode increases, the half width of the microcavity OLED becomes narrow. The relational expression representing the intensity of light emitted in the front direction can be expressed as follows by the Fabry–Perot model [25]. T I ∝ 2   √ √ 1 − R1 R2 + 4 R1 R2 sin2 φ 2

(2)

Here, I is the intensity of emitting light, T is the transmittance, and R1 , R2 is the reflectivity of the anode and cathode, respectively. ΔΦ is the round-trip phase term. According to this Eq. (2), the transmittance of the microcavity OLED is reduced by the semi-transmissive cathode, but the front emission intensity increases due to the increased reflectivity of the anode and cathode. However, the stronger the microcavity, the worse the viewing angle characteristics. Tessler et al. of Cambridge University expressed the change in emission wavelength according to the viewing angle in the following simple formula [26]. λ = 2π m L e f f · n · cos θ

(3)

Here, λ is the resonance wavelength, m is the mode number, n is the refractive index, θ is the internal angle which is related to the external viewing angle through Snell’s law, and L eff is the cavity effective length. According to this Eq. (3), the microcavity OLED has a shortened emission wavelength, that is, blue shifted as the viewing angle increases. Therefore, the spectrum formed by the microcavity has the advantage of forming a high color gamut by narrowing the width and increasing the luminance according to the high efficiency, but the spectrum obtained by the microcavity has the disadvantage that the color changes according to the viewing angle.

Large-Size OLED TVs with White OLED

215

4.1 Microcavity Top-Emitting White OLED Devices The following considerations should be considered when developing top emission white OLEDs for large-area displays. First, since a large-area display is viewed by several people at the same time, the change in luminance or color according to the viewing angle should be small. Second, in order to reduce power consumption or increase luminance, four sub-pixels of WRGB should be used, and the spectrum of the white sub-pixel should implement a white color with a relatively wide spectrum. In this section, we would like to introduce examples of how the top-emitting white OLEDs for large-area displays have been developed and applied to actual products. In 2004, Sony introduced a top-emitting white OLED with microcavity, and reconstructed it based on the contents introduced in the paper and patent as shown in Fig. 19 [27, 28]. One-stack white OLED was used to stack the RGB layer, and a semitransparent metal was used for the microcavity. To adjust the cavity length, different anode ITO thickness was applied for each RGB pixel. In addition, an optimal color filter was applied to present a 12.5-in. prototype. Each color coordinate of the RGB color showed high color saturation characteristics such as (0.67, 0.33), (0.28, 0.63), and (0.14, 0.07), respectively. In addition, since the reflection of external light is prevented by a color filter without using a circular polarizer, a reflectance of 2.5% is Fig. 19 Schematic cross-sectional view of WOLED + Microcavity + CF structure. Reproduced with permission [27]. Copyright 2004, John Wiley & Sons, Inc.

216

C. W. Han et al.

Fig. 20 Hybrid tandem white OLED structure with fluorescent blue and phosphorescent red and green. Reproduced with permission [29]. Copyright 2010, John Wiley & Sons, Inc.

secured. Also, the viewing angle showed a value of 0.017 in the direction of 45° from the front. The microcavity and color filter were optimized to obtain good characteristics, but the viewing angle was limited to 45°, and the viewing angle was slightly narrow to apply to a large-area TV. In addition, the application of RGB 3 sub-pixels is expected to result in high-power consumption. Therefore, it was not applied to real display products. In 2010, Samsung introduced the hybrid tandem white OLED as shown in Fig. 20 [29]. The anode reflector used Ag, the cathode also used a vapor-deposited semitransparent silver and magnesium compound, the cavity length was adjusted to the anode ITO thickness, the red and blue sub-pixels used the same thickness ITO, and the green sub-pixel used an ITO with a thickness different from that of the red and blue sub-pixels. The maximum values of the wavelengths were 452 nm, 529 nm, and 628 nm, respectively. By applying optimized color filters, the color coordinates of RGB colors were (0.673, 0.317), (0.244, 0.706), (0.106, 0.55), showing high color saturation characteristics. The voltage of the white OLED device was 6.25 V, and regarding the color filter, a conversion white efficiency of 15.7 cd/A was obtained. Even in this case, the viewing angle is limited by the microcavity. Therefore, it cannot be used for a large-area display, so in the published paper, it was proposed for a micro-display, a near-to-eye application with less viewing angle limitations. In 2012 and 2014, SEL introduced a top-emitting white OLED device as shown in Fig. 21 [30, 31]. The 13.5-in. and 5.2-in. prototypes were announced by applying hybrid 2-stack white OLEDs with a Al/Ti/ITO reflective anode and semi-transparent cathode. In particular, in the case of the 5.2-in. device, four sub-pixels were used to reduce power consumption. The color coordinates of the white sub-pixel were (0.44, 0.38) by the microcavity, and the color temperature was 2780 k, resulting in a very low color temperature. This color temperature is suitable for lighting, but TVs require a cool white with a display color temperature as high as 10,000 K. Therefore, in order

Large-Size OLED TVs with White OLED

217

Fig. 21 Hybrid tandem white OLED structure with fluorescent blue and phosphorescent green and red. Reproduced with permission [30]. Copyright 2012, John Wiley & Sons, Inc.

to meet the display color temperature, the high luminance of the blue sub-pixel is required when the color temperature of the white sub-pixel is low. Therefore, even if four sub-pixels are used, this effect cannot be fully utilized, and there is a limit to reducing power consumption.

4.2 Non-microcavity Top-Emitting White OLED Devices We will look at the contribution of each power consumption when using a microcavity and when using four sub-pixels of WRGB through a published paper by Cox et al. [32]. As shown in Fig. 22, Kodak manufactured one-stack white OLEDs in a bottomemitting structure and a top-emitting structure. Also, as shown in Fig. 23, two-stack white OLEDs were manufactured with a bottom-emitting structure and a top-emitting structure. And for each, microcavity and non-microcavity structures were also fabricated and compared. Then, 13,000 standard still images were constructed to analyze power consumption and color gamut, and the results were summarized as shown in the table. As shown in Table 2, when microcavity is applied, the color reproducibility is improved by about 10%, and the average power consumption is improved by 25– 62% compared to the non-microcavity structure. However, when applying WRGB 4 sub-pixel compared to RGB 3 sub-pixel, the average power consumption is improved by 72% depending on the structure. If microcavity is applied, the color gamut is improved and efficiency is increased, resulting in reduced power consumption, but applying a non-microcavity structure and using four WRGB sub-pixel structures is much more advantageous in terms

218 Fig. 22 Left: bottom-emitter single-stack microcavity OLED structure, Right: Top-emitter single-stack microcavity OLED structure. Reproduced with permission [32]. Copyright 2009, John Wiley & Sons, Inc.

Fig. 23 Left: Bottom-emitter tandem microcavity OLED structure, Right: top-emitter tandem microcavity OLED structure. Reproduced with permission [32]. Copyright 2009, John Wiley & Sons, Inc.

C. W. Han et al.

Large-Size OLED TVs with White OLED

219

Table 2 Device performance summary for various design options

Reproduced with permission [32]. Copyright 2009, John Wiley & Sons, Inc

of power consumption. Other papers have reported power consumption measured from 2.16-in. WRGB and RGB displays [33]. The average power consumption of WRGB and RGB displays showing the same color gamut was 180 mW and 340 mW, respectively. The power consumption of the WRGB format is almost 1/2 of that of RGB. In addition, since it is possible to secure the viewing angle required for large-area display products, it can be considered more advantageous to apply a non-microcavity structure to large-area display products. The problem is that in order to realize a top-emitting white OLED of a non-microcavity structure, it is necessary to suppress the microcavity effect by using a transparent electrode as a cathode. In general, ITO, which is a transparent electrode, must be deposited on organic materials via sputtering. However, the sputtering method was not actively applied due to the problem of organic materials being damaged by sputtering. Liao et al. of City University of Hong Kong conducted an experiment to irradiate an argon ion beam on the Alq3 organic thin film to simulate ion irradiation in the sputtering process on Alq3 , and conducted surface analysis using XPS and UPS [34]. Through the XPS, it was confirmed that the binding of Alq3 was broken through the appearance of another compound. Also, it was confirmed that a metal-like conductive layer was formed on the surface through changes in the Alq3 valence band structure due to UPS. When the surface of Alq3 is conductive, electrons are injected from the cathode to Alq3 , non-emission quenching occurs, and electrical shorts can occur. Based on these studies, Kodak continued research to reduce sputter damage on organic materials. In particular, as the research results showing CuPc having stable properties on the sputter became known [35], the research on CuPc proceeded further. However, another problem was discovered when CuPc was used as a buffer layer. In other words, the CuPc forms an electron-injection barrier at the interface with the ETL, causing electron–hole recombination at the interface and lowering efficiency due to non-lighting emission [36]. Therefore, in order to solve this problem, an aluminum alloy material doped with Li was used for sputtering.

220

C. W. Han et al.

Table 3 Buffer structure and performance of sputtered cathode and a control (evaporated cathode) device

Reproduced with permission [37]. Copyright 2001, John Wiley & Sons, Inc

As a result, as lithium in the cathode diffused into the barrier layer CuPc, Li was present on the CuPc and ETL interfaces, thereby lowering the electron barrier and improving efficiency and lifetime. In general, diffused lithium deteriorates the lifetime of the OLED device, and it can be estimated that lithium does not diffuse to the light-emitting region through the improvement of the lifetime. In order to develop this research and solve sputter damage and electrical shorts, Kodak proposed a dual buffer structure of LiF and CuPc in 2001 by adding LiF as a buffer layer [37]. As a result, the grain size, the surface roughness, and the orientation of the CuPc were improved to secure a stable film quality, and the OLED device that deposited the cathode by sputter obtained characteristics comparable to those of OLED devices with evaporated Mg:Ag cathodes as shown in Table 3. The voltage of the OLED device deposited by sputtering Al:Li alloy was 6.2 V, and the current efficiency was 2.4 cd/A. This result confirmed the potential of the sputtered cathode by showing characteristics similar to a voltage of 6.5 V and current efficiency of 2.5 cd/A for an OLED device deposited with an evaporation of Mg:Ag.

4.3 Transparent OLED Display The transparent display shows a variety of information on the display panel to make it look like it is displayed on a clear glass. Transparent displays can be divided into a projection display [38] and a transmissive display. Transmissive displays are a way to change the transmittance of a luminous material itself. Transmissive displays are classified into LCDs and OLEDs. In the case of LCDs, light is lost as it passes through various components such as two polarizers, color filters, and backlight. As a result, a transmittance of only about 20% can be achieved, and there is the disadvantage that it is difficult to achieve a clear image quality outdoors [39, 40]. However, unlike the LCDs, OLEDs do not

Large-Size OLED TVs with White OLED

221

require polarizers and backlight, and can realize a high transmittance. Transparent displays can be applied to various applications such as show windows, refrigerators, billboards, and public displays. The initial transparent OLEDs were based on aSi:H TFTs and realized transparent double-sided transparent OLEDs with a lower transparency of 20% [41, 42]. In 2016, LG Display revealed a transparent OLED display with top-emitting white OLEDs. The WOLED consisted of a reflective anode layer, multi-organic layers, and a transparent cathode layer [43]. The reflective anode layer used an Ag alloy with a high reflectance as a reflector and forms an indium tin oxide with a high work function to inject a hole into hole-injection layer (HIL) easily. All organic layers consisting of three independent unit elements connected in series by two CGLs were deposited by thermal evaporation. The cathode used transparent conductive oxide (TCO) which was sputtered on an organic layer to increase the transmission of the transparent display and to realize non-microcavity white OLEDs. An organic buffer layer was inserted before the cathode layer to prevent damage to the underlying organic layer by the sputtering process of the cathode. In order to inject electrons easily into the electron-transport layer, electron-injection material with strong electron-injection properties was adopted. Based on optical simulation, a top-emitting WOLED device was composed of three organic stacks between the Ag alloy reflective anode and the TCO cathode. As shown in Fig. 24, this device is composed of one phosphorescent yellow-green stack between two fluorescent blue stacks in the order of blue, yellow-green and blue, Fig. 24 Structure of non-microcavity top-emitting three-stack tandem WOLED. Reproduced with permission [43]. Copyright 2016, John Wiley & Sons, Inc.

222

C. W. Han et al.

and each unit is connected to CGLs. By varying the thickness of the hole-transport layer in each unit device, a high efficiency was obtained by optical simulation that optimizes the position of the emitting layer (EML) and the total thickness of the organic layer. Table 4 summarizes the electro-optical performance of the top-emitting WOLED compared to the bottom-emitting WOLED, and the electro-optical characteristics including the JV curve, EL spectrum, and viewing angle characteristics are shown in Fig. 25. The three-stack top-emitting WOLED device exhibits a white point (0.304, Table 4 Performance comparison between top-emitting and bottom-emitting WOLED

Reproduced with permission [43]. Copyright 2016, John Wiley & Sons, Inc

Fig. 25 Electro-optical properties of top- and bottom-emitting WOLEDs. a J-V curve, b EL spectra (@10 mA/cm2 ), and c color shift depending on the viewing angle. Reproduced with permission [43]. Copyright 2016, John Wiley & Sons, Inc.

Large-Size OLED TVs with White OLED

223

0.349) corresponding to a CCT of 6,800 K with a current efficiency of 95 cd/A at a current density of 10 mA/cm2 . As shown in the EL spectrum graph in Fig. 25 (a), top-emitting WOLED has high blue strength and is an advantage of the 3-stacked device architecture that adopts dual blue units, as mentioned for the bottom-emitting WOLED on OLED TVs in the previous section. Top-emitting WOLEDs have a similar blue strength as bottom-emitting WOLEDs, as shown in Fig. 25b, but show higher strength in yellow-green, which results in somewhat lower CCT for top-emitting WOLEDs. This result is attributed to Ag electrodes with a higher reflectivity and TCO with a higher refractivity. As expected from optical simulations, the device represents good color viewing angle characteristics defined as a color shift of u’v’ from the color in the normal direction. The u’v’ at each 60° is 0.010 in the top-emitting WOLED with nonmicrocavity, compared with the bottom-emitting WOLED shown in Fig. 25c. Top-emitting WOLEDs are indispensable for transparent OLED displays. LG Display has demonstrated 55-in. FHD transparent OLED displays as shown in Fig. 26. In addition, top-emitting WOLEDs were applied on a transparent polyimide Fig. 26 Photograph of a 55-in. transparent OLED display

224

C. W. Han et al.

Fig. 27 Photograph of a 77-in. transparent flexible OLED display

substrate to demonstrate a 77-in. UHD transparent flexible OLED display with a curvature radius of 80 mm and a transmittance of 40% as shown in Fig. 27 [44]. Transparent OLED displays consist of a light-emitting part and a transparent part as shown in Fig. 28 [44]. When using a transparent OLED display as a showroom window, people want to see the image clearly in the background while they want to get information from the transparent display in the front. In order to do so, the transparent display must have high transparency and contrast ratio. The transparency of a transparent OLED display is determined by the opening area of the transparent part and its transmittance. LG Display has implemented transparent OLED displays with a transmittance of 40% through top-emitting white OLEDs with a long lifetime.

5 Conclusion Large OLED displays, including OLED TVs that are currently being produced, use WOLED technology that can be produced using large-sized substrates. In this chapter, the history of the development of bottom-emitting WOLED applied to OLED TVs and the performance improvement of OLED TVs were discussed. By increasing the number of stacks and adding a light-emitting color layer, it was possible to increase the luminance and color gamut of the display. In addition, this chapter introduced features such as wallpaper OLED, cinema sound OLED, rollable OLED, and 8 K OLED, which are applications that can only be used as a thin and light OLED display and that can change people’s lifestyle. Finally, the necessity of the nonmicrocavity structure of the top-emitting WOLED for realizing a signage display, such as a transparent product, and the process of development were introduced in detail. In the future, these are expected to quickly replace the existing LCD TV market by reducing the price through mass production and various types of products that utilize the superior image quality and freedom of design of the OLED TV.

Large-Size OLED TVs with White OLED

225

Fig. 28 Cross section of a transparent OLED panel. Modified from Park et al. Reproduced with permission [44]. Copyright 2018, John Wiley

References 1. www.lgdisplay.com 2. J.J. Lih, C.F. Sung, C.H. Li, T.H. Hsiao, H.H. Lee, Comparison of a-Si and Poly-Si for AMOLEDs, in SID Proceeding (2004), pp. 1505–1507 3. W.J. Nam, J.S. Shim, H.J. Shin, J.M. Kim, W.S. Ha, K.H. Park, H.G. Kim, B.S. Kim, C.H. Oh, B.C. Ahn, B.C. Kim, S.Y. Cha, 55-inch OLED TV using InGaZnO TFTs with WRGB pixel design, in SID Proceeding (2013), pp. 243–246 4. J. Hamer, A.D. Arnold, M.L. Boroson, M. Itoh, T.K. Hatwar, M.J. Helber, K. Miwa, C.I. Levey, M. Long, J.E. Ludwicki, D.C. Scheirer, J.P. Spindler, S.A. Van Slyke, System design for a wide-color-gamut TV-sized AMOLED display. J. SID 16(1), 3–14 (2008) 5. C.W. Han, Y.H. Tak, B.C. Ahn, 15-in. RGBW panel using two-stacked white OLED and color filters for large-sized display applications. J. SID 19(2), 190–195 (2011) 6. C.H. Oh, H.J. Shin, W.J. Nam, B.C. Ahn, S.Y. Cha, S.D. Yeo, Technological Progress and Commercialization of OLED TV, in SID Proceeding (2013), pp. 239–242 7. H.J. Shin, S. Takasugi, K.M. Park, S.H. Choi, Y.S. Jeong, B.C. Song, H.S. Kim, C.H. Oh, B.C. Ahn, Novel OLED display technologies for large-size UHD OLED TVs, in SID Proceeding (2015), pp. 53–56 8. I.H. Han, B.U. Kang, K.H. Oh, B.S. Kim, I.B. Kang, 8 K OLED TV technology, in IMID Proceeding (2019)

226

C. W. Han et al.

9. T.A. Ali, A.P. Ghosh, W.E. Howard, High efficiency white organic light emitting devices, in SID Proceeding (1999), pp. 442–445 10. T. Matsumoto, T. Nakada, J. Endo, K. Mori, N. Kawamura, A. Yokoi, J. Kido, Multiphoton organic EL device having charge generation layer, in SID Proceeding (2003), pp. 979–981 11. T.K. Hatwar, J.P. Spindler, S.A. Van Slyke, High performance tandem OLEDs for large area full color AM displays and lighting applications, in IMID/IDMC Proceeding (2006), pp. 1577–1582 12. T.K. Hatwar, J.P. Spindler, Novel approach to stabilize blue OLEDs and fabrication of highefficiency tandem white OLEDs for large-area display applications, in SID Proceeding (2008), pp. 814–817 13. S. Ishihara, K. Masuda, Y. Sakaki, H. Kotaki, S. Aratani, High-efficiency white organic lightemitting diodes with a two-stack multi-photon emission structure, in SID Proceeding (2007), pp. 1501–1503 14. T.K. Hatwar, J.P. Spindler, M. Kondakova, D. Giesen, J. Deaton, J.R. Vargas, Hybrid tandem white OLEDs with high efficiency and long life-time for AMOLED displays and solid-state lighting, in SID Proceeding (2010), pp. 778–781 15. C.W. Han, J.S. Park, H.S. Choi, T.S. Kim, Y.H. Shin, H.J. Shin, M.J. Lim, B.C. Kim, H.S. Kim, B.S. Kim, Y.H. Tak, C.H. Oh, S.Y. Cha, B.C. Ahn, Advanced technologies for UHD curved OLED TV. J. SID 22(11), 552–563 (2015) 16. H.S. Choi, T.S. Kim, C.W. Han, H.C. Choi, B.C. Ahn, I.B. Kang, C.H Oh, S.D. Yeo, Recent progress of white light-emitting diodes for an application to new models of OLED TV, in SID Proceeding (2016), pp. 65–68 17. D. Jackson, Improving Color on the Web (Webkit, 2016) 18. C.W. Han, M.Y. Han, S.R. Joung, J.S. Park, Y.K. Jung, J.M. Lee, H.S. Choi, G.J. Cho, D.H. Kim, M.K. Yee, H.G. Kim, H.C. Choi, C.H. Oh, I.B. Kang, 3 stack-3 color white OLEDs for 4 K premium OLED TV, in SID Proceeding (2017), pp. 1–4 19. Y.F. Liew, Investigation of the sites of dark spot in organic light-emitting devices. Appl. Phys. Lett. 77, 2650–2652 (2000) 20. H.J. Shin, K.M. Park, S. Takasugi, S.H. Choi, Y.S. Jeong, H.D. Choi, I.J. Kim, H.S. Kim, H.W. Lee, C.H. Oh, Advanced OLED display technologies for large-size semi-flexible TVs, in SID Proceeding (2016), pp. 609–612 21. www.oledspace.com 22. J.Y. Yang, Rollable Display and Technical Challenges (2nd Techweek, 2019) 23. P.K. Raychaudhuri, J.K. Madathil, J.D. Shore, S.A. Van Slyke, Performance enhancement of top- and bottom-emitting organic light-emitting devices using microcavity structures. J. SID 12(3), 315–321 (2004) 24. E.F. Schubert, N.E.J. Hunt, M. Micovic, R.J. Malik, D.L. Sivco, A.Y. Cho, G.J. Zydzik, Highly efficient light-emitting diodes with microcavities. Science 265, 943–945 (1994) 25. M. Thomschke, R. Nitsche, M. Furno, K. Leo, Optimized efficiency and angular emission characteristics of white top-emitting organic electroluminescent diodes. Appl. Phys. Lett. 94, 083303-3 (2009) 26. N. Tessler, S. Burns, H. Becker, R.H. Friend, Suppressed angular color dispersion in planar microcavities. Appl. Phys. Lett. 70, 556–558 (1997) 27. M. Kashiwabara, K. Hanawa, R. Asaki, I. Kobori, R. Matsuura, H. Yamada, T. Yamamoto, A. Ozawa, Y. Sato, S. Terada, J. Yamada, T. Sasaoka, S. Tamura, T. Urabe, Advanced AMOLED display based on white emitter with microcavity structure, in SID Proceeding (2004), pp. 1017–1019 28. M. Kashiwabara, J. Yamada, S. Yokoyama, K. Hanawa, Organic light emitting device, manufacturing method thereof and display device using the organic light emitting device, EP1672962 B1 29. S. Lee, C.Y. Park, J.B. Song, S.K. Kang, S. Tamura, H.K. Chung, S.S. Kim, Highly efficient top emitting white OLED architecture for micro-display applications, in SID Proceeding (2010), pp. 1796–1799 30. S. Eguchi, H. Shinoda, T. Isa, H. Miyake, S. Kawashima, M. Takahashi, Y. Hirakata, S. Yamazaki, M. Katayama, K. Okazaki, A. Nakamura, K. Kikuchi, M. Niboshi, Y. Tsukamoto,

Large-Size OLED TVs with White OLED

31.

32. 33.

34.

35. 36. 37.

38. 39. 40. 41. 42.

43.

44.

227

S. Mitsui, 13.5-inch quad-FHD top-emission OLED display using crystalline-OS-FET, in SID Proceeding (2012), pp. 367–370 R. Kataish, T. Sasaki, K. Toyotaka, H. Miyake, Y. Yangisawa, H. Ikeda, H. Nakashima, N. Ohsawa, S. Eguchi, S. Seo, Y. Hirakata, S. Yamazaki, C. Bower, D. Cotton, A. Matthews, P. Andrew, C. Gheorghiu, J. Berqquist, Development of side-roll and top-roll panels for an RGBW high-resolution flexible display using a white OLED with microcavity structure, in SID Proceeding (2014), pp. 187–190 R.S. Cok, J.D. Shore, Microcavity white-emitting OLED devices. J. SID 17(8), 617–627 (2009) A.D. Arnold, T.K. Hatwar, M.V. Hettel, P.J. Kane, M.E. Miller, M.J. Murdoch, J.P. Spindler, S.A. Van Slyke, K. Mameno, R. Nishikawa, T. Omura, S. Matsumoto, Full-color AMOLED with RGBW pixel pattern, in IMID Proceeding (2004), pp. 808–811 L.S. Liao, L.S. Hung, W.C. Chan, X.M. Ding, T.K. Sham, I. Bello, C.S. Lee, S.T. Lee, Ionbeam-induced surface damages on tris-8-hydroxyquinoline… aluminum. Appl. Phys. Lett. 75, 1619–1621 (1999) G. Parthasarathy, P.E. Burrows, V. Khalfin, V. G. Kozlov, S.R. Forrest, A metal-free cathode for organic semiconductor devices. Appl. Phys. Lett. 72, 2138–2140 (1998) L.S. Hung, L.S. Liao, C.S. Lee, S.T. Lee, Sputter deposition of cathodes in organic light emitting diodes. J. Appl. Phys. 86, 4607–4612 (1999) P.K. Raychaudhuri, C.W. Tang, J.K. Madathil, Fabrication of lithium-based alloy cathodes for organic light-emitting diodes by D C magnetron sputtering, in SID Proceeding, pp. 526–529 (2001) C.W. Hsu, B. Zhen, W. Qiu, O. Shapira, B.G. DeLacy, J.D. Joannopoulos, M. Soljacic, Transparent displays enabled by resonant nanoparticle scattering. Nat. Commun. 5, 1–6 (2014) C.W. Kuo, C.H. Lin, Y.Y. Liao, Y.H. Lai, C.T. Chuang, C.N. Yeh, J.K. Lu, N. Sugiura, Blur-Free transparent LCD with hybrid transparency, in SID Proceeding (2013), pp. 70–73 T. Riedl, P. Görrn, W. Kowalsky, Transparent electronics for see-through AMOLED displays. J. Disp. Technol. 5, 501–508 (2009) K.H. Lee, S.-Y. Ryu, J.H. Kwon, S.W. Kim, H.K. Chung, 2.2inch QCIF full color transparent AMOLED display, in SID Proceeding (2003), pp. 104–107 Y.-J. Tung, R. Hewitt, A. Chwang, M. Hack, J. Brown, K.-M. Kim, D. S. Kim, J.H. Hur, J. Jang, A 200-dpi transparent a-Si TFT active-matrix phosphorescent OLED display, in SID Proceeding (2003), pp. 1546–1549 Y.K. Jung, H.S Choi, S.Y. Ahn, S.H. Kim, H.D. Choi, C.W. Han, B. Kim, S.J. Kim, J.M. Kim, J.H. Choi, S.Y. Yoon, Y.H. Tak, H.C. Choi, B.C. Ahn, I.B. Kang, 3 stacked top emitting white OLED for high resolution OLED TV, in SID Proceeding (2016), pp. 707–710 C.I. Park, M.R. Seong, M.A. Kim, D.J. Kim, H.J. Jung, M.G. Cho, S.H. Lee, H.K. Lee, S.J. Min, J.Y. Kim, M.S. Kim, J.H. Park, S.Y. Kwon, B. Kim, S.J. Kim, W.S. Park, J.Y. Yang, S.Y. Yoon, I.B. Kang, World’s first large size 77-inch transparent flexible OLED display. J. SID 26(5), 287–295 (2018)

Quantum Dot-Enabled Displays Charlie Hotz and Jeff Yurek

Abstract Quantum dot emitters (QDs) have become common in wide color gamut LCD displays using quantum dot enhancement film. Several new display architectures using QDs are also under development. QDs have the potential to impact many different future display designs including LCD backlight units as is in the marketplace today; pixel-based color conversion of OLED, LCD, or micro-LED technologies; or as electroluminescent emitters in “OLED-like” displays. This chapter will describe the structure, benefits, and development status of each type of QD display, including the challenges each one faces relative to competitive technologies. A comparison of the emitter properties of QDs and alternative display emitters will also be discussed.

1 Introduction 1.1 Commercial History In the fall of 2013, the Kindle Fire HDX 7 introduced the display industry to quantum dot enhancement film, one of the first commercial uses of quantum dots in displays [1]. Two years later, Samsung brought the first cadmium-free quantum dots to the premium television market [2, 3]. Since that time, quantum dot technology has grown steadily with over 100 unique models available today from most of the world’s display makers and analyst projections of more than 10 million unit sales in 2020 [4]. As a result, consumers can now (early 2020) experience the benefits of quantum dot technology for less than one third the cost of a comparably sized OLED set [5]. In addition to televisions, monitor makers have also begun adoption with a number of models serving the gaming and creative professional markets [6]. C. Hotz · J. Yurek (B) Nanosys Inc., Milpitas, CA 95035, USA e-mail: [email protected] C. Hotz e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_10

229

230

C. Hotz and J. Yurek

With QD enhancement film having become mainstream in televisions and monitors, display manufacturers are developing new implementations of quantum dots to further improve the performance and quality of displays. Along with QD film, these new implementations will be the subject of this chapter.

1.2 Quantum Dots as Emitters Quantum dots are nanometer-scale semiconductor particles that emit light with a narrow spectral shape and at a wavelength dependent on their size. These two properties make quantum dots an ideal material for use in displays: Narrow, neargaussian emission spectra means their emission is closer to monochromatic, leading to increased color gamut, and Wavelength tunability enables the emission peak to be moved over a considerable range allowing optimization of the wavelength for the display. Together these properties make quantum dot displays capable of reproducing larger color gamuts than competitive displays. Commercial QDs today are based on cadmium selenide (CdSe) or indium phosphide (InP, often referred to as Cd-free or heavy metal free QDs). CdSe QDs have narrower emission spectra than InP QDs and can achieve larger color gamuts in displays. Figure 1 shows typical wavelength ranges and spectral widths for both types of QDs used today. The broader spectra of InP compared to CdSe are apparent. Compared to other competitive emitter materials, QDs have a number of advantages. Most other emitters are considerably broader in their emission spectrum and

Fig. 1 Spectra of common display emitters. Blue (pump) emitter: InGaN LED; Green emitters: CdSe QD, InP QD, β-sialon phosphor; Yellow (combined red, green) emitter: YAG phosphor; Red emitters: CdSe QD, InP QD, KSF phosphor

Quantum Dot-Enabled Displays

231

have a fixed emission. Some common display emitters are shown in Fig. 1. It is important to note that some phosphors can be placed directly on an LED chip, enhancing the LED light output with the phosphor’s emission spectrum. This has not yet been broadly accomplished with QDs due to limitations at high flux excitation, although a product has been released for lighting using a red CdSe QD [7]. Another difference between phosphors and QDs is that the radiative lifetime (time duration between emission and excitation) is several orders of magnitude longer for phosphors. This may not impact performance unless rapid cycling of the phosphor creates color anomalies, such as when local dimming is used with phosphors of different radiative lifetimes [8]. The spectrum of KSF phosphor is shown in Fig. 1, and can be seen as a quite narrow red emitter. Like most emitters, its wavelength is fixed, so while it is a valuable emitter for DCI-P3 color gamut, it is not well matched for the BT.2020 gamut.

1.3 Benefit of QDs in Displays An objective of a color display is to reproduce all the colors of the natural world and do so at brightness levels that recreate visual experiences. To deliver such a viewing experience, there are two factors that need to be addressed. The first is high dynamic range (HDR) which requires sufficiently high peak brightness and sufficiently low black level in a scene such that it can convey the same visual experience that a viewer would perceive in real life. According to the research done by Dolby Labs, a typical peak brightness of 2000–4000 nits and a black level of 0.01nit are required to truly convey the intended visual experience to viewers [9]. We will see in subsequent sections that QDs enable higher brightness than existing emitters. The second factor is color reproduction. A color display should accurately reproduce all of the surface, or reflected colors that human eyes can see in nature. The display color gamut that can fully cover all these colors is called the BT.2020 color standard [10]. This suggests that for a display to truly reproduce the color experience that one has in real life, it needs to have very high coverage of the BT.2020 color gamut. QD displays today can deliver over 90% BT.2020 color gamut coverage. Figure 2 shows the color gamut of a 2019 quantum dot LCD monitor with greater than 90% coverage of the BT.2020 color gamut standard.

2 QD-Enhanced LCD Backlight Displays 2.1 Overview All displays produce color at each pixel by combining red, green, and blue subpixels. Existing LCD displays rely on a white light source to produce color by filtering white

232

C. Hotz and J. Yurek

Fig. 2 Color gamut of a 2019 model quantum dot LCD monitor and the BT.2020 color gamut standard, plotted in the CIE1931 color space. The 2019 quantum dot display covers greater than 90% of the BT.2020 color gamut

light at each subpixel to reproduce the unique colors called for by the content on display (Fig. 3). The color and intensity of each subpixel in an LCD display are a function of two factors; the spectrum of the white light from the backlight and the color filter transmission spectrum at the subpixel. The product of the white light spectrum and the transmission spectrum of the color filter produce the output of that subpixel.

Fig. 3 Schematic of an LCD display utilizing QD enhancement film

Quantum Dot-Enabled Displays

233

If the light source is weak in the component color or the color filter spectrum is narrow, the display will be dim. The use of broader spectrum filters increases the brightness, but at the cost of color gamut as the broader filters produce light that is further from monochromatic. QDs create a white light spectrum that is composed of three component colors (red, green, and blue) that are well matched to the color filters enabling the display to have both high brightness and gamut coverage. Typical LED light sources used in the LCD displays are broad; a common white display LED is based on YAG phosphor (yttrium–aluminum-garnet phosphor pumped by an InGaN blue source). When this light is filtered into the component RGB colors at the subpixels, either the gamut coverage is low, or the brightness is compromised.

2.2 Technology Status An early version of QD technology in LCDs used QDs in a capillary at the edge of the display. Immediately behind the QD/capillary were LEDs directing the QD-emitted light onto a light guide plate [11]. While the technical problems that accompanied this implementation (cracking, alignment) were ultimately addressed, the display market has moved entirely to QD film which is simpler to integrate and compatible with local dimming required for high dynamic range (HDR). Another implementation reported recently is for the QD film to be coated directly on a glass light guide plate, requiring encapsulation only on the top side [12, 13].

2.2.1

Quantum Dot Enhancement Film

QDs are incorporated into LCD displays today using QD enhancement film. The film is a three-layer laminate of a QD-thermoset resin matrix between two sheets of the optical film with a barrier layer to limit the ingress of oxygen and water vapor. The active layer is typically 50–100 um in thickness and 200–300 um for the entire laminate. These films are made at high volume in roll-to-roll coating lines and cut to the dimensions of the display. Both cadmium selenide and indium phosphide QD films are made similarly. The film contains red and green QDs and is designed at a concentration and film thickness to transmit a portion of the illuminating blue light and convert the remaining to red and green, thereby creating white light (Table 1). The first QDs to be incorporated into consumer displays were based on cadmium selenide QDs. Since cadmium is a regulated substance under RoHS [14], some manufacturers were unwilling to commercialize quantum dots despite an exemption stemming from the increased efficiency of these displays. This led to the development of cadmium-free indium phosphide QDs. Samsung’s SUHD line of televisions (currently branded QLED) was the first to use indium phosphide quantum dots. However, the emission spectra for these QDs are somewhat broader than for

234

C. Hotz and J. Yurek

Table 1 Properties of commercial QDs and BT.2020 gamut coverage for an LCD display using QDEF (Data from Nanosys Inc.) QD material system Color Cadmium selenide

Green 18 nm/23 nm Red

Indium phosphide

Emission FWHM BT.2020 gamut coverage using typical color (min/typ) filters (%)

Green 34 nm/39 nm Red

>90

20 nm/25 nm ~85

38 nm/40 nm

cadmium-based dots so these displays produce somewhat smaller gamut coverage, although they still cover nearly 100% of DCI-P3 gamut. Continuing improvements to QD stability enable QD film to be used with increasingly bright backlights leading to higher brightness displays. QD displays as high as 3,000 nits are available at the time of this review [15]. This improves the appearance of high dynamic range (HDR) content. Furthermore, these stability improvements have also been used to reduce the QD quantity needed and the packaging (barrier) requirements; both lowering the cost of quantum dot implementation.

2.3 Market for the Quantum Dot LCDs Today With performance levels acceptable for commercial displays, quantum dot manufacturers such as Nanosys and Hansol Chemical have installed capacity to produce enough QDs to supply millions of square meters of display area annually.

2.3.1

Market Segments

Quantum Dot technology has been widely deployed in most LCD market segments including mobile, tablet, notebook, monitor, and TV. Today, the TV market is by far the largest segment both in terms of unit volume and total display area. However, improvements in the performance, cost, and form factor of quantum dot components are now enabling the technology to gain traction in other segments. This section provides an overview of the benefits and challenges of quantum dot technology in each of these market segments.

Television Market The world’s first quantum dot TV was launched by Sony in early 2013 [16]. The set was praised for its vivid color reproduction, and other leading brands launched

Quantum Dot-Enabled Displays

235

quantum dot LCD sets of their own in the years that followed. However, rapid growth in the adoption of quantum dots did not begin until five years later, in late 2018. QD LCD TVs enjoyed a panel cost advantage of 37–52% compared to WOLEDs of the same size. Despite the lower cost of production, TV brands initially priced the technology into the premium segment (above $1,500), comparable to OLED [17]. This high pricing limited the initial market adoption of quantum dot technology by the mainstream market. Televisions priced above $2,000 make up just 1% of the market, while TVs priced below $500 comprise roughly 80% of all units sold in the US market [18]. Beginning in 2019, QD TVs began to enter the mainstream market with prices roughly equal to standard, non-QD, LCD sets [19]. An increasing number of brands including Vizio, Hisense, TCL, Konka, and others have joined Samsung in launching mainstream QD products in markets around the world. Today, consumers have a wide array of choices when shopping for a QD TV with SKUs priced from $379 to $9,999. A number of market factors contributed to the change in pricing strategy. These included wider availability of lower cost QD films, improvements in QD film performance that enabled cost reductions in other TV components, such as LEDs or brightness enhancing films, and increased competition in the LCD market as new Gen 10.5 fabs came online in China, driving LCD panel prices lower [20].

Gaming and Creative Monitors The brightness and color benefits offered by QD technology have been well received by the gaming and creative professional monitor segments. For gamers, quantum dots offer vivid color and fast response time. In the creative professional segment, the wavelength tunability of quantum dots enables monitor makers to precisely target color gamut specifications that are most important to their users such as DCI-P3 for movie makers and Adobe RGB 1998 for print graphics and photography. The creative professional segment remains a small but influential niche market while gaming monitors is one of the fastest growing monitor segments. According to industry analysts, gaming monitor unit sales grew 57% year over year in 2019 and are projected to ship over 11 million units in 2020 [21]. There are several quantum dot gaming monitor products on the market in 2020 from brands including Acer, Agon, Asus, HP, and Samsung.

Notebook The notebook segment is an area of emerging growth for quantum dots. Thinner quantum dot enhancement film has enabled display makers to begin integrating the technology into the very thin form factors that today’s notebook consumers expect.

236

C. Hotz and J. Yurek

Just as in the TV market, quantum dot enhancement film will offer LCD notebook makers an important technology differentiator as OLED technology begins to penetrate the market over the next few years [22]. The first mainstream quantum dot LCD notebooks were launched by Samsung in late 2019: Galaxy Book Flex and Galaxy Book Ion, with quantum dot LCD displays. According to Samsung, both laptops offer full screen brightness of 600 nits, about 50% more than a typical notebook, and can last up to 20 h on a single charge [23]. These significant performance and efficiency gains could lead to future rapid growth for quantum dot technology in the notebook segment.

Mobile/Tablet The mobile and tablet market has not been an area of growth of quantum dot technology to date. The Kindle Fire HDX 7, a 7 tablet launched in late 2013, was the first commercialized quantum dot enhancement film display product [24]. Since that time, a limited number of tablets and no smartphones have shipped with the technology. A key limiting factor for quantum dot technology in mobile/tablet devices has been the thickness of available QD enhancement films which have been in the range of 200–300 μm. Films thicker than 100 μmμm have proven challenging for mobile device makers to adopt. Recent advancements enabling thinner quantum dot enhancement films have been reported and could change this situation, opening the door to further growth in QDEF LCD tablets and phones [25].

2.3.2

Market Forecast

The market for quantum dot displays is expected to grow in all major segments over the next 5 years (Fig. 4). Analysts at DSCC forecast over 10 million displays, including televisions, monitors, notebooks, and tablets, will ship in 2020. That number is expected to grow to approximately 26 million units by 2025. The TV segment is projected to be the main driver of growth during this time period. Next generation QD technologies such as quantum dot-OLED (QD-OLED) hybrid displays are making progress toward commercialization and these will become the premium market revenue driver, especially for television brands. These emerging QD technologies offer new avenues for growth going forward.

3 QD Color Conversion While QD technology is established in the LCD market today, several new technology implementations are under development in order to further improve display characteristics. In this section, QD color conversion technology will be described, where the color conversion moves from the backlight to the front of the display.

Quantum Dot-Enabled Displays

237

Fig. 4 Recent market history and projections for QD display panels in the television, monitor, notebook, and tablet segments. Note that “QDEF” refers to QD enhancement film for LCD displays, “QDOG” refers to QD on glass light guide plate for LCD displays, and “QD-OLED” refers to the emerging use of QD as a color conversion layer for OLED displays. Chart courtesy of DSCC. (Source DSCC Annual Quantum Dot Display Technology and Market Outlook, April 2020)

3.1 Overview While display color performance and efficiency are improved through the use of QD film, these displays still rely on conventional LCD modules which are inherently inefficient. LCD backlights generate white light which the red, green, and blue subpixels filter to remove the undesired component colors. This design, while quite optimized, still leaves approximately two thirds of the light unused. To avert this inefficiency, a new implementation of QDs in displays can be used where the color conversion is moved from the backlight to the subpixel, termed quantum dot color conversion (QDCC) [26]. In this implementation, a thin film of quantum dots replaces the color filters in a conventional LCD module and generates light in the plane where the image is reproduced (See Fig. 5). In this design, the excitation light is blue only, which can be absorbed by both red and green QDs. If the blue light wavelength is appropriate for the desired color gamut, theblue subpixel can simply pass the blue light with no filter. This design removes the need for a color filter since, in each subpixel, only red or green light is emitted provided the blue excitation light is completely absorbed by the QD film. In addition to the potential for efficiency improvement, an LCD with QDCC layers has

238

C. Hotz and J. Yurek

Fig. 5 Typical QD color conversion structure (in this case using blue LED for excitation)

a wider viewing angle since the light generated by the QDCC layer is at the front of the display.

3.2 Technology Status 3.2.1

QD, Film, and Deposition Requirements

The QDCC approach necessitates several physical and optical requirements beyond what is required for the QD film discussed in the previous section. As a color filter replacement in LCD displays, the QDCC layer must be thin (on the order of six to fifteen microns) to be compatible with current technology. This is significantly thinner than the ca. 100 μm films used for QD enhancement film. Furthermore, the QDCC layer must absorb virtually all (>99%) of the blue light used for excitation which was not necessary for QD film. It is also necessary that the quantum efficiency be very high to avoid reabsorption losses in the high absorbing films. Since the film absorption must be very high, the loading of QDs must be high as well. In order to comply with the RoHS limitation requiring that any homogenous layer of the system contain less than 100 parts per million of cadmium, the QDCC layer must be made from completely cadmium-free quantum dots (indium phosphide typically). Additionally, the QDCC layer must be patterned at the pitch of the subpixel, either by a photolithography process or through inkjet printing. Photolithography can produce much smaller features than inkjet printing (sub-micron versus ~50 μm), although inkjet printing is generally more material efficient. In either patterning approach, the quantum dots will need to be compatible with the process, unimpacted by the various thermal and chemical processing steps that are needed in photolithography (UV exposure, development, baking), or printing (bake steps, jetting). It is also expected that the film optical properties do not change during processing, whether by inkjet printing or photolithography. Most development efforts today for large displays involve QD inks for inkjet printing. It is useful to note that 65 televisions at 4 K or 8 K resolution have minimum subpixel dimensions in the 50–100 μm range— suitable for inkjet print technology. Inkjet printed QDCC film demonstrations have been reported by several groups [27, 28], one of which is shown in Fig. 7. Numerous

Quantum Dot-Enabled Displays

239

demonstrations have been made using QD photoresist [29, 30]. Currently, Nanosys and Sumitomo Chemical are supplying red and green QD photoresist to customers [31].

3.2.2

Optical Requirements

In terms of optical properties, the QDCC layer must produce a highly efficient, single color light output. The figure of merit for this is photon conversion efficiency (PCE), which is the ratio of emitted red or green photons in the forward direction to the number of incident blue photons. A minimum PCE is expected to be approaching 40% for this application, as measured in a one-pass setup. Additionally, the layer must absorb virtually all of the incident blue excitation light, as any leakage of blue light into the red or green channels results in a decreased color gamut. Expected film performance is >99% blue absorption, although it is possible to achieve this using some additional optical filtering, albeit with some reduction of PCE [32].

Impact on Color Gamut The requirement for high blue absorption of the QDCC layer is due to the out-ofchannel light contamination caused by the unabsorbed blue light in the red or green subpixel. Even a small amount of blue light contaminating the red or green channels creates a significant reduction in the color gamut. With optimized peak emission wavelengths, a display with QDCC layers can achieve greater than 95% BT.2020 color gamut coverage if all the blue excitation light is absorbed. However, with 1% of the blue light leaking through each red and green layer, the BT.2020 color gamut coverage is reduced to 86% (see Fig. 6). The change in the color point is especially pronounced for the red color conversion layer. Although the green color conversion layer can tolerate more blue light leakage while maintaining high color saturation, the blue light leakage still must be under 1% for a display using QDCC layers to achieve color gamut coverages at least as high as conventional LCD displays with QD enhancement film. The maximum absorption of blue light in a film is a function of the film thickness, the QD and scattering agent loading, and the inherent properties of the QDs [33]. Since the film thickness is determined by the application, the effort is mainly to maximize the QD and scattering agent loading while maintaining other film physical requirements. Generally, red QDs being physically larger have significantly higher inherent absorption of blue light than green QDs. In particular, indium phosphide green has low inherent absorption at 450 nm (the wavelength of blue excitation typically). Efforts are underway to develop new green-emitting materials with improved absorption. The maximum concentration of QDs in a film is limited by several factors. For QD ink formulations, the concentration of quantum dots has a strong impact on the

240

C. Hotz and J. Yurek

Fig. 6 Impact of blue light leakage on the color gamut. For films with 100% blue light absorption, BT.2020 coverage is high (>95%). When 1% blue light is transmitted in both red and green channels, coverage is reduced significantly (to 86%). Note the larger impact on the red primary

Fig. 7 Picture of a white RGB array with inkjet printed QDCC film. Subpixel dimensions are 280 × 80 μm

viscosity of the ink. This will affect the compatibility with existing inkjet nozzles, printing equipment, and processing techniques. QDs strongly absorb ultraviolet light, so the curing properties of UV-curable ink are also affected at high quantum dot concentrations. Quantum dot photoresists also have the same UV-curing constraints. In addition, high quantum dot concentrations can adversely affect the development and pattern formation, thereby necessitating extra efforts to prevent undercutting and ensure vertical sidewalls in the patterned films. Nanosys and collaborators have demonstrated patterned QDCC layers using both QDPR and inkjet printing [29]. Figure 7 shows an RGB printed array with 280 μm ×

Quantum Dot-Enabled Displays

241

80 μm subpixels demonstrated by Nanosys and ink maker DIC [34]. The green and red subpixels contain thermal cure quantum dot ink while the blue subpixel contains scattering media and resin only.

3.3 Applications 3.3.1

LCD Displays

In order for the QDCC technology to be suitable for LCD displays, the QD cannot simply replace the color filter in the existing LCD design. This is due to the need for polarization to be maintained for the LC array to function. QDs are isotropic emitters and therefore depolarize coming from the LC array. The solution is to move the polarizer “in-cell”, or under the QD layer. This has been demonstrated for small area displays, but so far is not manufacturable for monitors or televisions [35].

3.3.2

Blue OLED Displays

QDCC layers may be better suited to technologies other than LCD. In principle, any blue light source can be used to excite QDs including blue OLEDs. This type of hybrid display combines the benefits of electroluminescence (low black levels and wide viewing angles) with QD color performance. In order for OLED-based displays to use QDCC technology, new blue-emitting OLED materials may be required. Currently, blue OLED emitters have the lowest efficiency and shortest lifetime among all colors [36]. Despite this, Samsung Display Corp. has announced that they plan to invest $11 billion to bring a display to market in 2021 based on QDCC using a blue OLED light source [37].

3.3.3

Micro-LED Displays

Compared to OLED blue emitters, inorganic LED sources are highly efficient and significantly more stable. Thus, the combination of QDCC layers and a blue microLED array could be a powerful combination for display applications. Using QDCC layers eliminates the need for separate red and green LEDs to make afull color display, which is one of the major technical challenges for micro-LED displays [38]. A limitation on their use in micro-LEDs is QD degradation caused by the high local flux of very small micro-LEDs. A recent report suggests that Samsung Display Corp. is considering blue gallium nitride (GaN) nanorod LED as an improved blue emitter for future devices based on the QD-OLED architecture that is widely expected to be commercialized in 2021 [39, 40]. The technology, called Quantum Nano Emitting Diode or QNED, is reported

242

C. Hotz and J. Yurek

to have significant advantages over blue OLED excitation. Aside from higher efficiency and brightness compared to blue OLED, the QNED approach also provides a longer lifetime and elimination of burn-in associated with OLED displays. Improved environmental stability over OLED may also remove the need for an encapsulation layer. The increased surface area of the nanorod array results in an increased area compared to a planar electrode and the increased brightness. The GaN nanorod technology, while under development for over 10 years, may soon be found in displays [41]. In summary, quantum dot color conversion layers offer the potential for high efficiency, better color, and low-cost implementation for LCD, OLED, and microLED displays. Improvements in both the optical properties and stability of cadmiumfree quantum dots along with advancements in the process technology have made QDCC layers very close to commercialization.

4 Electroluminescent QD Displays QD electroluminescent (QDEL) displays have the potential to revolutionize the display industry in near future by improving on some of the shortcomings of OLED displays.

4.1 Overview The first report of QD electroluminescence was made in 1994 [42]. Since then, considerable effort has been made by numerous groups to improve the performance of QDEL devices toward commercial performance levels. White OLED TV, first commercially released by LG in 2013, represents the benchmark for performance for this type of emissive display. QDEL devices have not yet matched this level of performance. It is worth noting that unlike QD color conversion materials, the effort is being made to develop both CdSe- and InP-based devices. QDEL devices are thin-film stacks similar to OLED devices. Specific published structures vary, but most report using a device structure similar to the one shown in Fig. 8. As shown in Fig. 8, the active device layers are only 10 s of nanometers in thickness. A significant key difference between OLED and QDEL devices is that OLED emitters are deposited by evaporation, which is not possible for QDs as they must be solution processed. Most laboratory QDEL development work is based on the spin-coat deposition; however, a number of publications have demonstrated functional devices made by printing, including inkjet [43], transfer printing [44], and 3D printing [45]. While QDEL devices are still being developed, some of their presumed benefits over OLED devices include the following: Lower operating voltage which leads to

Quantum Dot-Enabled Displays

243

Fig. 8 Typical QDEL device stack. On left is a schematic for the FIB-TEM image on the right

lower power consumption and possibly longer lifetime; Larger color gamut and color volume due to the QDs narrower emission; Brighter emission.

4.2 Status While commercial devices have numerous requirements, at the development level, there are two key figures of merit for QDEL devices: Efficiency represented by external quantum efficiency (EQE) and lifetime (LT95). Like the analogous OLED devices, efficiency (EQE) is expected to be similar (approximately 20%, varies with color), with a LT95 at display brightness (also varies with color) of approximately 10 k hours. Currently, reported efficiencies for all three colors are close to meeting commercial levels, as shown in Fig. 9. With regard to LT95, reports indicate that red and green lifetimes are acceptable for CdSe devices, but behind for blue. A recent report describes an InP red device with an LT95 of 615 h at 1000 nits; [46] however, no reports describe lifetimes near this level for Cd-free green or blue.

4.3 Efficiency The external quantum efficiency (EQE) of an EL device can be expressed by the following equation: η E Q E = η I Q E ηoc = γ ηs η P L ηoc

244

C. Hotz and J. Yurek

Fig. 9 Historical data showing efficiency and lifetime improvement

where η I Q E is the internal quantum efficiency (IQE), γ is the charge balance factor, ηs is the percentage of excitons that contribute to the radiative process dictated by their spin states, η P L is the photoluminescence quantum efficiency, and ηoc is the outcoupling efficiency. In fluorescent OLEDs, ηs is theoretically limited at 25% although could be higher with the help of the triplet–triplet annihilation process. In QD-LEDs, the energy split between different QD energy levels is very small due to the effect of intrinsic crystal field and crystalline lattice asymmetry in addition to the effect of the electron–hole exchange interaction, and can be easily overcome by thermal fluctuation [47]. As a result, ηs is not a limiting factor for QD-LEDs. The efficiency calculation can be simplified to η E Q E = γ η P L ηoc Note η E Q E only addresses electron to photon conversion efficiency or current efficiency. To consider power efficiency, the voltage needs to be included in the calculation. Outcoupling is quite similar between QD-LED and OLED and will not be discussed here.

4.3.1

Charge Balance (γ )

To better understand the operation of QD-LEDs, it is important to note that there is no host material used in the EML layer. QD materials are not only the emitter, but also the charge transport material in the EML layer. This unique property is responsible for the low drive voltage, and thus high power efficiency, of QD-LEDs. However, this makes it more challenging to balance the charges in the EML layer. Some of the effective charge balance and exciton management techniques used in OLEDs, such as mixed hosts, cannot be applied to the QD-LED system.

Quantum Dot-Enabled Displays

245

In the normal QD-LED structure, the range of available solution processable HTL materials is very limited. TFB is one of the most widely used HTL materials in literature because of its high mobility. However, its HOMO level is shallow (~5.3 eV), making it problematic for hole injection into QDs. Cao et al. compared red CdSe QDs with different shell compositions and found the inject barriers were 0.32 ± 0.05 and 0.08 ± 0.02 eV for ZnS and ZnSe shells, respectively [48]. With hole injection barrier significantly reduced, devices using QDs with ZnSe shell showed a T95 operation lifetime of more than 2300 h with an initial luminance of 1000 cd/m2 , and an equivalent T50 lifetime of >2,200,000 h at 100 cd/m2 initial luminance. Lim et al. proposed a different way to achieve charge balance [49]. Instead of reducing injection barriers, they added a wide bandgap outer shell which slowed down charge flows of both electrons and holes. The slowing down was much more effective on electrons than holes, leading to a better-balanced system. This team demonstrated QD-LEDs operating with high internal efficiencies (up to 70%) virtually droop-free up to impressive brightness of >100,000 cd/m2 (at ~ 500 mA/cm2 ).

4.3.2

PL Efficiency (η P L )

Solution quantum yield (QY) is used to track the photoluminescent efficiency of QDs in solution. CdSe-based quantum dots have been well optimized over the years. In the last couple of years, the most interesting results are around Cd-free QD development. Through stoichiometry control, Li et al. synthesized InP/ZnSe/ZnS QDs with nearunity QY, mono-exponential decay dynamics, narrow line width, and nonblinking at a single dot level. Devices made with these dots achieved 12.2% peak EQE and maximum brightness of >10,000 cd/m2 [50]. Won et al. reported QD developed for EL application with QY of approximately 100%. The techniques applied include adding hydrofluoric acid to etch out oxidative InP core surface during the growth of the initial ZnSe shell and high-temperature ZnSe growth at 340 C. High solution QY does not always lead to high device EQE. When QDs are closely packed in EL devices, film QY is a more accurate indicator for device EQE. This is demonstrated by Ippen et al. in their study on the shell thickness effect in red devices using InP-based QDs [51]. Although five QDs with different shell thicknesses showed similar solution QYs, the devices showed dramatic different EQE performance, as shown in Fig. 10. EQE tracked well with film QY, indicating film QY is the limiting factor in these devices. When quantum dots are closely packed in films, new quenching channels form. The first one is due to the Forster resonance energy transfer (FRET)—the energy of an exciton can be transferred to a nearby non-luminescent QD. This quenching effect depends on the solution QY and also the energy transfer efficiency. As Fig. 10 shows, increasing shell thickness is an effective way to reduce FRET efficiency and thus improve film QY. When the QD film is deposited next to CTL layers, two other quenching pathways are formed: (1) excitons in QDs energy transfer to CTL materials and (2) excitons disassociate and charges transfer to CTLs. During device operation, the high electrical field can also break up excitons leading to field-induced

246

C. Hotz and J. Yurek

Fig. 10 QD solution QY, QD film QY, and device EQE versus QD shell thickness

exciton disassociation. Finally, whenever there are unpaired charges in QD, caused by either charge transfer with CTL or exciton disassociation, trions are formed which can lead to the Auger process.

4.4 Lifetime An unbalanced charge has a consequence in lifetime. Chang et al. studied operational instability of QD-LEDs with inverted structure and discovered two mechanisms as a result of unbalanced charge: (1) nonradiative Auger recombination process due to accumulation of excess electrons in QD layer and (2) electron leakage toward HTL that leads to irreversible physical damage to the HTL [52]. Chen et al. studied red and blue CdSe QDs to understand the degradation mechanism [53]. They found that the lifetime of red devices is primarily limited by the slow degradation of HTL materials due to the hole injection barrier. However, the conduction band level is much shallower in blue devices, making the electron injection relatively more problematic. As a result, QD-ETL junction degrades quickly in blue devices. Unlike photoluminescent QD applications, in QDEL devices, electrons and holes have to be injected into the QD to form excitons. During this process, QD surface and ligands experience charge flow. Pu et al. studied different ligand systems and pointed out that the operando electrochemical reactions of surface ligands with injected charge carriers are an important degradation channel in EL devices. After applying electrochemically inert ligands to quantum dots, the device operational lifetimes were significantly boosted for both red-emitting light-emitting diodes (T95 > 3800 h at 1000 cd/m2 ) and blue-emitting light-emitting diodes (T50 > 10,000 h at 100 cd/m2 ) [54].

Quantum Dot-Enabled Displays

247

4.5 Challenges While considerable effort over many years has been made to reveal numerous mechanisms of degradation in OLED devices [55], it is not clear what degradation mechanism is dominant in QDEL devices. As mentioned earlier, a significant difference is that QDs are printed (as they cannot be evaporated), so solvents and drying are necessary components of material deposition. Additionally, QDs are not analytically pure materials (QDs are distributions of similar materials), nor can they be purified in the same manner as small (OLED) molecules, so it is likely impurities play a role in the degradation. Despite these challenges, numerous groups continue to develop and publish improvements to these devices and the QDs used to make them. With their highly saturated color, simple device structure, and low-cost printing and patterning process, QD-LEDs have great potential to become the next generation display technology.

5 Conclusions and Outlook Since its appearance in the display market in 2013, QD enhancement film has made significant penetration in the television market. It is clear that this technology will continue to grow for the foreseeable future and possibly become the standard architecture in the LCD market. QD color conversion is likely to become commercial in 2021, as a higher performance display moving QD technology beyond LCD. The most advanced form of QD display, QDEL display, must overcome lifetime limitations to reach commercial viability; however, continual progress is being made Acknowledgements The authors would like to thank Drs. R. Ma and R. Tangirala for their assistance in the preparation of portions of this manuscript; as well as the entire technical team at Nanosys for their contributions to the technology reported here. Nanosys would also like to thank all of its collaborators who have contributed to the development of products based on QD technology.

References 1. J. Chen, V. Hardev, J. Hartlove, J. Hofler, E. Lee, A high-efficiency wide-color-gamut solid-state backlight system for LCDs using quantum dot enhancement film, in SID Symposium Digest of Technical Papers, vol. 43 (2012), pp. 895–896 2. E. Jang, Environmentally friendly quantum dots for display applications, in 2018 IEEE International Electron Devices Meeting (IEDM) (San Francisco, CA, 2018), pp. 38.2.1–38.2.4 3. T. Moynihan, What are quantum dots, and why do i want them in my TV? Wired. https://www. wired.com/2015/01/primer-quantum-dot/ 4. Nanosys Inc. Press Release, Nanosys on track for record shipments of quantum dots for displays in 2019 (2019). https://www.prnewswire.com/news-releases/nanosys-on-track-for-record-shi pments-of-quantum-dots-for-displays-in-2019-300937397.html

248

C. Hotz and J. Yurek

5. Walmart.com, 2019 LG 65C9PUA OLED = $2,196.99, 2019 Vizio M658-G1 = $598.00 https://www.walmart.com/browse/electronics/tvs/3944_1060825_447913. Accessed 16 April 2020 6. Wired Staff, The best gaming monitors for any budget in 2020. Wired (2020). https://www. wired.co.uk/article/best-gaming-monitors 7. Osram Corp. Press Release (2019). https://www.osram.us/cb/press/press-releases/05_20_19_ quantum_dots.jsp 8. H.F. Sijbom, R. Verstraete, J.J. Joos, D. Poelman, P.F. Smet, K2 SiF6 :Mn4+ as a red phosphor for displays and warm-white LEDs: a review of properties and perspectives. Opt. Mater. Express 7, 3332–3365 (2017) 9. S. Farrell, T. Kunkel, S. Daly, A cinema luminance range by the people, for the people: viewer preferences on luminance limits for a large screen environment, SMPTE 2014 Annual Technical Conference & Exhibition. Hollywood, CA, USA 2014, 1–9 (2014) 10. K. Masaoka, Y. Nishida, M. Sugawara, Designing display primaries with currently available light sources for UHDTV wide-gamut system colorimetry Opt. Express 22, 19069–19077 (2014) 11. K. Bullis, Quantum dots get commercial debut in more colorful TVs. MIT Technol. Rev. 11 (2013) 12. HP Pavilion Quantum Dot Display datasheet (2019). https://www8.hp.com/h20195/V2/Get PDF.aspx/c06257847 13. B. O’Brien, Display daily (2018). https://www.displaydaily.com/article/display-daily/willqdog-have-its-day 14. Öko-Institut eV, Joint revaluation of two requests for exemption, first reviewed in 2013–2014, related to cadmium quantum dot applications (2014). https://rohs.exemptions.oeko.info/index. php?id=265 15. C. Welsh, Vizio’s 2020 lineup includes its biggest TV yet and first-ever OLED. The Verge (2020). https://www.theverge.com/2020/1/5/21049756/vizio-new-4k-tvs-oled-p-seriesquantum-x-ces-2020. Accessed 16 April 2020. 16. J. Blagdon, Sony’s new Triluminos TVs pursue vibrant hues with quantum dots. The Verge (2013). https://www.theverge.com/2013/1/16/3881546/sonys-new-triluminous-tvs-pursue-vib rant-hues-with-quantum-dots. Accessed 16 April 2020. 17. Display Daily, High cost of WOLED TV panels detailed in DSCC TV cost report (2019). https://www.displaydaily.com/paid-news/ldm/ldm-market-news/high-cost-of-woledtv-panels-detailed-in-dscc-tv-cost-report. Accessed 16 April 2020 18. The NPD Group, U.S. retail tracking service, LCD TVs (2018) 19. IHS Markit, Inc. Press Release (2018). https://news.ihsmarkit.com/prviewer/release_only/slug/ technology-8k-tv-shipments-reach-more-400000-units-2019-ihs-markit-says 20. IHS Markit, Inc. Press Release (2017). https://news.ihsmarkit.com/prviewer/release_only/slug/ technology-gen-10-and-larger-flat-panel-display-capacity-grow-59-percent-cagr-2022-ihs 21. Trend Force Corp. Press Release (2020). https://press.trendforce.com/node/view/3324.html 22. B. O’Brien, Record OLED panel revenues of $28 Billion in 2019, will grow to $50 Billion by 2024. DSCC (2020). https://www.displaysupplychain.com/blog/record-oled-panel-revenuesof-28-billion-in-2019-will-grow-to-50-billion-by-2024. Accessed 16 April 2020 23. C. Gartenberg, Samsung’s new galaxy book flex and ion laptops bring better designs to the lineup. The Verge (2019). https://www.theverge.com/circuitbreaker/2019/10/29/20938145/ samsung-galaxy-book-flex-ion-laptops-design-wireless-charging-qled-hands-on. Accessed 16 April 2020 24. R. Soneira, Mini tablet display technology shoot-out. Display Mate (2020). https://www.dis playmate.com/Tablet_ShootOut_4.htm. Accessed 16–4–2020. 25. H. Toyama, T. Oba, A. Watano, N. Chikushi, T. Yonemoto, S. Hikita, H. Nishikawa, Y. Suga, Y. Ito, 41–2: novel thinnest free form QD film with honeycomb structure, in SID Symposium Digest of Technical Papers, vol. 49 (2018), pp. 518–521. https://doi.org/10.1002/sdtp.12449 26. E. Lee et al., Quantum dot conversion layers through inkjet printing, in SID Symposium Digest of Technical Papers, vol. 49, no. 1 (2018), pp. 525–527

Quantum Dot-Enabled Displays

249

27. R. Tangirala et al., Displays using quantum dot color conversion by inkjet printing of quantum dot inks, in IDW Symposium (Japan, 2017) 28. Z. Hu et al., Inkjet printed uniform quantum dots as color conversion for active matrix microLED displays, in SID Display Week (2019) 29. E. Lee, Ambient processing of quantum dot photoresist for emissive displays, in SID 2017 Digest (2017), p. 984 30. H. Kim et al., Enhancement of optical efficiency in white OLED display using the patterned photoresist film dispersed with quantum dot nanocrystals. J. Disp. Technol. 12(6), 526–531 (2016) 31. R. Tangirala et al., Color conversion using quantum dots for LCD, OLED and MicroLED displays, in SID Display Week 2020 (San Jose, CA, 2021) 32. C. Yang et al., High color purity emission from quantum dots by optimizing the full width at half the maximum and optical density to blue light. J. Soc. Inf. Disp. 2019, 1–6 (2019) 33. J. Osinski, P. Palomaki, Quantum dot design criteria for color conversion in microLED displays, in Society for Information Display International Symposium Digest of Technical Papers, vol. 50, no. 1 (2019), pp. 34–37 34. E. Lee. (2018) Quantum dot conversion layers through inkjet printing, in SID Display Week 2018 Presentation (2018) 35. C. Huang et al., A scheme to manufacture a high-color-purity quantum-dot display, in SID Display Week (San Jose, CA, 2020) 36. S. Kim et al., Degradation of blue-phosphorescent organic light-emitting devices involves exciton-induced generation of polaron pair within emitting layers. Nat. Commun. (2018) 37. R. Mertens, QD-OLED. OLED-info. https://www.oled-info.com/qd-oled. Accessed 16 April 2020 38. A. Paranjpe, J. Montgomery, S. Lee, C. Morath, Micro-LED displays: key manufacturing challenges and solutions, in Society for Information Display International Symposium Digest of Technical Papers, vol. 49, no. 1 (2018), 597–600 39. S. Harikrishna-Mohan (2020) Are quantum nano emitting diodes (QNEDs) the next big thing?, in Display Daily. https://www.displaydaily.com/article/display-daily/are-quantum-nano-emi tting-diodes-qneds-the-next-big-thing 40. Z. Liu, C. Lin, B. Hyun et al., Micro-light-emitting diodes with quantum dots in display technology. Light Sci. Appl. 9, 83 (2009). https://doi.org/10.1038/s41377-020-0268-1 41. H. Park et al., A facile method for highly uniform GaN-based nanorod light-emitting diodes with InGaN/GaN multi-quantum-wells. Opt. Express 21(10) (2013). DOI:https://doi.org/10. 1364/OE.21.012908 42. V. Colvin, M. Schlamp, A. Alivisatos, Light emitting diodes made from cadmium selenide nanocrystals and a semiconducting polymer. Nature 370, 354–357 (1994) 43. C. Jiang et al., Coffee-ring-free quantum dot thin film using inkjet printing from a mixedsolvent system on modified ZnO transport layer for light-emitting devices. ACS Appl. Mater. Interfaces 8, 39 (2016) 44. L. Kim et al., Contact printing of quantum dot light-emitting devices. Nano Lett. 8(12), 4513– 4517 (2008) 45. Y. Kong et al., Correction to 3D printed quantum dot light-emitting diodes. Nano Lett. 19, 2187 (2019) 46. Y. Won et al., Highly efficient and stable InP/ZnSe/ZnS quantum dot light-emitting diodes. Nature 575, 634–638 (2019) 47. L.E. Al, Luminescence polarization of CdSe microcrystals. Phya. Rev. B 46, 7448 (1992) 48. W. Cao et al., Highly stable QLEDs with improved hole injection via quantum dot structure tailoring. Nat. Commun. 9, 2608 (2018) 49. J. Lim et al., Droop-free colloidal quantum dot light emitting diodes. Nano Lett. 18(10), 6645 (2018) 50. Y. Li et al., Stoichiometry-controlled InP-based quantum dots: synthesis, photoluminescence, and electroluminescence. J. Am. Chem. Soc. 141(16), 6448–6452 (2019)

250

C. Hotz and J. Yurek

51. C. Ippen et al., High efficiency heavy metal free QD-LEDs for next generation displays. J. Soc. Inf. Disp. 1–9 (2013) 52. J. Chang et al., Unraveling the origin of operational instability of quantum dot based lightemitting diodes. ACS Nano 12, 10231 (2018) 53. S. Chen et al., On the degradation mechanisms of quantum-dot light-emitting diodes. Nat Commun. 10, 765 (2019) 54. C. Pu et al., Electrochemically-stable ligands bridge the photoluminescenceelectroluminescence gap of quantum dots. Nat Commun. 11, 93 (2020) 55. Z. Yang et al., Recent advances in quantum dot-based light-emitting devices: challenges and possible solutions. Mater. Today 24, 69–93 (2019)

Electroluminescence Devices with Colloidal Quantum Dots Seunghyun Rhee, Jeong Woo Park, and Wan Ki Bae

Abstract Colloidal quantum dots (QDs) are a few nanometer-sized semiconductor nanocrystals whose electronic states are subject to change depending on their dimension. QDs have been of great interest as light-emitting materials in future displays owing to their superb optical properties such as near-unity photoluminescence quantum yield and narrow emission spectra, as well as their solution processability. The present chapter focuses on the emerging display technologies based on quantum dots. Specifically, this chapter covers the brief introduction of quantum dots for lightemitting applications, photophysical properties of quantum dots relevant to lightemitting applications, and the state-of-the-art of, and the perspectives on QD-based display technologies.

1 Introduction of Quantum Dot Based Display Applications In modern society, individuals acquire most information through displays installed in TVs, computers, smartphones, and tablets in everyday life. The usage of display applications has continued to expand its territory to outdoor applications and portable displays, and the growing needs obligate lightweight display devices equipped with brighter and more vivid images. Colloidal semiconductor nanocrystals, referred to as quantum dots (QDs), have been regarded as the most promising emissive materials for next generation displays due to their superb optical properties, such as broad absorption but narrow emission bandwidth, tunable bandgap over the entire visible region, and near-unity luminescence efficiency [1–8]. Benefited from notable optical S. Rhee · J. W. Park · W. K. Bae (B) SKKU Advanced Institute of Nanotechnology (SAINT), Sungkyunkwan University, Suwon-si, Gyeonggi-do 16419, Korea e-mail: [email protected] S. Rhee e-mail: [email protected] J. W. Park e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_11

251

252

S. Rhee et al.

Fig. 1 Schematics illustrating the architectures of QD-LCD (QDEF), QD-LCD (QDCF), QDBOLED, and QD-LED display technologies

characteristics as well as the cost-effective solution processability, QDs succeeded to enter the commercial display products as down conversion luminophores [9, 10] in back-lit units of liquid crystal displays (QD-LCDs) and on the front glass of blue organic light-emitting diodes (QD-BOLEDs), and are now evolving to fit into the emissive-displays (QD-LEDs) (Fig. 1). Quantum down converters, such as quantum dot enhancement film (QDEF) and quantum dot color filter (QDCF), provides a successful improvement in the color gamut from 70% of the NTSC (National television system committee) to 110% of the NTSC standard without compromising current LCD manufacturing processes [11]. Various display companies (e.g. Samsung, Hisense, TCL, and Xiaomi) have produced QDEF-enhanced LCD displays. However, due to the low contrast ratio and difficulty on flexibility form factor issues in LCD, they are challenging to next generation display as QD-enhanced OLED display and toward QD electroluminescence (QDLED) display, which is a longstanding goal in the quantum dot display field. QD-LEDs are electroluminescence devices, in which charge carriers (holes and electrons) injected from two opposite electrodes recombine within QDs to generate photons. Due to the simple device structure and outstanding luminescence properties, QD-LEDs are expected to have better potential compared to other QD PL-based displays. Since QD-LEDs recruit the device architecture and materials (charge transport materials and electrodes) that are explored and proven in OLEDs, QD-LEDs are also awarded with advantageous features of OLEDs, such as thinness, lightweight, and flexibility [6, 12, 13]. In addition, QDs hold high color purity for all three primary colors, promising the realization of full-color displays with a wide color gamut meeting the BT2020 standard [7, 14, 15]. Furthermore, inorganic QDs are capable of being processed via conventional photo-patterning [16, 17] or inkjet printing [18, 19] methods, and thus are suited to realize high-resolution displays with various color pixels. Since the first demonstration by Colvin et al. [1], the last two decades have witnessed magnificent progress in the performance of QD-LEDs. The external quantum efficiency (EQE) close to the theoretical limit (20%) has been achieved in primary-colored QD-LEDs [5, 8, 20, 21] and the full width at half maximum (FWHM) reaches below 30 nm [15, 22] without application of the optical cavity

Electroluminescence Devices with Colloidal Quantum Dots

253

structure. Simultaneously, the luminescence exceeding 105 cd/m2 has been successfully realized with QD-LEDs [4, 8, 23, 24], which promises the expansion of their end applications to AR/VR, lightings, and lasers. This chapter will overview the progress made in QD-LEDs and explain the operation and degradation mechanism in the-state-of-the-art devices. As the device performance is governed by the dynamics of charge carriers in QDs, the photophysical properties of QDs will be also supported. Finally, we will address the remaining challenges that need to be overcome for the commercialization of QD-LEDs.

2 QD-Based Electroluminescence Device (QD-LED) 2.1 General Description of QD-LED QD-LEDs consist of the p-i-n heterojunction structure, in which the emission material layer (EML, QDs) is sandwiched by (p-type and n-type) charge transport layers (CTLs) and electrodes (Fig. 2). The state-of-the-art QD-LEDs are constructed in a multilayer diode structure of anode/hole injection layer (HIL)/hole transport layer (HTL)/QD EML/electron transport layer (ETL)/electron injection layer (EIL)/cathode, whereby QDs recruit charge carriers injected from electrodes to generate photons (Fig. 2b) [4, 24–28]. The structure of QD-LEDs has considerable similarity with that of OLEDs, particularly in the materials and the device architecture, except the emission material layer. In contrast to OLEDs having 40–100 nm thick EML of thermally evaporated organic molecules, QD-LEDs employ a thin compact inorganic QD layer (less than 2–3 monolayers) [6, 29, 30] prepared via solution-based processing methods (spin-casting or inkjet printing) (Fig. 2c). The difference in the EML processing methods inevitably constraints the choice of the materials for CTLs under EMLs, and hence the state-of-the-art QD-LEDs adopt inorganic charge transport layers or conducting polymers that hold structural robustness against the QD EML deposition [24, 31–34]. QDs can be optically or electrically excited, and the emitted luminescence is called photoluminescence (PL) and electroluminescence (EL), respectively. While optical excitations only produce excitons of singlet states, whose optical transition to the ground singlet state (S0 ) is allowed, the electrical excitations yield 25% of singlet states and 75% of triplet states, where the selection rule forbids triplet states to pair with ground singlet states (Fig. 3) [35]. Typical organic molecules have a large energy offset between the lowest excited states for singlet (S1 ) and triplet states (T1 ) (typically greater than 100 meV), and thus the intersystem crossing between these excited states is not allowed at room temperature (Fig. 3a) [36]. Therefore, finding a way to generate photons from this 75% of forbidden transition triplet states has been of utmost importance in the OLED community (Fig. 3b). However, in QDLEDs, such spin-forbidden transition can be relaxed by efficient intersystem crossing between the lowest excited states for singlet (S1 , bright exciton state) and triplet (T1 ,

254

S. Rhee et al.

Fig. 2 a TEM image of nanometer semiconductor QD. Inset: illustration of the structure of QD. b Schematic illustration of QD-LED. c A cross-sectional TEM of QD-LED image. d Schematic illustration of the energy band diagram of QD-LED

Fig. 3 Energy band diagrams for a typical organic molecules and b colloidal QD. kISC , rate of intersystem crossing; kRISC , rate of reverse intersystem crossing; S1 , singlet excited state; T1 , triplet excited state; S0 and GS, ground states; EST , singlet-triplet splitting; kB , Boltzmann constant; T, temperature

Electroluminescence Devices with Colloidal Quantum Dots

255

dark exciton state) with an energy offset less than kT (Thermal energy, 25.9 meV at 300 K) (Fig. 3c) [37–40]. As a result, QD-LEDs can achieve theoretically 100% of electrically pumped exciton to photon conversion efficiency at room temperature. QD-LEDs generate EL from electrically injected charge carriers into the QD EML. When QD-LEDs are electrically operated by forwarding bias, electrons and holes are injected from anode and cathode and transported into the QD EML passing through HIL/HTL and EIL/ETL, respectively. The injected carriers form excitons (electronhole pairs bound by the electrostatic Coulomb force) that recombine to emit photons via the radiative recombination process. As attested in p-i-n heterojunction diodes, the device performance of QD-LED is decided by (i) charge injection/transport processes and (ii) charge to photon conversion processes in QDs. Since QD-LEDs recruit CTLs and electrodes whose performance and stability are verified in OLEDs, directing the charge injection from CTLs into QDs and the charge to photon conversion processes in QDs within the working QD-LEDs in efficient ways are critical to the device characteristics of QD-LEDs. For QD-LEDs with the absence of trap sites for charge carriers, the device efficiency is directly linked to the luminescence efficiency of QDs in working devices, which in principle deals with the fate of electrically injected charge carriers within QDs under the presence of the external electric field and the heat. Considering the applied voltage (or the current density) to operate QD-LEDs for display applications (1,000–2,000 nit) do not supply significant electric field across QDs or heats to QDs that could alter dynamics of charge carriers within QD EML, the device efficiency of the QD EML in working devices can be interpreted with the collective luminescence efficiency of QD films with the same charge carrier distributions [41–43]. Therefore, the charge carrier distribution in QD EML within working devices and the luminescence efficiency of QDs for each charge species are two important parameters that decide the device efficiency of QD-LEDs.

2.2 Operation Mechanism of QD-LED The unique photophysical characteristics of QD come from the quantum confinement effect that arises from the spatial confinement of charge carriers by its dimension [7, 44]. Unlike its bulk counterparts having continuous energy bands, QD with a size smaller than its bulk Bohr radius has discrete atom-like electronic energy levels with energy offsets greater than kT (Fig. 4a). Therefore, QD shows well-defined excitonic features in absorbance and emits a narrow and symmetrical luminescence spectrum. The wavelength of the luminescence spectrum depends on the energy gap between the lowest excited states of electron and hole, which are equivalent to the conduction band minimum (CBM) and valence band maximum (VBM) of bulk materials, respectively. As the quantum confinement effect becomes greater for the smaller QD, the optical bandgap of QD can be enlarged by reducing the size of QDs (e.g., blue emission from 0.8 nm radius CdSe QDs and red emission from 2 nm radius QDs) [4, 6].

256

S. Rhee et al.

Fig. 4 a Discrete atomic-like electronic states in a QD. b The Auger recombination of a negative trion and positive trion. The negative trion can be described in terms of the Coulomb scattering between the two conduction band electrons when one of them undergoes an interband transition, while the other is excited within the conduction band. Similarly, the positive trion can be described as Coulomb scattering between the two valence-band holes. The Auger lifetimes of biexcitons (τ2A ), and negative (τ−1A ) and positive (τ+1A ) trions in thin-shell CdSe/ZnSe QD. Adapted with permission from [87]. c Contour plot of the exciton density inside quantum dot corresponding to operational condition (Current density and luminescence) of QD-LED

When a QD is excited by a photon with an energy greater than the optical bandgap, the electron and hole pair is created within the QD. High-energy charge carriers cool down to the lowest excited states by releasing their energy in a form of heat on a time scale of 100 fs–10 ps. Then, electron and hole at their lowest excited states (single exciton, X) recombine to yield a photon on a time scale of 10–100 ns [7, 38, 45]. Throughout the photon absorption and emission processes, the charge state of the QD remains as neutral (QD0 ). Simultaneously with the charge relaxation and recombination processes, one of the charge carriers could be ejected to the localized electronic states at the QD surface or in CTLs nearby the QD, leaving the counter charge carrier within the QD (QD+1 or QD−1 for the case with a hole or an electron stays in the QD, respectively). The QD becomes neutralized via either process, in which the counterpart charge carrier moves back into the QD and recombines to yield a photon or the charge carrier sitting in the QD escapes and leaves the QD empty. When the singly charged QD is excited by an extra photon, the QD accommodates three charge carriers (trions, one hole and two electrons (X− ), or two holes and one electron (X+ )) that promote the recombination rate of the electron and hole pair, statistically, by a factor of two. At the same time, the three charge carriers can undergo the non-radiative Auger recombination processes [27, 46, 47], in which the electron and hole pair recombines by transferring its energy to the third charge carrier. Auger recombination processes take place on a time scale of 10–100 ps, which is three orders of magnitude faster than its competitor (radiative recombination processes of charged excitons). The hot charge carrier resulting from the Auger recombination relaxes to its lowest electronic states and the QD remains charged that is subject

Electroluminescence Devices with Colloidal Quantum Dots

257

to undergo the Auger-involved non-radiative path upon photoexcitation, unless it escapes the QD or the counter charge carrier comes in (Fig. 4b). The charge to photon conversion processes (EL) within QDs in QD-LEDs follows similar processes except for the fact that the generation of electron and hole in QDs is not strictly paired. At the current density regime of 10–100 mA/cm2 , the average injection rate of electron and hole into each QDs in working devices is calculated to be an order of 1 μs−1 , which is two orders of magnitude slower than the recombination processes of excitons [7, 26, 32]. Therefore, the presence of multiexcitons (more than two pairs of excitons) can be neglected, grounded on the calculation that the electron and hole pair undergoes either via the radiative recombination processes or non-radiative Auger recombinant processes before additional charge carriers come into play. The average exciton density inside the QD emissive layer of QD-LED is calculated corresponding to the size of QD, current density, and luminescence of QDLEDs grounded on the same electron and hole injection rate conditions (Fig. 4c). At the current density regime of 10–30 mA/cm2 (Luminescence of 1,000–10,000 nit), the calculated exciton density is 0.01–0.02, which shows the negligible possibility of generation of biexciton inside the QD. Considering that the internal quantum efficiencies (IQEs) of devices at this current density are measured to be 90% (Out-coupling efficiency, nout = 0.2), which is almost same PL QYs of solution QDs (~90%), these results indicate that the balanced electron and hole injection rate in QD emissive layer can completely suppress the non-radiative Auger recombination process inside the QD-LEDs. In a simplified model with the absence of charge carrier ejection processes from QDs, the charge carrier population in QDs can be understood as the sequences of stepwise charge accumulation and the deduction of electron and hole pair via the radiative or Auger recombination processes [27]. The injection of an electron or hole into a ground-state neutral QD (QD0 ) produces a singly negatively or positively charged QD (QD−1 or QD+1 ). The alternated injection of charge species neutralizes the QD with electron and hole pair inside (QD* ), which cools down to the groundstate neutral QD (QD0 ) via the radiative recombination process in less than 1 μs. By contrast, the injection of same charge species after the first charge carrier injection yields doubly charged QDs (QD−2 or QD+2 ). The injection of opposite charge carriers into these QDs promotes the non-radiative Auger recombination processes, which yield singly charged QDs with one hot charge carrier (QD−1* or QD+1* ) in less than ~100 ps that are followed by carrier relaxation processes to be singly charged QDs (QD−1 or QD+1 ). The repetition of random charge carrier injection processes allows one to estimate the EL efficiency of a QD to be 50% of its photoluminescence quantum yield (PL QY) that solely allows for the paired generation of electron and hole in QDs. However, the EL efficiency of QD-LEDs reaches to PL QY of QD films in working devices, implying that the injection processes of electron and hole into QDs are not completely independent and paired to each other. Quantum mechanical calculation shows that the presence of charge carrier creates the Coulomb potential that impedes the addition of the same charge carrier and promotes the injection of counter charge carrier (Fig. 5)

258

S. Rhee et al.

Fig. 5 Sketch of the relationship among probable QD charged states in an operating QD-LED. Charge injection of an electron (blue arrow) and hole (red arrow), radiative recombination (yellow arrow), the Auger recombination (gray arrow), charge hopping (improbable), and energy transfer (ET, broken gray arrow) are noted. The electronic energy levels of neutral QD (gray dashed line) are superimposed on that of charged QDs for comparison. Adapted with permission from [26]

[26]. The Coulomb potential created by the presence of a charge carrier escalates the probability of paired generation of electron and hole in QDs in QD-LEDs, and enables one to achieve near-unity EL efficiency. The Coulomb potential created in singly charged QDs and doubly charged CdSe/CdZnS QDs are estimated to be 0.07–0.19 eV and 0.25–0.36 eV, respectively. As the doubly charged QDs attract the counter charge carrier and block the injection of the same charge carrier in at most three orders of magnitude different rates in working devices, the population of highly charged QDs with more than two charge carriers can be neglected at the QD charge scheme. Therefore, the luminescence efficiency of QDs in working devices is deduced to the probability of QDs accommodating single exciton (X) and trions (X− or X+ ) and their luminescence efficiency, which are in principle associated with the suppression of the Auger recombination processes in QDs in QD-LEDs.

2.3 QD Emissive Materials As discussed earlier, the ejection of charge carriers to the localized states nearby QDs is the non-radiative recombination process of single exciton (X), and the other charge carrier remaining in QDs opens the possibility of the non-radiative Auger recombination processes upon continuing photoexcitation, leading to the reduction of the photoluminescence quantum yield (PL QY) of QDs. The charge carrier ejection process as well as the charge carrier injection imbalance are also responsible for the

Electroluminescence Devices with Colloidal Quantum Dots

259

reduction of the electroluminescence (EL) efficiency of QDs in QD-LEDs. Therefore, the suppression of the two processes, the ejection of either one charge carrier of single exciton and the Auger recombination processes of trions, have been of the utmost priority in designing QDs and QD-LEDs [24, 26, 34, 48, 49]. Two approaches, the suppression of the Auger recombination processes in QDs [24, 27, 50] or the containment of the situation that provoke the Auger recombination processes [28, 34, 51], have been suggested to prevent non-radiative recombination processes in QD materials. From the viewpoint of QDs, the first is to inhibit the charge carrier escape from QDs to the localized states, which not only reduces the luminescence efficiency of a single exciton, but also remains a charge carrier in QDs that prompts the non-radiative Auger recombination processes upon addition of an exciton. The other is to impede the rate of the non-radiative Auger recombination processes in QDs by tailoring the electron and hole wavefunctions and their interactions. Both approaches are enabled by the structural engineering of QDs [9, 24, 52–56], specifically, core/shell heterostructuring (Fig. 6a, b). The nanometer-sized QDs have 10–70% of surface atoms depending on their size, and these hold broken bonds that act as the localized electronic states, which are the final destination of ejection processes of charge carriers. Passivating the

Fig. 6 a Schematic illustrations (up) and the corresponding potential profiles (down) of a coreonly QD, a core/shell QD with a sharp interface, and a core/shell QD with a smooth interface. b Shell-thickness-dependent PL QYs of QDs with different shell formulations. Adapted with permission from [53]. c Early time Auger dynamics of multicarrier states for reference core/shell (C/S) CdSe/CdS QDs with an abrupt interface and core/alloy/shell (C/A/S) CdSe/CdSex S1–x /CdS QDs with an alloyed smoothen interface. Both samples have the same core radius (r = 1.5 nm) and the same total radius (R = 7 nm). The thickness of the CdSex S1-x interfacial alloy is 1.5 nm. d Schematic illustration of synthesis of the C/S and C/A/S QD samples. Adapted with permission from [54]

260

S. Rhee et al.

surface atoms of QDs (core) with other semiconductor materials with greater bandgap (shell) aids to decouple the charge carriers of QDs from the surface trap states [57, 58]. In principle, the core/shell heterojunctions are classified into type-I and typeII depending on the band alignment across the core and shell materials. For type-I core/shell QDs, CBM and VBM of the core reside inside of these of the shell (e.g., CdSe/ZnS, CdSe/CdS, and InP/ZnS) [21, 57, 59–62]. In this configuration, both electron and hole wavefunction is confined in the core phase, and the interaction of the charge carriers with the surface trap states is diminished by the potential barrier, whose width and height are represented by the shell thickness and the energy offset in CBM and VBM between core and shell, respectively. As a result of inhibition of charge carrier ejection by the presence of the shell, type-I core/shell QDs show high luminescence efficiency of single exciton and hence they are widely accepted in display applications. Meanwhile, type-II core/shell QDs hold a staggered core/shell band alignment, in which electron and hole wavefunctions separately stay either in the core or the shell phase (e.g., CdTe/CdSe, CdSe/ZnTe, CdS/ZnSe, and InP/CdS) [53, 63–66], and thus show relatively low PL QY due to the reduced electron-hole wavefunction overlap, limiting their use in display applications. The delicate design of the band structure in core/shell QDs permits to engineer the rate of the Auger recombination processes (Fig. 6). Auger recombination processes involve inelastic scattering of three charge carriers (two electrons and one hole or one electron and two holes), and thus the rate of the Auger recombination processes is characterized to be inversely proportional to the volume of QDs that is linked with the overlap of electron and hole wavefunctions [50, 67]. Apart from the volume scaling, the rate of Auger decay is also affected by the shape of confinement potential that determines the strength of the intraband transition wherein the third charge carrier is excited to a higher energy. The comparative spectroscopic analysis on CdSe/CdS core/shell QDs having an abrupt interface versus a smooth interface unveiled that the potential profile at the core/shell interface affects the rate of the non-radiative Auger recombination processes and its impact far surpasses that of the volume scaling (Fig. 6d). An ultimate achievement is the giant core/shell QDs having a potential gradient, which allows for the complete decoupling of charge carriers residing in cores with the surface trap states as well as the significant suppression of the nonradiative Auger recombination processes.

2.4 QD-LED Structure The external quantum efficiency (ηEQE , EQE), the ratio of the total number of photons emanating from the QD-LED to the number of injected carriers, is obtained by multiplying the charge to photon conversion efficiency of devices (ηIQE , the internal quantum efficiency) and the extraction efficiency of generated photons out of the devices (ηout , the out-coupling efficiency) (ηEQE = ηIQE xηout ). Considering that the ηout is optical characteristics of the multilayers that consist of QD-LEDs (ηout = 20% for typical planar QD-LEDs) [68, 69], ηIQE is the key factor that defines the device

Electroluminescence Devices with Colloidal Quantum Dots

261

efficiency. As discussed in the previous chapter, the materials’ characteristics together with the charge injection balance into the QD EML determine the luminescence efficiency of the QD EML and the EQE of QD-LEDs. The charge injection balance into QDs not only governs the efficiency of QDLEDs, but also characterizes the operational stability of devices. Chang et al. [26] monitored operation-time-dependent optoelectronic performance and the photophysical properties of QD EMLs in working devices in relationship with the disparity between electron versus hole injection rates into QDs, and unveiled that the operational instability of QD-LEDs mainly originates from the charge injection imbalance of devices (Fig. 7a). Later, Rhee et al. [34] observed the linear relationship between the device degradation rate and the extent of charge injection imbalance. Interestingly, the degradation rate estimated from the fitted lines approximate to nearly 0 for the points where charge injection imbalance is 0, promising the complete suppression of device degradation when the QD-LED harmonizes the charge injection rate into QDs (Fig. 7b).

Fig. 7 a Operation-time-dependent traces of IQE (Black line) and the PL QY of the QD emissive layer in the corresponding device (Red circle). The initial fast decay is due to the Auger recombination of negativelycharged QD. Adapted with permission from [26]. b Degradation rate at stage I (Initial fast decay) vs the initial charge injection imbalance into QDs (Je0 − Jh ) with varying operating current density and hole injection properties of QD-LED. c Enhancing operational stability of QD-LED by suppressing hole injection barrier of device. Adapted with permission from [34]. d Schematic illustration of enhanced hole injection rate through reducing ligand length at the interface between QD EML and HTL. Adapted with permission from [72]

262

S. Rhee et al.

The charge injection imbalance in QD-LEDs arises from the differences in the charge carrier transport rate of ETL versus HTL and the injection properties of charge carriers at the interfaces between electrodes and CTLs and CTLs and QDs. For the state-of-the-art QD-LEDs having inorganic CTLs, the hole injection from organic HTL into the inorganic QD EML is known to be the main hurdle due to the relatively large energy offset (>0.5 eV) between the HOMO of HTL and the VBM of QD EML [8, 48, 70, 71]. The impediment of hole injection yields the presence of excess electrons in the QD EML, leading to the reduction of the device efficiency as well as the acceleration of device degradation. Various studies have been attempted to enhance the hole supply for balanced charge carrier injection into the QD EML. Cho et al. [28, 72] showed that reducing the ligand length at the interface between QD EML and HTL could enhance the hole supply into QDs and thereby the device efficiency (Fig. 7d). Rhee et al. [34] demonstrated that boosting the hole supply by eliminating the universal hole injection barrier (~0.3 eV) at the interface between Al/MoOx anode and organic HTL enhances the efficiency and operational stability of QD-LEDs (Fig. 7c). The structural engineering of QDs has been also taken into account to foster the hole injection. QDs with shallow bandgap shell materials (e.g., ZnSe) [8, 51] effectively lower the height of the energy barrier for hole injection from HTL into QDs, leading to the augmented hole injection into QDs and thereby the superior device performance. The combination of the structural engineering of QD materials and the optimization of device architecture allows one to effectively manage the non-radiative recombination processes within the QD EML in working devices. A notable achievement is QD-LEDs with reduced efficiency roll-off at the elevated current densities by Lim et al. [51, 67], which apply the Auger-suppressed QDs in the charge matched device architecture. These results not only promise the practicable use of QD-LEDs in displays, but also shed light on the use QD-LEDs in high power light sources including lightings, outdoor displays, and electrically pumped lasers.

3 Perspectives on Future Quantum Dot Displays 3.1 Recent Progress in Cd-Free QD-LEDs Despite significant progress made in QD-LEDs, a few more steps still remain to their practicable use in displays. The biggest concern is the materials’ toxicity mainly due to the presence of heavy metal elements in QDs (e.g., Cd and Pb), whose use in the industry is strictly regulated (Restricted Hazardous Substances, RoHS). QDs consisted of Cd-free elements have been rigorously explored, yet their performance in QD-LEDs is far falling behind that of Cd-based QDs. Hereafter, we address the current status of Cd-free QDs and QD-LEDs and the remaining challenges that need to be overcome for their practicable use.

Electroluminescence Devices with Colloidal Quantum Dots

263

Cd-free QDs can be categorized into four groups: III-V (e.g., InP [16, 73–76], InX Ga1−X P) [77], II-VI (e.g., ZnSe [78–81], ZnTe, and ZnSeX Te1−X ) [22], IV (e.g., Si), and I-III-VI (e.g., Cu-In-SeX S1−X [82, 83]) (Fig. 8). Among potential candidates, InP QDs that cover green to red emission have shown rapid progress in their optical and optoelectronic performance by utilizing the same structural design strategies that have been established in Cd-based QDs. Specifically, ZnS shell is adopted at the exterior of InP-based QDs for effective charge carrier localization within the InP core. Also, multi-shell structure (InP/ZnSe/ZnS [74] or InP/GaP/ZnS [75]) or compositional-gradient structure (InP/ZnSeX S1−X /ZnS) [73, 76] have been proposed to reduce the structural stress between InP core and ZnS exterior shell, enabling one to accomplish PL QYs of InP QDs comparable to these of Cd-based ones (PL QY > 90% for red- and green-emitting InP QDs). Along with the materials development, the device structural engineering to reach the charge carrier injection balance allows to achieve InP-based QD-LEDs showing the EQE close to the theoretical limit. In parallel with the materials development of green- and red-emitting QDs, IIVI ZnSe [79–81] and ZnSeX Te1−X QDs have been substantially scrutinized as the blue-emitters (Fig. 8a). By exploiting the chemistry and the structural designing

Fig. 8 a PL spectra of Cd-free QDs (ZnSe for violet to blue emission, ZnSex Te1−x for blue to green emission, and InP for green to red emission). Adapted with permission from [88]. b Schematic illustrating the ligand displacement from native oleate ligands (OA) to mono-2-(methacryloyloxy)ethyl succinate (MMES) ligands and a photograph of QD-LED and electroluminescence. Adapted with permission from [16]. c Schematic illustration of the fabrication process of ZnSe/ZnSe1−x Tex core/shell QD and change of PL QY as a function of ZnSe ratio

264

S. Rhee et al.

principle, the core/shell heterostructures made of ZnSe or ZnSex Te1−x core and ZnS shell, which display PL QYs greater than 80% in the wavelength range of 400– 480 nm, have been successfully demonstrated (ZnSex Te1−x QDs with 70% PL QY and QD-LEDs with 4.2% peak EQE and 1195 cd/m2 luminance) (Fig. 8c) [22]. Up to now, Cd-free QD-LED research has primarily focused on the device efficiencies, but a systematic study on the operation stability is lacking. Given that underlying physics that governs photophysical and optoelectronic processes are effective regardless of the composition of QD materials, the lessons learned from Cd-based QDs and devices also offer rational guidelines for CD-free QD-QD-LEDs. As witnessed from Cd-based QDs, the chemistry for Cd-free QDs with suppressed Auger decay and the optimization of device structure will enable Cd-free QD-LEDs equipped with the operation stability and the brightness meeting the requirements for practicable use in displays and lightings.

3.2 QD Patterning for Display Applications The patterning of primary-colored QDs is a prerequisite for practicable use of QDLEDs in nearly all displays including TVs, smartphones, AR and VR. The human nature that desires augmented realness from images has driven higher resolution displays, and now the ultra-high-definition displays having a total image dimension of 7680 × 4320 pixels (also referred to as 8 K displays) is leading the premium TV market. Even higher resolution greater than 3,000 ppi (pixel per inch) is necessary for AR and VR technologies. Following the display trend, significant efforts have been made to realize high-resolution full-color QD-LEDs. In the section, we briefly review the QD patterning technologies. QD patterning technologies are categorized into the wet-processing method or the dry-processing method depending on whether the deposition of QD solutions is directed on the devices or the sacrificial substrates, respectively. The wet-processing methods are either the deposition of QDs directly into desired patterns or the deposition of whole QD films followed by the removal of unnecessary parts. The dryprocessing method transfers the desired patterns of QD solids selectively from whole QD films pre-deposited on the sacrificial substrates. It is noteworthy that all of these patterning technologies have been widely scrutinized in fabricating other functional materials, and hence their advantages and shortcomings are also applicable to the case of QD patterning. A representative of the dry-processing is the transfer printing technique, which picks a selected area of QD films from the pre-deposited whole QD films on the mother substrate and places it to the target substrate [84, 85]. Transfer printing requires the complete relocation of patterns from the mother substrate to the stamp and to the target substrate without structural deformation (Fig. 9b). Controlling the pressure applied by the stamp and the interfacial energies between substrate/QD and QD/stamp is central for reducing the imperfection of the transferred QD patterns.

Electroluminescence Devices with Colloidal Quantum Dots

265

Fig. 9 Schematic representation of a the inkjet printing, b transfer printing, and c photolithograph patterning of QD. d Photographs of patterned QD films with the negative photolithography process. c Adapted with permission from [89]. d Adapted with permission from [16]

QD patterns with the line width 10,000:1 (HDR/local Very high 1,080,000:1 Very high >1,080,000:1 Very high >1,080,000:1 High dimming)

Contrast ratio

Laser projection

Long

Low

Life time

Cost

High

Short

Medium

Low

Shock resistance Medium

Medium

High

µs −100 to 70 °C

µs

ms −50 to 70 °C

Response time

Operating temperature 0–60 °C

3000 cd/m2 (full) 104 cd/m2 (green)

Medium

High

Long

High

−100 to 120 °C

µs

105 cd/m2 (full) 107 cd/m2 (blue/green)

Medium

High

Medium

Medium

0–60 °C

ms

1000 cd/m2 (full color)

High

Backlighting

Medium

Micro-LED Self-emissive

Luminance

QLED Self-emissive

Lum. efficacy

OLED

Backlighting/LED

Mechanism

Self-emissive

LCD

Technology

Table 1 The comparison of the different technologies [1]

274 D.-S. Lee and J.-H. Han

Micro-LED Technology for Display Applications

275

Fig. 1 Applications using micro-LEDs

Fig. 2 Market forecast of micro-LED-based applications [2]

Among many applications, it is expected that smartwatches will be the first to be commercialized as a beachhead for using micro-LED displays. Moreover, it is anticipated that micro-LEDs used in the alternative reality/virtual reality (AR/VR) market will blossom, whereas penetrating the OLED-based TV and smartphone markets will be very challenging in terms of not only technology but also price [3].

1.2 Development of Micro-LED-Based Displays R & D is being continuously conducted on micro-LEDs by directly using individual LEDs as pixels rather than in BLUs on displays of various sizes. The application of individual LED pixels is a distinct change from conventional LED TVs and is rather

276

D.-S. Lee and J.-H. Han

similar to OLED-based displays, where each pixel shines voluntarily and independently. Research on the structure and method of micro-LED displays related to thinfilm transistors (TFTs) or complementary metal-oxide semiconductors (CMOS) is also being carried out in a similar way to OLED-based displays. In the early days, LEDs were mainly used in LCD BLU due to various problems such as the efficiency difference among the red, green, and blue LEDs, the response time difference or efficiency droop, and their relatively high price. Meanwhile, InGaN-based microLEDs were developed in 2000 by H. X. Jiang’s group at Kansas State University (KSU) to increase light extraction efficiency (LEE) [4, 5]. After that, development of micro-LEDs as display pixels has continued [6, 7]. This is because the amount of light emitted from an LED greatly decreases as it passes through the LCD filter, and problems such as low image quality and color recognition in an outdoor environment can be addressed by using individual pixels the size of micro-LEDs. Since then, the development of micro-LED displays has proceeded rapidly as it has been confirmed that actual display operation is possible when applying LED array technology. MicroLED-based displays have been developed continuously since the group at KSU first produced monochromatic 10 × 10 resolution displays in 2001 by connecting 12-µmdiameter monochromatic micro-LEDs with a passive matrix. To improve problems when the resolution and brightness are not good in a display driven by a passive matrix, a group at Strathclyde University announced a micro-LED display with an active matrix in 2008. Since then, as the number of micro-LED arrays applied to displays has increased and it has become easier to connect with other electrical devices such as CMOS or drivers, display development has progressed significantly. The performance and characteristics of micro-LED-based displays developed so far are summarized in Table 1 compared with OLED and LCDs [8]. Many laboratories and companies are continuing to develop micro-LEDs for use as pixels in micro-displays. Typically, the 126 × 96 µm micro-LED array structure proposed by Choi et al. in 2004 has shown higher luminescence intensity than OLEDs [9]. Although the electrical and optical characteristics of micro-LEDs are not the same as those of conventional LEDs, it is possible to manufacture displays with micro-LED array structures connected to the driver matrix. In 2008, Gong et al. produced blue and ultraviolet flip-chip micro-LEDs with a size of 20 µm and an array with a pitch size of 50 µm. Micro-LEDs fabricated using their method showed high luminous efficiency, and problems such as low efficiency and electrical characteristics evident in previously fabricated micro-LEDs were improved to some extent. Beyond 2010, as interest in micro-LED-based displays has become focused and development has been accelerated, the reasons why the efficiency of micro-LEDs is low have been analyzed, technologies for improving their performance have been identified, and micro-LEDs in various structures have been announced one after the other.

Micro-LED Technology for Display Applications

277

Fig. 3 A flow diagram of the process steps for the development of micro-LED displays [13]

In 2020, Samsung’s 35 × 60 µm micro-LED adopted a 75-inch display, Plessey’s 6-µm-diameter micro-LED adopted a 0.7-inch display, and Jade Bird Display’s 1µm-diameter micro-LED for a 0.6-inch display were presented as prototypes [10– 12]. However, in spite of these possibilities and efforts, problems such as low efficiency, low luminous intensity, and yield reduction have persisted due to the miniaturization of LEDs. Furthermore, the optimization of mass production-related technologies (transfer, inspection, etc.) for display production remains incomplete. Currently, problems such as reduced efficiency due to reduced LED size require considerable improvement. Therefore, it is imperative to optimize various related technologies such as LED transfer, device integration, and heat dispersion, as well as development of light source production technology for the development of high-quality micro-LEDs of various wavelengths. Figure 3 shows the major processes in developing micro-LED-based displays [13]. The process for micro-LED display fabrication is divided into LED production, transfer, pixel assembly, defect management, and fabrication of the backplane, which includes TFT and CMOS development. Here, we briefly look into the new micro-LED technologies presented by various research institutes and companies and introduce their products for various applications to cover their R & D strategies. In this brief introduction, we look at the major issues for each technology, the most important being the pixel assembly step in Fig. 3. We focus more on studies about the important light source development steps, such as dividing the LEDs into appropriate sizes for pixel implementation, micro-LED transfer, and bonding technology with TFT. As shown in Fig. 4, micro-LED-based displays and applications have different LED chip sizes, which can be applied according to the resolution in PPI required for the product, and each production process requires suitable transfer technology. Currently, products needing high PPI such as AR/VR displays or projectors require a large number of LEDs that are closely arranged, and so a monolithic array integration method is suitable. However, applications such as TVs, cell phones, laptops, and smartwatches with relatively low PPI are suitable for pick-and-place transfer technology, which moves individual LEDs [14]. Monolithic array methods transfer a whole LED arrays from the growth substrate onto the desired TFT or CMOS and bonds them by soldering, etc. The size of the pitch formed on the growth substrate is

278

D.-S. Lee and J.-H. Han

Fig. 4 Classification and examples of transfer technology [14]

directly applied to the final product, so technology that precisely matches the TFT in the backplane and LEDs is also required for this process. Therefore, products fabricated by monolithic methods generally use a single color source and thus require perfect alignment, entailing that defect management becomes very important. On the other hand, with the pick-and-place method, the individual LEDs divided on the growth substrate are directly handled, so they can be set onto the display at the desired spacing and arrangement, while mismatching due to differences in thermal expansion does not occur during the bonding process. This means that the required transfer technology differs depending on the application being developed; not only the size, light quantity, and efficiency of the LED required from the light source development stage, but also the subsequent process stage and required backplane structure are all different. Therefore, one should determine the PPI level of the display before developing the light source for the product. In this chapter, we mainly focus on the light source development issues according to size, although the current state-of-the-art transfer technologies concerning micro-LEDs and the backplane, as well as defect management, are also briefly explained.

Micro-LED Technology for Display Applications

279

2 Issues Concerning Micro-LED Development Currently, commercialized LEDs for illumination and various other applications are normally 250 µm or larger, while those between 250 and 100 µm (mini-LEDs) are being rapidly developed and applied in response to the demand for better portable devices. No significant problems have been found in reducing the size of LEDs from 250 to 100 µm, but various problems have arisen during size reduction below 100 µm. Some problems that can be ignored in relatively large LEDs, such as an increase in the impact of LED surface defects, current crowding, and the sidewall effect, as well as wiring problems leading to a decrease in luminescence efficiency, are problematic in micro-LEDs. When reducing the size of micro-LEDs to less than 100 µm, it has been shown that the reduction in efficiency, especially at around 10 µm, where it is expected that micro-LED can yield very significant benefits, is much greater than researchers first thought. This makes the expected advantages of using micro-LEDs (five times lower power consumption, 100 times brighter, and faster response time than OLEDs with no afterimage) difficult to realize. In this chapter, we examine the current development status of micro-LEDs of various wavelengths for display applications and look at various problems that are preventing the development of micro-LED-based displays and the ongoing studies to improve them.

2.1 Micro-LED Development Status Group III–V-based LEDs generally emit light of a desired wavelength by combining inorganic materials such as Al, Ga, In, N, P, and As. In addition, when necessary, the light wavelength can be adjusted by combining short-wavelength light and color converters such as phosphors. However, in the process of manufacturing these inorganic LEDs with a size of a few tens of micrometers, the various problems mentioned previously have been discovered. This means that existing fabrication processes for conventional LEDs are unable to produce micro-LEDs with high efficiency, high brightness, and high yield. Therefore, studies to analyze the causes of these problems and to improve them require different approaches for each one depending on the material constituting the light source and the emission wavelength. To apply micro-LEDs to various application fields, it is necessary to develop light sources for each wavelength evenly. In particular, the implementation of RGB pixels is a very important issue not only in micro-displays used in AR/VR applications but also in relatively large displays based on micro-LEDs. The current research into micro-LED is largely divided into the size reduction of the LED itself, the implementation of a passive/active matrix to drive the ultra-small light source, the transfer technology to move the micro-LED to the matrix, and the defect-detection and errorchip replacement technology. Nowadays, more and more laboratories and companies

280

D.-S. Lee and J.-H. Han

Fig. 5 Prototype applications with micro-LEDs: a a monochromatic RGB micro-LED array and a micro-LED-based b display, and c projector [15–17]

Fig. 6 Micro-LED-based prototype displays by Samsung, LG, and Lumens (in that order from left to right)

are producing and presenting prototypes for various applications such as displays, projectors, and illuminating light sources by integrating the methodology of these studies, as shown in Fig. 5. At one of the world’s largest exhibitions, CES 2019, numerous companies such as Samsung, LG, Seoul Semiconductor, Glo, AUO, Lumens, Playnitride, and Konka showcased a variety of micro-LED-based products, as shown in Fig. 6. Display products such as Samsung’s The Wall (introduced during the exhibition) is an example of a 50-µm-level RGB array that many companies have already developed. Furthermore, many companies have developed high-quality micro-LEDs of less than 100 µm with related technologies such as TFT and backplane matrices. In 2016, Sony announced CLEDIS, which is a display prototype using 100 µmsized micro-LEDs, while companies such as Samsung, LG, and Lumens have announced large-scale displays using approximately 50-µm-sized micro-LEDs. Although 50-µm-level RGB arrays are a great advance in technology, mass production of these devices has not yet taken place in earnest. Moreover, to mass-produce 10-µm-level micro-LED-based displays for various applications such as AR/VR and ultra-small displays requires mass production, transfer, circuit integration, etc., and innovation is needed in these areas.

Micro-LED Technology for Display Applications

281

2.2 Efficiency of Micro-LEDs The external quantum efficiency (EQE) of conventional InGaN-based LEDs tends to increase depending on current density. It becomes maximized in the current density range of 0.1–1.0 A/cm2 and then decreases after that, as shown in Fig. 7 [18]. In most applications, micro-LEDs operate at a lower current density than the peak efficiency, which is unlike that applied to traditional lamps, and as the size of the LED decreases, the behavior of the lower current density appears to suffer from larger sidewall effects and non-radiative recombination compared to conventional LEDs, which will be discussed later [18]. When manufacturing displays, this characteristic becomes a bigger problem for applications with a smaller pixel size. As shown in Fig. 7a, it has been estimated that the pitch is almost six-fold lower for TVs than smartphones, and at 4 µm, it is at least 25-fold lower for AR applications. However, in the case of the micro-LEDs developed to date, the blue, green, and red LEDs show EQE levels of 40, 30, and 15%, respectively, based on a size of 10 µm in the laboratory; thus, engineers are having difficulty in developing LEDs that have the required peak brightness [19]. In particular, as shown in Fig. 7b, in applications that do not require ultra-high brightness, the operation of micro-LEDs is driven in a relatively low current region, although it should be pointed out that the EQE of micro-LEDs also decreases with size. As the size of the LED decreases, the peak efficiency is reached in the high current–density region, whereas in the low current– density region, the efficiency is very low due to the aforementioned problems. In other words, implementing a high PPI-based display using micro-LEDs with high brightness remains very difficult. Therefore, we look at the problems of micro-LEDs that are limiting their luminescence intensity and efficiency sufficiently to block their mass production, and we show various studies that are being conducted to solve these problems. In addition, based on these studies, we look at which research institutes and companies will ultimately implement which products and technologies.

Fig. 7 a Requirement and characteristics and b the relationship between the external quantum efficiency (EQE) and current density of micro-LEDs based on their applications [18]

282

2.2.1

D.-S. Lee and J.-H. Han

The Epi-Layer Growth Issue for Micro-LEDs

Blue, green, and red LEDs can be made through a combination of In, Ga, and N, and as the proportion of In increases, the emission wavelength of the LEDs moves from blue to green, and then to red. However, as the ratio of In increases, the quality of the epi-layer deteriorates, the strain increases, and the efficiency rapidly decreases, especially in the red region, especially in micro-LEDs, as shown in Fig. 8 [19–22]. Problems found during the development of blue and green micro-LEDs have been analyzed by many researchers. These can be largely classified into problems caused by defects, the relative size ratio, efficiency degradation by the sidewall effect, current crowding by reducing the LED size, and an increase in the area of separation compared to the LED size during the LED chip fabrication process. In addition, since the materials and substrates used differ depending on the wavelength, research on growth technologies for micro-LEDs according to the substrate is also required. As shown in Fig. 8, micro-LEDs have low EQE due to the various abovementioned problems occurring in the processing steps. EQE is highest for blue LEDs and lowest for red LEDs due to the gradual increase in the In ratio in the active layer when the LED wavelength varies from blue to red. As of 2018, the EQEs of blue and green conventional LEDs are higher than 50% and 20%, respectively, and these are reduced significantly to around 5% in 5-µm LEDs. These values were significantly lower than the commercialized conventional LEDs as of 2018 [19]. As shown in Fig. 8b, c, 10-µm-sized InGaN-based micro-LEDs and 15-µm-sized red AlGaInP-based currently have a low EQE (approximately 6%) and there have been lots of efforts to improve them, especially in epi-layer growth. As mentioned earlier, red LEDs can be grown not only by InGaN mainly on sapphire substrate but also by AlGaInP on GaAs substrates, and thus the composition and performance of the active layer differ. Hence, the problems and development directions in each case are different depending on the growth mechanism and process conditions. Currently, growth technology emphasizing the low manufacturing cost of InGaNbased active layers, and the advantage of large-scale substrate growth is being studied

Fig. 8 External quantum efficiency (EQE) of a conventional LEDs (2018), b InGaN-based blue, green, and c AlGaInP-based red micro-LEDs [19–21]. Reprinted with permission from a [19] and c [21] © The Optical Society. Reproduced from [20], with the permission of AIP Publishing

Micro-LED Technology for Display Applications

283

Fig. 9 Examples of prototype red micro-LEDs: a an InGaN-based red micro-LED by Plessey and b an AlGaInP-based red micro-LED [23, 25]

for micro-LED displays, and AlGaInP-based LEDs are also being developed for nextgeneration micro-LED-based displays based on conventional LED manufacturing technology. Conventional red LEDs with an AlGaInP active layer have the highest EQE, but companies such as Plessey and Glo have presented red micro-LEDs with an InGaN-based active layer grown on GaN or Si substrates, as shown in Fig. 9a [23, 24]. Moreover, AlGaInP-based micro-LEDs such as those shown in Fig. 9b are also continuously being studied at universities and research institutes such as the National Chiao Tung University and Alfaity, [25]. Although there have been reports about efficiency reduction at a high current density in conventional AlGaInP-based red LEDs, commercialized conventional LEDs still have around 50% EQE. Therefore, the fact that AlGaInP-based red LEDs have approximately 6% EQE at a size of 15 µm suggests that the internal quantum efficiency (IQE) and EQE reductions due to various problems caused by the LED size decrease are greater in the red micro-LEDs than in others. We analyze the phenomena and causes of this size-dependent efficiency reduction at all wavelengths later on. Typical problems in the development of efficient micro-LEDs at present include the formation of defects and dislocations due to chemical and structural problems occurring during growth, as well as the sidewall effect. The first problem to be considered is the formation of defects and dislocations that occur during the growth of the LED epi-layer, which is the most fundamental process in LED manufacturing [26]. Currently, the most efficient LEDs are made through InGaN-based epi-structures on sapphire substrates or AlGaInP-based epi-structures on GaAs substrates, depending on the wavelength. In this process, dislocation is caused by a lattice mismatch between the substrate and epi-layer [27, 28]: the smaller the LED size, the greater its influence and the consequential deterioration of the optical and electrical properties, resulting in major problems. In particular, the problems become even more serious when these efficiency-reducing factors are combined with the defects that occur during the fabrication process. Figure 10 shows the conventional LED structures. For both types, GaN is grown onto a substrate such as Si or sapphire and the structure is formed through various doping techniques according to purpose. During the growth process and each step in the manufacturing process, the wavelength and efficiency of the LEDs can vary widely with material composition and/or structural characteristics of the multiquantum well (MQW) area (i.e., the active layer). For example, blue or green LEDs

284

D.-S. Lee and J.-H. Han

Fig. 10 Conventional LED structures: a top-emitting and b bottom-emitting (flip-chip) LEDs

are manufactured by the growth of sequentially undoped GaN (u-GaN), n-type GaN (n-GaN), and MQWs (InGaN/GaN) on top of the sapphire substrate. During growth, defects and/or dislocations of various sizes and quantities are formed at random locations due to the chemical composition ratio, lattice mismatch, and/or strain in each layer. As shown in Fig. 11, it has been reported that issues related to epi-quality arise even before entering the follow-up process for LED manufacturing [29, 30].

Fig. 11 Examples of a defects and V-pits and b composition fluctuation in InGaN/GaN-based MQWs [29, 30]. This article is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/deed.en)

Micro-LED Technology for Display Applications

2.2.2

285

Epi-Layer Growth—Defect and Color Homogeneity

Even in the fabrication of the conventional LED, the formation of defects, V-pits, and/or compositional fluctuation can occur, which causes problems such as reduced optical output power and peak shift (Fig. 11a). However, they can cause more serious problems such as efficiency decrease when reducing the size of the LEDs. For instance, when the size of a defect formed on top of a 10-µm LED is around 3 µm, it will cover 30% of its surface, and when defects of a few hundreds of nm or several µm in size are concentrated in a certain area (Fig. 11a), this LED is very likely to be a dead pixel when used in the display. In addition, in the case of an epi-layer in which composition fluctuation has formed (Fig. 11b), there is also the problem that the color homogeneity becomes unequal. For example, in the case of InGaN/GaNbased LEDs, an In-rich area can form unevenly during growth and cause a peak shift or wavelength nonuniformity problem, which can create difficulties during application fabrication. Therefore, before starting a follow-up process for LED production, studies on minimizing the formation of defects and composition fluctuation during epi-layer growth are required. Minute color changes caused by partial variation in their composition are not a major factor affecting the overall color in conventional LED-based displays; however, the smaller the size of the LED, the larger the ratio of the composition fluctuation, which can result in color inhomogeneity in each pixel of a micro-LED-based display. In particular, minute wavelength inhomogeneity partially causes more serious problems in the monolithic array transfer method, in which a part of the wafer is lifted and transferred as a whole. Therefore, to manufacture displays using monolithic transfer technology, it has become very important to ensure that the LEDs formed on the same wafer have wavelength homogeneity during the epi-layer growth. To date, a method of excluding an LED chip deviating from the desired wavelength in a process of transferring a device has been used; thus, the problems of a decrease in yield and an increase in cost occur. Based on a 6-inch sapphire substrate, development is underway to ultimately reduce the color homogeneity of the current approximately 4 nm to the 1–2 nm level in the future [31].

2.2.3

Chip Manufacturing—The Sidewall Effect

The biggest problem that occurs after the epi-layer growth is the LED sidewall effect. This is a problem caused by the LED sidewall being exposed and damaged during the plasma process for making the LED shape, such as mesa-etching of the epi-layer. The exposed surface on the side of the LED through plasma etching creates defects such as dangling bonds that act as a non-radiative recombination center and leakage current path. In the case of conventional LEDs (i.e., with a size of 250 µm or larger), the sidewall area damaged by the plasma during the etching process is relatively small compared to the LED luminous area, and the distance between the luminous area and the sidewall is greater than the diffusion length of the carrier. Thus, it does not cause serious problems. However, as the size of the LED decreases to several µm

286

D.-S. Lee and J.-H. Han

or several tens of µm, the ratio of the sidewall perimeter gradually increases, and accordingly, the influence of the sidewall effect increases, resulting in a significant reduction in the LED efficiency. For example, when a 200 µm LED is reduced to 20 µm, the sidewall ratio increases tenfold [21, 32]. As briefly mentioned above, the sidewall effect is caused by damage generation during solution- or plasma-based inductively coupled plasma (ICP) etching or reactive-ion etching (RIE), which creates leakage current and/or increases Shockley– Read–Hall (SRH) recombination that is the main reason for the high non-radiative recombination rate, thereby resulting in low efficiency. As shown in Fig. 12, studies that have applied the ABC model (A, B, and C represent SRH non-radiative recombination, radiative recombination, and Auger non-radiative recombination coefficient, respectively) to analyze the cause of the reduction in efficiency proportional to the decrease in size of the LED shows that the cause of the efficiency decrease is mainly due to an increase in factor A (i.e., an increase in the SRH non-radiative recombination rate) [33, 34]. In low current–density areas where SRH recombination is the primary effector, the efficiency decreases as the size of the LED decreases. In addition, sidewall effects due to defects or plasma damage occurring during the etching process cause trapping or scattering of carriers, leading to deterioration of the electrical properties. This can also cause noise during operation and reduced device lifespan [35]. As the device size decreases, the electrodes and wires used in conventional LED devices are too large and can also cause heat generation and current crowding, even for the same current injection; thus, new concepts and designs of p-contacts are required [36]. Some studies have reported that a reduction in the size of a device results in edge region emission rather than full-area emission of the LED [32, 37]. The decline in IQE and EQE due to a decrease in size varies depending on the structure and composition of the LEDs. It has been reported that red micro-LEDs mainly of AlGaInP composition have more severe efficiency reduction according to the LED size decrease [21]. This is because AlGaInP-based red LEDs have much larger surface recombination velocity and carrier diffusion length than InGaN-based blue or green LEDs [21, 38]. In other words, while the ratio of the sidewall to the luminous area is the same for all LEDs of any size, the greater efficiency reduction

Fig. 12 LED size dependence of EQE on the effect of factors A (non-radiative recombination) and C (Auger recombination) in the ABC model [32, 33]. Reproduced from [33], with the permission from AIP Publishing

Micro-LED Technology for Display Applications

287

in AlGaInP-based red LEDs compared to InGaN-based blue or green LEDs has been cited as a major problem in the manufacture of the RGB-based full-color displays. Obviously, as the size of the LED decreases, the resistance decreases because the size of the current-spreading layer made from indium tin oxide (ITO) or p-GaN is also reduced, which is a major cause of resistance. Furthermore, the current density of the LED, even with the same amount of current injection, increases as the size of the LED decreases, so the droop decreases and the peak becomes located in EQE in the high current-intensity region [39]. There are also reports presenting the possibility of additional efficiency increase as the ratio of the sidewall increases, which naturally reduces the internal piezoelectric field (built-in potential) of GaN-based devices [40], and also the possibility of an increase in LEE [41]. However, despite the possibility of improving efficiency, these potential advantages do not offset the reduction of the efficiency of the LED due to the sidewall effect, and considering the recently discovered problems such as current injection, the reduction of the size of the LED generally causes serious problems concerning the device performance. Therefore, the latest studies have focused on curbing the aforementioned problems as much as possible and on maximizing any advantage.

2.3 The Current Status of Technological Development for Solving the Problems of Micro-LED Efficiency Studies to solve the various problems mentioned above are currently being conducted at the laboratory level, and research has been directed toward removing and suppressing sidewall damage or defects that occur in the etching process rather than on improving the growth technology.

2.3.1

High-Quality Epi-Layer Growth Technology for Micro-LEDs

Epi-layer growth studies for micro-LEDs including metal–organic chemical vapor deposition have recently been announced by a German LED growth equipment company. This technology has an advantage in that it is possible to form an average of 50 defects or less on a 6-inch wafer for blue and green LEDs [26]. To date, it has not been reported exactly how much the defect size and number of defects can be reduced. However, if the number of defects can be reduced, the number of micro-LEDs available on one substrate will significantly increase. Moreover, the overall chemical composition state of the substrate would be more stabilized, which means that relatively highly efficient LEDs can be manufactured even when they are small. High-speed, low-cost, superior growth with chemical stability, among other characteristics, should be studied to solve problems that may arise in the future.

288

2.3.2

D.-S. Lee and J.-H. Han

Treatment for Reducing the Sidewall Effect

Efficiency reduction is prevented in conventional LEDs by performing sidewall passivation using, for example, plasma-enhanced chemical vapor deposition (PECVD). However, it has been reported that conventional passivation technology is not suitable for solving such problems for small LEDs [38, 42]. In 2018, researchers at UCSB, including Matthew S. Wong, fabricated LEDs of various sizes and compared their characteristics after conventional SiO2 sidewall passivation by PECVD and atomic layer deposition (ALD) after either ICP or HF (hydrofluoric acid) etching [42]. The most stable light-emission conditions were identified through a combination of these processes (Fig. 13). The difference between ALD and PECVD passivation is the use of plasma, which is an important consideration when determining whether there is additional sidewall damage during passivation.

Fig. 13 Comparison of device luminescence intensity according to micro-LED size, and etching and sidewall passivation methods. LED-1 is a comparison group without sidewall passivation; LED2 underwent ALD passivation after ICP etching; LED-3 underwent PECVD passivated after HF etching; and LED-4 underwent ALD passivation after HF etching. Reprinted with permission from [42] © The Optical Society

Micro-LED Technology for Display Applications

289

It was confirmed that LEDs that had undergone ALD passivation emitted uniform light with high optical output power, and even when the LED size was reduced to 10 µm, they showed higher optical output power than those fabricated under the other conditions. In other words, they showed that sidewall damage leading to non-radiative surface recombination can be reduced when manufacturing the LED through wet chemical etching and ALD sidewall passivation. This means that dielectric sidewall passivation not only increases LEE but also suppresses the leakage current. However, despite trying to solve the instability and low efficiency of the device, the improvements in green and red micro-LEDs lag behind those for blue ones. The same UCSB group has recently reported a much-reduced sidewall effect from AlGaInP-based red micro-LEDs. With ALD passivation and wet treatment, they obtained 150% at 100 A/cm2 , which shows a big improvement than before but still not as good as with the blue LEDs [38]. Research on how to connect the micro-LEDs produced through this process to an active/passive matrix or a backplane composed of TFT or CMOS is also being conducted by researchers such as Dupré [43] and Bulashevic [37]. These studies include the development of TFTs, improvements in p-contact to minimize current crowding, and the optimization of electrode shapes and wiring arrangements.

2.3.3

Kerf Loss and Singulation

While the stable growth technology of micro-LEDs and the manufacturing technology for LEDs to improve efficiency are being researched, developing the transfer technology for mass production of displays first requires a solution to the problem of kerf loss in micro-LED manufacturing. To separate the LEDs from the growth substrate and divide them into individual chips, cutting them to the desired size using dry or wet etching is essentially required. However, the separation method using the laser scribing and dicing saw, which are mainly used in conventional LED production, can be factors that limit the number of LEDs per wafer in the manufacturing process for a micro-LED. As shown in Fig. 14, in the manufacturing process of conventional LEDs and micro-LEDs that are currently being produced, this 3–20 µm area causes serious wafer loss proportional to the size reduction of the LEDs. Based on 6-inch wafers, this cutting technique can produce 660 million and 165 million micro-LEDs (for 3-µm and 8-µm-sized, respectively). Kerf loss, which is more influential when using the pick-and-place method, is more strongly affected by the individual chip splitting technology than monolithic transfer technology. This refers to the area lost by cutting in the semiconductor fabrication process of dividing the wafer into individual chips using dicing sows, etc. Traditional laser cutting produces a kerf width of 20 µm [44], which is very large considering the size of micro-LEDs. LuxVue developed a wafer cutting technique using RIE and reported a kerf loss of 2 µm, which is the smallest produced by cutting techniques known to date but is also close to the micro-LED size of 5,000 dpi pixel density fabricated using monolithic hybrid integration process, in SID 2018 DIGEST, vol. 18 (2016), p. 1677 16. M. Meitl, E. Radauscher, S. Bonafede, D. Gomez, T. Moore, C. Prevatte, B. Raymond, B. Fisher, K. Ghosal, A. Fecioru, A.J. Trindade, D. Kneeburg, C.A. Bower, Passive matrix displays with transfer-printed microscale inorganic LEDs. SID Symp. Dig. Tech. Pap. 47 (2016) 17. C.W. Sun, C.H. Chao, H.Y. Chen, Y.H. Chiu, W.Y. Yeh, M.H. Wu, H.H. Yen, C.C. Liang, Development of micro-pixellated GaN LED array micro-display system. SID Symp. Dig. Tech. Pap. 42, 1042–1045 (2012) 18. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 83 19. Y. Zhao, H. Fu, G.T. Wang, S. Nakamura, Toward ultimate efficiency: progress and prospects on planar and 3D nanostructured nonpolar and semipolar InGaN light-emitting diodes. Adv. Opt. Photon. 10, 246–308 (2018) 20. J.M. Smith, R. Ley, M.S. Wong, Y.H. Baek, J.H. Kang, C.H. Kim, M.J. Gordon, S. Nakamura, J.S. Speck, S.P. DenBaars, Comparison of size-dependent characteristics of blue and green InGaN microLEDs down to 1 µm in diameter. Appl. Phys. Lett. 116, 071102 (2020). https:// doi.org/10.1063/1.5144819 21. J.T. Oh, S.Y. Lee, Y.T. Moon, J.H. Moon, S. Park, K.Y. Hong, K.Y. Song, C. Oh, J.I. Shim, H.H. Jeong, J.O. Song, H. Amano, T.Y. Seong, Light output performance of red AlGaInP-based light emitting diodes with different chip geometries and structures. Opt. Express 26(9), 11194 (2018) 22. S. Ishimoto, D.P. Han, K. Yamamoto, R. Mano, S. Kamiyama, T. Takeuchi, M. Iwaya, I. Akasaki, Enhanced device performance of GaInN-based green light-emitting diode with sputtered AlN buffer layer. Appl. Sci. 9, 788 (2019) 23. https://plesseysemiconductors.com/red-ingan-on-silicon-microleds/ 24. https://www.glo.se/ 25. R.H. Horng, H.Y. Chien, F.G. Tarntair, D.S. Wuu, Fabrication and study on red light micro-LED displays. IEEE J. Electron. Dev. 6, 1064–1069 (2018) 26. T. Jung, J.H. Choi, S.H. Jang, S.J. Han, Review of micro-light-emitting-diode technology for micro-display applications. SID Symp. Dig. Tech. Pap. 50, 442–446 (2019) 27. S. Zhou, H. Wang, Z. Lin, H. Yang, X. Hong, G. Li, Study of defects in LED epitaxial layers grown on the optimized hemispherical patterned sapphire substrates. Jpn. J. Appl. Phys. 53, 025503 (2014) 28. J. Yu, Z. Hao, L. Li, L. Wang, Y. Luo, J. Wang, C. Sun, Y. Han, B. Xiong, H. Li, Influence of dislocation density on internal quantum efficiency of GaN-based semiconductors. AIP Adv. 7, 035321 (2017) 29. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 108

304

D.-S. Lee and J.-H. Han

30. H. Jeong, H.J. Jeong, H.M. Oh, C.H. Hong, E.K. Suh, G. Lerondel, M.S. Jeong, Carrier localization in In-rich InGaN/GaN multiple quantum wells for green light-emitting diodes. Sci. Rep. 5, 9373 (2015), https://creativecommons.org/licenses/by/4.0/deed.en 31. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), pp. 111–112 32. F. Olivier, S. Tirano, L. Dupré, B. Aventurier, C. Largeron, F. Templier, Influence of sizereduction on the performances of GaN-based micro-LEDs for display application. J. Lumin. 191, 112–116 (2017) 33. F. Olivier, A. Daami, C. Licitra, F. Templier, Shockley-Read-Hall and Auger non-radiative recombination in GaN based LEDs: a size effect study. Appl. Phys. Lett. 111, 022104 (2017). https://doi.org/10.1063/1.4993741 34. D. Hwang, A. Mughal, C.D. Pynn, S. Nakamura, S.P. DenBaars, Sustained high external quantum efficiency in ultrasmall blue III–nitride micro-LEDs. Appl. Phys. Express 10, 032101 (2017) 35. Z.L. Li, K.H. Li, H.W. Choi, Mechanism of optical degradation in microstructured InGaN light-emitting diodes. J. Appl. Phys. 108, 114511 (2010) 36. K.A. Bulashevich, S.S. Konoplev, SYu. Karpov, Effect of die shape and size on performance of III-nitride micro-LEDs: a modeling study. Photonics 5, 41 (2018) 37. H.W. Choi, C.W. Jeon, M.D. Dawson, Mechanism of enhanced light output efficiency in InGaNbased microlight emitting diodes. J. Appl. Phys. 93(10), 5973 (2003) 38. M.S. Wong, J.A. Kearns, C. Lee, J.M. Smith, C. Lynsky, G. Lheureux, H. Choi, J. Kim, C. Kim, S. Nakamura, J.S. Speck, S.P. Denbaars, Improved performance of AlGaInP red micro-light-emitting diodes with sidewall treatments. Opt. Express 28(4), 5787 (2020) 39. P. Tian, J.J.D. McKendry, Z. Gong, B. Guilhabert, I.M. Watson, E. Gu, Z. Chen, G. Zhang, M.D. Dawson, Size-dependent efficiency and efficiency droop of blue InGaN micro-light emitting diodes. Appl. Phys. Lett. 101, 231110 (2012) 40. J. Zhan, Z. Chen, Q. Jiao, Y. Feng, C. Li, Y. Chen, Y. Chen, F. Jiao, X. Kang, S. Li, Q. Wwan, T. Yu, G. Zhang, B. Shen, Investigation on strain relaxation distribution in GaN-based µLEDs by Kelvin probe force microscopy and micro-photoluminescence. Opt. Express 26(5), 5265 (2018) 41. Y.C. Chu, M.H. Wu, C.J. Chung, Y.H. Fang, Y.K. Su, Micro-chip shaping for luminance enhancement of GaN micro-light-emitting diodes array. IEEE Electron Device Lett. 35(7) (2014) 42. M.S. Wong, D. Hwang, A.I. Alhassan, C. Lee, R. Ley, S. Nakamura, S.P. Denbaars, High efficiency of III-nitride micro-light emitting diodes by sidewall passivation using atomic layer deposition. Opt. Express 26(16), 21324 (2018) 43. L. Dupré, M. Marra, V. Verney, B. Aventurier, F. Henry, F. Olivier, S. Tirano, A. Daami, F. Templier, Processing and characterization of high resolution GaN/InGaN LED arrays at 10 micron pitch for micro display applications. SPIE Opto. 10104, 1010422–1010431 (2019) 44. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 116 45. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 34 46. J. Claypole, A. Holder, C. McCall, A. Winters, W. Ray, T. Claypole, Inorganic printed LEDs for wearable technology. Proceedings 32, 24 (2019) 47. US patent: US 9,722,145 B2 48. M. Choi, B. Jang, W. Lee, S. Lee, T.W. Kim, H.J. Lee, J.H. Kim, J.H. Ahn, Stretchable active matrix inorganic light-emitting diode display enabled by overlay-aligned roll-transfer printing. Adv. Funct. Mater. 27, 1606005 (2017) 49. V.R. Marinov, O. Swenson, Y. Atanasov, N. Schneck, Laser-assisted ultrathin bare die packaging: a route to a new class of microelectronic devices. Proc. SPIE 8608 (2013) 50. B.R. Tull, Z. Basaran, D. Gidony, A.B. Limanov, J.S. Im, I. Kymissis, V.W. Lee, High brightness, emissive microdisplay by integration of III-V LEDs with thin film silicon transistors. SID Symp. Dig. Tech. Pap. 46(1), 375–377 (2015) 51. US patent: US20160293586 52. US patent: WO2014053445

Micro-LED Technology for Display Applications

305

53. US patent: US9177992 54. J.J. Lee, Technology and latest results on fluidic assembly of micro LEDs, in Micro LED Display Conference 2019 (2019) 55. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 193 56. H.M. Kim, M. Ryu, J.H.J. Cha, H.S. Kim, T. Jeong, J. Jang, 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 (2019) 57. N. Sugiura, C.T. Chuang, C.T. Hsieh, C.T. Wu, C.Y. Tsai, C.H. Lin, C.C. Liu, C.Y. Liu, C.N. Yeh, C.Y. Liu, Y.C. Lin, 12.1-inch 169-ppi full-color micro-LED display using LTPS-TFT backplane. SID Symp. Dig. Tech. Pap. 50, 450–453 (2019) 58. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 141, 163 59. E. Virey, MicroLED displays: hype and reality, hopes and challenges. Yole report (2017), p. 208

Display Techniques for Augmented Reality and Virtual Reality Byoungho Lee and Youngjin Jo

Abstract Augmented reality (AR) and virtual reality (VR) have received a lot of attention and expectation with the experiences never possible before. The recent development of the display technology has resulted in high-resolution images indistinguishable from reality. However, unlike the table-top displays, the head-mounted AR and VR displays have additional factors to be considered such as the form factor of the device, the field of view, and the visual fatigue. In this chapter, some of the key technical issues for optically configuring the display in VR and AR are discussed.

1 Virtual Reality Display Virtual reality (VR) displays allow users to have more immersive experiences than conventional flat panel displays. Their wide field of view (FOV) and high resolution create an optical illusion that the users exist in the virtual space. To give the perception cues of three-dimensional (3D) depth, the VR display provides different images separately to the left and right eyes. As the display technology gradually evolves and the related hardware and software improve, the VR display is gaining more attention. The VR is realized by a variety of factors. Typically, they are a virtual environment simulated by a computer, an interface that connects it with the users, and a human perception process that accepts it. When all those things are provided in harmony, the users can have an immersive experience. With the advancement of technology, the VR content has changed from the passive experience like one-person theater to the active interaction with the virtual environment. Consequently, the performance of the sensors used in VR devices and the processing speed including position sensing and image rendering speed of the system have become very important. B. Lee (B) · Y. Jo School of Electrical and Computer Engineering, Seoul National University, Gwanak-gu Gwanakro 1, Seoul 08826, Korea e-mail: [email protected] Y. Jo e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 I. B. Kang et al. (eds.), Advanced Display Technology, Series in Display Science and Technology, https://doi.org/10.1007/978-981-33-6582-7_13

307

308

B. Lee and Y. Jo

Table 1 PC-based VR headsets [2–5] Name

Oculus Rift S

HTC Vive Pro

Playstation VR

Valve Index

Manufacturer

Lenovo

HTC

Sony

Valve

Field of view

90°

110°

100°

130°

Resolution

2560 × 1440

2880 × 1600

1920 × 1080

2880 × 1600

Refresh rate

80 Hz

90 Hz

120 Hz

144 Hz

Display

LCDa

AMOLEDb

OLED

LCD

a LCD:

liquid crystal display; b AMOLED: active-matrix organic light-emitting diode

Table 1 shows the personal computer (PC) based VR headsets and controllers. These systems have disadvantages that they require external computing devices, cables, sensors, and controllers for motion detection. Therefore, even if the device performance is limited, there is a strong demand for the standalone VR headsets that can be driven independently. For example, using the mobile VR display recently released with high resolution increases the accessibility with a lower price [1]. In particular, in the 5G era, as real-time video and data transmission are possible with low delay, the streaming type VR headsets are highly anticipated. The display panels used in VR systems typically have higher resolution than full high definition (FHD), but the display resolution is still insufficient. According to Advanced Micro Devices (AMD) Inc., a global semiconductor company, 16 K resolution was selected as a target for ‘true immersion’ in VR display. This surpasses the performance of current commercial VR headsets, and not only the displays, but also the processing speed must be improved. As a way to alleviate this requirement, the research has been conducted to provide high-resolution images only in the area responsible for sharp central vision [6]. Most commercial VR headsets in the market are commonly based on the simple optical system. It consists of a screen panel and eyepiece lenses. As shown in Fig. 1, a lens, which is often called an eyepiece, floats the virtual image of the screen, and the magnified screen is displayed to an observer. The image location L i is determined by the lens equation as follows: Li =

Ld f , f − Ld

(1)

where f and L d denote the focal length of the lens and the distance between the lens and the panel, respectively. Here L d is smaller than f . Normally, it is designed to be between 1.5 m and 3.0 m for commercial VR headsets. The image size Si is determined by geometric relations as follows: Si = Sd

Li , Ld

(2)

Display Techniques for Augmented Reality and Virtual Reality

309

Fig. 1 Schematic diagram of an optical system for typical VR headset

where Sd means the display size. One of the important design variables in VR optics is the FOV, which largely affects the immersion of the observer. The FOV θ is defined as follows:   Si −1 , (3) θ = 2 tan 2(L e + L i ) where L e is an eye relief, which means the distance between the observer’s eye and the eyepiece lens. Although the FOV can be extended by reducing the eye relief, it should be designed to maintain a proper distance considering users who wear glasses. It is normally set between 13 and 25 mm in typical VR headsets [7]. Although the magnifying optics can achieve moderate design parameters such as FOV and eye-box, they may cause user’s fatigue and visual discomfort, especially for long-time usage [8]. When the human eye locates a 3D object in the real world, not only binocular disparities, but also the focusing state of the crystalline lens, which is the focus-tunable eye lens, induce depth perception. The depth information of the object is inferred from the disparities, and it induces a change in the curvature of the eye lens and therefore, the focusing state of the eye lens is adjusted. However, typical VR headsets can display virtual images at a fixed depth plane, so that the focusing capability is not supported by them. Therefore, the individual depth cues of binocular disparities and accommodation of the eye result in different depth perceptions, and they become decoupled in the brain when using the VR headsets. It is known as vergence-accommodation conflict (VAC) [9]. A recent study has reported that the

310

B. Lee and Y. Jo

absence of focusing capability also deteriorates the depth perception in terms of ordering depth. Resolving VAC by supporting focus cues in the VR headsets has been widely studied, and various 3D display methodologies have shown to be effective for it [10].

2 Reflective Augmented Reality Display A typical method used to form an optical system for augmented reality (AR) near-eye displays (NEDs) is the use of a half-mirror as a beam splitter. Half-mirror is an optical element that can control the ratio between reflected light and transmitted light by a special metallic coating on the glass. As shown in Fig. 2a, it can combine two beams, or divide light into two beams. Normally the ratio of reflection and transmission of light is made at 50:50, but the reflectivity is controlled by different metal coatings according to the purpose of use. The basic principle of AR NED is the miniaturization of the half-mirror system. As shown in Fig. 2b, the image displayed from the micro-display device is reflected at a half-mirror tilted by 45° and visible to the observer. Since the light reflected from the actual object passes through the half-mirror, observers see the artificial image merged with the outer scene. In AR NED using half-mirror, the user observes that the displayed image is located close to the eye because the distances among the display unit, mirror, and eye are very small. Unless the actual object exists immediately in front of the eye, the actual object and displayed images cannot be seen at one sight. In practice, a lens is used to solve this problem. As shown in Fig. 3a, the displayed image can be placed far away from the observer by placing the lens closer than the focal length of the lens after the micro-display. The actual AR NED uses concave mirrors instead of convex lenses since they are optically identical, but mirrors are lighter and do not have chromatic aberration. In Fig. 3b, the system, called ‘bird bath’, is shown using a half-mirror and a concave mirror. When using the lens to locate the displayed image far from the observer, the displayed image is enlarged due to the optical property of the lens. The light rays

Fig. 2 a Scheme of half-mirror and b optical system for AR NED using half-mirror

Display Techniques for Augmented Reality and Virtual Reality

311

Fig. 3 a Principle of controlling the depth of the displayed image using lens, and b birdbath system using a concave mirror instead of a lens

emitted from the edge of the enlarged display form the angle defined as the FOV. Since the half-mirror is arranged in a tilted form in AR NED, the bigger FOV is achieved, the larger size it has. Due to this trade-off relationship, there are two types of onaxis reflective AR display [11, 12]. One, an immersive binocular display, which has wide FOV but large volume. The other one, a compact monocular display, which has small volume, but narrow FOV. Figure 4 shows the optical structure of the compact monocular displays. Both immersive binocular displays and compact monocular displays are implemented using birdbath designs. In order to reduce the size of the optical system, coated prism with a curved surface was used instead of a separate curved mirror. And the micro-display was embedded in the side so as not to interfere with the external eyesight. For micro-displays, liquid crystal on silicon (LCoS) or micro-organic lightemitting diode (micro-OLED) panels are mainly used. LCoS panels have the disadvantage of requiring additional light sources, while micro-OLED panels have the disadvantage that the brightness is not yet enough to be used under daylight.

Fig. 4 Design of compact monocular display using birdbath structure

312

B. Lee and Y. Jo

As described above, an on-axis reflective optical system, such as a birdbath, has a disadvantage in that its volume becomes large as the FOV increases. To compensate this problem, AR displays using off-axis reflective optical systems, multiple curved mirrors, or freeform optical systems have recently emerged. By placing appropriate mirrors in the path of multiple oblique light beams starting from the display, as shown in Fig. 5, a larger FOV can be obtained with a smaller thickness than conventional birdbath structures. However, the off-axis reflective optical system has the aberration due to the oblique incidence of light. Optical systems using freeform optics have been introduced recently to correct aberration [14]. In Fig. 6, the light from the micro-display enters through the optical surface S4 and is totally reflected at S2. At this time, the light is reflected from the semi-mirrorcoated plane S3 and enters the human eye. Through the prism E added in front, the outer scene also enters the human eye without distortion. Due to freeform surfaces, we can observe clean virtual and real images without aberrations caused by oblique incidence. In this way, using the off-axis freeform optical system can offset the

Fig. 5 Diagram of off-axis reflective AR optical system using multiple curved mirrors

Fig. 6 AR optical system using two freeform prisms, adapted with permission from [13], © The Optical Society

Display Techniques for Augmented Reality and Virtual Reality

313

drawbacks of the reflective optical system, but the high manufacturing cost and the large volume of the optical system remain as disadvantages.

3 Waveguide-Based Head-Mounted Displays The reflection-based head-mounted display (HMD) systems using a curved mirror or conventional lens described in the previous chapter have a serious problem that bulky and heavy imaging combiners should be placed in front of the viewer’s eye. This form-factor problem can cause discomfort when used in daily life. The waveguidebased or lightguide-based techniques are a very promising technology that can resolve the form-factor issue. The waveguide method uses a total internal reflection (TIR) principle to propagate the image light of the display through a thin waveguide to the front of the user’s eyes. By using a thin substrate for the waveguide, the display image can be theoretically delivered to the eye without loss of light. The display modules and various optical components can be shifted to the side face, and an unobstructed view of the real world can be achieved. The image combiners used in the waveguide systems utilize a thin and lightweight reflector and the diffractive optical element (DOE) differently from the conventional reflection-type system. These imaging combiners are generally attached to the guide substrate or engraved on the substrate surface. The compact form-factor has the great advantage of offering the possibility to utilize AR HMD devices in various fields, and AR development companies have launched commercial products or prototypes with the waveguide system. The Microsoft HoloLens and Magic Leap One are representative commercialized waveguide-based HMDs. These products are typical goggle-types to be mounted on the head, which are larger and heavier than spectacles [15]. The AR HMDs with high performance for immersive experiences are equipped with many cameras, sensors, and electronic systems for the mixing of a real scene and virtual images, and are bulky by non-optical elements such as batteries. Except for these components, the optical module is compact and lightweight. The smart-glasses, such as Vuzix’s Blade, which provide moderate visual performance while adequately reducing the product size and removing the several non-optic components, are also readily available [16]. Figure 7 shows the optical structure of the conventional waveguide systems. For compact form-factor, the micro-display or projection display system is adopted. For example, the display engine of the Microsoft HoloLens 1 consists of red/green/blue (RGB) light emitting diode (LED) sources and an LCoS micro-display with 0.57 in. The divergent lights emitted from the micro-display is converted into parallel beams by a collimation lens. This collimating method replaces spatially arranged display pixel information with the parallel beam propagating at a specific incidence angle. The virtual image incident on the waveguide is extracted to the eye in the form of parallel beam while maintaining the original input angle by the waveguide and the imaging combiner (in-coupler and out-coupler). Finally, the output beam is focused by the eye lens, and the virtual image is displayed at an infinite depth.

314

B. Lee and Y. Jo

Fig. 7 Schematic diagram of the waveguide-based HMD systems

The in-coupler is required to deflect the input beam from the display module to a particular angle satisfying TIR conditions. Similarly, the out-coupler redirects the light along the waveguide to the eyes. The image combiner can be divided into a reflection-type and a diffraction-type. The reflection-type waveguide system uses a semi-reflective mirror array inserted into the waveguide substrate. As shown in Fig. 8, the multiple reflections of the propagating beam by the mirror array are achieved, which can provide a wide viewing angle and eye-box [17]. Diffraction-type waveguide display on-the-shelf utilizes DOEs for their image combiners. DOEs have a precise grating structure, which diffracts the light to the intended direction. By delicately designing the structure, the arbitrary diffraction angle can be made, improving the degree of freedom in waveguide system design.

Fig. 8 The schematic diagram of the reflection-type waveguide

Display Techniques for Augmented Reality and Virtual Reality

315

DOEs can also be produced with small form factors and lightweight, which makes them more appealing to consumers. In commercial products, two types of DOEs are generally used; surface-relief gratings (SRGs) and holographic volume gratings (HVGs). The Microsoft HoloLens and Magic Leap One have SRGs for their image combiners. SRGs are manufactured by etching the diffractive gratings on the surface of the substrate. There are a variety of design variables, such as the shape, height, and period of the grating. The diffraction angle and efficiency for each wavelength of the input beam are highly dependent on those variables. Therefore, a total of three different layers of gratings is required to display RGB colors. Microsoft launched the first-generation model of HoloLens in 2015. The display module of HoloLens has HD-resolution per eye with a 16:9 ratio, and the diagonal FOV is 35°. The holographic density is higher than 2.5 K light points per radian [18]. In 2019, the upgraded model HoloLens 2 was released. The display of HoloLens 2 has 2 K resolution per eyeball with a ratio of 3:2, and the holographic density is similar to that of the first generation. The diagonal FOV of HoloLens 2 is 52°, which is highly improved compared to the first generation [7]. Magic Leap One is also a popular commercial waveguide-based display. Dissimilar to HoloLens, it provides two different depths of the virtual scene. Those depth planes are realized by inserting two layers of SRGs for each RGB color and depth. The display’s illumination module has several spatially separated exit pupils so that the proper image could be introduced to the intended layer’s in-coupler. The resolution is 1280 by 960 per eye, and the diagonal FOV is 50° [19, 20]. Although those products are among the most popular waveguide displays, the SRGs deteriorate the quality of the real scene by the grating on the surface. A periodic grating structure undesirably diffracts the light from the real scene, inducing duplicated ghost images. Also, the cost of the etching process and difficulty in inscribing the optimal grating structure limit the overall production yield. To resolve these issues, HVGs have been researched as a promising image combiner. HVGs diffract the light by the periodic index or absorption modulation in the medium. The volume grating can be recorded in the recording medium by interfering with two different coherent beams. Those beams are called the reference beam and the signal beam each. When the two beams interfere inside the medium, the refractive index of the medium is modulated by the intensity of the interference pattern, thus recording the grating structure. Once the reference beam is introduced to the recorded grating, several partial reflections occur by the modulated refractive index. Those partially reflected lights interfere with each other, and only the signal beam matching the Bragg condition can be extracted. When the incident beam slightly mismatches the Bragg condition, the input beam is diffracted with lower diffraction efficiency. Other beams that significantly mismatch the Bragg condition (such as the light from the real scene) remain intact. Therefore, the HVGs have a high transmittance for the real scene than that of the SRGs. The HVGs are manufactured by the optical recording, and its simple mass-production process can manage the shift to the required design specifications. Furthermore, the HVGs can be recorded in an ultra-thin film medium (less than 100 µm), reducing the total thickness of the waveguide.

316

B. Lee and Y. Jo

Table 2 Comparison table of various commercial waveguide-based HMDs [18, 21–24] Name

LUMUS OE Vision

HoloLens 1

HoloLens 2

Magic Leap One

Sony SED-E1

Type

Reflective

Diffractive (SRG)

Diffractive (SRG)

Diffractive (SRG)

Diffractive (HVG)

Resolution

1920 × 1080

1268 × 720

1440 × 936

1280 × 960

419 × 138

Aspect ratio

16:9

16:9

3:2

4:3

3.04:1

Number of depths

1

1

1

2

1

Diagonal FOV

40°

35°

52°

50°

20°

SONY SED-E1 is a commercial product with HVGs. The resolution of the display is 419 by 138 and the device only shows the green monochrome virtual images. The FOV is 20° diagonally. The FOV could be expanded by the multiplexing method of the HVGs. This product was released in 2015, and it still requires further development. Table 2 shows the major specifications of various commercial waveguide-based HMDs.

4 Pin-Light Augmented Reality Display In 2014, NVIDIA introduced a pin-light AR display, which has the strong advantage of having a wide FOV with a simple structure [25]. The demonstration of the system showed 110° of diagonal FOV, which is quite wider compared to 30–40° of commercialized products at that time. The system mainly consists of two optical elements: a transparent light source array using the form of a laser-etched waveguide shown in Fig. 9 and an LCD.

Fig. 9 a Defocused point light sources manufactured by b laser drilling and etching system

Display Techniques for Augmented Reality and Virtual Reality

317

Fig. 10 Basic principle of pin-light display

Figure 10 illustrates the basic principle of a pin-light display. Suppose a point light source is at close enough distance to the eye so the observer even cannot focus on it. Then, the bundle of light from the source will not converge into a single point at the retina but will be projected into a circular shape, which is the shape of the pupil. If we modulate this circular image on the retina by putting an LCD between the source and pupil, the observer will see an image in the circle instead of just a white circle. This kind of near-eye display is classified as a retinal-projection display since the image is literally projected on the retina. Retinal projection displays have all-in focus property that allows observation of clear images regardless of the focal state of the eye lens. Each beam for each pixel originates from the very small light source and passes through each small pixel of the LCD, so has a very thin beam-width compared to the human pupil diameter. Since the blur effect under the out-of-focus state is proportional to the beam-width, the thin beam for each pixel is projected clearly into the retina even in the out-of-focus state. This principle is similar to that of pinhole cameras which can capture all-in-focus images (Fig. 11). If we configure a pin-light display using only a single light source, the achievable FOV is limited. The FOV gets larger with shorter eye-relief and larger human pupil diameter, but both factors have practical limitations. Approximately one light source can generate a FOV of around five degrees. However, when multiple light sources are used instead of a single one, each light source can generate a circular image at different positions on the retina. Therefore, an array of multiple light sources needs to be used for large FOV. A transparent light source array is easily configurable using a waveguide as shown in Fig. 9. When small dots of cracks are arranged on the waveguide surface, light travels through the inner surface and leaks through the cracks, forming an array of light sources. Here, a number of parameters, such as arrangement pattern, spacing between the sources, distance from the eye, location of the LCD, and the displayed image, must be accurately designed to give observers a clear view of the desired image.

318

B. Lee and Y. Jo

Fig. 11 Pin-light array for enlarged FOV

The arrangement and spacing of the sources should be determined so that the circular projection generated by each source can cover the FOV at the retinal plane most efficiently. Therefore, it is selected to have a hexagonal grid array. And to avoid overlapping between adjacent circular projection images, only the hexagonal portion of each circle is used for display. In addition, the image displayed on the LCD is flipped by 180° within each circular projection on the retina, so the LCD should have a pre-inverted image displayed. As a result, when the image shown in Fig. 12b is displayed on the LCD, the user will observe the matched image as shown in Fig. 12c. The pin-light display projects the virtual image directly onto the retina. In general, the retinal projection type uses a projector for the display part and transmits it to the observer’s retina with optical elements such as lenses and beam splitter. However, the pin-light display does not use a projector for a virtual display. Instead, a light source (pin-light array) and a modulating LCD are located right in front of the eye pupil. This unique configuration makes the characteristics of the system different from most retinal projection techniques. First, an extremely wide FOV can be achieved. Since the virtual image represented by the LCD is observed to the user, the FOV of the

Fig. 12 a Target image. b Displayed on LCD. c Image on observer’s retina

Display Techniques for Augmented Reality and Virtual Reality

319

system can be calculated by a geometric angle between the pupil and the LCD. If the horizontal length of the LCD is c, the FOV is calculated as follows: F O V = 2 tan

−1



 c . 2dm

(4)

Also, any optical or physical element does not exist between the pupil and the LCD, so the FOV is determined by the above equation only. Therefore, by using a large-sized LCD and reducing the distance between the pupil and the LCD, the FOV can be increased. If an LCD with a width of 36.9 mm is placed at a distance of 15 mm from the pupil, a wide FOV of about 100° can be obtained. Although the pin-light display can provide a wide FOV, there is a limitation that it cannot provide a high-resolution image. In order to provide a high-resolution image, the pixel pitch of the LCD panel should be smaller. The half angle at which light emitted from one pin passes through one pixel of the LCD is given by θp =

p , 2(d p − dm )

(5)

where p is the pixel pitch of the LCD and the paraxial approximation is adopted. Therefore, when only the resolution of the virtual image is considered, it is possible to provide high-angular resolution by reducing the pixel pitch. However, in AR displays, light from the real scene also should be considered. In the pin-light display, light from the real scene is transmitted through the LCD. Since the LCD has a pixelated structure, the real scene undergoes diffraction. The maximum diffraction angle of LCD is calculated by θd =

λ . p

(6)

The smaller the pixel pitch, the larger the diffraction, so Eq. (5) for determining the resolution of VR and Eq. (6) for determining the resolution of the real world are in a trade-off relationship with a pixel pitch of p. The maximum angular resolution of a pin-light display can be found by the intersection of the two conditions. And the resolution is given by 2–5 pixels/degree from the proposed prototype [25]. In addition to these resolution issues, pin-light displays have the limitation that it is difficult to observe a complete virtual image due to its small eye-box. The design parameters of the pin-light display are related to the pupil size and the focus state of the observer. These two parameters vary depending on the environment. For example, the size of the pupil becomes smaller when the environment is bright and becomes larger in a dark environment. The focal length of the human eye changes depending on the focusing depth. However, the spacing of the designed pin array, the LCD, the observer, and the image to be displayed on the LCD are designed according to the specific pupil size and focal length condition. Therefore, the entire image can be seen when the pupil is located at a designed position, and the surrounding environment

320

B. Lee and Y. Jo

Fig. 13 Observed image when the design value and the actual user’s eye condition are different. a When the pupil size is smaller than the designed value. b The lens focal length is closer than the design value. c The pupil’s position deviates from the designed position. d The distance to the pupil is closer than the designed value

is limited. Figure 13 shows the result of simulating the image observed when these design parameters differ from the actual observer’s eye condition.

5 Light Field Display Light is an electromagnetic wave. Assuming the wavelength of the electromagnetic wave to be infinitesimally small, the corresponding wave can be assumed as a bundle of rays with specific propagation directions. The radiance function depending on the spatial coordinates is introduced as a light field. In the presentation of the light field, five dimensions are necessary. However, based on the assumption that the intensity profile is constant regardless of propagation, the light field can be represented with four variables. Hence, four-dimensional (4D) light field presentation is widely used in the application of 3D displays and 3D imaging.

Display Techniques for Augmented Reality and Virtual Reality

321

Light Field Display Human visual system can recognize 3D information when observing different images through binocular eyes. 3D displays can be classified into two types: stereoscopic and autostereoscopic display, depending on whether additional devices such as glasses are worn to provide binocular disparity. In the case of stereoscopic 3D display, different images are provided to binocular eyes through the operation of the head-worn device. This provides different images to two eyes through spatial multiplexing, polarization multiplexing, or temporal multiplexing without additional image processing. Alternatively, a representative autostereoscopic 3D display is implemented by reproducing a 4D light field. At least two physical planar devices are required to realize the autostereoscopic 3D display through the reproduction of a 4D light field. There are various types of autostereoscopic 3D displays depending on the type of device used and the light field mapping method, and they are collectively referred to as light field display in a wide range. Autostereoscopic 3D Display: Parallax Barrier, Integral Imaging The autostereoscopic 3D display is generally implemented through at least two or more physical devices as mentioned above. Among these, the most basic autostereoscopic 3D display is a one-dimensional multi-view display in which only one directional parallax exists. This is mainly achieved by placing a parallax barrier or a cylindrical lens array in front of the display panel [26]. Figure 14a, b shows an autostereoscopic 3D display that provides unidirectional parallax using (a) parallax barrier or (b) cylindrical lens array. After mapping the 4D light field based on the positions of the human eye and the parallax barrier (or the cylindrical lens array), the information is provided to the display panel. After this procedure, images suitable for each viewpoint are properly distributed to the display panel. The number of viewpoints varies according to the system specification and viewing condition. Unlike an autostereoscopic 3D display that provides parallax in one direction, the realization of a 3D object through integral imaging supports a parallax in every direction. In this approach, a pinhole array or a lens array serving as a twodimensional parallax barrier is placed in front of the display. Figure 14c, d shows the elemental image acquisition process and the 3D object reconstruction process through integral imaging, respectively. Acquisition of elemental image is conducted by mapping display patterns and object points. Unlike the parallax barrier method, it is advantageous as bi-directional parallax can be supported. Conversely, the maximum resolution of a single view image is degraded by the factor of the number of views. Tensor Display Tensor display [27] is a technology that reproduces a 4D light field using two or more display panels. The main feature of the tensor display is that it stacks the display panels in multiple layers to determine the 3D information of the light field. The key idea of a tensor display is that if the 2D information array for each display panel is properly combined, the information in each direction can be reproduced and hence form a 4D light field. The combination of two-dimensional information

322

B. Lee and Y. Jo

Fig. 14 Autostereoscopic 3D display based on one- and two-dimensional multi-view display. a The parallax barrier and b the cylindrical lens array provide different images for the left and right eyes. In integral imaging, c the direction of the element image acquisition process is the opposite of d the direction of 3D object reconstruction

arrays for reproducing an arbitrary four-dimensional light field can be obtained by an optimization process through numerical analysis. This is different from the method of determining the direction of light for each 2D pixel, such as a multi-view display or integral imaging. The optical structure of the tensor display differs depending on the purpose of use and the display combination method, but the core principle of expressing the 4D light field is not very different. The principle shared by most tensor displays is that the direction of light emitted from each 2D pixel of a display is not determined through a special optical system. The tensor display expresses a 4D light field by utilizing the fact that the light intensity in each direction is determined by different pixel combinations. Therefore, the spatial resolution of the light field reproduced by the tensor display is not clearly determined by the optical structure. This feature enables to improve the resolution by a numerical analysis process, differentiated from the other 3D display technologies such as multi-view display or integral imaging. Numerical optimization is like solving a least-squares problem that minimizes the error value between a target 4D lightfield and a 4D lightfield reproduced on a tensor display. The result of this optimization process is a 2D array of pixels, which are the images to be uploaded to each display. The specific optimization method depends on the optical structure of the tensor display, especially how each 2D

Display Techniques for Augmented Reality and Virtual Reality

323

array is combined. For example, if different displays are provided in a time-division manner, and pixels are combined by addition, this problem can be solved through an optimization algorithm used for computed tomography [27]. If different displays are stacked and each pixel is combined by multiplication, a matrix factorization algorithm may be applied. Depth-Scanning Display The depth scanning display uses a method of generating multiple depth planes through synchronization of a high-speed driving projection system and a scattering screen that reciprocates. The high-speed projection system can update images thousands of times per second, which has the advantage of effectively reproducing multiple depth planes through a time-division method. Using this high-speed driving projection system, it is possible to scatter the desired image at the desired depth by projecting an appropriate image according to the depth of the scattering surface. When images scattered from various depths are combined in a time-division form, it is possible to merge them as 3D images in the human brain. In addition, the advantage of this methodology is that the image spreads at a wide angle through scattering, so that it can produce a natural visual effect in a wide viewing range. Tomographic Display The tomographic display [28] is a method of generating a large number of depth planes by synchronizing a high-speed driving display system with a reciprocating variable focus lens. Similar to the depth scanning display, the tomographic display also provides images of multiple depth planes to the user through a time division method. The high-speed passive display system used in tomographic displays can be made by combining a special spatially modulated high-speed backlight with a typical display panel. The spatially modulated high-speed backlight has the feature of being able to locally illuminate the pixels of the display panel, and the number of frames can be increased because the operation is determined in binary form (on/off) according to the input signal. Using a digital micro-mirror device, frames up to 14 kHz can be obtained, and using a typical LED array, frames above 1 kHz can be achieved.

6 Sensor and Display Techniques As AR/VR devices are gaining attention as a new paradigm after smartphones, new interfaces are demanded. Since conventional interfaces are not adequate for AR/VR devices, interfaces that allow users to move freely with easy use in the virtual world are required. Eye-tracking sensors, inertial measurement units, and 3D depth sensors are typical sensors for AR/VR device interfaces that help us to provide immersive experiences for AR/VR devices [29].

324

B. Lee and Y. Jo

Eye-Tracking Sensor The eye-tracking sensor tracks the position and movement of the eye of the user. In AR/VR devices, eye-tracking sensors are used for foveated image processing and enlarging the eye-box. Applying these methods overcomes the limit of AR/VR devices [30]. Optical tracking is the most widely used technology for eye-tracking. By irradiating near-infrared light into the eye and analyzing reflected images, which are called Purkinje images, it is possible to track eye movement. Inertial Measurement Unit The motion of the user must be considered to provide a natural virtual world to users. For instance, a commercialized AR device ‘Hololens’ uses SLAM (simultaneous localization and mapping) algorithm that provides a virtual image considering the relative movement of the user [31]. Accelerometer and gyroscope sensor are typical components of the inertial measurement unit. The accelerometer is based on the mass-spring-damper model and measures the translational movement of the user by the change of length of the spring. The gyroscope sensor measures the rotational movement of the user. The most widely used gyroscope sensor is a vibrating structure gyroscope, which utilizes the Coriolis force of vibrating object on the rotating system. 3D Depth Sensor For a realistic virtual world image, 3D information of the real world must be obtained. For instance, the occlusion effect, which provides a 3D effect by covering objects behind, can be implemented only with depth information of the real world. There are various ways to measure 3D depth. ToF (time of flight) method calculates the depth of the object by irradiating the light and measures the phase or delay time of the reflected light [32]. The stereo method uses two image sensors in different positions and angles; therefore, 3D depth is calculated by the disparity of two images [33]. The structured light method illuminates the object by specially patterned light such as checkerboards or stripes. The depth of the object is calculated by measuring the distortion of the reflected pattern. Display Techniques for Augmented Reality and Virtual Reality In order for the observer to experience virtual images similar to the real world, images having a viewing angle 120° or higher and a frame rate of 120 Hz or higher should be provided from the AR and VR device. Also, considering a person with a visual acuity of 1.0, it is possible to distinguish two points at an angle of up to 1/60°, so it is necessary to provide virtual images with a resolution of 60 cpd (cycles per degree) or higher. Figure 15 shows the number of pixels per angle according to the viewing angle of the AR and VR device while changing the resolution of the display panel [34]. The figure shows that considering the resolution of the human eye, at least 8 K display panel is required. However, the panel size is limited to a very small size

Display Techniques for Augmented Reality and Virtual Reality

325

Fig. 15 Relation between the field of view and angular resolution. The relation is plotted separately for the display resolution of 1 k, 2 k, 4 k, and 8 k

because the display panel should be mounted on a wearable device. Assuming the panel size is 1 cm, the pixel pitch is 1.25 µm, which is a very small size that exceeds the current fabrication technology. In addition, there are issues in that the latency of the panels must be very short, and display panels for AR and VR require controllers and processors to process high-capacity information. Various researches on LCD, OLED, and micro-LED (µLED) are being conducted to address these issues. Liquid Crystal Display Incident light to the LCD panel is modulated by a liquid crystal which has a different refractive index according to polarization state. As shown in Fig. 16 [35], the LCD

Fig. 16 System configuration of LCD panel

326

B. Lee and Y. Jo

consists of a polarizer, a glass substrate containing a transparent electrode, and liquid– crystal molecules. The LCD does not contain a self-luminous element and modulates the light from the backlight unit. LC-based display panel is also called a spatial light modulator (SLM) in the sense that it modulates incident light from the outside, and it is an important characteristic for adopting LC-based display panel for holographic display. Various researches for holographic AR and VR devices are being conducted by using a miniaturized reflection-type LC-based display panel which is called liquid–crystal on silicon. Organic Light-Emitting Diode The OLED is an LED that contains an electrically controllable emissive layer of organic compound. In contrast to an LC-based display panel, OLED display panel does not contain a backlight unit. Hence it can be fabricated with the thin and light form factor. The form factor of the display panel for AR and VR devices is an important issue, so the OLED panel receives much attention for AR and VR devices. Especially, micro-OLED panel called OLED on silicon (OLEDoS) is a good candidate because it has not only a very small form factor, but also all other advantages of OLED. However, it still has a drawback of low brightness for use in AR devices that provide virtual images mixed with the real-world scene. Micro Light Emitting Diode µLED display panel consists of an array of micro-scale LED pixels which form the individually controllable pixel elements. Compared to LC-based display panel, µLED display provides a higher contrast ratio, faster response time, higher energy efficiency, and wider color gamut. The µLED is considered as a next-generation display technique to replace the micro-OLED because µLED does not contain organic materials, so does not cause screen burn-in. It has been attracting attention because it can take the advantages of OLED, and also has high reliability. However, since µLED is fabricated by a sapphire-based process, the nano-level fabrication is very difficult compared to other silicon-based fabrication, and it takes much time to transfer µLED on a plastic substrate, which makes it expensive.

7 Summary We have discussed the background of AR and VR display technologies in optical design. AR and VR play a significant role in enabling the fourth industrial revolution that connects the digital and physical worlds. In order for AR and VR to become a part of our life, further development of various components and system technology is still required. In particular, more compact and high-resolution displays will directly affect the head-mounted display system. In addition to the improvement of such hardware devices, when the understanding of human perception, the proper optical design, and effective interface are well mixed in harmony, AR and VR will provide a higher level of an immersion experience and expand the real world.

Display Techniques for Augmented Reality and Virtual Reality

327

Acknowledgements This work was supported by the Institute of Information and Communications Technology Planning and Evaluation (IITP) Grant funded by the Korean Government (No. 20170-00787).

References 1. 2. 3. 4. 5. 6.

7. 8. 9. 10.

11. 12. 13.

14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25.

26.

Samsung Gear VR. https://www.samsung.com/global/galaxy/wearables/gear-vr/ Oculus Rift S. https://www.oculus.com/rift-s/ HTC Vive Pro. https://www.vive.com/eu/product/vive-pro/ Playstation VR. https://www.playstation.com/en-us/explore/playstation-vr/ Valve Index. https://store.steampowered.com/valveindex/ A. Patney, J. Kim, M. Salvi, A. Kaplanyan, C. Wyman, N. Benty, A. Lefohn, D. Luebke, Perceptually-based foveated virtual reality, in ACM SIGGRAPH 2016 Emerging Technologies, vol. 17 (2016) B.C. Kress, Optical Architectures for Augmented-, Virtual-, and Mixed-Reality Headsets (SPIE Press, Bellingham, Washington, USA, 2020). G.-A. Koulieris, B. Bui, M.S. Banks, G. Drettakis, Accommodation and comfort in headmounted displays. ACM Trans. Graph. (TOG) 36, 87 (2017) M. Lambooij, M. Fortuin, I. Heynderickx, W. IJsselsteijn, Visual discomfort and visual fatigue of stereoscopic displays: a review. J. Imaging Sci. Technol. 53, 30201–1 (2009) G.A. Koulieris, K. Ak¸sit, M. Stengel, R.K. Mantiuk, K. Mania, C. Richardt, Near-eye display and tracking technologies for virtual and augmented reality, in Computer Graphics Forum, vol. 38 (Wiley Online Library, 2019), pp. 493–519. Google Glass. https://www.google.com/glass/start/ Lenovo Mirage AR headset. https://www.lenovo.com/us/en/jedichallenges/ Q. Wang, D. Cheng, Y. Wang, H. Hua, G. Jin, Design, tolerance, and fabrication of an optical see-through head-mounted display with free-form surface elements. Appl. Opt. 52, C88–C99 (2013) LENOVO DaystAR. https://research.lenovo.com/webapp/view_English/ResearchNews.html? id=236/ B.C. Kress, W.J. Cummings, Towards the ultimate mixed reality experience: hololens display architecture choices. SID Int. Symp. Digest Tech. Papers 48(1), 127–131 (2017) Vuzix Blade. https://www.vuzix.com/products/blade-smart-glasses/ Y. Amitai, P-21: Extremely compact high-performance HMDs based on substrate-guided optical element, in SID Symposium Digest of Technical Papers, vol. 35 (Wiley Online Library, 2004), pp. 310–313. Microsoft. https://docs.microsoft.com/en-us/hololens/hololens1-hardware Karl Guttag on Technology. https://www.kguttag.com/2018/10/22/magic-leap-hololens-andlumus-resolution-shootout-ml1-review-part-3/ M.A. Klug, S.C. Cahall, H. Chung, Separated pupil optical systems for virtual and augmented reality and methods for displaying images using same. US Patent App. 15/146,296 (2016) Lumus. https://lumusvision.com/products/oe-vision/ Microsoft. https://docs.microsoft.com/en-us/windows/mixed-reality/news Magic Leap One. https://www.magicleap.com/en-us/magic-leap-1/ SONY. https://developer.sony.com/develop/lmx-001-holographic-waveguide-display/ A. Maimone, D. Lanman, K. Rathinavel, K. Keller, D. Luebke, H. Fuchs, Pinlight displays: wide field of view augmented reality eyeglasses using defocused point light sources, in ACM SIGGRAPH 2014 Emerging Technologies, vol. 20 (2014) B. Lee, S.-G. Park, K. Hong, J. Hong, Design and Implementation of Autostereoscopic Displays (SPIE Press, Bellingham, Washington, USA, 2016).

328

B. Lee and Y. Jo

27. S. Lee, C. Jang, S. Moon, J. Cho, B. Lee, Additive light field displays: realization of augmented reality with holographic optical elements. ACM Trans. Graph. (TOG) 35, 60 (2016) 28. S. Lee, Y. Jo, D. Yoo, J. Cho, D. Lee, B. Lee, Tomographic near-eye displays. Nat. Comm. 10, 2497 (2019) 29. P. Riendeau, Next generation sensors and a potential new ecosystem for marketing and advertising in augmented and virtual reality. PRESSBOOKS, https://augmentedrealitymarketing.pre ssbooks.com/chapter/sensors-for-arvr/ (2017) 30. C. Jang, K. Bang, S. Moon, J. Kim, S. Lee, B. Lee, Retinal 3d: augmented reality near-eye display via pupil-tracked light field projection on retina. ACM Trans. Graph. (TOG) 36, 190 (2017) 31. Microsoft Research Blog. https://www.microsoft.com/en-us/research/blog/microsoft-hol olens-facilitates-computer-vision-research-by-providing-access-to-raw-image-sensor-str eams-with-research-mode/ 32. M. Hansard, S. Lee, O. Choi, R.P. Horaud, Time-of-flight Cameras: Principles, Methods and Applications (Springer Science & Business Media, 2012). 33. Teledyne E2V, 3D imaging technology-time of flight. AZO MATERIALS, https://www.azom. com/article.aspx?ArticleID=16003 (2018) 34. H.J. Jang, J.Y. Lee, J. Kwak, D. Lee, J.-H. Park, B. Lee, Y.Y. Noh, Progress of display performances: AR, VR, QLED, OLED, and TFT. J. Inf. Disp. 20, 1–8 (2019) 35. W.Y. Park, A. Phadke, N. Shah, Efficiency improvement opportunities for personal computer monitors: implications for market transformation programs. Energy Effi. 6, 545–569 (2013)