Advances in display technology: augmented reality, virtual reality, quantum dot-based light-emitting diodes, and organic light-emitting diodes

Abstract

Virtual reality, augmented reality, quantum dot light-emitting diodes, and organic light-emitting diodes have progressed over the last two years. Key achievements in these displays are discussed in terms of device performance.

KEYWORDS:

• Augmented reality
• virtual reality
• quantum dot light-emitting diode
• organic light-emitting diode

1. Introduction

Augmented reality (AR), virtual reality (VR), quantum dot light-emitting diode (QLED), and the organic light-emitting diode (OLED) are next-generation display technologies with remarkable progressive device performances during the last two years. OLEDs are the current leader in the display market, and the device performances of the high-efficiency blue OLEDs were dramatically improved by developing stable blue emitters and device architecture harnessing the excitons with little loss. QLEDs are still behind OLEDs in terms of efficiency and device lifetime, but the rapid advance of QLEDs shines a light on their commercialization in real products through the ink-jet printing process. AR and VR have emerged as the practical display although the market size is still quite small. Apple launched Vision Pro® using VR technology and triggered the commercialization of VR and AR technologies. It may take time to expand the AR and VR market, but there is no doubt that they will be dominant displays in the future.

In this progress report, recent developments and progress of the AR, VR, QLED, and OLED displays are summarized and the prospect of them is proposed based on the technology and market trend.

2. Recent advances in augmented reality and virtual reality

AR and VR are emerging fields that provide a new way for users to interact with other people and digital information. From a display technology perspective, the near-eye display (NED), also called a head-mounted display (HMD), is the main device that enables many AR and VR applications. For an ideal VR experience, NEDs need to provide high-resolution immersive virtual images with a wide field of view (FoV) that covers the entire human visual field. For the AR experience, NEDs should present virtual images along with the real scene in front of the user. In AR and VR NEDs, a slim form factor and a wide eye box are also required for a comfortable viewing experience.

In the traditional display sectors, which include the mobile phone, monitor, and TV sectors, the digital image on the display panel is directly seen by the user; thus, the display system’s performance is solely determined by the display panel. In AR and VR NEDs, however, the digital image on the display panel is observed through an optics that forms the virtual image of the display panel far from the eye. Therefore, the performance of AR and VR NEDs depends on not only the display panel itself but also on the associated optics of the NEDs. Hence, various NED display panel techniques, including liquid crystal on silicon (LCoS), organic light-emitting diode on silicon (OLEDoS), micro-LED (also called light-emitting diode on silicon, LEDoS), and laser beam scanning display (LBS) should be mated with proper optics and co-optimized according to the specific AR and VR applications to realize a high-performance system [1–3].

Recently, we have observed progress in AR and VR NED technologies. In the subsections that follow, we briefly review the major advances made in 2022 and 2023, focusing on the display panel and optics together. Along with the status of the micro-display panel, novel optical techniques that reduce the VR NED form factor and enhance the AR NED waveguide are discussed. Advances in focus cue support and foveated display are also reviewed. Finally, the recently emergent digital pass-through AR NED is compared with optical see-through AR NED. Table 1 shows the features of recently reported or commercialized NEDs.

Table 1. Features of recently reported or commercialized AR and VR NEDs.

Name	FOV	Resolution (pixels)	Focus cue	Image combiner	Year	Ref.
Hu et al.	33°×25°	1024 × 768	varifocal	Freeform half-mirror	2014	[4]
Google Glass	13°×7.3°	640 × 360	X	Birdbath	2014	[5]
Yeom et al.			holography	HOE	2015	[6]
HoloLens	30°×17.5°	1366 × 768	X	DOE	2016	[7]
Maimone et al.	80° (diag.)	1920 × 1080	holography	HOE	2017	[8]
Jang et al.	55°×40°	960 × 540	light field	HOE	2017	[9]
Aksit et al.	55° – 63°	1280 × 720	varifocal	Diffuser-type HOE	2017	[10]
Magic Leap	40°×30°	1280 × 960	two layers	DOE	2018	[11]
Lee et al.	90° (diag.)		X	BS + Metasurface lens	2018	[12]
Nreal	52° (diag)	1080 p	X	Birdbath	2019	[13]
Pimax 8K	170°×130°	3840 × 2160	X	VR	2019	[14]
Lee et al.	30°		tomographic	VR	2019	[15]
Kim et al.	75°×78°	60 cpd in fovea	varifocal	HOE	2019	[16]
HoloLens 2	52° (diag)	2048 × 1080	X	DOE	2019	[17]
Oculus Quest 2	89°(H)	1832 × 1920	X	VR	2020	[18]
Lee et al.	80° (diag) for peripheral, 21° (diag) for foveal	30 cpd in foveal; 5 cpd in peripheral	holographic	Polarization-selective optics + HOE	2020	[19]
Tilt Five	110°	1280 × 720	X	Polarized projection + retro-reflector screen	2021	[20]
Bang et al.	102°×102°		X	VR (lenslet array + Fresnel lens + polarized folding optics)	2021	[21]
Cakmakci et al.	29°×12°	1920 × 1080	X	Lightguide + HOE + polarized folding optics	2021	[22]
Yoshikaie et al.	44°×16°	980 × 560	Maxwellian	Free-space HOE	2023	[23]
Qin et al.	25° (diag)	2560 × 2560	multifocal	BS	2023	[24]
Jang et al.	12° (diag)	1300 × 1300	holographic	Waveguide + DOE	2024	[25]

HOE: Holographic Optical Element, DOE: Diffractive Optical Element, cpd: cycles per degree.

2.1. Micro-display panel

A micro-display panel is placed inside the NED that presents virtual images to the user through magnifying optics. In VR, the micro-display panel must have a high resolution to support a wide FoV with a high angular resolution (∼60 pixels per degree, ppd). The luminance must also be high (>1 K nit) to compensate for the optical efficiency loss of polarization-based path folding VR optics, also called pancake lenses, which have recently become popular. In AR, the micro-display panel needs to support integration into a small projection module and have very high luminance (>1M nit) to ensure high image visibility during outdoor use, even with the low optical efficiency of the AR optical combiners. Currently, the available micro-display panels include LCoS, OLEDoS, micro-LED (or LED on Silicon), and LBS [26].

LCoS is a mature display technique. Although it requires an additional light source and illumination optics, the luminance can be readily increased with a high-power light source, addressing the high luminance requirement of AR. The polarized light output of the LCoS matches well with diffraction-based waveguide couplers, such as surface relief grating (SRG) and holographic optical element (HOE), which increase optical efficiency. The etendue of the illumination optics also enhances the optical efficiency by projecting all the LCoS light into the waveguide without loss [27]. Therefore, LCoS remains a viable option for waveguide-type AR NEDs. The OLEDoS display is an emerging display type. As it is an emissive display, the OLEDoS does not require an additional light source, which is an advantage in terms of form factor. Recent advances in high-resolution and high-ppi OLEDoS also make it attractive, especially in VR NEDs [28]. However, the maximum luminance of the currently available OLEDoS is approximately 10,000 nits [28], which is insufficient for diffraction-coupler waveguide-type AR NEDs. Therefore, OLEDoS is mainly considered for VR NEDs. For AR, OLEDoS is usually mated with optical combiners having high optical efficiency, such as a Birdbath or reflection-coupler waveguide [26]. Micro-LED is self-emissive and extremely bright (>1M nit), making it ideal for VR and AR. However, a full-color red, green, and blue (RGB) micro-LED panel is still being developed, and many problems with the monolithic full-color LED fabrication process remain [29]. The current option is to combine three mono-color R, G, and B panels using an x-cube, which increases the system volume and makes it difficult to match pixels between the individual panels [26]. Therefore, despite its potential advantages, the micro-LED panel has yet to become a mainstream technique for AR and VR. LBS has a small form factor and high luminance. The narrow linewidth of the laser light source is also beneficial for obtaining high optical efficiency in the diffraction-based couplers. However, scanning optics that uses 1D or 2D micro-electromechanical systems (MEMS) mirrors is complex, and the coherence noise coming from the laser is also problematic [1,26]. Currently, LBS is usually applied to AR with free-space optical combiners.

2.2. VR NED form factor

A slim and lightweight form factor is key to making VR NEDs comfortable to wear. It will enhance the popularity of VR NEDs and expand their application from gaming to the recently emerging spatial computing. The form factor for the VR NED is mainly determined by the gap between the micro-display panel and the collimation lens. To form virtual images at a distance from the eye, the micro-display panel is placed at the focal length of the collimation lens. As the focal length of the collimation lens cannot be reduced arbitrarily at a given aperture size, the current VR NEDs have at least a few tens of millimeters thickness, which is far from a glasses-like compact form factor.

Current efforts to reduce the thickness of VR NEDs include using micro-lens arrays (MLAs) [21], pinhole arrays [30], and pancake lenses [31–33]. In the MLA-based approach, the display panel is placed at the focal length of the MLA. The MLA forms the intermediate virtual image of the display panel, which is magnified and collimated by the collimation lens to form the final virtual image at a far distance. Because the MLA can have a much shorter focal length than a large-aperture collimation lens, the overall thickness of conventional VR NEDs can be reduced to the short focal length of the MLA, thereby achieving a thin form factor. In the pancake lens approach, polarization-based path folding is used. The polarized light from the display panel undergoes double reflection inside the pancake optics, effectively reducing the physical gap between the display panel and the optics. Although the concept of pancake optics is quite old [32], pancake optics has recently become a mainstream approach to achieving a thinner form factor for VR NEDs [31,33].

Although the MLA and pancake lens approaches effectively reduce the thickness of the system, they still pose problems. For instance, in the MLA approach, the loss of resolution is problematic. The display panel is first imaged into the intermediate virtual image by the MLA at magnification m_MLA= f_{collimation_lens}/f_MLA, and then it is imaged again into the final virtual image by the collimation lens at magnification m_{collimation_lens}= d_{final_image}/f_{collimation_lens}. Therefore, the overall system magnification is given by m = m_MLAm_{collimation_lens}= d_{final_image}/f_MLA, which is significantly larger than the magnification used in conventional VR, i.e. m = m_{collimation_lens}=d_{final_image}/f_{collimation_lens}. This increased magnification results in a coarser angular resolution of the final image, which requires the display panel to have higher pixel density and resolution for its compensation. When a pinhole array is used instead of the MLA, the displayed images do not have a uniform brightness, which is problematic [30]. In terms of the pancake lens approach, the light efficiency loss and ghost image are problematic [31,33]. For a polarized display panel, the pancake lens only outputs at most 25% of the incident light, wasting the rest 75%. For a non-polarized display panel, the optical efficiency drops even further to 12.5%. Ghost images that are due to nonideal polarization optics can also degrade the image quality [34]. Hence, recent research on pancake lenses has attempted to overcome these limitations by recycling polarized light [33] and enhancing broadband characteristics of the polarization optics on curved surfaces [34] (Table 2).

Table 2. Slim VR NED techniques reported in 2022, 2023, and 2024.

Type	Feature	Ref.
Pinhole array	Image non-uniformity compensation by multiplexed retinal projection	[30]
Pancake	Stray light analysis and suppression of pancake optics	[31]
Pancake	Optical efficiency enhancement by recycling polarized light	[33]
Pancake	Stray light suppression by optimizing the broadband quarter-wave plate	[34]
Metasurface	Bi-facial metasurface for implementing the pancake lens	[35]

2.3. AR NED optical combiner

In AR NEDs, the optical combiner that adds light from the display panel to the light of the see-through view of the real scene is the main optical component that determines the form factor and the performance of the AR NEDs. Among the many optical combiners, including the Birdbath, free-space combiner, and waveguide [1], the waveguide has become a mainstream configuration because of its glasses-like compact form factor and easy 2D exit pupil expansion, which give a wide eye box. For the waveguide optical combiner, couplers that in – and out-couple the light into and out of the waveguide are crucial to obtaining adequate FoV and optical efficiency. Waveguide couplers can be classified as diffraction-based, including the SRG and the HOE, and reflection-based couplers, which use partial mirror arrays and are also called light optical elements (LOEs).

The diffraction-based couplers have advantages in form factor, as they are micrometer-thick components attached to or embedded inside the waveguide and they support a sub-millimeter waveguide slab. However, the diffraction-based operation inevitably results in significant wavelength dependency. It reduces the optical efficiency, especially when the linewidth of the display panel light is broad and requires a stack of multiple waveguides of individual colors [36–39]. Reflection-based couplers achieve good image quality free from these limitations, but they usually require a thicker waveguide for embedding the partial mirrors inside the waveguide, and the expansion of the 2D exit pupil is not trivial.

Past approaches commonly used diffraction-based couplers. But, recently, the reflection-based couplers are gaining new attention due to their superior image quality [26]. The recent development of the reflection-based couplers reduced the overall thickness of the waveguide to less than 2 mm and led to the expanded 2D exit pupil [40,41] (Table 3).

Table 3. AR NED optical combiners reported in 2022 and 2023.

Combiner Type	Feature	Ref.
Waveguide + HOE	2D exit pupil expansion with enhanced uniformity	[37]
Free space + HOE	HOE recording with off-axis aberration compensation using a freeform mirror. Full color by a single HOE	[38]
Free space + HOE	Chromatic and off-axis aberration compensation using dual HOEs and projection optics with a freeform mirror. Full color by a stack of 3 HOEs	[39]
Waveguide + LOE	2D exit pupil expansion, < 2 mm thickness	[40,41]

2.4. Focus cue support

Focus cue support is critical to providing visual comfort to AR and VR NEDs users. Conventional stereoscopic NEDs offer depth perception only through binocular disparity, which introduces vergence-accommodation conflict (VAC) to the user. VAC is the main cause of visual fatigue in AR and VR experiences. VAC becomes especially severe when the 3D images are at arm’s distance from the user. Because this arm’s distance is important in AR and VR applications for the interaction between the user and the image, solving the VAC is vital.

Studies that address the VAC issue in AR and VR NEDs have included the Maxwellian view display, multifocal display, varifocal display, light field display, and holographic display. In recent years, all of those display techniques have been actively studied. The Maxwellian display alleviates VAC by presenting always-focused images regardless of the focal power of the eye lens, which is achieved by limiting the effective beam width that enters the eye pupil [42]. Although it is simple to configure, its reduced eye box is problematic. To solve this problem, various techniques, including multiple beam spot formation [42] and dynamic steering of beam spots with eye pupil tracking [23,43–45], have recently been recommended. The varifocal display dynamically controls the virtual image distance according to the eye’s gaze direction. A notable recent study in this field is the work of META, which achieved a varifocal display by dynamically shifting the display panel’s axial position using a mechanical system [46]. For multifocal displays, techniques using a stack of fast switchable diffusers [47] or an electrically controllable split-Lohmann lens array [24] have been reported, gaining wide attention. Light field displays that use a high refresh rate display panel and steering optics have also been studied [48,49] and have shown progress.

Among the technologies that mitigate VAC, holographic displays have made the most significant advances in recent years. Holographic displays are considered to be a promising solution to the VAC problem because they can provide natural 3D images with per-pixel continuous depth to users. However, they still have problems such as low image quality due to the coherent noise of the light source, small etendue that stems from the insufficient space – bandwidth product (SBP) of the SLM, and complicated system configuration required to remove the DC and conjugate images. Fast-developing deep learning techniques have been applied to the synthesis of hologram content and have enabled high-quality and high-speed optimization of SLM modulation patterns [25,50,51]. The active feedback from the displayed holographic images to the SLM pattern further enhances the 3D image quality [52], compensates for the nonideal operations of the optical components in the holographic displays, and suppresses the coherent noise from the light source. Optimization techniques that consider eye pupil size and position present a wide FoV holographic image in a large eye box and overcome the conventional SBP limitation of the SLM [53,54]. The ability of holographic displays to directly reconstruct wavefronts has even led to the highly compact waveguide-type holographic AR NEDs without lenses [25]. The advancements in 2022 and 2023 are expected to further accelerate the development of holographic display techniques in the coming years (Table 4).

Table 4. Focus cue supporting NEDs reported in 2022 and 2023.

Focus cue type	Feature	Ref.
Maxwellian	Pin-mirror HOE array as an out-coupler for waveguide AR NED	[42]
Maxwellian	Discrete pupil steering by polarization-dependent combiner (cholesteric liquid crystal off-axis lenses)	[43]
Maxwellian	Pupil replication and steering using a computer-generated hologram	[44]
Maxwellian	Pupil switching using multiple waveguides and HOE out-couplers	[45]
Maxwellian	Free-space HOE combiner. 2D pupil steering by a 1D mechanical shifting of HOE (H) and projector (V)	[23]
Vari-focal	Mechanical shifting of micro-display panel	[46]
Multifocal	Pixel-wise depth using split-Lohman lens + SLM as a dynamic beam deflector	[24]
Multifocal	High-speed electrically controllable diffuser stack + layer image projection	[47]
Light field (super multi-view)	High-speed display panel + pupil steering MEMS mirror, Free-space combiner	[48]
Light field (super multi-view)	DMD + LED array, waveguide + HOE combiner	[49]
Holography	Learning-based aberration compensation	[50]
Holography	Narrow depth of field by mimicking incoherent blur kernel	[51]
Holography	Speckle reduction by hologram optimization using active feedback	[52]
Holography	Multi-illumination expanding FoV and eye box	[53,54]
Holography	FoV expansion using tunable liquid crystal grating	[55]
Holography	Waveguide holographic 3D AR NED without lens. Waveguide aberration pre-compensated in the computer-generated hologram	[25]
Holography	Incoherent holographic camera + holographic 3D display for real-time streaming	[56]

2.5. Foveated display

To implement an NED having a wide FoV and a high angular resolution simultaneously, the display panel must have a high resolution. For immersive VR in particular, where the FoV should be large enough to cover the entire visual field of the human visual system, the display panel resolution requirement is quite high, much higher than the current level of availability. Foveated displays alleviate this problem by utilizing the non-uniform resolution of the human visual system over the visual field. The human visual system has high resolution only in the foveal area, whereas it has low resolution in the periphery [1]. The foveated displays have spatially varying resolutions and provide high-resolution images in the foveal area and low-resolution images in the peripheral area. Thus, foveated displays reduce the total number of pixels in the system without losing the perceived image resolution. Note that unlike foveated rendering, which renders images having spatially varying resolution and displays them in a uniform pixel grid of the display panel [57], the foveated display aims to dynamically control the physical pixel pitch to present images with spatially varying pixel density over the FoV.

Although the concept is attractive, the actual implementation of a foveated display is not trivial. Different spatial pixel densities in a display panel are possible [58]. However, the dynamic movement of the high-resolution area according to the user’s direction of gaze is highly challenging. Previous studies have usually integrated two display panels (i.e. one for wide-FoV and low-resolution peripheral vision and the other for small-FoV and high-resolution foveal vision) [16,19]. For dynamic control, a mechanical system is also commonly used [16]. Using the two display panels and the mechanical system made the overall system bulky, thus indicating room for enhancement. The progress made in 2022 and 2023 includes the development of an electrically controllable prism array to steer the high-resolution area over the FoV [59] and a single-panel foveated display with optics of spatially nonuniform magnification [60].

2.6. Optical see-through vs. digital pass-through

For AR, the light from the real scene in front of the user can be transmitted to the user’s eye via an optical-see-through (OST) configuration or captured and streamed to the display via a video-see-through (VST) configuration. In the past, most AR NEDs used OST configuration for the natural delivery of the real scene without latency or distortion. The recent advance in real-time video capture and streaming to display technology, however, makes the VST configuration another feasible option. The VST approach also solves the highly demanding luminance issue of AR micro-display panels [1], which makes them more attractive. In 2022 and 2023, the emergence of the VST approach was observed when several commercial products adopted this technique [61,62]. Recently, the traditional term (i.e. VST) has been frequently replaced with the new term digital pass-through (DPT) highlighting its birth.

In the VST and DPT approaches, one problem is the visibility of the user’s eyes for people outside. For a natural interaction between the user wearing the AR NED and other people, the eyes of the user should be visible to other people. However, the VST and DPT have an optically opaque structure, which hinders visibility. Therefore, providing a view of the user’s eye to people in the outside world has become an essential topic for further research. Preliminary research using an outside light field display panel that presents an image of each eye at its original axial position has been reported [63]. Additional research results are expected in the coming years.

3. Progress in quantum dot light-emitting diodes

Colloidal quantum dots (QDs) have garnered considerable attention as one of the most promising emitters owing to their excellent optoelectronic properties, such as wide absorption, narrow emission with a near-unity photoluminescent (PL) quantum yield (QY), size-dependent bandgap tunability, and high stability. On these merits, QDs have been applied in various fields, such as display devices, photovoltaics, biomedicine, and lasers [64–72]. Among these applications, research on QD-based light-emitting diodes (QLEDs) has been widely conducted because of the vivid and bright color emission of QLEDs. To achieve high QLED performance, several strategies, such as material synthesis, device structure, and light uncoupling efficiency, have been proposed. Due to multilateral efforts, the performance of electroluminescent (EL) devices has been significantly improved using Cd-containing and Cd-free QDs and has approached the theoretical limit value (∼20%) of external quantum efficiency (EQE) and showed an impressive half-lifetime (T₅₀) exceeding a million hours at an initial brightness of 100 cd/m². The progress of the Cd-based and Cd-free QLEDs in terms of EQE is plotted in Figure 1.

Figure 1. Peak EQEs of QLEDs using Cd-based and Cd-free QDs [64–91].

Despite their excellent device performance, however, state-of-the-art commercialized devices have adopted color-conversion-type QD displays consisting of red (R), and green (G) QDs integrated onto blue (B)-emitting organic light-emitting diode (OLED) sub-pixels. To make QDs adoptable in self-emissive-type full-color and high-resolution QLED displays, patterning technologies for RGB QDs at pixels and device performance should be developed. To date, various patterning technologies have been introduced, including inkjet printing (IJP) and direct patterning. The following section summarizes our efforts on QD patterning and the progress in QLEDs that use Cd-containing QDs and Cd-free QDs.

3.1. Cd-based QLEDs

Recently, the progress in Cd-containing QDs has been reported using various strategies, including optimal device architecture, engineered QD structure, and a modulated charge transport layer. Owing to their high PL QY and mature synthesis techniques, the device performance and operational stability of QLEDs have increased remarkably. In this regard, the highest EQEs for R, G, and B QLEDs reached 44.5%, 28.7%, and 21.9%, respectively [82,92,93]. These record-high values were realized using a top-emission structure with the IZO electrode, suppressing Auger recombination, and eliminating electron leakage at the organic/inorganic interface. From these methods, the maximum current efficiency (CE) of R – and G-QLEDs was approximately 100 cd/A [92,93], which is advantageous for implementing active-matrix QLED displays. The state-of-the-art performances achieved by the Cd-based QLEDs are specified in Table 5.

Table 5. Device performance of Cd-based QLEDs.

	QD structure	Peak wavelength (nm)	FWHM (nm)	Max. Luminance	Max. EQE	Max. CE	T50 @100 cd/m^2 (h)	Ref.
				(cd/m^2)	(%)	(cd/A)
R	CdSe/CdZnSe-ZnSe/ZnS	615	16	3,300,000	20.7	75.6	125,000,000	[84]
	CdZnSe/ZnS	623	22	67,190	32.5	30.3	–	[94]
	CdSe/ZnS	628	–	∼400,000	44.5	93.7	–	[92]
	CdSe/CdZnSe/ZnS	644	–	∼50,000	23.8	–	48,000 T95@1,000 cd/m^2	[95]
	CdSe/CdZnSe/ZnS	609	18	430,000	21.6	51.9	133,100,000	[96]
G	CdSe/ZnS	515	38	218,800	5.8	19.2	600 @500 cd/m^2	[64]
	CdZnSeS/ZnS	538	20	1,680,000	16.6	75.3		[74]
	CdSe/ZnS/ZnS	522	19	106,400	24.8	98.2	–	[68]
	CdSe/ZnS/ZnS	535	27	496,980	22.9	94.7	–	[82]
	CdZnSeS/ZnS	533	25	190,400	19.2	68.3	–	[94]
	CdSe/CdZnSe/ZnS	537	26	∼100,000	28.7	127.00	2,570,000	[93]
	CdZnSeS/ZnS	527	22	∼80,000	20.10	79.80	145 @3,400 cd/m^2	[97]
B	CdZnSe/ZnSe	465	21	2,322	10.1	–	–	[83]
	CdZnSeS/ZnS	475	22	16,370	10.3	8.1	–	[92]
	CdZnSe/ZnS	479	23	∼50,000	21.9	20.00	24,000	[82]
	ZnCdSe/ZnSeS/ ZnCdS/ZnS	471	20	∼10,000	20.80	–	80,377	[98]

^a Estimated from the graph.

3.2. Cd-free QLEDs

The development of QDs without heavy metal atoms has become prominent due to environmental restrictions on the mass production of consumer electronic devices. To replace Cd-containing QDs, nontoxic alternatives, such as InP and ZnSe (or ZnSeTe), have been extensively investigated. Although the device performance of InP-based QLEDs is continuously narrowing the performance gap with Cd-based devices, the surface traps surrounding the InP QDs and the lattice mismatch between core and shell layers should be further examined to improve performance [89,99]. Recently, the modified shell thickness of InP/ZnSe QDs was studied to reduce exciton quenching, resulting in the highest EQE of 22.5% and luminance of 136,090 cd/m² in red-emitting InP QLEDs [99]. As for green emissive QLEDs, the difficulty in synthesizing smaller QD cores has slowed any improvement in device performance from the EQE of 16.3% reported in 2021 [100]. For the blue-emitting Cd-free QLEDs, highly pure ZnSeTe/ZnSe/ZnS B-QLEDs were obtained by controlling the internal shell thickness, exhibiting an EQE of 18.2% with a spectral full-width-at-half-maximum (FWHM) of 22 nm [101]. The performances of the Cd-free QLEDs are listed in Table 6.

Table 6. Device performance of Cd-free QLEDs.

	QD structure	Peak wavelength (nm)	FWHM	Max. Luminance	Max. EQE	Max. CE	T50 @100 cd/m^2 (h)	Ref.
			(nm)	(cd/m^2)	(%)	(cd/A)
R	InP/ZnSeS	623	38	27,800	4.4	8.5		[87]
	InP/ZnSe/ZnS	618	42	10,000	12.2	14.7		[88]
	InP/ZnSe/ZnS	630	35	100,000	21.4	-	T75 of 4,300 @ 985 cd/m^2	[89]
	InP/ZnSe	621	38	107,160	22.5	26.7	112,765	[99]
Yellow	InP/ZnSeS/ZnS	545	56	10,490	1.5	4.44	-	[90]
	InP/ZnSe/ZnS	559	59	3,586	7.76	22.5	290	[102]
G	InP/ZnSeS	539	37	17,400	3.4	21.6	2.5 @ 300 cd/m^2	[87]
	InP/ZnSeS/ZnS	545	39	12,646	16.3	57.5	1033.4	[100]
	InP/ZnS	535	39	52,730	7.93	-	280.75	[103]
B	ZnSeTe	445	27	2,904	9.5	5.3	-	[104]
	ZnSeTe	457	35	88,900	20.2	-	15,850	[105]
	ZnSeTe	452	22	-	18.0	-	49.1	[101]

^a Estimated from the graph.

3.3. Inkjet-printed QLEDs

To achieve full-color QLEDs, a QD layer should be formed in each sub-pixel. In line with this, IJP has been widely adopted in academia and industry because of its low material consumption and high throughput for large-area QD patterning [106,107]. Despite these advantages, the non-uniform surface morphology of the inkjet-printed pixels is a drawback, and it is related to the ink’s viscosity, the surface energy of underlying layers, and drying conditions. To overcome these problems, several strategies have been proposed, such as using co-solvents to tune the boiling point and vapor pressure, mixing additives to control viscosity, and treating the substrate surface to enhance wettability [106,108]. As a result, the EQEs of RGB IJP QLEDs have been reported as 19.3%, 18.0%, and 6.82%, respectively [109,110], as enumerated in Table 7.

Table 7. Device performance of inkjet-printed QLEDs.

	Peak wavelength (nm)	FWHM	Max. Luminance	Turn-on Voltage	Max. EQE	Max. CE	Ref.
		(nm)	(cd/m^2)	(V)	(%)	(cd/A)
R	630	30	73,360	2.3	2.8	3.3	[106]
	624	27	10,000	2.2	18.3	26.6	[111]
	630	–	104,826	2.2	19.3	–	[109]
	625	42	17,559	–	8.1	11.1	[112]
	628	29	–	3.9	18.4	24.4	[113]
G	548	33	3,000	6	2.4	2.8	[107]
	525	–	13,445	4	1.29	4.21	[114]
	530	–	283,996	3	18	–	[109]
	542	28	–	3.3	13.3	60.6	[113]
B	465	–	2,367	3.5	4.4	–	[109]
	475	25	–	3.1	6.82	5.69	[110]
	475	26	–	4.5	5.6	4.9	[113]

^a Estimated from the graph.

3.4. Photolithography-based QD patterning

To utilize QD-based displays as microdisplays, higher-resolution QD patterning techniques are required. Compared with the IJP method, photolithography-based QD patterning allows precise patterning of QDs at the sub-micrometer scale using well-established process schemes and high fidelity. However, to prevent the likely damage to the QDs by exposure to several chemicals used in conventional photolithography processes, researchers have developed a spin-off technique, so-called direct lithography, which does not use a photoresist. Representative direct lithography techniques can be classified into two types: one with photosensitive additives, and the other using cross-linkable ligands. First, photosensitive additives that contain multiple photocrosslinkable units change the solubility of QDs when exposed to light [115]. In general, as-synthesized QDs can be used without ligand modification. Second, cross-linkable ligands functionalized on QDs form a rigid pattern by holding the QDs with crosslinked ligands when exposed to UV light [116]. The direct patterning of QDs reported so far is detailed in Table 8.

Table 8. QD patterning with direct photolithography.

Methods	QD	Color	Materials	Pattern size	Ref.
Photosensitive additive	CdSe	R, G, B	ethane-1,2-dial bis (4-azido-2,3,5,6-tetrafluorobenzoate)	4 µm × 16 µm	[117]
	CdSe	G	2-(4-methoxyethyl)-4,6-bis(trichloromethyl)−1,3,5-triazine 2-diazo-1-naphthol-4-sulfonic acid	1.5 µm (line)	[118]
	CdSe	R, G, B	ZnCl2	10 µm (line)	[119]
	InP	R, G, B	ethane-1,2-diyl bis (4-azido-2,3,5,6-tetrafluorobenzoate)	1 µm (line)	[115]
	InP	R, G, B	2-(4-methoxyethyl)−4,6-bis(trichloromethyl)−1,3,5-triazine	1 µm (line)	[120]
Crosslinkable ligand	CdSe	R	2-phenyl-2-(5-((tosyloxy)-imino) thiophen-2-ylidene)acetonitrile	100 nm (line)	[121]
	CdSe	R	poly(vinyltriphenylamine-random-azidostyrene)	5 µm (line)	[122]
	CdSe InP	R, G, B	Benzophenone	0.8 µm × 0.8 µm	[116]

4. Progress in the OLEDs

The performance of OLEDs has advanced through new material design and device engineering over the last few years. Various development strategies are suggested and applied to enhance device parameters, such as external quantum efficiency (EQE), device lifetime, and color purity, by enhancing internal quantum efficiency (IQE) and optical out-coupling efficiency. The latest research efforts are directed toward designing advanced triplet exciton harvesting systems, such as triplet–triplet fusion (TTF) OLEDs, thermally activated delayed fluorescence (TADF) OLEDs, hyperfluorescence (HF) OLEDs using sensitizers and organic emitters, and phosphorescent OLEDs (PHOLEDs). In particular, the EQE, device lifetime, and color purity of the HF OLEDs and PHOLEDs have been notably enhanced due to narrow emitting and stable emitters. The state-of-the-art EQE data for the OLEDs reported to date are summarized in Figure 2. Detailed information about the devices developed over the last two years is synthesized in Table 9.

Figure 2. EQEs of fluorescent, phosphorescent, and TADF OLEDs.

Table 9. Device performance of fluorescent, TADF, hyperfluorescence, and phosphorescent OLEDs reported in 2022 and 2023.

Type	Color	λEL^a	FWHM^b	EQE^c (%)		CE^d (cd/A)		Color coordinates		Lifetime	Ref.
		(nm)	(nm)	(1000 cd/m^2)	(Max)	(1000 cd/m^2)	(Max)	x	y	(h)
FL	Blue	453	19	8.7	9.1	–	–	0.142	0.066	230 (LT80, 2000 cd/m^2)	[123]
		471	22	10.9	11.4	10.5	11.4	0.116	0.133	208 (LT95, 1000 cd/m^2)	[124]
		457	26	–	10.4	–	–	0.13	0.07	79 (LT95, 1000 cd/m^2)	[125]
		459	18	–	9.1	–	–	0.135	0.079	346 (LT95, 1525 cd/m^2)	[126]
TADF	Red	610	39	31.1	35.7	–	59.4	0.64	0.34	–	[127]
		618	39	29.8	34.4	–	46.5	0.65	0.35	–	[127]
		625	40	32.1	36.1	–	43.4	0.66	0.34	–	[127]
	Green	520	34	32.3	35.8	111.8	112.5	0.16	0.67	–	[128]
		517	34	28.1	40.1	–	141.2	0.19	0.70	35 (LT50, 1000 cd/m^2)	[129]
		516	34	18.6	37.3	–	131.0	0.19	0.70	354 (LT50, 1000 cd/m^2)	[129]
		548	35	31.3	34.9	–	145.8	0.35	0.63	–	[130]
		524	33	–	37.3	–	129.8	0.23	0.71	–	[131]
		528	35	26.3	36.5	–	130.0	0.27	0.70	–	[131]
		520	29	20.4	35.2	127.0	132.9	0.19	0.74	–	[132]
	Blue	472	28	16.2	30.8	25.0	47.1	0.14	0.22	–	[133]
		471	18	20.4 (400 cd/m^2)	36.2	18.6 (400 cd/m^2)	32.1	0.12	0.12	–	[134]
		464	18	16.7 (400 cd/m^2)	35.8	14.7 (400 cd/m^2)	26.8	0.13	0.08	–	[134]
		461	18	17.9 (400 cd/m^2)	33.7	13.2 (400 cd/m^2)	24.9	0.13	0.06	–	[134]
		455	29	17.8	30.7	9.2	15.7	0.14	0.06	–	[135]
		460	29	20.5	32.5	12.9	19.4	0.14	0.07	–	[135]
		473	28	14.8 (100 cd/m^2)	33.8	–	43.4	0.111	0.188	–	[136]
		466	38	–	30.8	–	34.1	0.13	0.14	–	[137]
HF	Red	643	42	31.4	34.4	–	25.2	0.708	0.292	605 (LT95, 5000 cd/m^2)	[138]
	Green	523	23	17.2	30.5	–	–	0.22	0.74	82.3 (LT95, 5000 cd/m^2)	[139]
		545	32	34.4	37.1	–	153.4	0.35	0.63	–	[130]
	Blue	457	28	9.3	31.2	–	20.9	0.14	0.08	–	[140]
		467	23	15.5	33.2	–	29.7	0.13	0.11	–	[140]
		458	23	26.2	37.6	–	27.5	0.14	0.08	–	[140]
		474	33	7.7	30.5	–	42.6	0.116	0.213	–	[136]
		–	–	25.8	29.1^e	27.7	–	–	0.165	72.9 (LT95, 1000 cd/m^2)	[141]
PH	Blue	–	43	23.4	25.4	28.2	31.8	0.141	0.197	150/1113 (LT95/LT70, 1000 cd/m^2)	[142]
		473	52	22.4	26.5	41.7	50.7	0.144	0.294	370 (LT75, 100 cd/m^2)	[143]

^a Wavelength of electroluminescence emission spectrum.

^bFull-width-at-half-maximum of the electroluminescence emission spectrum.

^cExternal quantum efficiency.

^dCurrent efficiency.

^eEstimated value from reported efficiency roll-off.

4.1. Fluorescent OLEDs

Research on fluorescent OLEDs (FLOLEDs) has centered on realizing improved device characteristics using a triplet exciton harvesting system. Mostly, multi-resonance (MR)-TADF OLEDs and TADF – or phosphor-sensitized FLOLEDs employing MR-TADF final emitters have been extensively developed. In particular, the development of high EQE and long-lifetime phosphors accelerated the development of phosphor-sensitized FLOLEDs, along with narrow-emitting MR-TADF emitters with a high photoluminescence quantum yield (>90%), small FWHM (<30 nm), small Stokes shift, and high molar absorption coefficient. Until now, the combination of a phosphorescent sensitizer and an MR-TADF final emitter has been the best choice for high EQE and long device lifetime.

Researchers are currently interested in the molecular design method applying multiple-resonance (MR) characteristics in designing red, green, and blue emitters. MR characteristics can be constructed by regularly organizing the local electron density of the conjugated π-system within the molecule. The properties of the MR-type emitters can be freely tuned or constructed depending on the combination, number, and arrangement of hetero atoms (e.g. boron, nitrogen, and oxygen) introduced into the conjugated π-system. The versatility of the MR-based design strategy has allowed it to become mainstream in state-of-the-art emitter design methodologies.

FLOLEDs that use a TTF system are studied for blue-only OLED applications. In 2022, a high EQE and long lifetime were reported using a boron–nitrogen derivative as a fluorophore. The FLOLED realized a high EQE of 11.4%, an EL peak wavelength (λ_EL) of 471 nm, an FWHM of 22 nm, a CIE_y value of 0.133, and a lifetime up to 95% of the initial luminance (LT₉₅) of 208 h [124].

Dramatic progress in the device performance of the MR-TADF emitters has been made for RGB OLEDs over the last couple of years. Short-range charge transfer character and conjugation length of emitters were carefully manipulated to control the emission color of the MR-TADF emitters by employing multiple boron and nitrogen atoms in the molecular structure. In the case of red MR-TADF emitters, para-oriented double boron-based structures were featured. Red MR-TADF devices demonstrated a λ_EL of 625 nm, a CIE_x value of 0.66, an FWHM of 40 nm, and a high EQE of 36.1% [127]. For the green MR-TADF emitters, various attempts have been made to substitute the peripheral position on the boron–nitrogen-based sky blue emitters to stabilize the emission energy and take advantage of the effective radiative transition. Among them, selenoxanthone moiety, which strengthens spin–orbit coupling interaction through the heavy-atom effect, was effective. In the case of green MR-TADF OLEDs, a λ_EL of 517 nm, a CIE_y value of 0.70, a high EQE of 40.1%, and an LT₅₀ of 35 h were recorded [129]. Moreover, in the case of blue MR-TADF emitters, an extended π-system along with meta-oriented double boron-based structures enabled a high horizontal emitting dipole orientation ratio and short-range charge transfer. In the case of blue MR-TADF OLEDs, a dimerized boron–nitrogen compound achieved a high EQE of 36.2%, an FWHM of 18 nm, and a CIE_y value of 0.12 [134]. However, the blue MR-TADF OLEDs’ lifetime was quite short.

4.2. Phosphorescent OLEDs

Research on PHOLEDs focused on improving the performance of blue devices because red and green PHOLEDs have already been commercialized. Generally, Ir (III) complexes are popular as blue phosphors, but Pt (II) complexes having a 5/6/6 ring configuration-based tetradentate ligand have been popular for the last couple of years due to the merit of a narrow emission spectrum. The blue PHOLEDs with the Pt (II) complex recorded a high EQE of 25.4%, CIE_y value of 0.197, and a high current efficiency of 31.8 cd/A along with a remarkable LT₉₅ of 150 h and an LT₇₀ of 1,113 h [142]. This is the best device lifetime data ever reported for blue PHOLEDs. The Pt (II) complex was also effective as a sensitizer of the phosphor-sensitized FLOLEDs.

5. Outlook

Over the past two years, AR and VR display technologies have significantly improved their system form factor, resolution, focus cue support, and pass-through capabilities. DP advancement has enabled the presentation of the real scene in front of the user not only in the AR display but also in the VR displays. Thus, in the future, AR and VR display will be driven by two distinct categories of devices – smart glasses, where the compactness of the form factor is paramount, and spatial computing devices, where high image resolution and quality are crucial.

Developments in QDs and QLEDs have greatly improved the device performance of Cd and Cd-free QD-based QLEDs, providing competitive device performances. Moreover, the advanced photolithography-based QLEDs may make it possible to use them as high-resolution display for AR and VR applications if the device performance can reach the commercial specification level. In the case of OLEDs, the development of high-performance emitters and device architectures successfully constructed an advanced triplet exciton harvesting system. Through this, the commercialization of highly efficient, long lifetime and high color purity blue OLEDs would be possible in the near future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Jihoon Kang

Jihoon Kang received his B.S. degree in 2020 from the Department of Chemical Engineering at Sungkyunkwan University in South Korea. He is now a candidate for a Ph.D. degree at Sungkyunkwan University’s School of Chemical Engineering. His main research areas are the synthesis and evaluation of OLED materials designed for hosts and emitters.

Geun Woo Baek

Geun Woo Baek received his Ph.D. degree in Electrical and Computer Engineering from SNU in 2021. He is currently a postdoctoral researcher at SNU. His current research interests include quantum dot light-emitting diodes, 1D/2D thin-film transistors, and neuromorphic devices.

Jun Yeob Lee

Jun Yeob Lee received his Ph.D. from Seoul National University, Korea, in 1998. After postdoctoral studies at the Rensselaer Polytechnic Institute (1998–1999), he joined Samsung SDI and in six years developed the active matrix organic light-emitting diode. He worked thereafter as a professor at the Department of Polymer Science and Engineering of Dankook University, and in 2015 onward, he joined the faculty of Sungkyunkwan University’s School of Chemical Engineering as a professor. His main research areas are the synthesis of organic electronic materials and the development of novel device structures for organic electronic devices.

Jeonghun Kwak

Jeonghun Kwak received his B.S. (2005) and Ph.D. (2010) degrees in Electrical Engineering from SNU. After working as a postdoctoral researcher at SNU for a year, he worked as an assistant/associate professor at Dong-A University, Republic of Korea (2011–2015) and at the University of Seoul, Republic of Korea (2015–2019). Since March 2019, he has been an associate professor at the Department of Electrical and Computer Engineering, SNU. His current research interests focus on opto – and nano-electronic devices, such as QLEDs, organic thermoelectric devices, and neuromorphic devices based on organic molecules and low-dimensional materials.

Jae-Hyeung Park

Jae-Hyeung Park is an associate professor at Seoul National University, Korea. He earned his B.S., M.S., and Ph. D. degrees from the same institution in 2000, 2002, and 2005, respectively. His career includes being a senior engineer at Samsung Electronics from 2005 to 2007, followed by faculty positions at Chungbuk National University (2007–2013), and Inha University (2013–2024). Since March 2024, he has rejoined Seoul National University. His research interests include the acquisition, processing, and display of three-dimensional information using holography and light field techniques as well as AR and VR near-eye displays.