research paper about emerging technologies

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Review Article
• Open access
• Published: 25 October 2021

Augmented reality and virtual reality displays: emerging technologies and future perspectives

• Jianghao Xiong 1 ,
• En-Lin Hsiang 1 ,
• Ziqian He 1 ,
• Tao Zhan   ORCID: orcid.org/0000-0001-5511-6666 1 &
• Shin-Tson Wu   ORCID: orcid.org/0000-0002-0943-0440 1  

Light: Science & Applications volume  10 , Article number:  216 ( 2021 ) Cite this article

116k Accesses

455 Citations

36 Altmetric

Metrics details

• Liquid crystals

With rapid advances in high-speed communication and computation, augmented reality (AR) and virtual reality (VR) are emerging as next-generation display platforms for deeper human-digital interactions. Nonetheless, to simultaneously match the exceptional performance of human vision and keep the near-eye display module compact and lightweight imposes unprecedented challenges on optical engineering. Fortunately, recent progress in holographic optical elements (HOEs) and lithography-enabled devices provide innovative ways to tackle these obstacles in AR and VR that are otherwise difficult with traditional optics. In this review, we begin with introducing the basic structures of AR and VR headsets, and then describing the operation principles of various HOEs and lithography-enabled devices. Their properties are analyzed in detail, including strong selectivity on wavelength and incident angle, and multiplexing ability of volume HOEs, polarization dependency and active switching of liquid crystal HOEs, device fabrication, and properties of micro-LEDs (light-emitting diodes), and large design freedoms of metasurfaces. Afterwards, we discuss how these devices help enhance the AR and VR performance, with detailed description and analysis of some state-of-the-art architectures. Finally, we cast a perspective on potential developments and research directions of these photonic devices for future AR and VR displays.

Similar content being viewed by others

Advanced liquid crystal devices for augmented reality and virtual reality displays: principles and applications

Achromatic diffractive liquid-crystal optics for virtual reality displays

Metasurface wavefront control for high-performance user-natural augmented reality waveguide glasses

Introduction.

Recent advances in high-speed communication and miniature mobile computing platforms have escalated a strong demand for deeper human-digital interactions beyond traditional flat panel displays. Augmented reality (AR) and virtual reality (VR) headsets 1 , 2 are emerging as next-generation interactive displays with the ability to provide vivid three-dimensional (3D) visual experiences. Their useful applications include education, healthcare, engineering, and gaming, just to name a few 3 , 4 , 5 . VR embraces a total immersive experience, while AR promotes the interaction between user, digital contents, and real world, therefore displaying virtual images while remaining see-through capability. In terms of display performance, AR and VR face several common challenges to satisfy demanding human vision requirements, including field of view (FoV), eyebox, angular resolution, dynamic range, and correct depth cue, etc. Another pressing demand, although not directly related to optical performance, is ergonomics. To provide a user-friendly wearing experience, AR and VR should be lightweight and ideally have a compact, glasses-like form factor. The above-mentioned requirements, nonetheless, often entail several tradeoff relations with one another, which makes the design of high-performance AR/VR glasses/headsets particularly challenging.

In the 1990s, AR/VR experienced the first boom, which quickly subsided due to the lack of eligible hardware and digital content 6 . Over the past decade, the concept of immersive displays was revisited and received a new round of excitement. Emerging technologies like holography and lithography have greatly reshaped the AR/VR display systems. In this article, we firstly review the basic requirements of AR/VR displays and their associated challenges. Then, we briefly describe the properties of two emerging technologies: holographic optical elements (HOEs) and lithography-based devices (Fig. 1 ). Next, we separately introduce VR and AR systems because of their different device structures and requirements. For the immersive VR system, the major challenges and how these emerging technologies help mitigate the problems will be discussed. For the see-through AR system, we firstly review the present status of light engines and introduce some architectures for the optical combiners. Performance summaries on microdisplay light engines and optical combiners will be provided, that serve as a comprehensive overview of the current AR display systems.

The left side illustrates HOEs and lithography-based devices. The right side shows the challenges in VR and architectures in AR, and how the emerging technologies can be applied

Key parameters of AR and VR displays

AR and VR displays face several common challenges to satisfy the demanding human vision requirements, such as FoV, eyebox, angular resolution, dynamic range, and correct depth cue, etc. These requirements often exhibit tradeoffs with one another. Before diving into detailed relations, it is beneficial to review the basic definitions of the above-mentioned display parameters.

Definition of parameters

Taking a VR system (Fig. 2a ) as an example. The light emitting from the display module is projected to a FoV, which can be translated to the size of the image perceived by the viewer. For reference, human vision’s horizontal FoV can be as large as 160° for monocular vision and 120° for overlapped binocular vision 6 . The intersection area of ray bundles forms the exit pupil, which is usually correlated with another parameter called eyebox. The eyebox defines the region within which the whole image FoV can be viewed without vignetting. It therefore generally manifests a 3D geometry 7 , whose volume is strongly dependent on the exit pupil size. A larger eyebox offers more tolerance to accommodate the user’s diversified interpupillary distance (IPD) and wiggling of headset when in use. Angular resolution is defined by dividing the total resolution of the display panel by FoV, which measures the sharpness of a perceived image. For reference, a human visual acuity of 20/20 amounts to 1 arcmin angular resolution, or 60 pixels per degree (PPD), which is considered as a common goal for AR and VR displays. Another important feature of a 3D display is depth cue. Depth cue can be induced by displaying two separate images to the left eye and the right eye, which forms the vergence cue. But the fixed depth of the displayed image often mismatches with the actual depth of the intended 3D image, which leads to incorrect accommodation cues. This mismatch causes the so-called vergence-accommodation conflict (VAC), which will be discussed in detail later. One important observation is that the VAC issue may be more serious in AR than VR, because the image in an AR display is directly superimposed onto the real-world with correct depth cues. The image contrast is dependent on the display panel and stray light. To achieve a high dynamic range, the display panel should exhibit high brightness, low dark level, and more than 10-bits of gray levels. Nowadays, the display brightness of a typical VR headset is about 150–200 cd/m 2 (or nits).

a Schematic of a VR display defining FoV, exit pupil, eyebox, angular resolution, and accommodation cue mismatch. b Sketch of an AR display illustrating ACR

Figure 2b depicts a generic structure of an AR display. The definition of above parameters remains the same. One major difference is the influence of ambient light on the image contrast. For a see-through AR display, ambient contrast ratio (ACR) 8 is commonly used to quantify the image contrast:

where L on ( L off ) represents the on (off)-state luminance (unit: nit), L am is the ambient luminance, and T is the see-through transmittance. In general, ambient light is measured in illuminance (lux). For the convenience of comparison, we convert illuminance to luminance by dividing a factor of π, assuming the emission profile is Lambertian. In a normal living room, the illuminance is about 100 lux (i.e., L am  ≈ 30 nits), while in a typical office lighting condition, L am  ≈ 150 nits. For outdoors, on an overcast day, L am  ≈ 300 nits, and L am  ≈ 3000 nits on a sunny day. For AR displays, a minimum ACR should be 3:1 for recognizable images, 5:1 for adequate readability, and ≥10:1 for outstanding readability. To make a simple estimate without considering all the optical losses, to achieve ACR = 10:1 in a sunny day (~3000 nits), the display needs to deliver a brightness of at least 30,000 nits. This imposes big challenges in finding a high brightness microdisplay and designing a low loss optical combiner.

Tradeoffs and potential solutions

Next, let us briefly review the tradeoff relations mentioned earlier. To begin with, a larger FoV leads to a lower angular resolution for a given display resolution. In theory, to overcome this tradeoff only requires a high-resolution-display source, along with high-quality optics to support the corresponding modulation transfer function (MTF). To attain 60 PPD across 100° FoV requires a 6K resolution for each eye. This may be realizable in VR headsets because a large display panel, say 2–3 inches, can still accommodate a high resolution with acceptable manufacture cost. However, for a glasses-like wearable AR display, the conflict between small display size and the high solution becomes obvious as further shrinking the pixel size of a microdisplay is challenging.

To circumvent this issue, the concept of the foveated display is proposed 9 , 10 , 11 , 12 , 13 . The idea is based on that the human eye only has high visual acuity in the central fovea region, which accounts for about 10° FoV. If the high-resolution image is only projected to fovea while the peripheral image remains low resolution, then a microdisplay with 2K resolution can satisfy the need. Regarding the implementation method of foveated display, a straightforward way is to optically combine two display sources 9 , 10 , 11 : one for foveal and one for peripheral FoV. This approach can be regarded as spatial multiplexing of displays. Alternatively, time-multiplexing can also be adopted, by temporally changing the optical path to produce different magnification factors for the corresponding FoV 12 . Finally, another approach without multiplexing is to use a specially designed lens with intended distortion to achieve non-uniform resolution density 13 . Aside from the implementation of foveation, another great challenge is to dynamically steer the foveated region as the viewer’s eye moves. This task is strongly related to pupil steering, which will be discussed in detail later.

A larger eyebox or FoV usually decreases the image brightness, which often lowers the ACR. This is exactly the case for a waveguide AR system with exit pupil expansion (EPE) while operating under a strong ambient light. To improve ACR, one approach is to dynamically adjust the transmittance with a tunable dimmer 14 , 15 . Another solution is to directly boost the image brightness with a high luminance microdisplay and an efficient combiner optics. Details of this topic will be discussed in the light engine section.

Another tradeoff of FoV and eyebox in geometric optical systems results from the conservation of etendue (or optical invariant). To increase the system etendue requires a larger optics, which in turn compromises the form factor. Finally, to address the VAC issue, the display system needs to generate a proper accommodation cue, which often requires the modulation of image depth or wavefront, neither of which can be easily achieved in a traditional geometric optical system. While remarkable progresses have been made to adopt freeform surfaces 16 , 17 , 18 , to further advance AR and VR systems requires additional novel optics with a higher degree of freedom in structure design and light modulation. Moreover, the employed optics should be thin and lightweight. To mitigate the above-mentioned challenges, diffractive optics is a strong contender. Unlike geometric optics relying on curved surfaces to refract or reflect light, diffractive optics only requires a thin layer of several micrometers to establish efficient light diffractions. Two major types of diffractive optics are HOEs based on wavefront recording and manually written devices like surface relief gratings (SRGs) based on lithography. While SRGs have large design freedoms of local grating geometry, a recent publication 19 indicates the combination of HOE and freeform optics can also offer a great potential for arbitrary wavefront generation. Furthermore, the advances in lithography have also enabled optical metasurfaces beyond diffractive and refractive optics, and miniature display panels like micro-LED (light-emitting diode). These devices hold the potential to boost the performance of current AR/VR displays, while keeping a lightweight and compact form factor.

Formation and properties of HOEs

HOE generally refers to a recorded hologram that reproduces the original light wavefront. The concept of holography is proposed by Dennis Gabor 20 , which refers to the process of recording a wavefront in a medium (hologram) and later reconstructing it with a reference beam. Early holography uses intensity-sensitive recording materials like silver halide emulsion, dichromated gelatin, and photopolymer 21 . Among them, photopolymer stands out due to its easy fabrication and ability to capture high-fidelity patterns 22 , 23 . It has therefore found extensive applications like holographic data storage 23 and display 24 , 25 . Photopolymer HOEs (PPHOEs) have a relatively small refractive index modulation and therefore exhibits a strong selectivity on the wavelength and incident angle. Another feature of PPHOE is that several holograms can be recorded into a photopolymer film by consecutive exposures. Later, liquid-crystal holographic optical elements (LCHOEs) based on photoalignment polarization holography have also been developed 25 , 26 . Due to the inherent anisotropic property of liquid crystals, LCHOEs are extremely sensitive to the polarization state of the input light. This feature, combined with the polarization modulation ability of liquid crystal devices, offers a new possibility for dynamic wavefront modulation in display systems.

The formation of PPHOE is illustrated in Fig. 3a . When exposed to an interfering field with high-and-low intensity fringes, monomers tend to move toward bright fringes due to the higher local monomer-consumption rate. As a result, the density and refractive index is slightly larger in bright regions. Note the index modulation δ n here is defined as the difference between the maximum and minimum refractive indices, which may be twice the value in other definitions 27 . The index modulation δ n is typically in the range of 0–0.06. To understand the optical properties of PPHOE, we simulate a transmissive grating and a reflective grating using rigorous coupled-wave analysis (RCWA) 28 , 29 and plot the results in Fig. 3b . Details of grating configuration can be found in Table S1 . Here, the reason for only simulating gratings is that for a general HOE, the local region can be treated as a grating. The observation of gratings can therefore offer a general insight of HOEs. For a transmissive grating, its angular bandwidth (efficiency > 80%) is around 5° ( λ  = 550 nm), while the spectral band is relatively broad, with bandwidth around 175 nm (7° incidence). For a reflective grating, its spectral band is narrow, with bandwidth around 10 nm. The angular bandwidth varies with the wavelength, ranging from 2° to 20°. The strong selectivity of PPHOE on wavelength and incident angle is directly related to its small δ n , which can be adjusted by controlling the exposure dosage.

a Schematic of the formation of PPHOE. Simulated efficiency plots for b1 transmissive and b2 reflective PPHOEs. c Working principle of multiplexed PPHOE. d Formation and molecular configurations of LCHOEs. Simulated efficiency plots for e1 transmissive and e2 reflective LCHOEs. f Illustration of polarization dependency of LCHOEs

A distinctive feature of PPHOE is the ability to multiplex several holograms into one film sample. If the exposure dosage of a recording process is controlled so that the monomers are not completely depleted in the first exposure, the remaining monomers can continue to form another hologram in the following recording process. Because the total amount of monomer is fixed, there is usually an efficiency tradeoff between multiplexed holograms. The final film sample would exhibit the wavefront modulation functions of multiple holograms (Fig. 3c ).

Liquid crystals have also been used to form HOEs. LCHOEs can generally be categorized into volume-recording type and surface-alignment type. Volume-recording type LCHOEs are either based on early polarization holography recordings with azo-polymer 30 , 31 , or holographic polymer-dispersed liquid crystals (HPDLCs) 32 , 33 formed by liquid-crystal-doped photopolymer. Surface-alignment type LCHOEs are based on photoalignment polarization holography (PAPH) 34 . The first step is to record the desired polarization pattern in a thin photoalignment layer, and the second step is to use it to align the bulk liquid crystal 25 , 35 . Due to the simple fabrication process, high efficiency, and low scattering from liquid crystal’s self-assembly nature, surface-alignment type LCHOEs based on PAPH have recently attracted increasing interest in applications like near-eye displays. Here, we shall focus on this type of surface-alignment LCHOE and refer to it as LCHOE thereafter for simplicity.

The formation of LCHOEs is illustrated in Fig. 3d . The information of the wavefront and the local diffraction pattern is recorded in a thin photoalignment layer. The volume liquid crystal deposited on the photoalignment layer, depending on whether it is nematic liquid crystal or cholesteric liquid crystal (CLC), forms a transmissive or a reflective LCHOE. In a transmissive LCHOE, the bulk nematic liquid crystal molecules generally follow the pattern of the bottom alignment layer. The smallest allowable pattern period is governed by the liquid crystal distortion-free energy model, which predicts the pattern period should generally be larger than sample thickness 36 , 37 . This results in a maximum diffraction angle under 20°. On the other hand, in a reflective LCHOE 38 , 39 , the bulk CLC molecules form a stable helical structure, which is tilted to match the k -vector of the bottom pattern. The structure exhibits a very low distorted free energy 40 , 41 and can accommodate a pattern period that is small enough to diffract light into the total internal reflection (TIR) of a glass substrate.

The diffraction property of LCHOEs is shown in Fig. 3e . The maximum refractive index modulation of LCHOE is equal to the liquid crystal birefringence (Δ n ), which may vary from 0.04 to 0.5, depending on the molecular conjugation 42 , 43 . The birefringence used in our simulation is Δ n  = 0.15. Compared to PPHOEs, the angular and spectral bandwidths are significantly larger for both transmissive and reflective LCHOEs. For a transmissive LCHOE, its angular bandwidth is around 20° ( λ  = 550 nm), while the spectral bandwidth is around 300 nm (7° incidence). For a reflective LCHOE, its spectral bandwidth is around 80 nm and angular bandwidth could vary from 15° to 50°, depending on the wavelength.

The anisotropic nature of liquid crystal leads to LCHOE’s unique polarization-dependent response to an incident light. As depicted in Fig. 3f , for a transmissive LCHOE the accumulated phase is opposite for the conjugated left-handed circular polarization (LCP) and right-handed circular polarization (RCP) states, leading to reversed diffraction directions. For a reflective LCHOE, the polarization dependency is similar to that of a normal CLC. For the circular polarization with the same handedness as the helical structure of CLC, the diffraction is strong. For the opposite circular polarization, the diffraction is negligible.

Another distinctive property of liquid crystal is its dynamic response to an external voltage. The LC reorientation can be controlled with a relatively low voltage (<10 V rms ) and the response time is on the order of milliseconds, depending mainly on the LC viscosity and layer thickness. Methods to dynamically control LCHOEs can be categorized as active addressing and passive addressing, which can be achieved by either directly switching the LCHOE or modulating the polarization state with an active waveplate. Detailed addressing methods will be described in the VAC section.

Lithography-enabled devices

Lithography technologies are used to create arbitrary patterns on wafers, which lays the foundation of the modern integrated circuit industry 44 . Photolithography is suitable for mass production while electron/ion beam lithography is usually used to create photomask for photolithography or to write structures with nanometer-scale feature size. Recent advances in lithography have enabled engineered structures like optical metasurfaces 45 , SRGs 46 , as well as micro-LED displays 47 . Metasurfaces exhibit a remarkable design freedom by varying the shape of meta-atoms, which can be utilized to achieve novel functions like achromatic focus 48 and beam steering 49 . Similarly, SRGs also offer a large design freedom by manipulating the geometry of local grating regions to realize desired optical properties. On the other hand, micro-LED exhibits several unique features, such as ultrahigh peak brightness, small aperture ratio, excellent stability, and nanosecond response time, etc. As a result, micro-LED is a promising candidate for AR and VR systems for achieving high ACR and high frame rate for suppressing motion image blurs. In the following section, we will briefly review the fabrication and properties of micro-LEDs and optical modulators like metasurfaces and SRGs.

Fabrication and properties of micro-LEDs

LEDs with a chip size larger than 300 μm have been widely used in solid-state lighting and public information displays. Recently, micro-LEDs with chip sizes <5 μm have been demonstrated 50 . The first micro-LED disc with a diameter of about 12 µm was demonstrated in 2000 51 . After that, a single color (blue or green) LED microdisplay was demonstrated in 2012 52 . The high peak brightness, fast response time, true dark state, and long lifetime of micro-LEDs are attractive for display applications. Therefore, many companies have since released their micro-LED prototypes or products, ranging from large-size TVs to small-size microdisplays for AR/VR applications 53 , 54 . Here, we focus on micro-LEDs for near-eye display applications. Regarding the fabrication of micro-LEDs, through the metal-organic chemical vapor deposition (MOCVD) method, the AlGaInP epitaxial layer is grown on GaAs substrate for red LEDs, and GaN epitaxial layers on sapphire substrate for green and blue LEDs. Next, a photolithography process is applied to define the mesa and deposit electrodes. To drive the LED array, the fabricated micro-LEDs are transferred to a CMOS (complementary metal oxide semiconductor) driver board. For a small size (<2 inches) microdisplay used in AR or VR, the precision of the pick-and-place transfer process is hard to meet the high-resolution-density (>1000 pixel per inch) requirement. Thus, the main approach to assemble LED chips with driving circuits is flip-chip bonding 50 , 55 , 56 , 57 , as Fig. 4a depicts. In flip-chip bonding, the mesa and electrode pads should be defined and deposited before the transfer process, while metal bonding balls should be preprocessed on the CMOS substrate. After that, thermal-compression method is used to bond the two wafers together. However, due to the thermal mismatch of LED chip and driving board, as the pixel size decreases, the misalignment between the LED chip and the metal bonding ball on the CMOS substrate becomes serious. In addition, the common n-GaN layer may cause optical crosstalk between pixels, which degrades the image quality. To overcome these issues, the LED epitaxial layer can be firstly metal-bonded with the silicon driver board, followed by the photolithography process to define the LED mesas and electrodes. Without the need for an alignment process, the pixel size can be reduced to <5 µm 50 .

a Illustration of flip-chip bonding technology. b Simulated IQE-LED size relations for red and blue LEDs based on ABC model. c Comparison of EQE of different LED sizes with and without KOH and ALD side wall treatment. d Angular emission profiles of LEDs with different sizes. Metasurfaces based on e resonance-tuning, f non-resonance tuning and g combination of both. h Replication master and i replicated SRG based on nanoimprint lithography. Reproduced from a ref. 55 with permission from AIP Publishing, b ref. 61 with permission from PNAS, c ref. 66 with permission from IOP Publishing, d ref. 67 with permission from AIP Publishing, e ref. 69 with permission from OSA Publishing f ref. 48 with permission from AAAS g ref. 70 with permission from AAAS and h , i ref. 85 with permission from OSA Publishing

In addition to manufacturing process, the electrical and optical characteristics of LED also depend on the chip size. Generally, due to Shockley-Read-Hall (SRH) non-radiative recombination on the sidewall of active area, a smaller LED chip size results in a lower internal quantum efficiency (IQE), so that the peak IQE driving point will move toward a higher current density due to increased ratio of sidewall surface to active volume 58 , 59 , 60 . In addition, compared to the GaN-based green and blue LEDs, the AlGaInP-based red LEDs with a larger surface recombination and carrier diffusion length suffer a more severe efficiency drop 61 , 62 . Figure 4b shows the simulated result of IQE drop in relation with the LED chip size of blue and red LEDs based on ABC model 63 . To alleviate the efficiency drop caused by sidewall defects, depositing passivation materials by atomic layer deposition (ALD) or plasma enhanced chemical vapor deposition (PECVD) is proven to be helpful for both GaN and AlGaInP based LEDs 64 , 65 . In addition, applying KOH (Potassium hydroxide) treatment after ALD can further reduce the EQE drop of micro-LEDs 66 (Fig. 4c ). Small-size LEDs also exhibit some advantages, such as higher light extraction efficiency (LEE). Compared to an 100-µm LED, the LEE of a 2-µm LED increases from 12.2 to 25.1% 67 . Moreover, the radiation pattern of micro-LED is more directional than that of a large-size LED (Fig. 4d ). This helps to improve the lens collection efficiency in AR/VR display systems.

Metasurfaces and SGs

Thanks to the advances in lithography technology, low-loss dielectric metasurfaces working in the visible band have recently emerged as a platform for wavefront shaping 45 , 48 , 68 . They consist of an array of subwavelength-spaced structures with individually engineered wavelength-dependent polarization/phase/ amplitude response. In general, the light modulation mechanisms can be classified into resonant tuning 69 (Fig. 4e ), non-resonant tuning 48 (Fig. 4f ), and combination of both 70 (Fig. 4g ). In comparison with non-resonant tuning (based on geometric phase and/or dynamic propagation phase), the resonant tuning (such as Fabry–Pérot resonance, Mie resonance, etc.) is usually associated with a narrower operating bandwidth and a smaller out-of-plane aspect ratio (height/width) of nanostructures. As a result, they are easier to fabricate but more sensitive to fabrication tolerances. For both types, materials with a higher refractive index and lower absorption loss are beneficial to reduce the aspect ratio of nanostructure and improve the device efficiency. To this end, titanium dioxide (TiO 2 ) and gallium nitride (GaN) are the major choices for operating in the entire visible band 68 , 71 . While small-sized metasurfaces (diameter <1 mm) are usually fabricated via electron-beam lithography or focused ion beam milling in the labs, the ability of mass production is the key to their practical adoption. The deep ultraviolet (UV) photolithography has proven its feasibility for reproducing centimeter-size metalenses with decent imaging performance, while it requires multiple steps of etching 72 . Interestingly, the recently developed UV nanoimprint lithography based on a high-index nanocomposite only takes a single step and can obtain an aspect ratio larger than 10, which shows great promise for high-volume production 73 .

The arbitrary wavefront shaping capability and the thinness of the metasurfaces have aroused strong research interests in the development of novel AR/VR prototypes with improved performance. Lee et al. employed nanoimprint lithography to fabricate a centimeter-size, geometric-phase metalens eyepiece for full-color AR displays 74 . Through tailoring its polarization conversion efficiency and stacking with a circular polarizer, the virtual image can be superimposed with the surrounding scene. The large numerical aperture (NA~0.5) of the metalens eyepiece enables a wide FoV (>76°) that conventional optics are difficult to obtain. However, the geometric phase metalens is intrinsically a diffractive lens that also suffers from strong chromatic aberrations. To overcome this issue, an achromatic lens can be designed via simultaneously engineering the group delay and the group delay dispersion 75 , 76 , which will be described in detail later. Other novel and/or improved near-eye display architectures include metasurface-based contact lens-type AR 77 , achromatic metalens array enabled integral-imaging light field displays 78 , wide FoV lightguide AR with polarization-dependent metagratings 79 , and off-axis projection-type AR with an aberration-corrected metasurface combiner 80 , 81 , 82 . Nevertheless, from the existing AR/VR prototypes, metasurfaces still face a strong tradeoff between numerical aperture (for metalenses), chromatic aberration, monochromatic aberration, efficiency, aperture size, and fabrication complexity.

On the other hand, SRGs are diffractive gratings that have been researched for decades as input/output couplers of waveguides 83 , 84 . Their surface is composed of corrugated microstructures, and different shapes including binary, blazed, slanted, and even analogue can be designed. The parameters of the corrugated microstructures are determined by the target diffraction order, operation spectral bandwidth, and angular bandwidth. Compared to metasurfaces, SRGs have a much larger feature size and thus can be fabricated via UV photolithography and subsequent etching. They are usually replicated by nanoimprint lithography with appropriate heating and surface treatment. According to a report published a decade ago, SRGs with a height of 300 nm and a slant angle of up to 50° can be faithfully replicated with high yield and reproducibility 85 (Fig. 4g, h ).

Challenges and solutions of VR displays

The fully immersive nature of VR headset leads to a relatively fixed configuration where the display panel is placed in front of the viewer’s eye and an imaging optics is placed in-between. Regarding the system performance, although inadequate angular resolution still exists in some current VR headsets, the improvement of display panel resolution with advanced fabrication process is expected to solve this issue progressively. Therefore, in the following discussion, we will mainly focus on two major challenges: form factor and 3D cue generation.

Form factor

Compact and lightweight near-eye displays are essential for a comfortable user experience and therefore highly desirable in VR headsets. Current mainstream VR headsets usually have a considerably larger volume than eyeglasses, and most of the volume is just empty. This is because a certain distance is required between the display panel and the viewing optics, which is usually close to the focal length of the lens system as illustrated in Fig. 5a . Conventional VR headsets employ a transmissive lens with ~4 cm focal length to offer a large FoV and eyebox. Fresnel lenses are thinner than conventional ones, but the distance required between the lens and the panel does not change significantly. In addition, the diffraction artifacts and stray light caused by the Fresnel grooves can degrade the image quality, or MTF. Although the resolution density, quantified as pixel per inch (PPI), of current VR headsets is still limited, eventually Fresnel lens will not be an ideal solution when a high PPI display is available. The strong chromatic aberration of Fresnel singlet should also be compensated if a high-quality imaging system is preferred.

a Schematic of a basic VR optical configuration. b Achromatic metalens used as VR eyepiece. c VR based on curved display and lenslet array. d Basic working principle of a VR display based on pancake optics. e VR with pancake optics and Fresnel lens array. f VR with pancake optics based on purely HOEs. Reprinted from b ref. 87 under the Creative Commons Attribution 4.0 License. Adapted from c ref. 88 with permission from IEEE, e ref. 91 and f ref. 92 under the Creative Commons Attribution 4.0 License

It is tempting to replace the refractive elements with a single thin diffractive lens like a transmissive LCHOE. However, the diffractive nature of such a lens will result in serious color aberrations. Interestingly, metalenses can fulfil this objective without color issues. To understand how metalenses achieve achromatic focus, let us first take a glance at the general lens phase profile \(\Phi (\omega ,r)\) expanded as a Taylor series 75 :

where \(\varphi _0(\omega )\) is the phase at the lens center, \(F\left( \omega \right)\) is the focal length as a function of frequency ω , r is the radial coordinate, and \(\omega _0\) is the central operation frequency. To realize achromatic focus, \(\partial F{{{\mathrm{/}}}}\partial \omega\) should be zero. With a designed focal length, the group delay \(\partial \Phi (\omega ,r){{{\mathrm{/}}}}\partial \omega\) and the group delay dispersion \(\partial ^2\Phi (\omega ,r){{{\mathrm{/}}}}\partial \omega ^2\) can be determined, and \(\varphi _0(\omega )\) is an auxiliary degree of freedom of the phase profile design. In the design of an achromatic metalens, the group delay is a function of the radial coordinate and monotonically increases with the metalens radius. Many designs have proven that the group delay has a limited variation range 75 , 76 , 78 , 86 . According to Shrestha et al. 86 , there is an inevitable tradeoff between the maximum radius of the metalens, NA, and operation bandwidth. Thus, the reported achromatic metalenses at visible usually have limited lens aperture (e.g., diameter < 250 μm) and NA (e.g., <0.2). Such a tradeoff is undesirable in VR displays, as the eyepiece favors a large clear aperture (inch size) and a reasonably high NA (>0.3) to maintain a wide FoV and a reasonable eye relief 74 .

To overcome this limitation, Li et al. 87 proposed a novel zone lens method. Unlike the traditional phase Fresnel lens where the zones are determined by the phase reset, the new approach divides the zones by the group delay reset. In this way, the lens aperture and NA can be much enlarged, and the group delay limit is bypassed. A notable side effect of this design is the phase discontinuity at zone boundaries that will contribute to higher-order focusing. Therefore, significant efforts have been conducted to find the optimal zone transition locations and to minimize the phase discontinuities. Using this method, they have demonstrated an impressive 2-mm-diameter metalens with NA = 0.7 and nearly diffraction-limited focusing for the designed wavelengths (488, 532, 658 nm) (Fig. 5b ). Such a metalens consists of 681 zones and works for the visible band ranging from 470 to 670 nm, though the focusing efficiency is in the order of 10%. This is a great starting point for the achromatic metalens to be employed as a compact, chromatic-aberration-free eyepiece in near-eye displays. Future challenges are how to further increase the aperture size, correct the off-axis aberrations, and improve the optical efficiency.

Besides replacing the refractive lens with an achromatic metalens, another way to reduce system focal length without decreasing NA is to use a lenslet array 88 . As depicted in Fig. 5c , both the lenslet array and display panel adopt a curved structure. With the latest flexible OLED panel, the display can be easily curved in one dimension. The system exhibits a large diagonal FoV of 180° with an eyebox of 19 by 12 mm. The geometry of each lenslet is optimized separately to achieve an overall performance with high image quality and reduced distortions.

Aside from trying to shorten the system focal length, another way to reduce total track is to fold optical path. Recently, polarization-based folded lenses, also known as pancake optics, are under active development for VR applications 89 , 90 . Figure 5d depicts the structure of an exemplary singlet pancake VR lens system. The pancake lenses can offer better imaging performance with a compact form factor since there are more degrees of freedom in the design and the actual light path is folded thrice. By using a reflective surface with a positive power, the field curvature of positive refractive lenses can be compensated. Also, the reflective surface has no chromatic aberrations and it contributes considerable optical power to the system. Therefore, the optical power of refractive lenses can be smaller, resulting in an even weaker chromatic aberration. Compared to Fresnel lenses, the pancake lenses have smooth surfaces and much fewer diffraction artifacts and stray light. However, such a pancake lens design is not perfect either, whose major shortcoming is low light efficiency. With two incidences of light on the half mirror, the maximum system efficiency is limited to 25% for a polarized input and 12.5% for an unpolarized input light. Moreover, due to the existence of multiple surfaces in the system, stray light caused by surface reflections and polarization leakage may lead to apparent ghost images. As a result, the catadioptric pancake VR headset usually manifests a darker imagery and lower contrast than the corresponding dioptric VR.

Interestingly, the lenslet and pancake optics can be combined to further reduce the system form. Bang et al. 91 demonstrated a compact VR system with a pancake optics and a Fresnel lenslet array. The pancake optics serves to fold the optical path between the display panel and the lenslet array (Fig. 5e ). Another Fresnel lens is used to collect the light from the lenslet array. The system has a decent horizontal FoV of 102° and an eyebox of 8 mm. However, a certain degree of image discontinuity and crosstalk are still present, which can be improved with further optimizations on the Fresnel lens and the lenslet array.

One step further, replacing all conventional optics in catadioptric VR headset with holographic optics can make the whole system even thinner. Maimone and Wang demonstrated such a lightweight, high-resolution, and ultra-compact VR optical system using purely HOEs 92 . This holographic VR optics was made possible by combining several innovative optical components, including a reflective PPHOE, a reflective LCHOE, and a PPHOE-based directional backlight with laser illumination, as shown in Fig. 5f . Since all the optical power is provided by the HOEs with negligible weight and volume, the total physical thickness can be reduced to <10 mm. Also, unlike conventional bulk optics, the optical power of a HOE is independent of its thickness, only subject to the recording process. Another advantage of using holographic optical devices is that they can be engineered to offer distinct phase profiles for different wavelengths and angles of incidence, adding extra degrees of freedom in optical designs for better imaging performance. Although only a single-color backlight has been demonstrated, such a PPHOE has the potential to achieve full-color laser backlight with multiplexing ability. The PPHOE and LCHOE in the pancake optics can also be optimized at different wavelengths for achieving high-quality full-color images.

Vergence-accommodation conflict

Conventional VR displays suffer from VAC, which is a common issue for stereoscopic 3D displays 93 . In current VR display modules, the distance between the display panel and the viewing optics is fixed, which means the VR imagery is displayed at a single depth. However, the image contents are generated by parallax rendering in three dimensions, offering distinct images for two eyes. This approach offers a proper stimulus to vergence but completely ignores the accommodation cue, which leads to the well-known VAC that can cause an uncomfortable user experience. Since the beginning of this century, numerous methods have been proposed to solve this critical issue. Methods to produce accommodation cue include multifocal/varifocal display 94 , holographic display 95 , and integral imaging display 96 . Alternatively, elimination of accommodation cue using a Maxwellian-view display 93 also helps to mitigate the VAC. However, holographic displays and Maxwellian-view displays generally require a totally different optical architecture than current VR systems. They are therefore more suitable for AR displays, which will be discussed later. Integral imaging, on the other hand, has an inherent tradeoff between view number and resolution. For current VR headsets pursuing high resolution to match human visual acuity, it may not be an appealing solution. Therefore, multifocal/varifocal displays that rely on depth modulation is a relatively practical and effective solution for VR headsets. Regarding the working mechanism, multifocal displays present multiple images with different depths to imitate the original 3D scene. Varifocal displays, in contrast, only show one image at each time frame. The image depth matches the viewer’s vergence depth. Nonetheless, the pre-knowledge of the viewer’s vergence depth requires an additional eye-tracking module. Despite different operation principles, a varifocal display can often be converted to a multifocal display as long as the varifocal module has enough modulation bandwidth to support multiple depths in a time frame.

To achieve depth modulation in a VR system, traditional liquid lens 97 , 98 with tunable focus suffers from the small aperture and large aberrations. Alvarez lens 99 is another tunable-focus solution but it requires mechanical adjustment, which adds to system volume and complexity. In comparison, transmissive LCHOEs with polarization dependency can achieve focus adjustment with electronic driving. Its ultra-thinness also satisfies the requirement of small form factors in VR headsets. The diffractive behavior of transmissive LCHOEs is often interpreted by the mechanism of Pancharatnam-Berry phase (also known as geometric phase) 100 . They are therefore often called Pancharatnam-Berry optical elements (PBOEs). The corresponding lens component is referred as Pancharatnam-Berry lens (PBL).

Two main approaches are used to switch the focus of a PBL, active addressing and passive addressing. In active addressing, the PBL itself (made of LC) can be switched by an applied voltage (Fig. 6a ). The optical power of the liquid crystal PBLs can be turned-on and -off by controlling the voltage. Stacking multiple active PBLs can produce 2 N depths, where N is the number of PBLs. The drawback of using active PBLs, however, is the limited spectral bandwidth since their diffraction efficiency is usually optimized at a single wavelength. In passive addressing, the depth modulation is achieved through changing the polarization state of input light by a switchable half-wave plate (HWP) (Fig. 6b ). The focal length can therefore be switched thanks to the polarization sensitivity of PBLs. Although this approach has a slightly more complicated structure, the overall performance can be better than the active one, because the PBLs made of liquid crystal polymer can be designed to manifest high efficiency within the entire visible spectrum 101 , 102 .

Working principles of a depth switching PBL module based on a active addressing and b passive addressing. c A four-depth multifocal display based on time multiplexing. d A two-depth multifocal display based on polarization multiplexing. Reproduced from c ref. 103 with permission from OSA Publishing and d ref. 104 with permission from OSA Publishing

With the PBL module, multifocal displays can be built using time-multiplexing technique. Zhan et al. 103 demonstrated a four-depth multifocal display using two actively switchable liquid crystal PBLs (Fig. 6c ). The display is synchronized with the PBL module, which lowers the frame rate by the number of depths. Alternatively, multifocal displays can also be achieved by polarization-multiplexing, as demonstrated by Tan et al. 104 . The basic principle is to adjust the polarization state of local pixels so the image content on two focal planes of a PBL can be arbitrarily controlled (Fig. 6d ). The advantage of polarization multiplexing is that it does not sacrifice the frame rate, but it can only support two planes because only two orthogonal polarization states are available. Still, it can be combined with time-multiplexing to reduce the frame rate sacrifice by half. Naturally, varifocal displays can also be built with a PBL module. A fast-response 64-depth varifocal module with six PBLs has been demonstrated 105 .

The compact structure of PBL module leads to a natural solution of integrating it with above-mentioned pancake optics. A compact VR headset with dynamic depth modulation to solve VAC is therefore possible in practice. Still, due to the inherent diffractive nature of PBL, the PBL module face the issue of chromatic dispersion of focal length. To compensate for different focal depths for RGB colors may require additional digital corrections in image-rendering.

Architectures of AR displays

Unlike VR displays with a relatively fixed optical configuration, there exist a vast number of architectures in AR displays. Therefore, instead of following the narrative of tackling different challenges, a more appropriate way to review AR displays is to separately introduce each architecture and discuss its associated engineering challenges. An AR display usually consists of a light engine and an optical combiner. The light engine serves as display image source, while the combiner delivers the displayed images to viewer’s eye and in the meantime transmits the environment light. Some performance parameters like frame rate and power consumption are mainly determined by the light engine. Parameters like FoV, eyebox and MTF are primarily dependent on the combiner optics. Moreover, attributes like image brightness, overall efficiency, and form factor are influenced by both light engine and combiner. In this section, we will firstly discuss the light engine, where the latest advances in micro-LED on chip are reviewed and compared with existing microdisplay systems. Then, we will introduce two main types of combiners: free-space combiner and waveguide combiner.

Light engine

The light engine determines several essential properties of the AR system like image brightness, power consumption, frame rate, and basic etendue. Several types of microdisplays have been used in AR, including micro-LED, micro-organic-light-emitting-diodes (micro-OLED), liquid-crystal-on-silicon (LCoS), digital micromirror device (DMD), and laser beam scanning (LBS) based on micro-electromechanical system (MEMS). We will firstly describe the working principles of these devices and then analyze their performance. For those who are more interested in final performance parameters than details, Table 1 provides a comprehensive summary.

Working principles

Micro-LED and micro-OLED are self-emissive display devices. They are usually more compact than LCoS and DMD because no illumination optics is required. The fundamentally different material systems of LED and OLED lead to different approaches to achieve full-color displays. Due to the “green gap” in LEDs, red LEDs are manufactured on a different semiconductor material from green and blue LEDs. Therefore, how to achieve full-color display in high-resolution density microdisplays is quite a challenge for micro-LEDs. Among several solutions under research are two main approaches. The first is to combine three separate red, green and blue (RGB) micro-LED microdisplay panels 106 . Three single-color micro-LED microdisplays are manufactured separately through flip-chip transfer technology. Then, the projected images from three microdisplay panels are integrated by a trichroic prism (Fig. 7a ).

a RGB micro-LED microdisplays combined by a trichroic prism. b QD-based micro-LED microdisplay. c Micro-OLED display with 4032 PPI. Working principles of d LCoS, e DMD, and f MEMS-LBS display modules. Reprinted from a ref. 106 with permission from IEEE, b ref. 108 with permission from Chinese Laser Press, c ref. 121 with permission from Jon Wiley and Sons, d ref. 124 with permission from Spring Nature, e ref. 126 with permission from Springer and f ref. 128 under the Creative Commons Attribution 4.0 License

Another solution is to assemble color-conversion materials like quantum dot (QD) on top of blue or ultraviolet (UV) micro-LEDs 107 , 108 , 109 (Fig. 7b ). The quantum dot color filter (QDCF) on top of the micro-LED array is mainly fabricated by inkjet printing or photolithography 110 , 111 . However, the display performance of color-conversion micro-LED displays is restricted by the low color-conversion efficiency, blue light leakage, and color crosstalk. Extensive efforts have been conducted to improve the QD-micro-LED performance. To boost QD conversion efficiency, structure designs like nanoring 112 and nanohole 113 , 114 have been proposed, which utilize the Förster resonance energy transfer mechanism to transfer excessive excitons in the LED active region to QD. To prevent blue light leakage, methods using color filters or reflectors like distributed Bragg reflector (DBR) 115 and CLC film 116 on top of QDCF are proposed. Compared to color filters that absorb blue light, DBR and CLC film help recycle the leaked blue light to further excite QDs. Other methods to achieve full-color micro-LED display like vertically stacked RGB micro-LED array 61 , 117 , 118 and monolithic wavelength tunable nanowire LED 119 are also under investigation.

Micro-OLED displays can be generally categorized into RGB OLED and white OLED (WOLED). RGB OLED displays have separate sub-pixel structures and optical cavities, which resonate at the desirable wavelength in RGB channels, respectively. To deposit organic materials onto the separated RGB sub-pixels, a fine metal mask (FMM) that defines the deposition area is required. However, high-resolution RGB OLED microdisplays still face challenges due to the shadow effect during the deposition process through FMM. In order to break the limitation, a silicon nitride film with small shadow has been proposed as a mask for high-resolution deposition above 2000 PPI (9.3 µm) 120 .

WOLED displays use color filters to generate color images. Without the process of depositing patterned organic materials, a high-resolution density up to 4000 PPI has been achieved 121 (Fig. 7c ). However, compared to RGB OLED, the color filters in WOLED absorb about 70% of the emitted light, which limits the maximum brightness of the microdisplay. To improve the efficiency and peak brightness of WOLED microdisplays, in 2019 Sony proposed to apply newly designed cathodes (InZnO) and microlens arrays on OLED microdisplays, which increased the peak brightness from 1600 nits to 5000 nits 120 . In addition, OLEDWORKs has proposed a multi-stacked OLED 122 with optimized microcavities whose emission spectra match the transmission bands of the color filters. The multi-stacked OLED shows a higher luminous efficiency (cd/A), but also requires a higher driving voltage. Recently, by using meta-mirrors as bottom reflective anodes, patterned microcavities with more than 10,000 PPI have been obtained 123 . The high-resolution meta-mirrors generate different reflection phases in the RGB sub-pixels to achieve desirable resonant wavelengths. The narrow emission spectra from the microcavity help to reduce the loss from color filters or even eliminate the need of color filters.

LCoS and DMD are light-modulating displays that generate images by controlling the reflection of each pixel. For LCoS, the light modulation is achieved by manipulating the polarization state of output light through independently controlling the liquid crystal reorientation in each pixel 124 , 125 (Fig. 7d ). Both phase-only and amplitude modulators have been employed. DMD is an amplitude modulation device. The modulation is achieved through controlling the tilt angle of bi-stable micromirrors 126 (Fig. 7e ). To generate an image, both LCoS and DMD rely on the light illumination systems, with LED or laser as light source. For LCoS, the generation of color image can be realized either by RGB color filters on LCoS (with white LEDs) or color-sequential addressing (with RGB LEDs or lasers). However, LCoS requires a linearly polarized light source. For an unpolarized LED light source, usually, a polarization recycling system 127 is implemented to improve the optical efficiency. For a single-panel DMD, the color image is mainly obtained through color-sequential addressing. In addition, DMD does not require a polarized light so that it generally exhibits a higher efficiency than LCoS if an unpolarized light source is employed.

MEMS-based LBS 128 , 129 utilizes micromirrors to directly scan RGB laser beams to form two-dimensional (2D) images (Fig. 7f ). Different gray levels are achieved by pulse width modulation (PWM) of the employed laser diodes. In practice, 2D scanning can be achieved either through a 2D scanning mirror or two 1D scanning mirrors with an additional focusing lens after the first mirror. The small size of MEMS mirror offers a very attractive form factor. At the same time, the output image has a large depth-of-focus (DoF), which is ideal for projection displays. One shortcoming, though, is that the small system etendue often hinders its applications in some traditional display systems.

Comparison of light engine performance

There are several important parameters for a light engine, including image resolution, brightness, frame rate, contrast ratio, and form factor. The resolution requirement (>2K) is similar for all types of light engines. The improvement of resolution is usually accomplished through the manufacturing process. Thus, here we shall focus on other three parameters.

Image brightness usually refers to the measured luminance of a light-emitting object. This measurement, however, may not be accurate for a light engine as the light from engine only forms an intermediate image, which is not directly viewed by the user. On the other hand, to solely focus on the brightness of a light engine could be misleading for a wearable display system like AR. Nowadays, data projectors with thousands of lumens are available. But the power consumption is too high for a battery-powered wearable AR display. Therefore, a more appropriate way to evaluate a light engine’s brightness is to use luminous efficacy (lm/W) measured by dividing the final output luminous flux (lm) by the input electric power (W). For a self-emissive device like micro-LED or micro-OLED, the luminous efficacy is directly determined by the device itself. However, for LCoS and DMD, the overall luminous efficacy should take into consideration the light source luminous efficacy, the efficiency of illumination optics, and the efficiency of the employed spatial light modulator (SLM). For a MEMS LBS engine, the efficiency of MEMS mirror can be considered as unity so that the luminous efficacy basically equals to that of the employed laser sources.

As mentioned earlier, each light engine has a different scheme for generating color images. Therefore, we separately list luminous efficacy of each scheme for a more inclusive comparison. For micro-LEDs, the situation is more complicated because the EQE depends on the chip size. Based on previous studies 130 , 131 , 132 , 133 , we separately calculate the luminous efficacy for RGB micro-LEDs with chip size ≈ 20 µm. For the scheme of direct combination of RGB micro-LEDs, the luminous efficacy is around 5 lm/W. For QD-conversion with blue micro-LEDs, the luminous efficacy is around 10 lm/W with the assumption of 100% color conversion efficiency, which has been demonstrated using structure engineering 114 . For micro-OLEDs, the calculated luminous efficacy is about 4–8 lm/W 120 , 122 . However, the lifetime and EQE of blue OLED materials depend on the driving current. To continuously display an image with brightness higher than 10,000 nits may dramatically shorten the device lifetime. The reason we compare the light engine at 10,000 nits is that it is highly desirable to obtain 1000 nits for the displayed image in order to keep ACR>3:1 with a typical AR combiner whose optical efficiency is lower than 10%.

For an LCoS engine using a white LED as light source, the typical optical efficiency of the whole engine is around 10% 127 , 134 . Then the engine luminous efficacy is estimated to be 12 lm/W with a 120 lm/W white LED source. For a color sequential LCoS using RGB LEDs, the absorption loss from color filters is eliminated, but the luminous efficacy of RGB LED source is also decreased to about 30 lm/W due to lower efficiency of red and green LEDs and higher driving current 135 . Therefore, the final luminous efficacy of the color sequential LCoS engine is also around 10 lm/W. If RGB linearly polarized lasers are employed instead of LEDs, then the LCoS engine efficiency can be quite high due to the high degree of collimation. The luminous efficacy of RGB laser source is around 40 lm/W 136 . Therefore, the laser-based LCoS engine is estimated to have a luminous efficacy of 32 lm/W, assuming the engine optical efficiency is 80%. For a DMD engine with RGB LEDs as light source, the optical efficiency is around 50% 137 , 138 , which leads to a luminous efficacy of 15 lm/W. By switching to laser light sources, the situation is similar to LCoS, with the luminous efficacy of about 32 lm/W. Finally, for MEMS-based LBS engine, there is basically no loss from the optics so that the final luminous efficacy is 40 lm/W. Detailed calculations of luminous efficacy can be found in Supplementary Information .

Another aspect of a light engine is the frame rate, which determines the volume of information it can deliver in a unit time. A high volume of information is vital for the construction of a 3D light field to solve the VAC issue. For micro-LEDs, the device response time is around several nanoseconds, which allows for visible light communication with bandwidth up to 1.5 Gbit/s 139 . For an OLED microdisplay, a fast OLED with ~200 MHz bandwidth has been demonstrated 140 . Therefore, the limitation of frame rate is on the driving circuits for both micro-LED and OLED. Another fact concerning driving circuit is the tradeoff between resolution and frame rate as a higher resolution panel means more scanning lines in each frame. So far, an OLED display with 480 Hz frame rate has been demonstrated 141 . For an LCoS, the frame rate is mainly limited by the LC response time. Depending on the LC material used, the response time is around 1 ms for nematic LC or 200 µs for ferroelectric LC (FLC) 125 . Nematic LC allows analog driving, which accommodates gray levels, typically with 8-bit depth. FLC is bistable so that PWM is used to generate gray levels. DMD is also a binary device. The frame rate can reach 30 kHz, which is mainly constrained by the response time of micromirrors. For MEMS-based LBS, the frame rate is limited by the scanning frequency of MEMS mirrors. A frame rate of 60 Hz with around 1 K resolution already requires a resonance frequency of around 50 kHz, with a Q-factor up to 145,000 128 . A higher frame rate or resolution requires a higher Q-factor and larger laser modulation bandwidth, which may be challenging.

Form factor is another crucial aspect for the light engines of near-eye displays. For self-emissive displays, both micro-OLEDs and QD-based micro-LEDs can achieve full color with a single panel. Thus, they are quite compact. A micro-LED display with separate RGB panels naturally have a larger form factor. In applications requiring direct-view full-color panel, the extra combining optics may also increase the volume. It needs to be pointed out, however, that the combing optics may not be necessary for some applications like waveguide displays, because the EPE process results in system’s insensitivity to the spatial positions of input RGB images. Therefore, the form factor of using three RGB micro-LED panels is medium. For LCoS and DMD with RGB LEDs as light source, the form factor would be larger due to the illumination optics. Still, if a lower luminous efficacy can be accepted, then a smaller form factor can be achieved by using a simpler optics 142 . If RGB lasers are used, the collimation optics can be eliminated, which greatly reduces the form factor 143 . For MEMS-LBS, the form factor can be extremely compact due to the tiny size of MEMS mirror and laser module.

Finally, contrast ratio (CR) also plays an important role affecting the observed images 8 . Micro-LEDs and micro-OLEDs are self-emissive so that their CR can be >10 6 :1. For a laser beam scanner, its CR can also achieve 10 6 :1 because the laser can be turned off completely at dark state. On the other hand, LCoS and DMD are reflective displays, and their CR is around 2000:1 to 5000:1 144 , 145 . It is worth pointing out that the CR of a display engine plays a significant role only in the dark ambient. As the ambient brightness increases, the ACR is mainly governed by the display’s peak brightness, as previously discussed.

The performance parameters of different light engines are summarized in Table 1 . Micro-LEDs and micro-OLEDs have similar levels of luminous efficacy. But micro-OLEDs still face the burn-in and lifetime issue when driving at a high current, which hinders its use for a high-brightness image source to some extent. Micro-LEDs are still under active development and the improvement on luminous efficacy from maturing fabrication process could be expected. Both devices have nanosecond response time and can potentially achieve a high frame rate with a well-designed integrated circuit. The frame rate of the driving circuit ultimately determines the motion picture response time 146 . Their self-emissive feature also leads to a small form factor and high contrast ratio. LCoS and DMD engines have similar performance of luminous efficacy, form factor, and contrast ratio. In terms of light modulation, DMD can provide a higher 1-bit frame rate, while LCoS can offer both phase and amplitude modulations. MEMS-based LBS exhibits the highest luminous efficacy so far. It also exhibits an excellent form factor and contrast ratio, but the presently demonstrated 60-Hz frame rate (limited by the MEMS mirrors) could cause image flickering.

Free-space combiners

The term ‘free-space’ generally refers to the case when light is freely propagating in space, as opposed to a waveguide that traps light into TIRs. Regarding the combiner, it can be a partial mirror, as commonly used in AR systems based on traditional geometric optics. Alternatively, the combiner can also be a reflective HOE. The strong chromatic dispersion of HOE necessitates the use of a laser source, which usually leads to a Maxwellian-type system.

Traditional geometric designs

Several systems based on geometric optics are illustrated in Fig. 8 . The simplest design uses a single freeform half-mirror 6 , 147 to directly collimate the displayed images to the viewer’s eye (Fig. 8a ). This design can achieve a large FoV (up to 90°) 147 , but the limited design freedom with a single freeform surface leads to image distortions, also called pupil swim 6 . The placement of half-mirror also results in a relatively bulky form factor. Another design using so-called birdbath optics 6 , 148 is shown in Fig. 8b . Compared to the single-combiner design, birdbath design has an extra optics on the display side, which provides space for aberration correction. The integration of beam splitter provides a folded optical path, which reduces the form factor to some extent. Another way to fold optical path is to use a TIR-prism. Cheng et al. 149 designed a freeform TIR-prism combiner (Fig. 8c ) offering a diagonal FoV of 54° and exit pupil diameter of 8 mm. All the surfaces are freeform, which offer an excellent image quality. To cancel the optical power for the transmitted environmental light, a compensator is added to the TIR prism. The whole system has a well-balanced performance between FoV, eyebox, and form factor. To release the space in front of viewer’s eye, relay optics can be used to form an intermediate image near the combiner 150 , 151 , as illustrated in Fig. 8d . Although the design offers more optical surfaces for aberration correction, the extra lenses also add to system weight and form factor.

a Single freeform surface as the combiner. b Birdbath optics with a beam splitter and a half mirror. c Freeform TIR prism with a compensator. d Relay optics with a half mirror. Adapted from c ref. 149 with permission from OSA Publishing and d ref. 151 with permission from OSA Publishing

Regarding the approaches to solve the VAC issue, the most straightforward way is to integrate a tunable lens into the optical path, like a liquid lens 152 or Alvarez lens 99 , to form a varifocal system. Alternatively, integral imaging 153 , 154 can also be used, by replacing the original display panel with the central depth plane of an integral imaging module. The integral imaging can also be combined with varifocal approach to overcome the tradeoff between resolution and depth of field (DoF) 155 , 156 , 157 . However, the inherent tradeoff between resolution and view number still exists in this case.

Overall, AR displays based on traditional geometric optics have a relatively simple design with a decent FoV (~60°) and eyebox (8 mm) 158 . They also exhibit a reasonable efficiency. To measure the efficiency of an AR combiner, an appropriate measure is to divide the output luminance (unit: nit) by the input luminous flux (unit: lm), which we note as combiner efficiency. For a fixed input luminous flux, the output luminance, or image brightness, is related to the FoV and exit pupil of the combiner system. If we assume no light waste of the combiner system, then the maximum combiner efficiency for a typical diagonal FoV of 60° and exit pupil (10 mm square) is around 17,000 nit/lm (Eq. S2 ). To estimate the combiner efficiency of geometric combiners, we assume 50% of half-mirror transmittance and the efficiency of other optics to be 50%. Then the final combiner efficiency is about 4200 nit/lm, which is a high value in comparison with waveguide combiners. Nonetheless, to further shrink the system size or improve system performance ultimately encounters the etendue conservation issue. In addition, AR systems with traditional geometric optics is hard to achieve a configuration resembling normal flat glasses because the half-mirror has to be tilted to some extent.

Maxwellian-type systems

The Maxwellian view, proposed by James Clerk Maxwell (1860), refers to imaging a point light source in the eye pupil 159 . If the light beam is modulated in the imaging process, a corresponding image can be formed on the retina (Fig. 9a ). Because the point source is much smaller than the eye pupil, the image is always-in-focus on the retina irrespective of the eye lens’ focus. For applications in AR display, the point source is usually a laser with narrow angular and spectral bandwidths. LED light sources can also build a Maxwellian system, by adding an angular filtering module 160 . Regarding the combiner, although in theory a half-mirror can also be used, HOEs are generally preferred because they offer the off-axis configuration that places combiner in a similar position like eyeglasses. In addition, HOEs have a lower reflection of environment light, which provides a more natural appearance of the user behind the display.

a Schematic of the working principle of Maxwellian displays. Maxwellian displays based on b SLM and laser diode light source and c MEMS-LBS with a steering mirror as additional modulation method. Generation of depth cues by d computational digital holography and e scanning of steering mirror to produce multiple views. Adapted from b, d ref. 143 and c, e ref. 167 under the Creative Commons Attribution 4.0 License

To modulate the light, a SLM like LCoS or DMD can be placed in the light path, as shown in Fig. 9b . Alternatively, LBS system can also be used (Fig. 9c ), where the intensity modulation occurs in the laser diode itself. Besides the operation in a normal Maxwellian-view, both implementations offer additional degrees of freedom for light modulation.

For a SLM-based system, there are several options to arrange the SLM pixels 143 , 161 . Maimone et al. 143 demonstrated a Maxwellian AR display with two modes to offer a large-DoF Maxwellian-view, or a holographic view (Fig. 9d ), which is often referred as computer-generated holography (CGH) 162 . To show an always-in-focus image with a large DoF, the image can be directly displayed on an amplitude SLM, or using amplitude encoding for a phase-only SLM 163 . Alternatively, if a 3D scene with correct depth cues is to be presented, then optimization algorithms for CGH can be used to generate a hologram for the SLM. The generated holographic image exhibits the natural focus-and-blur effect like a real 3D object (Fig. 9d ). To better understand this feature, we need to again exploit the concept of etendue. The laser light source can be considered to have a very small etendue due to its excellent collimation. Therefore, the system etendue is provided by the SLM. The micron-sized pixel-pitch of SLM offers a certain maximum diffraction angle, which, multiplied by the SLM size, equals system etendue. By varying the display content on SLM, the final exit pupil size can be changed accordingly. In the case of a large-DoF Maxwellian view, the exit pupil size is small, accompanied by a large FoV. For the holographic display mode, the reduced DoF requires a larger exit pupil with dimension close to the eye pupil. But the FoV is reduced accordingly due to etendue conservation. Another commonly concerned issue with CGH is the computation time. To achieve a real-time CGH rendering flow with an excellent image quality is quite a challenge. Fortunately, with recent advances in algorithm 164 and the introduction of convolutional neural network (CNN) 165 , 166 , this issue is gradually solved with an encouraging pace. Lately, Liang et al. 166 demonstrated a real-time CGH synthesis pipeline with a high image quality. The pipeline comprises an efficient CNN model to generate a complex hologram from a 3D scene and an improved encoding algorithm to convert the complex hologram to a phase-only one. An impressive frame rate of 60 Hz has been achieved on a desktop computing unit.

For LBS-based system, the additional modulation can be achieved by integrating a steering module, as demonstrated by Jang et al. 167 . The steering mirror can shift the focal point (viewpoint) within the eye pupil, therefore effectively expanding the system etendue. When the steering process is fast and the image content is updated simultaneously, correct 3D cues can be generated, as shown in Fig. 9e . However, there exists a tradeoff between the number of viewpoint and the final image frame rate, because the total frames are equally divided into each viewpoint. To boost the frame rate of MEMS-LBS systems by the number of views (e.g., 3 by 3) may be challenging.

Maxwellian-type systems offer several advantages. The system efficiency is usually very high because nearly all the light is delivered into viewer’s eye. The system FoV is determined by the f /# of combiner and a large FoV (~80° in horizontal) can be achieved 143 . The issue of VAC can be mitigated with an infinite-DoF image that deprives accommodation cue, or completely solved by generating a true-3D scene as discussed above. Despite these advantages, one major weakness of Maxwellian-type system is the tiny exit pupil, or eyebox. A small deviation of eye pupil location from the viewpoint results in the complete disappearance of the image. Therefore, to expand eyebox is considered as one of the most important challenges in Maxwellian-type systems.

Pupil duplication and steering

Methods to expand eyebox can be generally categorized into pupil duplication 168 , 169 , 170 , 171 , 172 and pupil steering 9 , 13 , 167 , 173 . Pupil duplication simply generates multiple viewpoints to cover a large area. In contrast, pupil steering dynamically shifts the viewpoint position, depending on the pupil location. Before reviewing detailed implementations of these two methods, it is worth discussing some of their general features. The multiple viewpoints in pupil duplication usually mean to equally divide the total light intensity. In each time frame, however, it is preferable that only one viewpoint enters the user’s eye pupil to avoid ghost image. This requirement, therefore, results in a reduced total light efficiency, while also conditioning the viewpoint separation to be larger than the pupil diameter. In addition, the separation should not be too large to avoid gap between viewpoints. Considering that human pupil diameter changes in response to environment illuminance, the design of viewpoint separation needs special attention. Pupil steering, on the other hand, only produces one viewpoint at each time frame. It is therefore more light-efficient and free from ghost images. But to determine the viewpoint position requires the information of eye pupil location, which demands a real-time eye-tracking module 9 . Another observation is that pupil steering can accommodate multiple viewpoints by its nature. Therefore, a pupil steering system can often be easily converted to a pupil duplication system by simultaneously generating available viewpoints.

To generate multiple viewpoints, one can focus on modulating the incident light or the combiner. Recall that viewpoint is the image of light source. To duplicate or shift light source can achieve pupil duplication or steering accordingly, as illustrated in Fig. 10a . Several schemes of light modulation are depicted in Fig. 10b–e . An array of light sources can be generated with multiple laser diodes (Fig. 10b ). To turn on all or one of the sources achieves pupil duplication or steering. A light source array can also be produced by projecting light on an array-type PPHOE 168 (Fig. 10c ). Apart from direct adjustment of light sources, modulating light on the path can also effectively steer/duplicate the light sources. Using a mechanical steering mirror, the beam can be deflected 167 (Fig. 10d ), which equals to shifting the light source position. Other devices like a grating or beam splitter can also serve as ray deflector/splitter 170 , 171 (Fig. 10e ).

a Schematic of duplicating (or shift) viewpoint by modulation of incident light. Light modulation by b multiple laser diodes, c HOE lens array, d steering mirror and e grating or beam splitters. f Pupil duplication with multiplexed PPHOE. g Pupil steering with LCHOE. Reproduced from c ref. 168 under the Creative Commons Attribution 4.0 License, e ref. 169 with permission from OSA Publishing, f ref. 171 with permission from OSA Publishing and g ref. 173 with permission from OSA Publishing

Nonetheless, one problem of the light source duplication/shifting methods for pupil duplication/steering is that the aberrations in peripheral viewpoints are often serious 168 , 173 . The HOE combiner is usually recorded at one incident angle. For other incident angles with large deviations, considerable aberrations will occur, especially in the scenario of off-axis configuration. To solve this problem, the modulation can be focused on the combiner instead. While the mechanical shifting of combiner 9 can achieve continuous pupil steering, its integration into AR display with a small factor remains a challenge. Alternatively, the versatile functions of HOE offer possible solutions for combiner modulation. Kim and Park 169 demonstrated a pupil duplication system with multiplexed PPHOE (Fig. 10f ). Wavefronts of several viewpoints can be recorded into one PPHOE sample. Three viewpoints with a separation of 3 mm were achieved. However, a slight degree of ghost image and gap can be observed in the viewpoint transition. For a PPHOE to achieve pupil steering, the multiplexed PPHOE needs to record different focal points with different incident angles. If each hologram has no angular crosstalk, then with an additional device to change the light incident angle, the viewpoint can be steered. Alternatively, Xiong et al. 173 demonstrated a pupil steering system with LCHOEs in a simpler configuration (Fig. 10g ). The polarization-sensitive nature of LCHOE enables the controlling of which LCHOE to function with a polarization converter (PC). When the PC is off, the incident RCP light is focused by the right-handed LCHOE. When the PC is turned on, the RCP light is firstly converted to LCP light and passes through the right-handed LCHOE. Then it is focused by the left-handed LCHOE into another viewpoint. To add more viewpoints requires stacking more pairs of PC and LCHOE, which can be achieved in a compact manner with thin glass substrates. In addition, to realize pupil duplication only requires the stacking of multiple low-efficiency LCHOEs. For both PPHOEs and LCHOEs, because the hologram for each viewpoint is recorded independently, the aberrations can be eliminated.

Regarding the system performance, in theory the FoV is not limited and can reach a large value, such as 80° in horizontal direction 143 . The definition of eyebox is different from traditional imaging systems. For a single viewpoint, it has the same size as the eye pupil diameter. But due to the viewpoint steering/duplication capability, the total system eyebox can be expanded accordingly. The combiner efficiency for pupil steering systems can reach 47,000 nit/lm for a FoV of 80° by 80° and pupil diameter of 4 mm (Eq. S2 ). At such a high brightness level, eye safety could be a concern 174 . For a pupil duplication system, the combiner efficiency is decreased by the number of viewpoints. With a 4-by-4 viewpoint array, it can still reach 3000 nit/lm. Despite the potential gain of pupil duplication/steering, when considering the rotation of eyeball, the situation becomes much more complicated 175 . A perfect pupil steering system requires a 5D steering, which proposes a challenge for practical implementation.

Pin-light systems

Recently, another type of display in close relation with Maxwellian view called pin-light display 148 , 176 has been proposed. The general working principle of pin-light display is illustrated in Fig. 11a . Each pin-light source is a Maxwellian view with a large DoF. When the eye pupil is no longer placed near the source point as in Maxwellian view, each image source can only form an elemental view with a small FoV on retina. However, if the image source array is arranged in a proper form, the elemental views can be integrated together to form a large FoV. According to the specific optical architectures, pin-light display can take different forms of implementation. In the initial feasibility demonstration, Maimone et al. 176 used a side-lit waveguide plate as the point light source (Fig. 11b ). The light inside the waveguide plate is extracted by the etched divots, forming a pin-light source array. A transmissive SLM (LCD) is placed behind the waveguide plate to modulate the light intensity and form the image. The display has an impressive FoV of 110° thanks to the large scattering angle range. However, the direct placement of LCD before the eye brings issues of insufficient resolution density and diffraction of background light.

a Schematic drawing of the working principle of pin-light display. b Pin-light display utilizing a pin-light source and a transmissive SLM. c An example of pin-mirror display with a birdbath optics. d SWD system with LBS image source and off-axis lens array. Reprinted from b ref. 176 under the Creative Commons Attribution 4.0 License and d ref. 180 with permission from OSA Publishing

To avoid these issues, architectures using pin-mirrors 177 , 178 , 179 are proposed. In these systems, the final combiner is an array of tiny mirrors 178 , 179 or gratings 177 , in contrast to their counterparts using large-area combiners. An exemplary system with birdbath design is depicted in Fig. 11c . In this case, the pin-mirrors replace the original beam-splitter in the birdbath and can thus shrink the system volume, while at the same time providing large DoF pin-light images. Nonetheless, such a system may still face the etendue conservation issue. Meanwhile, the size of pin-mirror cannot be too small in order to prevent degradation of resolution density due to diffraction. Therefore, its influence on the see-through background should also be considered in the system design.

To overcome the etendue conservation and improve see-through quality, Xiong et al. 180 proposed another type of pin-light system exploiting the etendue expansion property of waveguide, which is also referred as scanning waveguide display (SWD). As illustrated in Fig. 11d , the system uses an LBS as the image source. The collimated scanned laser rays are trapped in the waveguide and encounter an array of off-axis lenses. Upon each encounter, the lens out-couples the laser rays and forms a pin-light source. SWD has the merits of good see-through quality and large etendue. A large FoV of 100° was demonstrated with the help of an ultra-low f /# lens array based on LCHOE. However, some issues like insufficient image resolution density and image non-uniformity remain to be overcome. To further improve the system may require optimization of Gaussian beam profile and additional EPE module 180 .

Overall, pin-light systems inherit the large DoF from Maxwellian view. With adequate number of pin-light sources, the FoV and eyebox can be expanded accordingly. Nonetheless, despite different forms of implementation, a common issue of pin-light system is the image uniformity. The overlapped region of elemental views has a higher light intensity than the non-overlapped region, which becomes even more complicated considering the dynamic change of pupil size. In theory, the displayed image can be pre-processed to compensate for the optical non-uniformity. But that would require knowledge of precise pupil location (and possibly size) and therefore an accurate eye-tracking module 176 . Regarding the system performance, pin-mirror systems modified from other free-space systems generally shares similar FoV and eyebox with original systems. The combiner efficiency may be lower due to the small size of pin-mirrors. SWD, on the other hand, shares the large FoV and DoF with Maxwellian view, and large eyebox with waveguide combiners. The combiner efficiency may also be lower due to the EPE process.

Waveguide combiner

Besides free-space combiners, another common architecture in AR displays is waveguide combiner. The term ‘waveguide’ indicates the light is trapped in a substrate by the TIR process. One distinctive feature of a waveguide combiner is the EPE process that effectively enlarges the system etendue. In the EPE process, a portion of the trapped light is repeatedly coupled out of the waveguide in each TIR. The effective eyebox is therefore enlarged. According to the features of couplers, we divide the waveguide combiners into two types: diffractive and achromatic, as described in the followings.

Diffractive waveguides

As the name implies, diffractive-type waveguides use diffractive elements as couplers. The in-coupler is usually a diffractive grating and the out-coupler in most cases is also a grating with the same period as the in-coupler, but it can also be an off-axis lens with a small curvature to generate image with finite depth. Three major diffractive couplers have been developed: SRGs, photopolymer gratings (PPGs), and liquid crystal gratings (grating-type LCHOE; also known as polarization volume gratings (PVGs)). Some general protocols for coupler design are that the in-coupler should have a relatively high efficiency and the out-coupler should have a uniform light output. A uniform light output usually requires a low-efficiency coupler, with extra degrees of freedom for local modulation of coupling efficiency. Both in-coupler and out-coupler should have an adequate angular bandwidth to accommodate a reasonable FoV. In addition, the out-coupler should also be optimized to avoid undesired diffractions, including the outward diffraction of TIR light and diffraction of environment light into user’s eyes, which are referred as light leakage and rainbow. Suppression of these unwanted diffractions should also be considered in the optimization process of waveguide design, along with performance parameters like efficiency and uniformity.

The basic working principles of diffractive waveguide-based AR systems are illustrated in Fig. 12 . For the SRG-based waveguides 6 , 8 (Fig. 12a ), the in-coupler can be a transmissive-type or a reflective-type 181 , 182 . The grating geometry can be optimized for coupling efficiency with a large degree of freedom 183 . For the out-coupler, a reflective SRG with a large slant angle to suppress the transmission orders is preferred 184 . In addition, a uniform light output usually requires a gradient efficiency distribution in order to compensate for the decreased light intensity in the out-coupling process. This can be achieved by varying the local grating configurations like height and duty cycle 6 . For the PPG-based waveguides 185 (Fig. 12b ), the small angular bandwidth of a high-efficiency transmissive PPG prohibits its use as in-coupler. Therefore, both in-coupler and out-coupler are usually reflective types. The gradient efficiency can be achieved by space-variant exposure to control the local index modulation 186 or local Bragg slant angle variation through freeform exposure 19 . Due to the relatively small angular bandwidth of PPG, to achieve a decent FoV usually requires stacking two 187 or three 188 PPGs together for a single color. The PVG-based waveguides 189 (Fig. 12c ) also prefer reflective PVGs as in-couplers because the transmissive PVGs are much more difficult to fabricate due to the LC alignment issue. In addition, the angular bandwidth of transmissive PVGs in Bragg regime is also not large enough to support a decent FoV 29 . For the out-coupler, the angular bandwidth of a single reflective PVG can usually support a reasonable FoV. To obtain a uniform light output, a polarization management layer 190 consisting of a LC layer with spatially variant orientations can be utilized. It offers an additional degree of freedom to control the polarization state of the TIR light. The diffraction efficiency can therefore be locally controlled due to the strong polarization sensitivity of PVG.

Schematics of waveguide combiners based on a SRGs, b PPGs and c PVGs. Reprinted from a ref. 85 with permission from OSA Publishing, b ref. 185 with permission from John Wiley and Sons and c ref. 189 with permission from OSA Publishing

The above discussion describes the basic working principle of 1D EPE. Nonetheless, for the 1D EPE to produce a large eyebox, the exit pupil in the unexpanded direction of the original image should be large. This proposes design challenges in light engines. Therefore, a 2D EPE is favored for practical applications. To extend EPE in two dimensions, two consecutive 1D EPEs can be used 191 , as depicted in Fig. 13a . The first 1D EPE occurs in the turning grating, where the light is duplicated in y direction and then turned into x direction. Then the light rays encounter the out-coupler and are expanded in x direction. To better understand the 2D EPE process, the k -vector diagram (Fig. 13b ) can be used. For the light propagating in air with wavenumber k 0 , its possible k -values in x and y directions ( k x and k y ) fall within the circle with radius k 0 . When the light is trapped into TIR, k x and k y are outside the circle with radius k 0 and inside the circle with radius nk 0 , where n is the refractive index of the substrate. k x and k y stay unchanged in the TIR process and are only changed in each diffraction process. The central red box in Fig. 13b indicates the possible k values within the system FoV. After the in-coupler, the k values are added by the grating k -vector, shifting the k values into TIR region. The turning grating then applies another k -vector and shifts the k values to near x -axis. Finally, the k values are shifted by the out-coupler and return to the free propagation region in air. One observation is that the size of red box is mostly limited by the width of TIR band. To accommodate a larger FoV, the outer boundary of TIR band needs to be expanded, which amounts to increasing waveguide refractive index. Another important fact is that when k x and k y are near the outer boundary, the uniformity of output light becomes worse. This is because the light propagation angle is near 90° in the waveguide. The spatial distance between two consecutive TIRs becomes so large that the out-coupled beams are spatially separated to an unacceptable degree. The range of possible k values for practical applications is therefore further shrunk due to this fact.

a Schematic of 2D EPE based on two consecutive 1D EPEs. Gray/black arrows indicate light in air/TIR. Black dots denote TIRs. b k-diagram of the two-1D-EPE scheme. c Schematic of 2D EPE with a 2D hexagonal grating d k-diagram of the 2D-grating scheme

Aside from two consecutive 1D EPEs, the 2D EPE can also be directly implemented with a 2D grating 192 . An example using a hexagonal grating is depicted in Fig. 13c . The hexagonal grating can provide k -vectors in six directions. In the k -diagram (Fig. 13d ), after the in-coupling, the k values are distributed into six regions due to multiple diffractions. The out-coupling occurs simultaneously with pupil expansion. Besides a concise out-coupler configuration, the 2D EPE scheme offers more degrees of design freedom than two 1D EPEs because the local grating parameters can be adjusted in a 2D manner. The higher design freedom has the potential to reach a better output light uniformity, but at the cost of a higher computation demand for optimization. Furthermore, the unslanted grating geometry usually leads to a large light leakage and possibly low efficiency. Adding slant to the geometry helps alleviate the issue, but the associated fabrication may be more challenging.

Finally, we discuss the generation of full-color images. One important issue to clarify is that although diffractive gratings are used here, the final image generally has no color dispersion even if we use a broadband light source like LED. This can be easily understood in the 1D EPE scheme. The in-coupler and out-coupler have opposite k -vectors, which cancels the color dispersion for each other. In the 2D EPE schemes, the k -vectors always form a closed loop from in-coupled light to out-coupled light, thus, the color dispersion also vanishes likewise. The issue of using a single waveguide for full-color images actually exists in the consideration of FoV and light uniformity. The breakup of propagation angles for different colors results in varied out-coupling situations for each color. To be more specific, if the red and the blue channels use the same in-coupler, the propagating angle for the red light is larger than that of the blue light. The red light in peripheral FoV is therefore easier to face the mentioned large-angle non-uniformity issue. To acquire a decent FoV and light uniformity, usually two or three layers of waveguides with different grating pitches are adopted.

Regarding the system performance, the eyebox is generally large enough (~10 mm) to accommodate different user’s IPD and alignment shift during operation. A parameter of significant concern for a waveguide combiner is its FoV. From the k -vector analysis, we can conclude the theoretical upper limit is determined by the waveguide refractive index. But the light/color uniformity also influences the effective FoV, over which the degradation of image quality becomes unacceptable. Current diffractive waveguide combiners generally achieve a FoV of about 50°. To further increase FoV, a straightforward method is to use a higher refractive index waveguide. Another is to tile FoV through direct stacking of multiple waveguides or using polarization-sensitive couplers 79 , 193 . As to the optical efficiency, a typical value for the diffractive waveguide combiner is around 50–200 nit/lm 6 , 189 . In addition, waveguide combiners adopting grating out-couplers generate an image with fixed depth at infinity. This leads to the VAC issue. To tackle VAC in waveguide architectures, the most practical way is to generate multiple depths and use the varifocal or multifocal driving scheme, similar to those mentioned in the VR systems. But to add more depths usually means to stack multiple layers of waveguides together 194 . Considering the additional waveguide layers for RGB colors, the final waveguide thickness would undoubtedly increase.

Other parameters special to waveguide includes light leakage, see-through ghost, and rainbow. Light leakage refers to out-coupled light that goes outwards to the environment, as depicted in Fig. 14a . Aside from decreased efficiency, the leakage also brings drawback of unnatural “bright-eye” appearance of the user and privacy issue. Optimization of the grating structure like geometry of SRG may reduce the leakage. See-through ghost is formed by consecutive in-coupling and out-couplings caused by the out-coupler grating, as sketched in Fig. 14b , After the process, a real object with finite depth may produce a ghost image with shift in both FoV and depth. Generally, an out-coupler with higher efficiency suffers more see-through ghost. Rainbow is caused by the diffraction of environment light into user’s eye, as sketched in Fig. 14c . The color dispersion in this case will occur because there is no cancellation of k -vector. Using the k -diagram, we can obtain a deeper insight into the formation of rainbow. Here, we take the EPE structure in Fig. 13a as an example. As depicted in Fig. 14d , after diffractions by the turning grating and the out-coupler grating, the k values are distributed in two circles that shift from the origin by the grating k -vectors. Some diffracted light can enter the see-through FoV and form rainbow. To reduce rainbow, a straightforward way is to use a higher index substrate. With a higher refractive index, the outer boundary of k diagram is expanded, which can accommodate larger grating k -vectors. The enlarged k -vectors would therefore “push” these two circles outwards, leading to a decreased overlapping region with the see-through FoV. Alternatively, an optimized grating structure would also help reduce the rainbow effect by suppressing the unwanted diffraction.

Sketches of formations of a light leakage, b see-through ghost and c rainbow. d Analysis of rainbow formation with k-diagram

Achromatic waveguide

Achromatic waveguide combiners use achromatic elements as couplers. It has the advantage of realizing full-color image with a single waveguide. A typical example of achromatic element is a mirror. The waveguide with partial mirrors as out-coupler is often referred as geometric waveguide 6 , 195 , as depicted in Fig. 15a . The in-coupler in this case is usually a prism to avoid unnecessary color dispersion if using diffractive elements otherwise. The mirrors couple out TIR light consecutively to produce a large eyebox, similarly in a diffractive waveguide. Thanks to the excellent optical property of mirrors, the geometric waveguide usually exhibits a superior image regarding MTF and color uniformity to its diffractive counterparts. Still, the spatially discontinuous configuration of mirrors also results in gaps in eyebox, which may be alleviated by using a dual-layer structure 196 . Wang et al. designed a geometric waveguide display with five partial mirrors (Fig. 15b ). It exhibits a remarkable FoV of 50° by 30° (Fig. 15c ) and an exit pupil of 4 mm with a 1D EPE. To achieve 2D EPE, similar architectures in Fig. 13a can be used by integrating a turning mirror array as the first 1D EPE module 197 . Unfortunately, the k -vector diagrams in Fig. 13b, d cannot be used here because the k values in x-y plane no longer conserve in the in-coupling and out-coupling processes. But some general conclusions remain valid, like a higher refractive index leading to a larger FoV and gradient out-coupling efficiency improving light uniformity.

a Schematic of the system configuration. b Geometric waveguide with five partial mirrors. c Image photos demonstrating system FoV. Adapted from b , c ref. 195 with permission from OSA Publishing

The fabrication process of geometric waveguide involves coating mirrors on cut-apart pieces and integrating them back together, which may result in a high cost, especially for the 2D EPE architecture. Another way to implement an achromatic coupler is to use multiplexed PPHOE 198 , 199 to mimic the behavior of a tilted mirror (Fig. 16a ). To understand the working principle, we can use the diagram in Fig. 16b . The law of reflection states the angle of reflection equals to the angle of incidence. If we translate this behavior to k -vector language, it means the mirror can apply any length of k -vector along its surface normal direction. The k -vector length of the reflected light is always equal to that of the incident light. This puts a condition that the k -vector triangle is isosceles. With a simple geometric deduction, it can be easily observed this leads to the law of reflection. The behavior of a general grating, however, is very different. For simplicity we only consider the main diffraction order. The grating can only apply a k -vector with fixed k x due to the basic diffraction law. For the light with a different incident angle, it needs to apply different k z to produce a diffracted light with equal k -vector length as the incident light. For a grating with a broad angular bandwidth like SRG, the range of k z is wide, forming a lengthy vertical line in Fig. 16b . For a PPG with a narrow angular bandwidth, the line is short and resembles a dot. If multiple of these tiny dots are distributed along the oblique line corresponding to a mirror, then the final multiplexed PPGs can imitate the behavior of a tilted mirror. Such a PPHOE is sometimes referred as a skew-mirror 198 . In theory, to better imitate the mirror, a lot of multiplexed PPGs is preferred, while each PPG has a small index modulation δn . But this proposes a bigger challenge in device fabrication. Recently, Utsugi et al. demonstrated an impressive skew-mirror waveguide based on 54 multiplexed PPGs (Fig. 16c, d ). The display exhibits an effective FoV of 35° by 36°. In the peripheral FoV, there still exists some non-uniformity (Fig. 16e ) due to the out-coupling gap, which is an inherent feature of the flat-type out-couplers.

a System configuration. b Diagram demonstrating how multiplexed PPGs resemble the behavior of a mirror. Photos showing c the system and d image. e Picture demonstrating effective system FoV. Adapted from c – e ref. 199 with permission from ITE

Finally, it is worth mentioning that metasurfaces are also promising to deliver achromatic gratings 200 , 201 for waveguide couplers ascribed to their versatile wavefront shaping capability. The mechanism of the achromatic gratings is similar to that of the achromatic lenses as previously discussed. However, the current development of achromatic metagratings is still in its infancy. Much effort is needed to improve the optical efficiency for in-coupling, control the higher diffraction orders for eliminating ghost images, and enable a large size design for EPE.

Generally, achromatic waveguide combiners exhibit a comparable FoV and eyebox with diffractive combiners, but with a higher efficiency. For a partial-mirror combiner, its combiner efficiency is around 650 nit/lm 197 (2D EPE). For a skew-mirror combiner, although the efficiency of multiplexed PPHOE is relatively low (~1.5%) 199 , the final combiner efficiency of the 1D EPE system is still high (>3000 nit/lm) due to multiple out-couplings.

Table 2 summarizes the performance of different AR combiners. When combing the luminous efficacy in Table 1 and the combiner efficiency in Table 2 , we can have a comprehensive estimate of the total luminance efficiency (nit/W) for different types of systems. Generally, Maxwellian-type combiners with pupil steering have the highest luminance efficiency when partnered with laser-based light engines like laser-backlit LCoS/DMD or MEM-LBS. Geometric optical combiners have well-balanced image performances, but to further shrink the system size remains a challenge. Diffractive waveguides have a relatively low combiner efficiency, which can be remedied by an efficient light engine like MEMS-LBS. Further development of coupler and EPE scheme would also improve the system efficiency and FoV. Achromatic waveguides have a decent combiner efficiency. The single-layer design also enables a smaller form factor. With advances in fabrication process, it may become a strong contender to presently widely used diffractive waveguides.

Conclusions and perspectives

VR and AR are endowed with a high expectation to revolutionize the way we interact with digital world. Accompanied with the expectation are the engineering challenges to squeeze a high-performance display system into a tightly packed module for daily wearing. Although the etendue conservation constitutes a great obstacle on the path, remarkable progresses with innovative optics and photonics continue to take place. Ultra-thin optical elements like PPHOEs and LCHOEs provide alternative solutions to traditional optics. Their unique features of multiplexing capability and polarization dependency further expand the possibility of novel wavefront modulations. At the same time, nanoscale-engineered metasurfaces/SRGs provide large design freedoms to achieve novel functions beyond conventional geometric optical devices. Newly emerged micro-LEDs open an opportunity for compact microdisplays with high peak brightness and good stability. Further advances on device engineering and manufacturing process are expected to boost the performance of metasurfaces/SRGs and micro-LEDs for AR and VR applications.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper. Additional data related to this paper may be requested from the authors.

Cakmakci, O. & Rolland, J. Head-worn displays: a review. J. Disp. Technol. 2 , 199–216 (2006).

Article   ADS   Google Scholar  

Zhan, T. et al. Augmented reality and virtual reality displays: perspectives and challenges. iScience 23 , 101397 (2020).

Rendon, A. A. et al. The effect of virtual reality gaming on dynamic balance in older adults. Age Ageing 41 , 549–552 (2012).

Article   Google Scholar  

Choi, S., Jung, K. & Noh, S. D. Virtual reality applications in manufacturing industries: past research, present findings, and future directions. Concurrent Eng. 23 , 40–63 (2015).

Li, X. et al. A critical review of virtual and augmented reality (VR/AR) applications in construction safety. Autom. Constr. 86 , 150–162 (2018).

Kress, B. C. Optical Architectures for Augmented-, Virtual-, and Mixed-Reality Headsets (Bellingham: SPIE Press, 2020).

Cholewiak, S. A. et al. A perceptual eyebox for near-eye displays. Opt. Express 28 , 38008–38028 (2020).

Lee, Y. H., Zhan, T. & Wu, S. T. Prospects and challenges in augmented reality displays. Virtual Real. Intell. Hardw. 1 , 10–20 (2019).

Kim, J. et al. Foveated AR: dynamically-foveated augmented reality display. ACM Trans. Graph. 38 , 99 (2019).

Tan, G. J. et al. Foveated imaging for near-eye displays. Opt. Express 26 , 25076–25085 (2018).

Lee, S. et al. Foveated near-eye display for mixed reality using liquid crystal photonics. Sci. Rep. 10 , 16127 (2020).

Yoo, C. et al. Foveated display system based on a doublet geometric phase lens. Opt. Express 28 , 23690–23702 (2020).

Akşit, K. et al. Manufacturing application-driven foveated near-eye displays. IEEE Trans. Vis. Computer Graph. 25 , 1928–1939 (2019).

Zhu, R. D. et al. High-ambient-contrast augmented reality with a tunable transmittance liquid crystal film and a functional reflective polarizer. J. Soc. Inf. Disp. 24 , 229–233 (2016).

Lincoln, P. et al. Scene-adaptive high dynamic range display for low latency augmented reality. In Proc. 21st ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games . (ACM, San Francisco, CA, 2017).

Duerr, F. & Thienpont, H. Freeform imaging systems: fermat’s principle unlocks “first time right” design. Light.: Sci. Appl. 10 , 95 (2021).

Bauer, A., Schiesser, E. M. & Rolland, J. P. Starting geometry creation and design method for freeform optics. Nat. Commun. 9 , 1756 (2018).

Rolland, J. P. et al. Freeform optics for imaging. Optica 8 , 161–176 (2021).

Jang, C. et al. Design and fabrication of freeform holographic optical elements. ACM Trans. Graph. 39 , 184 (2020).

Gabor, D. A new microscopic principle. Nature 161 , 777–778 (1948).

Kostuk, R. K. Holography: Principles and Applications (Boca Raton: CRC Press, 2019).

Lawrence, J. R., O'Neill, F. T. & Sheridan, J. T. Photopolymer holographic recording material. Optik 112 , 449–463 (2001).

Guo, J. X., Gleeson, M. R. & Sheridan, J. T. A review of the optimisation of photopolymer materials for holographic data storage. Phys. Res. Int. 2012 , 803439 (2012).

Jang, C. et al. Recent progress in see-through three-dimensional displays using holographic optical elements [Invited]. Appl. Opt. 55 , A71–A85 (2016).

Xiong, J. H. et al. Holographic optical elements for augmented reality: principles, present status, and future perspectives. Adv. Photonics Res. 2 , 2000049 (2021).

Tabiryan, N. V. et al. Advances in transparent planar optics: enabling large aperture, ultrathin lenses. Adv. Optical Mater. 9 , 2001692 (2021).

Zanutta, A. et al. Photopolymeric films with highly tunable refractive index modulation for high precision diffractive optics. Optical Mater. Express 6 , 252–263 (2016).

Moharam, M. G. & Gaylord, T. K. Rigorous coupled-wave analysis of planar-grating diffraction. J. Optical Soc. Am. 71 , 811–818 (1981).

Xiong, J. H. & Wu, S. T. Rigorous coupled-wave analysis of liquid crystal polarization gratings. Opt. Express 28 , 35960–35971 (2020).

Xie, S., Natansohn, A. & Rochon, P. Recent developments in aromatic azo polymers research. Chem. Mater. 5 , 403–411 (1993).

Shishido, A. Rewritable holograms based on azobenzene-containing liquid-crystalline polymers. Polym. J. 42 , 525–533 (2010).

Bunning, T. J. et al. Holographic polymer-dispersed liquid crystals (H-PDLCs). Annu. Rev. Mater. Sci. 30 , 83–115 (2000).

Liu, Y. J. & Sun, X. W. Holographic polymer-dispersed liquid crystals: materials, formation, and applications. Adv. Optoelectron. 2008 , 684349 (2008).

Xiong, J. H. & Wu, S. T. Planar liquid crystal polarization optics for augmented reality and virtual reality: from fundamentals to applications. eLight 1 , 3 (2021).

Yaroshchuk, O. & Reznikov, Y. Photoalignment of liquid crystals: basics and current trends. J. Mater. Chem. 22 , 286–300 (2012).

Sarkissian, H. et al. Periodically aligned liquid crystal: potential application for projection displays. Mol. Cryst. Liq. Cryst. 451 , 1–19 (2006).

Komanduri, R. K. & Escuti, M. J. Elastic continuum analysis of the liquid crystal polarization grating. Phys. Rev. E 76 , 021701 (2007).

Kobashi, J., Yoshida, H. & Ozaki, M. Planar optics with patterned chiral liquid crystals. Nat. Photonics 10 , 389–392 (2016).

Lee, Y. H., Yin, K. & Wu, S. T. Reflective polarization volume gratings for high efficiency waveguide-coupling augmented reality displays. Opt. Express 25 , 27008–27014 (2017).

Lee, Y. H., He, Z. Q. & Wu, S. T. Optical properties of reflective liquid crystal polarization volume gratings. J. Optical Soc. Am. B 36 , D9–D12 (2019).

Xiong, J. H., Chen, R. & Wu, S. T. Device simulation of liquid crystal polarization gratings. Opt. Express 27 , 18102–18112 (2019).

Czapla, A. et al. Long-period fiber gratings with low-birefringence liquid crystal. Mol. Cryst. Liq. Cryst. 502 , 65–76 (2009).

Dąbrowski, R., Kula, P. & Herman, J. High birefringence liquid crystals. Crystals 3 , 443–482 (2013).

Mack, C. Fundamental Principles of Optical Lithography: The Science of Microfabrication (Chichester: John Wiley & Sons, 2007).

Genevet, P. et al. Recent advances in planar optics: from plasmonic to dielectric metasurfaces. Optica 4 , 139–152 (2017).

Guo, L. J. Nanoimprint lithography: methods and material requirements. Adv. Mater. 19 , 495–513 (2007).

Park, J. et al. Electrically driven mid-submicrometre pixelation of InGaN micro-light-emitting diode displays for augmented-reality glasses. Nat. Photonics 15 , 449–455 (2021).

Khorasaninejad, M. et al. Metalenses at visible wavelengths: diffraction-limited focusing and subwavelength resolution imaging. Science 352 , 1190–1194 (2016).

Li, S. Q. et al. Phase-only transmissive spatial light modulator based on tunable dielectric metasurface. Science 364 , 1087–1090 (2019).

Liang, K. L. et al. Advances in color-converted micro-LED arrays. Jpn. J. Appl. Phys. 60 , SA0802 (2020).

Jin, S. X. et al. GaN microdisk light emitting diodes. Appl. Phys. Lett. 76 , 631–633 (2000).

Day, J. et al. Full-scale self-emissive blue and green microdisplays based on GaN micro-LED arrays. In Proc. SPIE 8268, Quantum Sensing and Nanophotonic Devices IX (SPIE, San Francisco, California, United States, 2012).

Huang, Y. G. et al. Mini-LED, micro-LED and OLED displays: present status and future perspectives. Light.: Sci. Appl. 9 , 105 (2020).

Parbrook, P. J. et al. Micro-light emitting diode: from chips to applications. Laser Photonics Rev. 15 , 2000133 (2021).

Day, J. et al. III-Nitride full-scale high-resolution microdisplays. Appl. Phys. Lett. 99 , 031116 (2011).

Liu, Z. J. et al. 360 PPI flip-chip mounted active matrix addressable light emitting diode on silicon (LEDoS) micro-displays. J. Disp. Technol. 9 , 678–682 (2013).

Zhang, L. et al. Wafer-scale monolithic hybrid integration of Si-based IC and III–V epi-layers—A mass manufacturable approach for active matrix micro-LED micro-displays. J. Soc. Inf. Disp. 26 , 137–145 (2018).

Tian, P. F. et al. Size-dependent efficiency and efficiency droop of blue InGaN micro-light emitting diodes. Appl. Phys. Lett. 101 , 231110 (2012).

Olivier, F. et al. Shockley-Read-Hall and Auger non-radiative recombination in GaN based LEDs: a size effect study. Appl. Phys. Lett. 111 , 022104 (2017).

Konoplev, S. S., Bulashevich, K. A. & Karpov, S. Y. From large-size to micro-LEDs: scaling trends revealed by modeling. Phys. Status Solidi (A) 215 , 1700508 (2018).

Li, L. Z. et al. Transfer-printed, tandem microscale light-emitting diodes for full-color displays. Proc. Natl Acad. Sci. USA 118 , e2023436118 (2021).

Oh, J. T. et al. Light output performance of red AlGaInP-based light emitting diodes with different chip geometries and structures. Opt. Express 26 , 11194–11200 (2018).

Shen, Y. C. et al. Auger recombination in InGaN measured by photoluminescence. Appl. Phys. Lett. 91 , 141101 (2007).

Wong, M. S. et al. High efficiency of III-nitride micro-light-emitting diodes by sidewall passivation using atomic layer deposition. Opt. Express 26 , 21324–21331 (2018).

Han, S. C. et al. AlGaInP-based Micro-LED array with enhanced optoelectrical properties. Optical Mater. 114 , 110860 (2021).

Wong, M. S. et al. Size-independent peak efficiency of III-nitride micro-light-emitting-diodes using chemical treatment and sidewall passivation. Appl. Phys. Express 12 , 097004 (2019).

Ley, R. T. et al. Revealing the importance of light extraction efficiency in InGaN/GaN microLEDs via chemical treatment and dielectric passivation. Appl. Phys. Lett. 116 , 251104 (2020).

Moon, S. W. et al. Recent progress on ultrathin metalenses for flat optics. iScience 23 , 101877 (2020).

Arbabi, A. et al. Efficient dielectric metasurface collimating lenses for mid-infrared quantum cascade lasers. Opt. Express 23 , 33310–33317 (2015).

Yu, N. F. et al. Light propagation with phase discontinuities: generalized laws of reflection and refraction. Science 334 , 333–337 (2011).

Liang, H. W. et al. High performance metalenses: numerical aperture, aberrations, chromaticity, and trade-offs. Optica 6 , 1461–1470 (2019).

Park, J. S. et al. All-glass, large metalens at visible wavelength using deep-ultraviolet projection lithography. Nano Lett. 19 , 8673–8682 (2019).

Yoon, G. et al. Single-step manufacturing of hierarchical dielectric metalens in the visible. Nat. Commun. 11 , 2268 (2020).

Lee, G. Y. et al. Metasurface eyepiece for augmented reality. Nat. Commun. 9 , 4562 (2018).

Chen, W. T. et al. A broadband achromatic metalens for focusing and imaging in the visible. Nat. Nanotechnol. 13 , 220–226 (2018).

Wang, S. M. et al. A broadband achromatic metalens in the visible. Nat. Nanotechnol. 13 , 227–232 (2018).

Lan, S. F. et al. Metasurfaces for near-eye augmented reality. ACS Photonics 6 , 864–870 (2019).

Fan, Z. B. et al. A broadband achromatic metalens array for integral imaging in the visible. Light.: Sci. Appl. 8 , 67 (2019).

Shi, Z. J., Chen, W. T. & Capasso, F. Wide field-of-view waveguide displays enabled by polarization-dependent metagratings. In Proc. SPIE 10676, Digital Optics for Immersive Displays (SPIE, Strasbourg, France, 2018).

Hong, C. C., Colburn, S. & Majumdar, A. Flat metaform near-eye visor. Appl. Opt. 56 , 8822–8827 (2017).

Bayati, E. et al. Design of achromatic augmented reality visors based on composite metasurfaces. Appl. Opt. 60 , 844–850 (2021).

Nikolov, D. K. et al. Metaform optics: bridging nanophotonics and freeform optics. Sci. Adv. 7 , eabe5112 (2021).

Tamir, T. & Peng, S. T. Analysis and design of grating couplers. Appl. Phys. 14 , 235–254 (1977).

Miller, J. M. et al. Design and fabrication of binary slanted surface-relief gratings for a planar optical interconnection. Appl. Opt. 36 , 5717–5727 (1997).

Levola, T. & Laakkonen, P. Replicated slanted gratings with a high refractive index material for in and outcoupling of light. Opt. Express 15 , 2067–2074 (2007).

Shrestha, S. et al. Broadband achromatic dielectric metalenses. Light.: Sci. Appl. 7 , 85 (2018).

Li, Z. Y. et al. Meta-optics achieves RGB-achromatic focusing for virtual reality. Sci. Adv. 7 , eabe4458 (2021).

Ratcliff, J. et al. ThinVR: heterogeneous microlens arrays for compact, 180 degree FOV VR near-eye displays. IEEE Trans. Vis. Computer Graph. 26 , 1981–1990 (2020).

Wong, T. L. et al. Folded optics with birefringent reflective polarizers. In Proc. SPIE 10335, Digital Optical Technologies 2017 (SPIE, Munich, Germany, 2017).

Li, Y. N. Q. et al. Broadband cholesteric liquid crystal lens for chromatic aberration correction in catadioptric virtual reality optics. Opt. Express 29 , 6011–6020 (2021).

Bang, K. et al. Lenslet VR: thin, flat and wide-FOV virtual reality display using fresnel lens and lenslet array. IEEE Trans. Vis. Computer Graph. 27 , 2545–2554 (2021).

Maimone, A. & Wang, J. R. Holographic optics for thin and lightweight virtual reality. ACM Trans. Graph. 39 , 67 (2020).

Kramida, G. Resolving the vergence-accommodation conflict in head-mounted displays. IEEE Trans. Vis. Computer Graph. 22 , 1912–1931 (2016).

Zhan, T. et al. Multifocal displays: review and prospect. PhotoniX 1 , 10 (2020).

Shimobaba, T., Kakue, T. & Ito, T. Review of fast algorithms and hardware implementations on computer holography. IEEE Trans. Ind. Inform. 12 , 1611–1622 (2016).

Xiao, X. et al. Advances in three-dimensional integral imaging: sensing, display, and applications [Invited]. Appl. Opt. 52 , 546–560 (2013).

Kuiper, S. & Hendriks, B. H. W. Variable-focus liquid lens for miniature cameras. Appl. Phys. Lett. 85 , 1128–1130 (2004).

Liu, S. & Hua, H. Time-multiplexed dual-focal plane head-mounted display with a liquid lens. Opt. Lett. 34 , 1642–1644 (2009).

Wilson, A. & Hua, H. Design and demonstration of a vari-focal optical see-through head-mounted display using freeform Alvarez lenses. Opt. Express 27 , 15627–15637 (2019).

Zhan, T. et al. Pancharatnam-Berry optical elements for head-up and near-eye displays [Invited]. J. Optical Soc. Am. B 36 , D52–D65 (2019).

Oh, C. & Escuti, M. J. Achromatic diffraction from polarization gratings with high efficiency. Opt. Lett. 33 , 2287–2289 (2008).

Zou, J. Y. et al. Broadband wide-view Pancharatnam-Berry phase deflector. Opt. Express 28 , 4921–4927 (2020).

Zhan, T., Lee, Y. H. & Wu, S. T. High-resolution additive light field near-eye display by switchable Pancharatnam–Berry phase lenses. Opt. Express 26 , 4863–4872 (2018).

Tan, G. J. et al. Polarization-multiplexed multiplane display. Opt. Lett. 43 , 5651–5654 (2018).

Lanman, D. R. Display systems research at facebook reality labs (conference presentation). In Proc. SPIE 11310, Optical Architectures for Displays and Sensing in Augmented, Virtual, and Mixed Reality (AR, VR, MR) (SPIE, San Francisco, California, United States, 2020).

Liu, Z. J. et al. A novel BLU-free full-color LED projector using LED on silicon micro-displays. IEEE Photonics Technol. Lett. 25 , 2267–2270 (2013).

Han, H. V. et al. Resonant-enhanced full-color emission of quantum-dot-based micro LED display technology. Opt. Express 23 , 32504–32515 (2015).

Lin, H. Y. et al. Optical cross-talk reduction in a quantum-dot-based full-color micro-light-emitting-diode display by a lithographic-fabricated photoresist mold. Photonics Res. 5 , 411–416 (2017).

Liu, Z. J. et al. Micro-light-emitting diodes with quantum dots in display technology. Light.: Sci. Appl. 9 , 83 (2020).

Kim, H. M. et al. 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).

Xuan, T. T. et al. Inkjet-printed quantum dot color conversion films for high-resolution and full-color micro light-emitting diode displays. J. Phys. Chem. Lett. 11 , 5184–5191 (2020).

Chen, S. W. H. et al. Full-color monolithic hybrid quantum dot nanoring micro light-emitting diodes with improved efficiency using atomic layer deposition and nonradiative resonant energy transfer. Photonics Res. 7 , 416–422 (2019).

Krishnan, C. et al. Hybrid photonic crystal light-emitting diode renders 123% color conversion effective quantum yield. Optica 3 , 503–509 (2016).

Kang, J. H. et al. RGB arrays for micro-light-emitting diode applications using nanoporous GaN embedded with quantum dots. ACS Applied Mater. Interfaces 12 , 30890–30895 (2020).

Chen, G. S. et al. Monolithic red/green/blue micro-LEDs with HBR and DBR structures. IEEE Photonics Technol. Lett. 30 , 262–265 (2018).

Hsiang, E. L. et al. Enhancing the efficiency of color conversion micro-LED display with a patterned cholesteric liquid crystal polymer film. Nanomaterials 10 , 2430 (2020).

Kang, C. M. et al. Hybrid full-color inorganic light-emitting diodes integrated on a single wafer using selective area growth and adhesive bonding. ACS Photonics 5 , 4413–4422 (2018).

Geum, D. M. et al. Strategy toward the fabrication of ultrahigh-resolution micro-LED displays by bonding-interface-engineered vertical stacking and surface passivation. Nanoscale 11 , 23139–23148 (2019).

Ra, Y. H. et al. Full-color single nanowire pixels for projection displays. Nano Lett. 16 , 4608–4615 (2016).

Motoyama, Y. et al. High-efficiency OLED microdisplay with microlens array. J. Soc. Inf. Disp. 27 , 354–360 (2019).

Fujii, T. et al. 4032 ppi High-resolution OLED microdisplay. J. Soc. Inf. Disp. 26 , 178–186 (2018).

Hamer, J. et al. High-performance OLED microdisplays made with multi-stack OLED formulations on CMOS backplanes. In Proc. SPIE 11473, Organic and Hybrid Light Emitting Materials and Devices XXIV . Online Only (SPIE, 2020).

Joo, W. J. et al. Metasurface-driven OLED displays beyond 10,000 pixels per inch. Science 370 , 459–463 (2020).

Vettese, D. Liquid crystal on silicon. Nat. Photonics 4 , 752–754 (2010).

Zhang, Z. C., You, Z. & Chu, D. P. Fundamentals of phase-only liquid crystal on silicon (LCOS) devices. Light.: Sci. Appl. 3 , e213 (2014).

Hornbeck, L. J. The DMD TM projection display chip: a MEMS-based technology. MRS Bull. 26 , 325–327 (2001).

Zhang, Q. et al. Polarization recycling method for light-pipe-based optical engine. Appl. Opt. 52 , 8827–8833 (2013).

Hofmann, U., Janes, J. & Quenzer, H. J. High-Q MEMS resonators for laser beam scanning displays. Micromachines 3 , 509–528 (2012).

Holmström, S. T. S., Baran, U. & Urey, H. MEMS laser scanners: a review. J. Microelectromechanical Syst. 23 , 259–275 (2014).

Bao, X. Z. et al. Design and fabrication of AlGaInP-based micro-light-emitting-diode array devices. Opt. Laser Technol. 78 , 34–41 (2016).

Olivier, F. et al. Influence of size-reduction on the performances of GaN-based micro-LEDs for display application. J. Lumin. 191 , 112–116 (2017).

Liu, Y. B. et al. High-brightness InGaN/GaN Micro-LEDs with secondary peak effect for displays. IEEE Electron Device Lett. 41 , 1380–1383 (2020).

Qi, L. H. et al. 848 ppi high-brightness active-matrix micro-LED micro-display using GaN-on-Si epi-wafers towards mass production. Opt. Express 29 , 10580–10591 (2021).

Chen, E. G. & Yu, F. H. Design of an elliptic spot illumination system in LED-based color filter-liquid-crystal-on-silicon pico projectors for mobile embedded projection. Appl. Opt. 51 , 3162–3170 (2012).

Darmon, D., McNeil, J. R. & Handschy, M. A. 70.1: LED-illuminated pico projector architectures. Soc. Inf. Disp. Int. Symp . Dig. Tech. Pap. 39 , 1070–1073 (2008).

Essaian, S. & Khaydarov, J. State of the art of compact green lasers for mobile projectors. Optical Rev. 19 , 400–404 (2012).

Sun, W. S. et al. Compact LED projector design with high uniformity and efficiency. Appl. Opt. 53 , H227–H232 (2014).

Sun, W. S., Chiang, Y. C. & Tsuei, C. H. Optical design for the DLP pocket projector using LED light source. Phys. Procedia 19 , 301–307 (2011).

Chen, S. W. H. et al. High-bandwidth green semipolar (20–21) InGaN/GaN micro light-emitting diodes for visible light communication. ACS Photonics 7 , 2228–2235 (2020).

Yoshida, K. et al. 245 MHz bandwidth organic light-emitting diodes used in a gigabit optical wireless data link. Nat. Commun. 11 , 1171 (2020).

Park, D. W. et al. 53.5: High-speed AMOLED pixel circuit and driving scheme. Soc. Inf. Disp. Int. Symp . Dig. Tech. Pap. 41 , 806–809 (2010).

Tan, L., Huang, H. C. & Kwok, H. S. 78.1: Ultra compact polarization recycling system for white light LED based pico-projection system. Soc. Inf. Disp. Int. Symp. Dig. Tech. Pap. 41 , 1159–1161 (2010).

Maimone, A., Georgiou, A. & Kollin, J. S. Holographic near-eye displays for virtual and augmented reality. ACM Trans. Graph. 36 , 85 (2017).

Pan, J. W. et al. Portable digital micromirror device projector using a prism. Appl. Opt. 46 , 5097–5102 (2007).

Huang, Y. et al. Liquid-crystal-on-silicon for augmented reality displays. Appl. Sci. 8 , 2366 (2018).

Peng, F. L. et al. Analytical equation for the motion picture response time of display devices. J. Appl. Phys. 121 , 023108 (2017).

Pulli, K. 11-2: invited paper: meta 2: immersive optical-see-through augmented reality. Soc. Inf. Disp. Int. Symp . Dig. Tech. Pap. 48 , 132–133 (2017).

Lee, B. & Jo, Y. in Advanced Display Technology: Next Generation Self-Emitting Displays (eds Kang, B., Han, C. W. & Jeong, J. K.) 307–328 (Springer, 2021).

Cheng, D. W. et al. Design of an optical see-through head-mounted display with a low f -number and large field of view using a freeform prism. Appl. Opt. 48 , 2655–2668 (2009).

Zheng, Z. R. et al. Design and fabrication of an off-axis see-through head-mounted display with an x–y polynomial surface. Appl. Opt. 49 , 3661–3668 (2010).

Wei, L. D. et al. Design and fabrication of a compact off-axis see-through head-mounted display using a freeform surface. Opt. Express 26 , 8550–8565 (2018).

Liu, S., Hua, H. & Cheng, D. W. A novel prototype for an optical see-through head-mounted display with addressable focus cues. IEEE Trans. Vis. Computer Graph. 16 , 381–393 (2010).

Hua, H. & Javidi, B. A 3D integral imaging optical see-through head-mounted display. Opt. Express 22 , 13484–13491 (2014).

Song, W. T. et al. Design of a light-field near-eye display using random pinholes. Opt. Express 27 , 23763–23774 (2019).

Wang, X. & Hua, H. Depth-enhanced head-mounted light field displays based on integral imaging. Opt. Lett. 46 , 985–988 (2021).

Huang, H. K. & Hua, H. Generalized methods and strategies for modeling and optimizing the optics of 3D head-mounted light field displays. Opt. Express 27 , 25154–25171 (2019).

Huang, H. K. & Hua, H. High-performance integral-imaging-based light field augmented reality display using freeform optics. Opt. Express 26 , 17578–17590 (2018).

Cheng, D. W. et al. Design and manufacture AR head-mounted displays: a review and outlook. Light.: Adv. Manuf. 2 , 24 (2021).

Google Scholar  

Westheimer, G. The Maxwellian view. Vis. Res. 6 , 669–682 (1966).

Do, H., Kim, Y. M. & Min, S. W. Focus-free head-mounted display based on Maxwellian view using retroreflector film. Appl. Opt. 58 , 2882–2889 (2019).

Park, J. H. & Kim, S. B. Optical see-through holographic near-eye-display with eyebox steering and depth of field control. Opt. Express 26 , 27076–27088 (2018).

Chang, C. L. et al. Toward the next-generation VR/AR optics: a review of holographic near-eye displays from a human-centric perspective. Optica 7 , 1563–1578 (2020).

Hsueh, C. K. & Sawchuk, A. A. Computer-generated double-phase holograms. Appl. Opt. 17 , 3874–3883 (1978).

Chakravarthula, P. et al. Wirtinger holography for near-eye displays. ACM Trans. Graph. 38 , 213 (2019).

Peng, Y. F. et al. Neural holography with camera-in-the-loop training. ACM Trans. Graph. 39 , 185 (2020).

Shi, L. et al. Towards real-time photorealistic 3D holography with deep neural networks. Nature 591 , 234–239 (2021).

Jang, C. et al. Retinal 3D: augmented reality near-eye display via pupil-tracked light field projection on retina. ACM Trans. Graph. 36 , 190 (2017).

Jang, C. et al. Holographic near-eye display with expanded eye-box. ACM Trans. Graph. 37 , 195 (2018).

Kim, S. B. & Park, J. H. Optical see-through Maxwellian near-to-eye display with an enlarged eyebox. Opt. Lett. 43 , 767–770 (2018).

Shrestha, P. K. et al. Accommodation-free head mounted display with comfortable 3D perception and an enlarged eye-box. Research 2019 , 9273723 (2019).

Lin, T. G. et al. Maxwellian near-eye display with an expanded eyebox. Opt. Express 28 , 38616–38625 (2020).

Jo, Y. et al. Eye-box extended retinal projection type near-eye display with multiple independent viewpoints [Invited]. Appl. Opt. 60 , A268–A276 (2021).

Xiong, J. H. et al. Aberration-free pupil steerable Maxwellian display for augmented reality with cholesteric liquid crystal holographic lenses. Opt. Lett. 46 , 1760–1763 (2021).

Viirre, E. et al. Laser safety analysis of a retinal scanning display system. J. Laser Appl. 9 , 253–260 (1997).

Ratnam, K. et al. Retinal image quality in near-eye pupil-steered systems. Opt. Express 27 , 38289–38311 (2019).

Maimone, A. et al. Pinlight displays: wide field of view augmented reality eyeglasses using defocused point light sources. In Proc. ACM SIGGRAPH 2014 Emerging Technologies (ACM, Vancouver, Canada, 2014).

Jeong, J. et al. Holographically printed freeform mirror array for augmented reality near-eye display. IEEE Photonics Technol. Lett. 32 , 991–994 (2020).

Ha, J. & Kim, J. Augmented reality optics system with pin mirror. US Patent 10,989,922 (2021).

Park, S. G. Augmented and mixed reality optical see-through combiners based on plastic optics. Inf. Disp. 37 , 6–11 (2021).

Xiong, J. H. et al. Breaking the field-of-view limit in augmented reality with a scanning waveguide display. OSA Contin. 3 , 2730–2740 (2020).

Levola, T. 7.1: invited paper: novel diffractive optical components for near to eye displays. Soc. Inf. Disp. Int. Symp . Dig. Tech. Pap. 37 , 64–67 (2006).

Laakkonen, P. et al. High efficiency diffractive incouplers for light guides. In Proc. SPIE 6896, Integrated Optics: Devices, Materials, and Technologies XII . (SPIE, San Jose, California, United States, 2008).

Bai, B. F. et al. Optimization of nonbinary slanted surface-relief gratings as high-efficiency broadband couplers for light guides. Appl. Opt. 49 , 5454–5464 (2010).

Äyräs, P., Saarikko, P. & Levola, T. Exit pupil expander with a large field of view based on diffractive optics. J. Soc. Inf. Disp. 17 , 659–664 (2009).

Yoshida, T. et al. A plastic holographic waveguide combiner for light-weight and highly-transparent augmented reality glasses. J. Soc. Inf. Disp. 26 , 280–286 (2018).

Yu, C. et al. Highly efficient waveguide display with space-variant volume holographic gratings. Appl. Opt. 56 , 9390–9397 (2017).

Shi, X. L. et al. Design of a compact waveguide eyeglass with high efficiency by joining freeform surfaces and volume holographic gratings. J. Optical Soc. Am. A 38 , A19–A26 (2021).

Han, J. et al. Portable waveguide display system with a large field of view by integrating freeform elements and volume holograms. Opt. Express 23 , 3534–3549 (2015).

Weng, Y. S. et al. Liquid-crystal-based polarization volume grating applied for full-color waveguide displays. Opt. Lett. 43 , 5773–5776 (2018).

Lee, Y. H. et al. Compact see-through near-eye display with depth adaption. J. Soc. Inf. Disp. 26 , 64–70 (2018).

Tekolste, R. D. & Liu, V. K. Outcoupling grating for augmented reality system. US Patent 10,073,267 (2018).

Grey, D. & Talukdar, S. Exit pupil expanding diffractive optical waveguiding device. US Patent 10,073, 267 (2019).

Yoo, C. et al. Extended-viewing-angle waveguide near-eye display with a polarization-dependent steering combiner. Opt. Lett. 45 , 2870–2873 (2020).

Schowengerdt, B. T., Lin, D. & St. Hilaire, P. Multi-layer diffractive eyepiece with wavelength-selective reflector. US Patent 10,725,223 (2020).

Wang, Q. W. et al. Stray light and tolerance analysis of an ultrathin waveguide display. Appl. Opt. 54 , 8354–8362 (2015).

Wang, Q. W. et al. Design of an ultra-thin, wide-angle, stray-light-free near-eye display with a dual-layer geometrical waveguide. Opt. Express 28 , 35376–35394 (2020).

Frommer, A. Lumus: maximus: large FoV near to eye display for consumer AR glasses. In Proc. SPIE 11764, AVR21 Industry Talks II . Online Only (SPIE, 2021).

Ayres, M. R. et al. Skew mirrors, methods of use, and methods of manufacture. US Patent 10,180,520 (2019).

Utsugi, T. et al. Volume holographic waveguide using multiplex recording for head-mounted display. ITE Trans. Media Technol. Appl. 8 , 238–244 (2020).

Aieta, F. et al. Multiwavelength achromatic metasurfaces by dispersive phase compensation. Science 347 , 1342–1345 (2015).

Arbabi, E. et al. Controlling the sign of chromatic dispersion in diffractive optics with dielectric metasurfaces. Optica 4 , 625–632 (2017).

Download references

Acknowledgements

The authors are indebted to Goertek Electronics for the financial support and Guanjun Tan for helpful discussions.

Author information

Authors and affiliations.

College of Optics and Photonics, University of Central Florida, Orlando, FL, 32816, USA

Jianghao Xiong, En-Lin Hsiang, Ziqian He, Tao Zhan & Shin-Tson Wu

You can also search for this author in PubMed   Google Scholar

Contributions

J.X. conceived the idea and initiated the project. J.X. mainly wrote the manuscript and produced the figures. E.-L.H., Z.H., and T.Z. contributed to parts of the manuscript. S.W. supervised the project and edited the manuscript.

Corresponding author

Correspondence to Shin-Tson Wu .

Ethics declarations

Conflict of interest.

The authors declare no competing interests.

Supplementary information

Supplementary information, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Xiong, J., Hsiang, EL., He, Z. et al. Augmented reality and virtual reality displays: emerging technologies and future perspectives. Light Sci Appl 10 , 216 (2021). https://doi.org/10.1038/s41377-021-00658-8

Download citation

Received : 06 June 2021

Revised : 26 September 2021

Accepted : 04 October 2021

Published : 25 October 2021

DOI : https://doi.org/10.1038/s41377-021-00658-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Color liquid crystal grating based color holographic 3d display system with large viewing angle.

• Qiong-Hua Wang

Light: Science & Applications (2024)

Mass-produced and uniformly luminescent photochromic fibers toward future interactive wearable displays

Enhancing the color gamut of waveguide displays for augmented reality head-mounted displays through spatially modulated diffraction grating.

• Jae-Sang Lee
• Seong-Hyeon Cho
• Young-Wan Choi

Scientific Reports (2024)

Effects of virtual reality exposure therapy on state-trait anxiety in individuals with dentophobia

• Elham Majidi
• Gholamreza Manshaee

Current Psychology (2024)

A review of convolutional neural networks in computer vision

• Milan Parmar

Artificial Intelligence Review (2024)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

A perspective on embracing emerging technologies research for organizational behavior

Organization Management Journal

ISSN : 2753-8567

Article publication date: 24 October 2021

Issue publication date: 14 June 2022

Emerging technologies are capable of enhancing organizational- and individual-level outcomes. The organizational behavior (OB) field is beginning to pursue opportunities for researching emerging technologies. This study aims to describe a framework consisting of white, black and grey boxes to demonstrate the tight coupling of phenomena and paradigms in the field and discusses deconstructing OB’s white box to encourage data-driven phenomena to coexist in the spatial framework.

Design/methodology/approach

A scoping literature review was conducted to offer a preliminary assessment of technology-oriented research currently occurring in OB.

The literature search revealed two findings. First, the number of published papers on emerging technologies in top management journals has been increasing at a steady pace. Second, various theoretical perspectives at the micro- and macro- organizational level have been used so far for conducting technology-oriented research.

Originality/value

By conducting a scoping review of emerging technologies research in OB literature, this paper reveals a conceptual black box relating to technology-oriented research. The essay advocates for loosening OB’s tightly coupled white box to incorporate emerging technologies both as a phenomenon and as data analytical techniques.

• Emerging technologies
• Artificial intelligence
• Machine learning
• Organizational behavior

Philip, J. (2022), "A perspective on embracing emerging technologies research for organizational behavior", Organization Management Journal , Vol. 19 No. 3, pp. 88-98. https://doi.org/10.1108/OMJ-10-2020-1063

Emerald Publishing Limited

Copyright © 2021, Jestine Philip.

Published in Organization Management Journal . Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence maybe seen at http://creativecommons.org/licences/by/4.0/legalcode

Introduction

Emerging technologies such as artificial intelligence (AI), blockchain, virtual reality (VR), robotics, Internet of Things (IoT) and quantum computing powered by data analytics, machine learning (ML) algorithms and automations enhance organizational- and individual-level outcomes ( Alpaydin, 2020 ). Management literature appreciates the significant advancements in firm performance that emerging technologies can bring ( Eggers & Kaul, 2018 ) and encourages scholars to use them in research for theory development ( George, Osinga, Lavie, & Scott, 2016 ). Incorporating data science in management research enables scholars to develop better answers to old research questions by establishing causal relationships ( Tonidandel, King, & Cortina, 2015 ). Even with such and more thrilling opportunities for research using these technologies, organizational behavior (OB) is only just beginning to pursue this area.

Existing challenges in conducting OB research pertaining to emerging technologies may stem from limited training on data science and machines ( Barnes et al. , 2018 ) and the technicality involving coding and mathematical modeling. Moreover, current OB research programs are heavily theory-driven and deductive. A theory-focused approach is more influenced by the gaps and perplexities in a theory or phenomenon and less by the actual experiences of individuals ( Weick, 1992 ). With the vast amounts of data and enhanced algorithmic capabilities available today, an appropriate balance and integration of theory and data can help reveal the intricate complexities in employee behaviors. Weick (1992) recognized technology as a concept at the psychological-level that would benefit from a revisited research agenda in theory-heavy OB scholarship. In offering a discussion on this topic for OB, this essay poses the research question – How can the prevalent theory-driven mindset in OB be revised to conduct more technology-oriented research?

For the purposes of this essay, technology-oriented research in OB is defined as comprising of two aspects. The first involves researching various emerging technologies within the OB context such as in decision-making, identity, trust, bias, leadership and so on, while the second aspect entails incorporating advanced data analysis techniques. Subsequent to a review of pertinent literature, this essay describes a framework consisting of white, black and grey boxes and offers guidance to help condense the conceptual black box.

Literature review

As the goal of this literature review is to offer a preliminary understanding of technology-oriented research currently occurring in OB, a scoping review was conducted. Unlike structured or systematic reviews, which are done to produce interdisciplinary assessment of a topic or to answer a review-based research question and offer practical implications, a scoping review is conducted when researchers are interested in providing an overview of the evidence of a topic being investigated in a given field ( Munn et al. , 2018 ). Moreover, a scoping review was deemed suitable for the current paper as the use of emerging technologies is a relatively nascent area, just at the cusp of OB and human resource management (HRM) research.

Inclusion criteria

The search was performed on a select list of journals. To find papers on both aspects of technology-oriented research, journals in the OB field that publish theoretical as well as theory-anchored empirical papers were searched. To generate the list of journals, I first included the top 10 journals in I/O psychology and management sciences offered by Zickar and Highhouse’s (2001) impact analysis. Next, following existing practice of adding and/or replacing certain journals based on the topic area of the literature review (c.f. Anderson, De Dreu, & Nijstad, 2004 ), I arrived at the journals listed in Table 1 . With the date range for the search set from 2005 (year of big data emergence, Oracle, n.d.) to present, the abstracts of the listed journals were searched for the names of top emerging technologies and related terms ( CompTIA, 2020 ). No other search criteria were set and any paper that mentioned the technology terms in the abstract was included in the database. Upon searching all journals for each technology term, the compiled list was further refined by only retaining papers that discussed a specific OB, management, or organizational theory in the context of that technology. This resulted in a list of 88 papers, consisting of 78 research articles, 5 editorials and 5 book reviews.

Reported findings of the literature search

The scoping search revealed two crucial findings. First, the number of published papers that either conceptually discuss or apply the technology terms listed in Table 1 has been increasing at a steady pace. While averaging under 5 papers a year until 2018, a steep rise is seen from then on, going up to 16 in 2019, 31 in 2020 and 14 papers published by mid-2021. Second, various organizational theoretical perspectives at the micro- and macro-level ( Table 1 ) have been used so far to conduct technology-oriented research. Two instances are Newman, Fast, & Harmon’s (2020) article that extends procedural justice theory by discussing the role of algorithmic decision-making in HRM and Doornenbal, Spisak, & van der Laken’s (2021) application of ML models (like random forest) to conduct a predictive study for leader traits. These findings reveal that scholars are attempting to integrate behavioral topics and emerging technologies. The somewhat slower pace of progress of technology-oriented research occurring in OB, however, warrants further discussion.

The box framework

Figure 1 displays the framework consisting of white, black and grey boxes.

The white box

This essay credits current research programs in OB – where phenomena and paradigms are viewed and interpreted by offering theory to answer research questions – in the white box. Such a depiction and understanding of a white box is borrowed from computer science and software model testing as a space where there is sufficient conceptual knowledge ( Khan & Khan, 2012 ). The white box model has also been used by scientists in biology ( Lo-Thong, 2020 ), ocean research ( Leifsson, 2008 ) and cryptography ( Bock, 2019 ). In behavioral science, the white box was employed to depict the observer’s depth of knowledge regarding the internal functioning of a system ( Glanville, 1982 ). Consistent with these framings, the white box in OB research would consist of existing theoretical models and conceptual understanding of various management phenomena. The OB white box is tightly bound by theory, largely adheres to the positivist paradigm and consists of strong relationships and interdependencies between phenomena and paradigms ( Wardlow, 1989 ). As loosely coupled systems are advantageous in complex environments due to their weak interdependencies ( Orton & Weick, 1990 ), strong underpinnings of core phenomena and conceptualizations in OB lead to rigid mindsets that hinder newer approaches from being considered and explored. In Figure 1 , the solid line boundaries around the white box and circles depict the tight coupling of OB’s current research programs.

As new technological phenomena emerge in the workplace and affect employee attitudes and behaviors, a revision of the current theory driven agendas in behavioral research may be useful. Scholars advocate for renewed research programs to advance theory when studying technological innovations in firms ( Carter, 2020 ), like applying abductive reasoning in studying AI-based decision-making ( von Krogh, 2018 ). Contemporary phenomena affecting management including open innovation, AI, ML, IoT, VR, robotics and quantum computing ( Makadok, Burton, & Barney, 2018 ) stand to benefit from the proposed decoupling that facilitates the infusion of new paradigms. To advance technology-oriented research in OB, loosening of existing interdependencies - that fuel conventional phenomena-based theorizing - is justified. Scholars have previously encouraged conducting unconventional management research in new and varied contexts that focus on phenomena outside of management, thereby, setting fresh paradigms for the field ( Bamberger & Pratt, 2010 ). For example, Becker, Cropanzano, & Sanfey (2011) discuss organizational neuroscience as a new paradigm for exploration, providing directions to assimilate OB theories in the neural black box. As micro-organizational scholars continually advocate for addressing the research-practice gap in behavioral management ( Fisher, 1989 ; Tenhiälä et al. , 2016 ) and for the integration of micro- and macro-level theories, paradigms and methodologies for scientific advancement ( Aguinis, Boyd, Pierce, & Short, 2011 ), it is an opportune time for OB scholarship to embed AI and related technology research within its white box.

The black box

When prominent scholars promote non-traditional research as a means to advance the field, it signals an implied recognition that a black box exists within the white box, and that that black box would potentially expand if the principles of the field’s prevailing paradigms were not relaxed to encourage fresh perspectives. Applying the same understanding of the white box from computer science, the black box[ 1 ] in OB would comprise of a space with low granularity in which there is limited to no subject knowledge. In ML, the black box relates to highly complex algorithms that generate through repeated data feeds and learning cycles, which eventually become too complicated for the human brain to comprehend. After numerous data loops, the human coder is only capable of reading the input and output codes – what happens inside the machine is a mystery, hence the need to decode the black box ( Castelvecchi, 2016 ). By the same token, this essay posits that OB’s black box is implicit, unapparent, and not readily visible to OB scholars until they engage in some form of research that requires exploration of new approaches, methods, and techniques. In other words, so long as OB scholars continue to study traditional micro-organizational phenomena using fundamental paradigms (i.e. engage in tightly coupled research), they would not likely realize that a black box exists or be forced to understand what is inside it. That said, with companies investing in AI and other technologies to improve business processes and performance, corresponding discussions of these technologies should occur in scholarly investigations to better understand organizational and individual outcomes. To begin decoupling the white box – and subsequently condensing the black box – a flexibility and appreciation for studying phenomena surrounding AI and other technologies through varied paradigms is critical.

In Figure 1 , the black box is shown as encompassing some portion of OB’s white box, thereby illustrating the suffusing of emerging phenomena like AI in the field. The placement of the black box inside of the white box is consistent with Glanville’s (1982) prognostic integration of cybernetics and psychological concepts. To emphasize the decoupling of conventional phenomena and paradigms in OB, Figure 1 displays the white box boundary and circles in dashed lines rather than solid lines.

The grey box

Alongside attempting decoupling efforts for the white box, a parallel endeavor should be undertaken - when necessary - towards adopting a digitally driven mindset and encouraging diverse data analytical techniques in OB research. When illustrated pictorially, we find from Figure 1 , that the size of the black box could be reduced by extending the two white bubbles towards the grey box. While OB research engages in paradigm continuity, incorporating contemporary phenomena would facilitate “paradigm extension”. One way to do so is by using advanced research methodology paradigms ( Qiu, Donaldson, & Luo, 2012 ) to progress conventional organizational theory ( Donaldson, 2010 ). Advances in data science have enabled data-driven phenomena and paradigms to become increasingly normalized in today’s digital age. Thus, this essay regards advanced data-driven approaches as constituting the paradigm of research methodology. As displayed in Figure 1 , extending the dashed circles of data-driven phenomenon and paradigms (grey box) towards the white box will overlap OB’s black box, and consequently shrink the black box.

To elaborate further, the grey box in this context comprises of approaches in data science such as data-driven modelling (DDM), ML algorithms (like neural networks and deep learning) and other nuanced data-driven methods. Because DDM represents advances achieved through AI, ML and data mining, where relationships between the input and output variables can be drawn without detailed knowledge regarding the system’s behavior ( Solomatine, See, & Abrahart, 2009 ), applying such methods in congruence with established conceptual knowledge produces refined understanding of constructs and relationships in a field. As examples, big data can capture patterns in a construct in real-time and social media data - like those from Twitter – can reveal the existence of socially sensitive behaviors like biases and discrimination in workplaces. Such theory extension, propelled by data, can occur through multiple nested-levels of analyses ( Barnes et al. , 2018 ). In taking these initiatives, OB research would be aligning itself with and drawing a page from advances in its sister field of psychology, where scholars are engaged in decoupling their white box. Jack, Crivelli, & Wheatley (2018) , for example, advocate for using data-driven methods alongside relaxing Darwin’s theoretical constraints when studying facial expressions associated with various emotions. Advanced methodologies have shown to improve researchers’ understanding of how people in different cultures use facial expressions during verbal communication.

A preliminary course of action for methodology paradigm extension could involve narrowing the research question ( Makadok et al. , 2018 ) toward quantitative precision. With much attention being presently given to AI’s ability to enhance managerial decision-making, a sample open-ended question could be, “How can AI blend with human reasoning to advance managerial decision-making?” The empirical approaches employed to answer such a research question should appropriate the technology being studied. This is emphasized through Figure 1 , which illustrates that in order to shrink the black box, we must include some portion of the grey box. Hence, using decision-making theory would encompass the white box and applying ML algorithms as the methodology to answer the research question in real-time in a variety of managerial settings suggests inclusion of the grey box. Furthermore, we might even propose a close-ended question like, “ Can AI blend with human reasoning to advance managerial decision-making?” This could be sufficiently answered through ML-based decision tree and clustering algorithms. Machine learning techniques can help reveal under what specific situations which manager(s) in the organization are able to effectively integrate their own reasoning abilities with algorithmic recommendations offered to them. Such research questions offer even personality researchers an avenue for exploration. These attempts represent theory extension by identifying highly specific boundary conditions for human decision-making theory in the age of machines. If the ML algorithm provides a positive recommendation, a follow-up research question, “ To what degree can AI advance managerial decision-making?” could be asked. In doing so, we are not only framing a theory-driven research question in a distinct and specific context, but also utilizing real-time data to quantifiably answer it, thereby extending OB’s white box and using the data-driven grey box to condense the black box.

The “Triple A” potency

With emerging technologies reshaping management practices and changing the nature of work at all hierarchies in the organization ( Manyika, Chui, Madgavkar, & Lund, 2017 ), this essay recognizes that the influential combined driving force of analytics, algorithms and automation serves as a catalyst to initiate a decoupling of OB’s white box.

Moving forward with a boundary condition

In acknowledging that a black box exists, we can begin to consider how we move forward to advance the field. The simplest visualization of condensing the black box involves two steps – first, separating the white box circles, and second, expanding the areas of these circles to approach and converge with the grey box. As a guide to conducting technology-oriented research in OB, an initial item on the black box reduction checklist could involve identifying which OB theories have strong potential for decoupling and expansion within the white box and then, subsequently applying varied approaches to expand those theories to fit within the emerging technologies context. Nelson (2020) combines interpretive and inductive approaches through ML techniques to suggest a three-step process to generate computational grounded theory. McAbee, Landis, & Burke (2017) recommend integrating interpretivist and positivist approaches in grounded theory and the latest computational techniques to break away from epistemological conventions like deductivism. We should begin revising some of our existing deductive research programs and start by first observing and asking research questions from practice ( Mathieu, 2016 ). Researchers in the hard sciences and humanities assert that big data and emerging technologies are leading us to renew our epistemology and make paradigm shifts ( Kitchin, 2014 ). Similarly, it is timely for OB scholars to encourage a mindset that creates opportunities for data-driven phenomena and paradigms (the grey box) to coexist alongside OB’s white box. In doing so, we augment both phenomenon-based DDM’s and data-driven theorizing.

The black box in ML is being tackled by scientists using “explainable AI” ( Zednik, 2019 ), where experts develop rules to make the opaque box of unknown codes as transparent as humanly comprehendible. A similar approach could be taken in OB as a way to condense our black box, where we generate a set of guiding principles for researching emerging technologies. Categorizing projects based on the interrogative word of the research question, such as the what , why , how , how much, and when of open- and close-ended questions ( Rajagopalan, 2020 ) and offering researchers a layout of OB phenomena and paradigms that complement data-driven approaches (i.e. specifying what parts of the white box combine with what parts of the grey box) might be helpful. AI and related technologies can aid research methods by offering nuance and improving predictability. Using ML in data analysis in behavioral studies would help build better predictive models and enable scholars to conduct more predictive style research alongside engaging in conventional hypothesis-driven research ( Doornenbal et al. , 2021 ). AI would also allow us to answer research questions quantitively with high precision (e.g.: by how much, to what degree).

In recommending so, this essay identifies a boundary condition in the methodological application of AI technologies in OB. As ML techniques are capable of enriching our understanding of intricate variable relationships, DDM and conventional methods should thrive simultaneously for methodological rigor in OB. Methodological monism in ML results in the loss of dialogue among paradigms ( Lindebaum & Ashraf, 2021 ). Hence, in keeping with the perspectives of positivism and interpretivism and in adopting methodological pluralism for OB research programs, this essay argues that while conventional quantitative and qualitative methods continue to be utilized to explain and understand behaviors ( Buchanan, 1998 ; Bryman & Bell, 2003 ), applying ML-based algorithms is recommended when the goals are exploration of patterns, investigation of complex relationships among variables, and precision (in research models and in answers to research questions). Using AI in research methodologies adheres to abductive inquiry, wherein unexpected observations made by ML add value for inductive theorizing and subsequent hypothesis generation ( Doornenbal et al. , 2021 ). In acknowledging concerns relating to the lack of transparency in decision-making in ML ( Lindebaum & Ashraf, 2021 ), the conditional applications of grey box techniques would enrich and complement existing core methodologies rather than competing with them ( Leavitt, Schabram, Hariharan, & Barnes, 2020 ).

In conclusion, this essay revealed a conceptual black box relating to technology-oriented research in OB by conducting a scoping review. From the findings of the review, opportunities for OB scholarship to engage in technology-oriented research are discussed. In answering the research question, the essay advocates for loosening OB’s tightly coupled white box so that emerging technologies could be incorporated both as a phenomenon and as data analytical techniques.

The box framework: white box, black box and grey box

Literature search: Management journals, key terms and findings

Beyond Zickar and Highhouse’s (2001) journal list, 4 Academy of Management journals (Annals, AMD, AMLE, AMP), 2 high impact HRM journals (HRMJ, HRMR), and 5 OB-relevant journals with an impact factor of at least 3.00 (JBP, JBE, JBR, JMS, LQ) were included. JMP was included as it had a special issue on technology in 2020

The usage of the term black box in this essay is intended to reflect the common understanding of the black box as it is used in computer programming, data science, neuroscience, and more recently in management research ( Castelvecchi, 2016 ; Becker, Cropanzano, & Sanfey, 2011 ; Doornenbal, Spisak, & van der Laken, 2021 ). While consideration was given to developing alternate terminology to data “boxes,” the goal of this essay was not intended to rename, but to draw attention to these technologies. Therefore, this essay adheres to the widely recognized terms. 

Aguinis , H. , Boyd , B.K. , Pierce , C.A. & Short , J.C. ( 2011 ). “ Walking new avenues in management research methods and theories: bridging micro and macro domains ”. Journal of Management , 37 ( 2 ), 395 - 403 . doi: 10.1177/0149206310382456 .

Alpaydin , E. ( 2020 ). Introduction to Machine Learning , MIT press .

Anderson , N. , De Dreu , C.K. & Nijstad , B.A. ( 2004 ). “ The routinization of innovation research: a constructively critical review of the state‐of‐the‐science ”. Journal of Organizational Behavior , 25 ( 2 ), 147 - 173 . doi: 10.1002/job.236 .

Bamberger , P.A. & Pratt , M.G. ( 2010 ). “ Moving forward by looking back: Reclaiming unconventional research contexts and samples in organizational scholarship ”. Academy of Management Journal , 53 ( 4 ), 665 - 671 . doi: 10.5465/amj.2010.52814357 .

Barnes , C.M. , Dang , C.T. , Leavitt , K. , Guarana , C.L. & Uhlmann , E.L. ( 2018 ). “ Archival data in micro-organizational research: a toolkit for moving to a broader set of topics ”. Journal of Management , 44 ( 4 ), 1453 - 1478 . doi: 10.1177/0149206315604188 .

Becker , W.J. , Cropanzano , R. & Sanfey , A.G. ( 2011 ). “ Organizational neuroscience: Taking organizational theory inside the neural black box ”. Journal of Management , 37 ( 4 ), 933 - 961 . doi: 10.1177/0149206311398955 .

Bock , E.A. , Bos , J.W. , Brzuska , C. , Hubain , C. , Michiels , W. , Mune , C. & Treff , A. ( 2019 ). “ White-box cryptography: don’t forget about grey-box attacks ”. Journal of Cryptology , 32 ( 4 ), 1095 - 1143 . doi: 10.1007/s00145-019-09315-1 .

Bryman , A. & Bell , E. ( 2003 ). Business research methods , Oxford : Oxford University Press .

Buchanan , D.R. ( 1998 ). “ Beyond positivism: humanistic perspectives on theory and research in health education ”. Health Education Research , 13 ( 3 ), 439 - 450 . doi: 10.1093/her/13.3.439 .

Carter , W. ( 2020 ). “ Putting choice in the spotlight to advance theory on organizational adaptation to technological discontinuities ”. Organization Management Journal , 17 ( 2 ) doi: 10.1108/OMJ-04-2019-0720 .

Castelvecchi , D. ( 2016 ). “ Can we open the black box of AI? ” Nature News , 538 ( 7623 ), 20 doi: 10.1038/538020a .

CompTIA. ( 2020 ). “ 2020 Emerging technology top 10 list ”. Retrieved from; https://connect.comptia.org/content/infographic/2020-emerging-technology-top-10-list

Donaldson , L. ( 2010 ). The Meta-analytic organization: Introducing Statistico-Organizational Theory , Armonk, New York, NY : M.E. Sharpe .

Doornenbal , B.M. , Spisak , B. R. & van der Laken , P.A. ( 2021 ). “ Opening the black box: uncovering the leader trait paradigm through machine learning ”. The Leadership Quarterly , doi: 10.1016/j.leaqua.2021.101515 .

Eggers , J.P. & Kaul , A. ( 2018 ). “ Motivation and ability? A behavioral perspective on the pursuit of radical invention in multi-technology incumbents ”. Academy of Management Journal , 61 ( 1 ), 67 - 93 . doi: 10.5465/amj.2015.1123 .

Fisher , C.D. ( 1989 ). “ Current and recurrent challenges in HRM ”. Journal of Management , 15 ( 2 ), 157 - 180 . doi: 10.1177/014920638901500203 .

George , G. , Osinga , E.C. , Lavie , D. & Scott , B.A. ( 2016 ). “ Big data and data science methods for management research ”. Academy of Management Journal , 59 ( 5 ), 1493 - 1507 . doi: 10.5465/amj.2016.4005 .

Glanville , R. ( 1982 ). “ Inside every white box there are two black boxes trying to get out ”. Behavioral Science , 27 ( 1 ), 1 - 11 . doi: 10.1002/bs.3830270102 .

Jack , R.E. , Crivelli , C. & Wheatley , T. ( 2018 ). “ Data-driven methods to diversify knowledge of human psychology ”. Trends in Cognitive Sciences , 22 ( 1 ), 1 - 5 . doi: 10.1016/j.tics.2017.10.002 .

Khan , M.E. & Khan , F. ( 2012 ). “ A comparative study of white box, black box and grey box testing techniques ”. International Journal of Advanced Computer Science and Applications , 3 ( 6 ) doi: 10.14569/IJACSA.2012.030603 .

Kitchin , R. ( 2014 ). “ Big data, new epistemologies and paradigm shifts ”. Big Data & Society , 1 ( 1 ), 2053951714528481 , doi: 10.1177/2053951714528481 .

Leavitt , K. , Schabram , K. , Hariharan , P. & Barnes , C.M. ( 2020 ). “ Ghost in the machine: on organizational theory in the age of machine learning ”. Academy of Management Review , doi: 10.5465/amr.2019.0247 .

Leifsson , L.Þ. , Sævarsdóttir , H. , Sigurðsson , S.Þ. & Vésteinsson , A. ( 2008 ). “ Grey-box modeling of an ocean vessel for operational optimization ”. Simulation Modelling Practice and Theory , 16 ( 8 ), 923 - 932 . doi: 10.1016/j.simpat.2008.03.006 .

Lindebaum , D. & Ashraf , M. ( 2021 ). “ The ghost in the machine, or the ghost in organizational theory? A complementary view on the use of machine learning ”. Academy of Management Review , doi: 10.5465/amr.2021.0036

Lo-Thong , O. , Charton , P. , Cadet , X.F. , Grondin-Perez , B. , Saavedra , E. , Damour , C. & Cadet , F. ( 2020 ). “ Identification of flux checkpoints in a metabolic pathway through white-box, grey-box and black-box modeling approaches ”. Scientific Reports , 10 ( 1 ), 1 - 19 . doi: 10.1038/s41598-020-70295-5 .

Makadok , R. , Burton , R. & Barney , J. ( 2018 ). “ A practical guide for making theory contributions in strategic management ”. Strategic Management Journal , 39 ( 6 ), 1530 - 1545 . doi: 10.1002/smj.2789 .

Manyika , J. , Chui , M. , Madgavkar , A. & Lund , S. ( 2017 ). What’s now and next in analytics, AI, and automation , McKiinsey Global Institute . Retrived from: www.mckinsey.com/featured-insights/digital-disruption/whats-now-and-next-in-analytics-ai-and-automation#

Mathieu , J.E. ( 2016 ). “ The problem with [in] management theory ”. Journal of Organizational Behavior , 37 ( 8 ), 1132 - 1141 . doi: 10.1002/job.2114 .

McAbee , S.T. , Landis , R.S. & Burke , M.I. ( 2017 ). “ Inductive reasoning: the promise of big data ”. Human Resource Management Review , 27 ( 2 ), 277 - 290 . doi: 10.1016/j.hrmr.2016.08.005 .

Munn , Z. , Peters , M.D. , Stern , C. , Tufanaru , C. , McArthur , A. & Aromataris , E. ( 2018 ). “ Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach ”. BMC Medical Research Methodology , 18 ( 1 ), 1 - 7 . doi: 10.1186/s12874-018-0611-x .

Nelson , L.K. ( 2020 ). “ Computational grounded theory: a methodological framework ”. Sociological Methods & Research , 49 ( 1 ), 3 - 42 . doi: 10.1177/0049124117729703 .

Newman , D.T. , Fast , N.J. & Harmon , D.J. ( 2020 ). “ When eliminating bias isn’t fair: Algorithmic reductionism and procedural justice in human resource decisions ”. Organizational Behavior and Human Decision Processes , 160 , 149 - 167 . doi: 10.1016/j.obhdp.2020.03.008 .

Oracle, (n.d.) , “ What is big data? ”, Retrieved from: www.oracle.com/big-data/what-is-big-data.html

Orton , J.D. & Weick , K.E. ( 1990 ). “ Loosely coupled systems: a reconceptualization ”. Academy of Management Review , 15 ( 2 ), 203 - 223 . doi: 10.5465/amr.1990.4308154 .

Qiu , J. , Donaldson , L. & Luo , B.N. ( 2012 ). “ The benefits of persisting with paradigms in organizational research ”. Academy of Management Perspectives , 26 ( 1 ) doi: 10.5465/amp.2011.0125 .

Rajagopalan , N. ( 2020 ). “ Rigor, relevance, and resilience in management research ”. Journal of Management Inquiry , 29 ( 2 ), 150 - 153 . doi: 10.1177/1056492619861690 .

Solomatine , D. , See , L.M. & Abrahart , R.J. ( 2009 ). “ Data-driven modelling: concepts, approaches and experiences ”. Practical HydroinFormatics , pp. 17 - 30 . Berlin, Heidelberg : Springer . In

Tenhiälä , A. , Giluk , T.L. , Kepes , S. , Simon , C. , Oh , I.S. & Kim , S. ( 2016 ). “ The research‐practice gap in human resource management: a cross‐cultural study ”. Human Resource Management , 55 ( 2 ), 179 - 200 . doi: 10.1002/hrm.21656 .

Tonidandel , S. King , E.B. & Cortina , J.M. (Eds), ( 2015 ). Big Data at Work: the Data Science Revolution and Organizational Psychology , Routledge .

Von Krogh , G. ( 2018 ). “ Artificial intelligence in organizations: New opportunities for phenomenon-based theorizing ”. Academy of Management Discoveries , 4 ( 4 ), doi: 10.5465/amd.2018.0084 .

Wardlow , G. ( 1989 ). “ Alternative modes of inquiry for agricultural education ”. Journal of Agricultural Education , 30 ( 4 ), 2 - 7 . doi: 10.5032/jae.1989.04002 .

Weick , K.E. ( 1992 ). “ Agenda setting in organizational behavior: a theory-focused approach ”. Journal of Management Inquiry , 1 ( 3 ), 171 - 182 . doi: 10.1177/105649269213001 .

Zednik , C. ( 2019 ). “ Solving the black box problem: a normative framework for explainable artificial intelligence ”. Philosophy Technology , 1 - 24 .

Zickar , M.J. & Highhouse , S. ( 2001 ). “ Measuring prestige of journals in industrial-organizational psychology ”. The Industrial-Organizational Psychologist , 38 ( 4 ), 29 - 36 .

Corresponding author

Related articles, we’re listening — tell us what you think, something didn’t work….

Report bugs here

All feedback is valuable

Please share your general feedback

Join us on our journey

Platform update page.

Visit emeraldpublishing.com/platformupdate to discover the latest news and updates

Questions & More Information

Answers to the most commonly asked questions here

• Research article
• Open access
• Published: 15 February 2018

Blended learning: the new normal and emerging technologies

• Charles Dziuban 1 ,
• Charles R. Graham 2 ,
• Patsy D. Moskal   ORCID: orcid.org/0000-0001-6376-839X 1 ,
• Anders Norberg 3 &
• Nicole Sicilia 1  

International Journal of Educational Technology in Higher Education volume  15 , Article number:  3 ( 2018 ) Cite this article

551k Accesses

355 Citations

118 Altmetric

Metrics details

This study addressed several outcomes, implications, and possible future directions for blended learning (BL) in higher education in a world where information communication technologies (ICTs) increasingly communicate with each other. In considering effectiveness, the authors contend that BL coalesces around access, success, and students’ perception of their learning environments. Success and withdrawal rates for face-to-face and online courses are compared to those for BL as they interact with minority status. Investigation of student perception about course excellence revealed the existence of robust if-then decision rules for determining how students evaluate their educational experiences. Those rules were independent of course modality, perceived content relevance, and expected grade. The authors conclude that although blended learning preceded modern instructional technologies, its evolution will be inextricably bound to contemporary information communication technologies that are approximating some aspects of human thought processes.

Introduction

Blended learning and research issues.

Blended learning (BL), or the integration of face-to-face and online instruction (Graham 2013 ), is widely adopted across higher education with some scholars referring to it as the “new traditional model” (Ross and Gage 2006 , p. 167) or the “new normal” in course delivery (Norberg et al. 2011 , p. 207). However, tracking the accurate extent of its growth has been challenging because of definitional ambiguity (Oliver and Trigwell 2005 ), combined with institutions’ inability to track an innovative practice, that in many instances has emerged organically. One early nationwide study sponsored by the Sloan Consortium (now the Online Learning Consortium) found that 65.2% of participating institutions of higher education (IHEs) offered blended (also termed hybrid ) courses (Allen and Seaman 2003 ). A 2008 study, commissioned by the U.S. Department of Education to explore distance education in the U.S., defined BL as “a combination of online and in-class instruction with reduced in-class seat time for students ” (Lewis and Parsad 2008 , p. 1, emphasis added). Using this definition, the study found that 35% of higher education institutions offered blended courses, and that 12% of the 12.2 million documented distance education enrollments were in blended courses.

The 2017 New Media Consortium Horizon Report found that blended learning designs were one of the short term forces driving technology adoption in higher education in the next 1–2 years (Adams Becker et al. 2017 ). Also, blended learning is one of the key issues in teaching and learning in the EDUCAUSE Learning Initiative’s 2017 annual survey of higher education (EDUCAUSE 2017 ). As institutions begin to examine BL instruction, there is a growing research interest in exploring the implications for both faculty and students. This modality is creating a community of practice built on a singular and pervasive research question, “How is blended learning impacting the teaching and learning environment?” That question continues to gain traction as investigators study the complexities of how BL interacts with cognitive, affective, and behavioral components of student behavior, and examine its transformation potential for the academy. Those issues are so compelling that several volumes have been dedicated to assembling the research on how blended learning can be better understood (Dziuban et al. 2016 ; Picciano et al. 2014 ; Picciano and Dziuban 2007 ; Bonk and Graham 2007 ; Kitchenham 2011 ; Jean-François 2013 ; Garrison and Vaughan 2013 ) and at least one organization, the Online Learning Consortium, sponsored an annual conference solely dedicated to blended learning at all levels of education and training (2004–2015). These initiatives address blended learning in a wide variety of situations. For instance, the contexts range over K-12 education, industrial and military training, conceptual frameworks, transformational potential, authentic assessment, and new research models. Further, many of these resources address students’ access, success, withdrawal, and perception of the degree to which blended learning provides an effective learning environment.

Currently the United States faces a widening educational gap between our underserved student population and those communities with greater financial and technological resources (Williams 2016 ). Equal access to education is a critical need, one that is particularly important for those in our underserved communities. Can blended learning help increase access thereby alleviating some of the issues faced by our lower income students while resulting in improved educational equality? Although most indicators suggest “yes” (Dziuban et al. 2004 ), it seems that, at the moment, the answer is still “to be determined.” Quality education presents a challenge, evidenced by many definitions of what constitutes its fundamental components (Pirsig 1974 ; Arum et al. 2016 ). Although progress has been made by initiatives, such as, Quality Matters ( 2016 ), the OLC OSCQR Course Design Review Scorecard developed by Open SUNY (Open SUNY n.d. ), the Quality Scorecard for Blended Learning Programs (Online Learning Consortium n.d. ), and SERVQUAL (Alhabeeb 2015 ), the issue is by no means resolved. Generally, we still make quality education a perceptual phenomenon where we ascribe that attribute to a course, educational program, or idea, but struggle with precisely why we reached that decision. Searle ( 2015 ), summarizes the problem concisely arguing that quality does not exist independently, but is entirely observer dependent. Pirsig ( 1974 ) in his iconic volume on the nature of quality frames the context this way,

“There is such thing as Quality, but that as soon as you try to define it, something goes haywire. You can’t do it” (p. 91).

Therefore, attempting to formulate a semantic definition of quality education with syntax-based metrics results in what O’Neil (O'Neil 2017 ) terms surrogate models that are rough approximations and oversimplified. Further, the derived metrics tend to morph into goals or benchmarks, losing their original measurement properties (Goodhart 1975 ).

Information communication technologies in society and education

Blended learning forces us to consider the characteristics of digital technology, in general, and information communication technologies (ICTs), more specifically. Floridi ( 2014 ) suggests an answer proffered by Alan Turing: that digital ICTs can process information on their own, in some sense just as humans and other biological life. ICTs can also communicate information to each other, without human intervention, but as linked processes designed by humans. We have evolved to the point where humans are not always “in the loop” of technology, but should be “on the loop” (Floridi 2014 , p. 30), designing and adapting the process. We perceive our world more and more in informational terms, and not primarily as physical entities (Floridi 2008 ). Increasingly, the educational world is dominated by information and our economies rest primarily on that asset. So our world is also blended, and it is blended so much that we hardly see the individual components of the blend any longer. Floridi ( 2014 ) argues that the world has become an “infosphere” (like biosphere) where we live as “inforgs.” What is real for us is shifting from the physical and unchangeable to those things with which we can interact.

Floridi also helps us to identify the next blend in education, involving ICTs, or specialized artificial intelligence (Floridi 2014 , 25; Norberg 2017 , 65). Learning analytics, adaptive learning, calibrated peer review, and automated essay scoring (Balfour 2013 ) are advanced processes that, provided they are good interfaces, can work well with the teacher— allowing him or her to concentrate on human attributes such as being caring, creative, and engaging in problem-solving. This can, of course, as with all technical advancements, be used to save resources and augment the role of the teacher. For instance, if artificial intelligence can be used to work along with teachers, allowing them more time for personal feedback and mentoring with students, then, we will have made a transformational breakthrough. The Edinburg University manifesto for teaching online says bravely, “Automation need not impoverish education – we welcome our robot colleagues” (Bayne et al. 2016 ). If used wisely, they will teach us more about ourselves, and about what is truly human in education. This emerging blend will also affect curricular and policy questions, such as the what? and what for? The new normal for education will be in perpetual flux. Floridi’s ( 2014 ) philosophy offers us tools to understand and be in control and not just sit by and watch what happens. In many respects, he has addressed the new normal for blended learning.

Literature of blended learning

A number of investigators have assembled a comprehensive agenda of transformative and innovative research issues for blended learning that have the potential to enhance effectiveness (Garrison and Kanuka 2004 ; Picciano 2009 ). Generally, research has found that BL results in improvement in student success and satisfaction, (Dziuban and Moskal 2011 ; Dziuban et al. 2011 ; Means et al. 2013 ) as well as an improvement in students’ sense of community (Rovai and Jordan 2004 ) when compared with face-to-face courses. Those who have been most successful at blended learning initiatives stress the importance of institutional support for course redesign and planning (Moskal et al. 2013 ; Dringus and Seagull 2015 ; Picciano 2009 ; Tynan et al. 2015 ). The evolving research questions found in the literature are long and demanding, with varied definitions of what constitutes “blended learning,” facilitating the need for continued and in-depth research on instructional models and support needed to maximize achievement and success (Dringus and Seagull 2015 ; Bloemer and Swan 2015 ).

Educational access

The lack of access to educational technologies and innovations (sometimes termed the digital divide) continues to be a challenge with novel educational technologies (Fairlie 2004 ; Jones et al. 2009 ). One of the promises of online technologies is that they can increase access to nontraditional and underserved students by bringing a host of educational resources and experiences to those who may have limited access to on-campus-only higher education. A 2010 U.S. report shows that students with low socioeconomic status are less likely to obtain higher levels of postsecondary education (Aud et al. 2010 ). However, the increasing availability of distance education has provided educational opportunities to millions (Lewis and Parsad 2008 ; Allen et al. 2016 ). Additionally, an emphasis on open educational resources (OER) in recent years has resulted in significant cost reductions without diminishing student performance outcomes (Robinson et al. 2014 ; Fischer et al. 2015 ; Hilton et al. 2016 ).

Unfortunately, the benefits of access may not be experienced evenly across demographic groups. A 2015 study found that Hispanic and Black STEM majors were significantly less likely to take online courses even when controlling for academic preparation, socioeconomic status (SES), citizenship, and English as a second language (ESL) status (Wladis et al. 2015 ). Also, questions have been raised about whether the additional access afforded by online technologies has actually resulted in improved outcomes for underserved populations. A distance education report in California found that all ethnic minorities (except Asian/Pacific Islanders) completed distance education courses at a lower rate than the ethnic majority (California Community Colleges Chancellor’s Office 2013 ). Shea and Bidjerano ( 2014 , 2016 ) found that African American community college students who took distance education courses completed degrees at significantly lower rates than those who did not take distance education courses. On the other hand, a study of success factors in K-12 online learning found that for ethnic minorities, only 1 out of 15 courses had significant gaps in student test scores (Liu and Cavanaugh 2011 ). More research needs to be conducted, examining access and success rates for different populations, when it comes to learning in different modalities, including fully online and blended learning environments.

Framing a treatment effect

Over the last decade, there have been at least five meta-analyses that have addressed the impact of blended learning environments and its relationship to learning effectiveness (Zhao et al. 2005 ; Sitzmann et al. 2006 ; Bernard et al. 2009 ; Means et al. 2010 , 2013 ; Bernard et al. 2014 ). Each of these studies has found small to moderate positive effect sizes in favor of blended learning when compared to fully online or traditional face-to-face environments. However, there are several considerations inherent in these studies that impact our understanding the generalizability of outcomes.

Dziuban and colleagues (Dziuban et al. 2015 ) analyzed the meta-analyses conducted by Means and her colleagues (Means et al. 2013 ; Means et al. 2010 ), concluding that their methods were impressive as evidenced by exhaustive study inclusion criteria and the use of scale-free effect size indices. The conclusion, in both papers, was that there was a modest difference in multiple outcome measures for courses featuring online modalities—in particular, blended courses. However, with blended learning especially, there are some concerns with these kinds of studies. First, the effect sizes are based on the linear hypothesis testing model with the underlying assumption that the treatment and the error terms are uncorrelated, indicating that there is nothing else going on in the blending that might confound the results. Although the blended learning articles (Means et al. 2010 ) were carefully vetted, the assumption of independence is tenuous at best so that these meta-analysis studies must be interpreted with extreme caution.

There is an additional concern with blended learning as well. Blends are not equivalent because of the manner on which they are configured. For instance, a careful reading of the sources used in the Means, et al. papers will identify, at minimum, the following blending techniques: laboratory assessments, online instruction, e-mail, class web sites, computer laboratories, mapping and scaffolding tools, computer clusters, interactive presentations and e-mail, handwriting capture, evidence-based practice, electronic portfolios, learning management systems, and virtual apparatuses. These are not equivalent ways in which to configure courses, and such nonequivalence constitutes the confounding we describe. We argue here that, in actuality, blended learning is a general construct in the form of a boundary object (Star and Griesemer 1989 ) rather than a treatment effect in the statistical sense. That is, an idea or concept that can support a community of practice, but is weakly defined fostering disagreement in the general group. Conversely, it is stronger in individual constituencies. For instance, content disciplines (i.e. education, rhetoric, optics, mathematics, and philosophy) formulate a more precise definition because of commonly embraced teaching and learning principles. Quite simply, the situation is more complicated than that, as Leonard Smith ( 2007 ) says after Tolstoy,

“All linear models resemble each other, each non nonlinear system is unique in its own way” (p. 33).

This by no means invalidates these studies, but effect size associated with blended learning should be interpreted with caution where the impact is evaluated within a particular learning context.

Study objectives

This study addressed student access by examining success and withdrawal rates in the blended learning courses by comparing them to face-to-face and online modalities over an extended time period at the University of Central Florida. Further, the investigators sought to assess the differences in those success and withdrawal rates with the minority status of students. Secondly, the investigators examined the student end-of-course ratings of blended learning and other modalities by attempting to develop robust if-then decision rules about what characteristics of classes and instructors lead students to assign an “excellent” value to their educational experience. Because of the high stakes nature of these student ratings toward faculty promotion, awards, and tenure, they act as a surrogate measure for instructional quality. Next, the investigators determined the conditional probabilities for students conforming to the identified rule cross-referenced by expected grade, the degree to which they desired to take the course, and course modality.

Student grades by course modality were recoded into a binary variable with C or higher assigned a value of 1, and remaining values a 0. This was a declassification process that sacrificed some specificity but compensated for confirmation bias associated with disparate departmental policies regarding grade assignment. At the measurement level this was an “on track to graduation index” for students. Withdrawal was similarly coded by the presence or absence of its occurrence. In each case, the percentage of students succeeding or withdrawing from blended, online or face-to-face courses was calculated by minority and non-minority status for the fall 2014 through fall 2015 semesters.

Next, a classification and regression tree (CART) analysis (Brieman et al. 1984 ) was performed on the student end-of-course evaluation protocol ( Appendix 1 ). The dependent measure was a binary variable indicating whether or not a student assigned an overall rating of excellent to his or her course experience. The independent measures in the study were: the remaining eight rating items on the protocol, college membership, and course level (lower undergraduate, upper undergraduate, and graduate). Decision trees are efficient procedures for achieving effective solutions in studies such as this because with missing values imputation may be avoided with procedures such as floating methods and the surrogate formation (Brieman et al. 1984 , Olshen et al. 1995 ). For example, a logistic regression method cannot efficiently handle all variables under consideration. There are 10 independent variables involved here; one variable has three levels, another has nine, and eight have five levels each. This means the logistic regression model must incorporate more than 50 dummy variables and an excessively large number of two-way interactions. However, the decision-tree method can perform this analysis very efficiently, permitting the investigator to consider higher order interactions. Even more importantly, decision trees represent appropriate methods in this situation because many of the variables are ordinally scaled. Although numerical values can be assigned to each category, those values are not unique. However, decision trees incorporate the ordinal component of the variables to obtain a solution. The rules derived from decision trees have an if-then structure that is readily understandable. The accuracy of these rules can be assessed with percentages of correct classification or odds-ratios that are easily understood. The procedure produces tree-like rule structures that predict outcomes.

The model-building procedure for predicting overall instructor rating

For this study, the investigators used the CART method (Brieman et al. 1984 ) executed with SPSS 23 (IBM Corp 2015 ). Because of its strong variance-sharing tendencies with the other variables, the dependent measure for the analysis was the rating on the item Overall Rating of the Instructor , with the previously mentioned indicator variables (college, course level, and the remaining 8 questions) on the instrument. Tree methods are recursive, and bisect data into subgroups called nodes or leaves. CART analysis bases itself on: data splitting, pruning, and homogeneous assessment.

Splitting the data into two (binary) subsets comprises the first stage of the process. CART continues to split the data until the frequencies in each subset are either very small or all observations in a subset belong to one category (e.g., all observations in a subset have the same rating). Usually the growing stage results in too many terminate nodes for the model to be useful. CART solves this problem using pruning methods that reduce the dimensionality of the system.

The final stage of the analysis involves assessing homogeneousness in growing and pruning the tree. One way to accomplish this is to compute the misclassification rates. For example, a rule that produces a .95 probability that an instructor will receive an excellent rating has an associated error of 5.0%.

Implications for using decision trees

Although decision-tree techniques are effective for analyzing datasets such as this, the reader should be aware of certain limitations. For example, since trees use ranks to analyze both ordinal and interval variables, information can be lost. However, the most serious weakness of decision tree analysis is that the results can be unstable because small initial variations can lead to substantially different solutions.

For this study model, these problems were addressed with the k-fold cross-validation process. Initially the dataset was partitioned randomly into 10 subsets with an approximately equal number of records in each subset. Each cohort is used as a test partition, and the remaining subsets are combined to complete the function. This produces 10 models that are all trained on different subsets of the original dataset and where each has been used as the test partition one time only.

Although computationally dense, CART was selected as the analysis model for a number of reasons— primarily because it provides easily interpretable rules that readers will be able evaluate in their particular contexts. Unlike many other multivariate procedures that are even more sensitive to initial estimates and require a good deal of statistical sophistication for interpretation, CART has an intuitive resonance with researcher consumers. The overriding objective of our choice of analysis methods was to facilitate readers’ concentration on our outcomes rather than having to rely on our interpretation of the results.

Institution-level evaluation: Success and withdrawal

The University of Central Florida (UCF) began a longitudinal impact study of their online and blended courses at the start of the distributed learning initiative in 1996. The collection of similar data across multiple semesters and academic years has allowed UCF to monitor trends, assess any issues that may arise, and provide continual support for both faculty and students across varying demographics. Table  1 illustrates the overall success rates in blended, online and face-to-face courses, while also reporting their variability across minority and non-minority demographics.

While success (A, B, or C grade) is not a direct reflection of learning outcomes, this overview does provide an institutional level indication of progress and possible issues of concern. BL has a slight advantage when looking at overall success and withdrawal rates. This varies by discipline and course, but generally UCF’s blended modality has evolved to be the best of both worlds, providing an opportunity for optimizing face-to-face instruction through the effective use of online components. These gains hold true across minority status. Reducing on-ground time also addresses issues that impact both students and faculty such as parking and time to reach class. In addition, UCF requires faculty to go through faculty development tailored to teaching in either blended or online modalities. This 8-week faculty development course is designed to model blended learning, encouraging faculty to redesign their course and not merely consider blended learning as a means to move face-to-face instructional modules online (Cobb et al. 2012 ; Lowe 2013 ).

Withdrawal (Table  2 ) from classes impedes students’ success and retention and can result in delayed time to degree, incurred excess credit hour fees, or lost scholarships and financial aid. Although grades are only a surrogate measure for learning, they are a strong predictor of college completion. Therefore, the impact of any new innovation on students’ grades should be a component of any evaluation. Once again, the blended modality is competitive and in some cases results in lower overall withdrawal rates than either fully online or face-to-face courses.

The students’ perceptions of their learning environments

Other potentially high-stakes indicators can be measured to determine the impact of an innovation such as blended learning on the academy. For instance, student satisfaction and attitudes can be measured through data collection protocols, including common student ratings, or student perception of instruction instruments. Given that those ratings often impact faculty evaluation, any negative reflection can derail the successful implementation and scaling of an innovation by disenfranchised instructors. In fact, early online and blended courses created a request by the UCF faculty senate to investigate their impact on faculty ratings as compared to face-to-face sections. The UCF Student Perception of Instruction form is released automatically online through the campus web portal near the end of each semester. Students receive a splash page with a link to each course’s form. Faculty receive a scripted email that they can send to students indicating the time period that the ratings form will be available. The forms close at the beginning of finals week. Faculty receive a summary of their results following the semester end.

The instrument used for this study was developed over a ten year period by the faculty senate of the University of Central Florida, recognizing the evolution of multiple course modalities including blended learning. The process involved input from several constituencies on campus (students, faculty, administrators, instructional designers, and others), in attempt to provide useful formative and summative instructional information to the university community. The final instrument was approved by resolution of the senate and, currently, is used across the university. Students’ rating of their classes and instructors comes with considerable controversy and disagreement with researchers aligning themselves on both sides of the issue. Recently, there have been a number of studies criticizing the process (Uttl et al. 2016 ; Boring et al. 2016 ; & Stark and Freishtat 2014 ). In spite of this discussion, a viable alternative has yet to emerge in higher education. So in the foreseeable future, the process is likely to continue. Therefore, with an implied faculty senate mandate this study was initiated by this team of researchers.

Prior to any analysis of the item responses collected in this campus-wide student sample, the psychometric quality (domain sampling) of the information yielded by the instrument was assessed. Initially, the reliability (internal consistency) was derived using coefficient alpha (Cronbach 1951 ). In addition, Guttman ( 1953 ) developed a theorem about item properties that leads to evidence about the quality of one’s data, demonstrating that as the domain sampling properties of items improve, the inverse of the correlation matrix among items will approach a diagonal. Subsequently, Kaiser and Rice ( 1974 ) developed the measure of sampling adequacy (MSA) that is a function of the Guttman Theorem. The index has an upper bound of one with Kaiser offering some decision rules for interpreting the value of MSA. If the value of the index is in the .80 to .99 range, the investigator has evidence of an excellent domain sample. Values in the .70s signal an acceptable result, and those in the .60s indicate data that are unacceptable. Customarily, the MSA has been used for data assessment prior to the application of any dimensionality assessments. Computation of the MSA value gave the investigators a benchmark for the construct validity of the items in this study. This procedure has been recommended by Dziuban and Shirkey ( 1974 ) prior to any latent dimension analysis and was used with the data obtained for this study. The MSA for the current instrument was .98 suggesting excellent domain sampling properties with an associated alpha reliability coefficient of .97 suggesting superior internal consistency. The psychometric properties of the instrument were excellent with both measures.

The online student ratings form presents an electronic data set each semester. These can be merged across time to create a larger data set of completed ratings for every course across each semester. In addition, captured data includes course identification variables including prefix, number, section and semester, department, college, faculty, and class size. The overall rating of effectiveness is used most heavily by departments and faculty in comparing across courses and modalities (Table  3 ).

The finally derived tree (decision rules) included only three variables—survey items that asked students to rate the instructor’s effectiveness at:

Helping students achieve course objectives,

Creating an environment that helps students learn, and

Communicating ideas and information.

None of the demographic variables associated with the courses contributed to the final model. The final rule specifies that if a student assigns an excellent rating to those three items, irrespective of their status on any other condition, the probability is .99 that an instructor will receive an overall rating of excellent. The converse is true as well. A poor rating on all three of those items will lead to a 99% chance of an instructor receiving an overall rating of poor.

Tables  4 , 5 and 6 present a demonstration of the robustness of the CART rule for variables on which it was not developed: expected course grade, desire to take the course and modality.

In each case, irrespective of the marginal probabilities, those students conforming to the rule have a virtually 100% chance of seeing the course as excellent. For instance, 27% of all students expecting to fail assigned an excellent rating to their courses, but when they conformed to the rule the percentage rose to 97%. The same finding is true when students were asked about their desire to take the course with those who strongly disagreed assigning excellent ratings to their courses 26% of the time. However, for those conforming to the rule, that category rose to 92%. When course modality is considered in the marginal sense, blended learning is rated as the preferred choice. However, from Table  6 we can observe that the rule equates student assessment of their learning experiences. If they conform to the rule, they will see excellence.

This study addressed increasingly important issues of student success, withdrawal and perception of the learning environment across multiple course modalities. Arguably these components form the crux of how we will make more effective decisions about how blended learning configures itself in the new normal. The results reported here indicate that blending maintains or increases access for most student cohorts and produces improved success rates for minority and non-minority students alike. In addition, when students express their beliefs about the effectiveness of their learning environments, blended learning enjoys the number one rank. However, upon more thorough analysis of key elements students view as important in their learning, external and demographic variables have minimal impact on those decisions. For example college (i.e. discipline) membership, course level or modality, expected grade or desire to take a particular course have little to do with their course ratings. The characteristics they view as important relate to clear establishment and progress toward course objectives, creating an effective learning environment and the instructors’ effective communication. If in their view those three elements of a course are satisfied they are virtually guaranteed to evaluate their educational experience as excellent irrespective of most other considerations. While end of course rating protocols are summative the three components have clear formative characteristics in that each one is directly related to effective pedagogy and is responsive to faculty development through units such as the faculty center for teaching and learning. We view these results as encouraging because they offer potential for improving the teaching and learning process in an educational environment that increases the pressure to become more responsive to contemporary student lifestyles.

Clearly, in this study we are dealing with complex adaptive systems that feature the emergent property. That is, their primary agents and their interactions comprise an environment that is more than the linear combination of their individual elements. Blending learning, by interacting with almost every aspect of higher education, provides opportunities and challenges that we are not able to fully anticipate.

This pedagogy alters many assumptions about the most effective way to support the educational environment. For instance, blending, like its counterpart active learning, is a personal and individual phenomenon experienced by students. Therefore, it should not be surprising that much of what we have called blended learning is, in reality, blended teaching that reflects pedagogical arrangements. Actually, the best we can do for assessing impact is to use surrogate measures such as success, grades, results of assessment protocols, and student testimony about their learning experiences. Whether or not such devices are valid indicators remains to be determined. We may be well served, however, by changing our mode of inquiry to blended teaching.

Additionally, as Norberg ( 2017 ) points out, blended learning is not new. The modality dates back, at least, to the medieval period when the technology of textbooks was introduced into the classroom where, traditionally, the professor read to the students from the only existing manuscript. Certainly, like modern technologies, books were disruptive because they altered the teaching and learning paradigm. Blended learning might be considered what Johnson describes as a slow hunch (2010). That is, an idea that evolved over a long period of time, achieving what Kaufmann ( 2000 ) describes as the adjacent possible – a realistic next step occurring in many iterations.

The search for a definition for blended learning has been productive, challenging, and, at times, daunting. The definitional continuum is constrained by Oliver and Trigwell ( 2005 ) castigation of the concept for its imprecise vagueness to Sharpe et al.’s ( 2006 ) notion that its definitional latitude enhances contextual relevance. Both extremes alter boundaries such as time, place, presence, learning hierarchies, and space. The disagreement leads us to conclude that Lakoff’s ( 2012 ) idealized cognitive models i.e. arbitrarily derived concepts (of which blended learning might be one) are necessary if we are to function effectively. However, the strong possibility exists that blended learning, like quality, is observer dependent and may not exist outside of our perceptions of the concept. This, of course, circles back to the problem of assuming that blending is a treatment effect for point hypothesis testing and meta-analysis.

Ultimately, in this article, we have tried to consider theoretical concepts and empirical findings about blended learning and their relationship to the new normal as it evolves. Unfortunately, like unresolved chaotic solutions, we cannot be sure that there is an attractor or that it will be the new normal. That being said, it seems clear that blended learning is the harbinger of substantial change in higher education and will become equally impactful in K-12 schooling and industrial training. Blended learning, because of its flexibility, allows us to maximize many positive education functions. If Floridi ( 2014 ) is correct and we are about to live in an environment where we are on the communication loop rather than in it, our educational future is about to change. However, if our results are correct and not over fit to the University of Central Florida and our theoretical speculations have some validity, the future of blended learning should encourage us about the coming changes.

Adams Becker, S., Cummins, M., Davis, A., Freeman, A., Hall Giesinger, C., & Ananthanarayanan, V. (2017). NMC horizon report: 2017 higher Education Edition . Austin: The New Media Consortium.

Google Scholar  

Alhabeeb, A. M. (2015). The quality assessment of the services offered to the students of the College of Education at King Saud University using (SERVQUAL) method. Journal of Education and Practice , 6 (30), 82–93.

Allen, I. E., & Seaman, J. (2003). Sizing the opportunity: The quality and extent of online education in the United States, 2002 and 2003. Retrieved from http://files.eric.ed.gov/fulltext/ED530060.pdf

Allen, I. E., Seaman, J., Poulin, R., & Straut, T. T. (2016). Online report card: Tracking online education in the United States, 1–4. Retrieved from http://onlinelearningsurvey.com/reports/onlinereportcard.pdf

Arum, R., Roksa, J., & Cook, A. (2016). Improving quality in American higher education: Learning outcomes and assessments for the 21st century . San Francisco: Jossey-Bass.

Aud, S., Hussar, W., Planty, M., Snyder, T., Bianco, K., Fox, M. A., & Drake, L. (2010). The condition of education - 2010. Education, 4–29. https://doi.org/10.1037/e492172006-019

Balfour, S. P. (2013). Assessing writing in MOOCs: Automated essay scoring and calibrated peer review. Research and Practice in Assessment , 2013 (8), 40–48.

Bayne, S., Evans, P., Ewins, R.,Knox, J., Lamb, J., McLeod, H., O’Shea, C., Ross, J., Sheail, P. & Sinclair, C, (2016) Manifesto for teaching online. Digital Education at Edinburg University. Retrieved from https://onlineteachingmanifesto.wordpress.com/the-text/

Bernard, R. M., Abrami, P. C., Borokhovski, E., Wade, C. A., Tamim, R. M., Surkes, M. A., & Bethel, E. C. (2009). A meta-analysis of three types of interaction treatments in distance education. Review of Educational Research , 79 (3), 1243–1289. https://doi.org/10.3102/0034654309333844 .

Article   Google Scholar  

Bernard, R. M., Borokhovski, E., Schmid, R. F., Tamim, R. M., & Abrami, P. C. (2014). A meta-analysis of blended learning and technology use in higher education: From the general to the applied. Journal of Computing in Higher Education , 26 (1), 87–122.

Bloemer, W., & Swan, K. (2015). Investigating informal blending at the University of Illinois Springfield. In A. G. Picciano, C. D. Dziuban, & C. R. Graham (Eds.), Blended learning: Research perspectives , (vol. 2, pp. 52–69). New York: Routledge.

Bonk, C. J., & Graham, C. R. (2007). The handbook of blended learning: Global perspectives, local designs . San Francisco: Pfeiffer.

Boring, A., Ottoboni, K., & Stark, P.B. (2016). Student evaluations of teaching (mostly) do not measure teaching effectiveness. EGERA.

Brieman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and regression trees . New York: Chapman & Hall.

California Community Colleges Chancellor’s Office. (2013). Distance education report.

Cobb, C., deNoyelles, A., & Lowe, D. (2012). Influence of reduced seat time on satisfaction and perception of course development goals: A case study in faculty development. The Journal of Asynchronous Learning , 16 (2), 85–98.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika , 16 (3), 297–334 Retrieved from http://psych.colorado.edu/~carey/courses/psyc5112/readings/alpha_cronbach.pdf .

Article   MATH   Google Scholar  

Dringus, L. P., and A. B. Seagull. 2015. A five-year study of sustaining blended learning initiatives to enhance academic engagement in computer and information sciences campus courses. In Blended learning: Research perspectives. Vol. 2. Edited by A. G. Picciano, C. D. Dziuban, and C. R. Graham, 122-140. New York: Routledge.

Dziuban, C. D., & Shirkey, E. C. (1974). When is a correlation matrix appropriate for factor analysis? Some decision rules. Psychological Bulletin , 81(6), 358. https://doi.org/10.1037/h0036316 .

Dziuban, C., Hartman, J., Cavanagh, T., & Moskal, P. (2011). Blended courses as drivers of institutional transformation. In A. Kitchenham (Ed.), Blended learning across disciplines: Models for implementation , (pp. 17–37). Hershey: IGI Global.

Chapter   Google Scholar  

Dziuban, C., & Moskal, P. (2011). A course is a course is a course: Factor invariance in student evaluation of online, blended and face-to-face learning environments. The Internet and Higher Education , 14 (4), 236–241.

Dziuban, C., Moskal, P., Hermsdorfer, A., DeCantis, G., Norberg, A., & Bradford, G., (2015) A deconstruction of blended learning. Presented at the 11 th annual Sloan-C blended learning conference and workshop

Dziuban, C., Picciano, A. G., Graham, C. R., & Moskal, P. D. (2016). Conducting research in online and blended learning environments: New pedagogical frontiers . New York: Routledge, Taylor & Francis Group.

Dziuban, C. D., Hartman, J. L., & Moskal, P. D. (2004). Blended learning. EDUCAUSE Research Bulletin , 7 , 1–12.

EDUCAUSE. (2017) 2017 key issues in teaching & learning. Retrieved from https://www.EDUCAUSE.edu/eli/initiatives/key-issues-in-teaching-and-learning

Fairlie, R. (2004). Race and the digital divide. The B.E. Journal of Economic Analysis & Policy , 3 (1). https://doi.org/10.2202/1538-0645.1263 .

Fischer, L., Hilton, J., Robinson, T. J., & Wiley, D. (2015). A Multi-institutional Study of the Impact of Open Textbook Adoption on the Learning Outcomes of Post-secondary Students . Journal of Computing in Higher Education. https://doi.org/10.1007/s12528-015-9101-x .

Floridi, L. (2008). A defence of informational structural realism. Synthese , 161 (2), 219–253.

Article   MathSciNet   Google Scholar  

Floridi, L. (2014). The 4th revolution: How the infosphere is reshaping human reality . Oxford: Oxford University Press.

Garrison, D. R., & Vaughan, N. D. (2013). Blended learning in higher education , (1st ed., ). San Francisco: Jossey-Bass Print.

Garrison, D. R., & Kanuka, H. (2004). Blended learning: Uncovering its transformative potential in higher education. The Internet and Higher Education , 7 , 95–105.

Goodhart, C.A.E. (1975). “Problems of monetary management: The U.K. experience.” Papers in Monetary Economics. Reserve Bank of Australia. I.

Graham, C. R. (2013). Emerging practice and research in blended learning. In M. G. Moore (Ed.), Handbook of distance education , (3rd ed., pp. 333–350). New York: Routledge.

Guttman, L. (1953). Image theory for the structure of quantitative variates. Psychometrika , 18 , 277–296.

Article   MathSciNet   MATH   Google Scholar  

Hilton, J., Fischer, L., Wiley, D., & Williams, L. (2016). Maintaining momentum toward graduation: OER and the course throughput rate. International Review of Research in Open and Distance Learning , 17 (6) https://doi.org/10.19173/irrodl.v17i6.2686 .

IBM Corp. Released (2015). IBM SPSS statistics for windows, version 23.0 . Armonk: IBM Corp.

Jean-François, E. (2013). Transcultural blended learning and teaching in postsecondary education . Hershey: Information Science Reference.

Book   Google Scholar  

Jones, S., Johnson-Yale, C., Millermaier, S., & Pérez, F. S. (2009). U.S. college students’ internet use: Race, gender and digital divides. Journal of Computer-Mediated Communication , 14 (2), 244–264 https://doi.org/10.1111/j.1083-6101.2009.01439.x .

Kaiser, H. F., & Rice, J. (1974). Little Jiffy, Mark IV. Journal of Educational and Psychological Measurement , 34(1), 111–117.

Kaufmann, S. (2000). Investigations . New York: Oxford University Press.

Kitchenham, A. (2011). Blended learning across disciplines: Models for implementation . Hershey: Information Science Reference.

Lakoff, G. (2012). Women, fire, and dangerous things: What categories reveal about the mind . Chicago: The University of Chicago Press.

Lewis, L., & Parsad, B. (2008). Distance education at degree-granting postsecondary institutions : 2006–07 (NCES 2009–044) . Washington: Retrieved from http://nces.ed.gov/pubs2009/2009044.pdf .

Liu, F., & Cavanaugh, C. (2011). High enrollment course success factors in virtual school: Factors influencing student academic achievement. International Journal on E-Learning , 10 (4), 393–418.

Lowe, D. (2013). Roadmap of a blended learning model for online faculty development. Invited feature article in Distance Education Report , 17 (6), 1–7.

Means, B., Toyama, Y., Murphy, R., & Baki, M. (2013). The effectiveness of online and blended learning: A meta-analysis of the empirical literature. Teachers College Record , 115 (3), 1–47.

Means, B., Toyama, Y., Murphy, R., Kaia, M., & Jones, K. (2010). Evaluation of evidence-based practices in online learning . Washington: US Department of Education.

Moskal, P., Dziuban, C., & Hartman, J. (2013). Blended learning: A dangerous idea? The Internet and Higher Education , 18 , 15–23.

Norberg, A. (2017). From blended learning to learning onlife: ICTs, time and access in higher education (Doctoral dissertation, Umeå University).

Norberg, A., Dziuban, C. D., & Moskal, P. D. (2011). A time-based blended learning model. On the Horizon , 19 (3), 207–216. https://doi.org/10.1108/10748121111163913 .

Oliver, M., & Trigwell, K. (2005). Can ‘blended learning’ be redeemed? e-Learning , 2 (1), 17–25.

Olshen, Stone , Steinberg , and Colla (1995). CART classification and regression trees. Tree-structured nonparametric data analysis. Statistical algorithms. Salford systems interface and documentation. Salford Systems .

O'Neil, C. (2017). Weapons of math destruction: How big data increases inequality and threatens democracy . Broadway Books.

Online Learning Consortium. The OLC quality scorecard for blended learning programs. Retrieved from https://onlinelearningconsortium.org/consult/olc-quality-scorecard-blended-learning-programs/

Open SUNY. The OSCQR course design review scorecard. Retrieved from https://onlinelearningconsortium.org/consult/oscqr-course-design-review/

Picciano, A. G. (2009). Blending with purpose: The multimodal model. Journal of Asynchronous Learning Networks , 13 (1), 7–18.

Picciano, A. G., Dziuban, C., & Graham, C. R. (2014). Blended learning: Research perspectives , (vol. 2). New York: Routledge.

Picciano, A. G., & Dziuban, C. D. (2007). Blended learning: Research perspectives . Needham: The Sloan Consortium.

Pirsig, R. M. (1974). Zen and the art of motorcycle maintenance: An inquiry into values . New York: Morrow.

Quality Matters. (2016). About Quality Matters. Retrieved from https://www.qualitymatters.org/research

Robinson, T. J., Fischer, L., Wiley, D. A., & Hilton, J. (2014). The Impact of Open Textbooks on Secondary Science Learning Outcomes . Educational Researcher. https://doi.org/10.3102/0013189X14550275 .

Ross, B., & Gage, K. (2006). Global perspectives on blended learning: Insight from WebCT and our customers in higher education. In C. J. Bonk, & C. R. Graham (Eds.), Handbook of blended learning: Global perspectives, local designs , (pp. 155–168). San Francisco: Pfeiffer.

Rovai, A. P., & Jordan, H. M. (2004). Blended learning and sense of community: A comparative analysis with traditional and fully online graduate courses. International Review of Research in Open and Distance Learning , 5 (2), 1–13.

Searle, J. R. (2015). Seeing things as they are: A theory of perception . Chicago: Oxford University Press.

Sharpe, R., Benfield, G., Roberts, G., & Francis, R. (2006). The undergraduate experience of blended learning: A review of UK literature and research. The Higher Education Academy, (October 2006).

Shea, P., & Bidjerano, T. (2014). Does online learning impede degree completion? A national study of community college students. Computers and Education , 75 , 103–111 https://doi.org/10.1016/j.compedu.2014.02.009 .

Shea, P., & Bidjerano, T. (2016). A National Study of differences between distance and non-distance community college students in time to first associate degree attainment, transfer, and dropout. Online Learning , 20 (3), 14–15.

Sitzmann, T., Kraiger, K., Stewart, D., & Wisher, R. (2006). The comparative effectiveness of web-based and classroom instruction: A meta-analysis. Personnel Psychology , 59 (3), 623–664.

Smith, L. A. (2007). Chaos: a very short introduction . Oxford: Oxford University Press.

Star, S. L., & Griesemer, J. R. (1989). Institutional ecology, translations and boundary objects: Amatuers and professionals in Berkely’s Museum of Vertebrate Zoology, 1907-39. Social Studies of Science , 19 (3), 387–420.

Stark, P. & Freishtat, R. (2014). An evaluation of course evaluations. ScienceOpen. Retrieved from https://www.stat.berkeley.edu/~stark/Preprints/evaluations14.pdf .

Tynan, B., Ryan, Y., & Lamont-Mills, A. (2015). Examining workload models in online and blended teaching. British Journal of Educational Technology , 46 (1), 5–15.

Uttl, B., White, C. A., & Gonzalez, D. W. (2016). Meta-analysis of faculty’s teaching effectiveness: Student evaluation of teaching ratings and student learning are not related. Studies in Educational Evaluation , 54 , 22–42.

Williams, J. (2016). College and the new class divide. Inside Higher Ed July 11, 2016.

Wladis, C., Hachey, A. C., & Conway, K. (2015). Which STEM majors enroll in online courses, and why should we care? The impact of ethnicity, gender, and non-traditional student characteristics. Computers and Education , 87 , 285–308 https://doi.org/10.1016/j.compedu.2015.06.010 .

Zhao, Y., Lei, J., Yan, B., Lai, C., & Tan, H. S. (2005). What makes the difference? A practical analysis of research on the effectiveness of distance education. Teachers College Record , 107 (8), 1836–1884. https://doi.org/10.1111/j.1467-9620.2005.00544.x .

Download references

Acknowledgements

The authors acknowledge the contributions of several investigators and course developers from the Center for Distributed Learning at the University of Central Florida, the McKay School of Education at Brigham Young University, and Scholars at Umea University, Sweden. These professionals contributed theoretical and practical ideas to this research project and carefully reviewed earlier versions of this manuscript. The Authors gratefully acknowledge their support and assistance.

Author information

Authors and affiliations.

University of Central Florida, Orlando, Florida, USA

Charles Dziuban, Patsy D. Moskal & Nicole Sicilia

Brigham Young University, Provo, Utah, USA

Charles R. Graham

Campus Skellefteå, Skellefteå, Sweden

Anders Norberg

You can also search for this author in PubMed   Google Scholar

Contributions

The Authors of this article are listed in alphabetical order indicating equal contribution to this article. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Patsy D. Moskal .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Student Perception of Instruction

Instructions: Please answer each question based on your current class experience. You can provide additional information where indicated.

All responses are anonymous. Responses to these questions are important to help improve the course and how it is taught. Results may be used in personnel decisions. The results will be shared with the instructor after the semester is over.

Please rate the instructor’s effectiveness in the following areas:

Organizing the course:

Excellent b) Very Good c) Good d) Fair e) Poor

Explaining course requirements, grading criteria, and expectations:

Communicating ideas and/or information:

Showing respect and concern for students:

Stimulating interest in the course:

Creating an environment that helps students learn:

Giving useful feedback on course performance:

Helping students achieve course objectives:

Overall, the effectiveness of the instructor in this course was:

What did you like best about the course and/or how the instructor taught it?

What suggestions do you have for improving the course and/or how the instructor taught it?

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Cite this article.

Dziuban, C., Graham, C.R., Moskal, P.D. et al. Blended learning: the new normal and emerging technologies. Int J Educ Technol High Educ 15 , 3 (2018). https://doi.org/10.1186/s41239-017-0087-5

Download citation

Received : 09 October 2017

Accepted : 20 December 2017

Published : 15 February 2018

DOI : https://doi.org/10.1186/s41239-017-0087-5

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Blended learning
• Higher education
• Student success
• Student perception of instruction

Ethical, Legal and Social Implications of Emerging Technology (ELSIET) Symposium

• Published: 24 June 2022
• Volume 19 , pages 363–370, ( 2022 )

Cite this article

• Evie Kendal   ORCID: orcid.org/0000-0002-8414-0427 1  

7403 Accesses

7 Citations

6 Altmetric

Explore all metrics

Avoid common mistakes on your manuscript.

Establishing ethical guidelines for the development and release of emerging technologies involves many practical challenges. Traditional methods of evaluating relevant ethical dimensions, such as Beauchamp and Childress’s ( 2001 ) Principlist framework, are often not fit for purpose: after all, how can one give autonomous, informed consent to the use of novel technologies whose effects are unknown? How can cost-benefit analyses be conducted in cases where there is a high degree of scientific uncertainty about the severity and likelihood of different risks, and potential benefits have not yet been demonstrated? Nevertheless, it is necessary to promote consideration of the ethical, legal, and social implications of emerging technologies to avoid them being released into what some commentators label a moral, policy, and/or legal vacuum (Moor 2005 ; Edwards 1991 ). Consequently, various methods for approaching the ethics of emerging technologies have arisen over the last few decades, some of the more common of which are summarized below.

Precautionary Approaches and the Precautionary Principle

Moor ( 2005 ) claims that the rapid emergence of new technologies “should give us a sense of urgency in thinking about the ethical (including social) implications” of these technologies (111), noting that when technological developments have significant social impact this is when “technological revolution occurs” (112). He notes, however, that these technological revolutions “do not arrive fully mature,” and their unpredictability yields many ethical concerns (112). For this reason, Wolff ( 2014 ) advocates for a precautionary approach to regulating emerging technologies with unknown risks, noting historical excitement over the benefits of new technologies that were far outweighed by harms that materialized later: asbestos used to fireproof buildings that led to high costs for removal and loss of human life, and chlorofluorocarbons used in refrigerants that caused significant damage to the ozone layer being his main examples (S27). He claims a precautionary approach to any new technology would always ask four questions: 1) is the technology known to have “intolerable risks”, 2) does it yield substantial benefits, 3) do these benefits “solve important problems”, and 4) could these problems be “solved in some other, less risky way” (S28)? According to this approach, unless the answers to questions 1 and 4 are no, and 2 and 3 are yes, technological development should not be permitted to proceed.

The Precautionary Principle (PP) formalizes a precautionary approach to emerging technologies, particularly those involving potential harms to human health and the environment (Hansson 2020 ). Bouchaut and Asveld ( 2021 ) note the PP originated in German domestic law in the 1970s before being adopted as the dominant approach to regulating new biotechnologies throughout Europe from the 1990s onward. One of the main regulatory areas where the PP is discussed is in European legislation surrounding genetically-modified foods, with many arguing this precedent will likely impact the treatment of more novel gene-editing techniques, such as CRISPR-Cas9 (Bouchaut and Asveld 2021 ). At its core, the PP translates any potential for harm in the face of scientific uncertainty into a positive duty for stakeholders to act to prevent or mitigate this harm (Guida 2021 ). Incorporating risk assessment, management, and communication, for Hansson ( 2020 ) the PP represents a “pattern of thought, namely that protective measures against a potential danger can be justified even if it not known for sure that the danger exists” (250). It is for this reason that use of the PP is often criticized for unreasonably blocking technological developments, including those that could yield significant health benefits for the population (Hester et al. 2015 ). However, in a study of nine jurisdictions with different levels of regulatory restrictions on biotechnological developments, Gouvea et al. ( 2012 ) found research productivity was not enhanced through the “absence of structural or ethical impediments,” but rather the presence of transparent guidelines (562):

While one might argue that an environment that lacks all constraints (including ethical barriers) may allow for rapid product development through the provision of an environment where anything goes, the opposite is found to be the case. It is likely that the presence of clearly defined rules, higher levels of disclosure, greater levels of trust, and reduced costs (associated to lower levels of corruption) results in the appropriate set of ethical rules and guidelines providing the best outcomes for development of commercial products (562–563).

This study drew comparisons between the European model applying the PP to advances in nanotechnology and the U.S. wait-and-see approach, which treats these new products the same as their more traditional counterparts (554). Applying Hester et al.’s ( 2015 ) logic to this situation, the European approach imposes restrictions and a requirement to avoid potential harms, while the United States takes a more conventional legal approach that would merely award damages if a harm is sustained.

Hansson ( 2020 ) claims “no other safety principle has been so vehemently contested” as the PP, with many arguing it “stifles innovation by imposing unreasonable demands on the safety of new technologies” (245). However, applying precautionary measures to novel situations with uncertain risks has proven essential in the global response to the COVID-19 crisis, where preventive actions had to be put in place while scientific data were still being collected (Guida 2021 ). For Hansson ( 2020 ), and many others, what the PP lacks is a method of adjusting to new knowledge as it becomes available. Wareham and Nardini ( 2015 ) similarly note that in its strongest formulation, the PP might ban entire research projects going ahead, due to risk, however small, of a significant harm. Mittelstadt, Stahl, and Fairweather ( 2015 ) also note that if the purpose of the PP is to avoid harm, and preventing scientific progress and the development of new technologies can be considered a harm, this results in a precautionary paradox where the principle “would instruct us to refrain from implementing itself” (1034). In all the above cases, the authors advocate instead for an iterative process that allows progressive steps of experimentation, proportional to their associated risk, with regulations adapting as scientific uncertainty “gives way to new scientific knowledge” (Hansson 2020 , 253). This relates to the next approach to be covered here: design ethics.

Design Ethics Versus the Collingridge Dilemma

Design ethics takes into account the social and ethical dimensions of the context in which a product is designed and will be used. One of the more common methods is “value-sensitive design,” which tries to identify relevant human values during technology research and development phases to ensure they are “promoted and respected by the design” (Umbrello and van de Poel 2021 , 283). These might include respect for privacy, environmental sustainability, accountability, and many other values that are at stake in the interaction between people, technology, and the environment (Friedman et al. 2021 ). When it comes to new technologies with unknown risks, this model would support adopting preventive measures to mitigate harm, but as Bouchaut and Asveld ( 2021 ) claim, could also allow for “controlled learning” experiments, where step-by-step potential risks can be explored as the technology develops. These authors refer to this process as responsible learning , noting it would also require a degree of regulatory flexibility that the PP does not currently support. In this way, it is a method of proceeding in the face of uncertainty, reflecting on the ethical and safety concerns of each new stage in development before a technology is fully realized. One such model is the “Safe-by-Design” approach, which these authors note is “associated with learning processes that aim for designing specifically for the notion of safety by iteratively integrating knowledge about the adverse effects of materials” (Bouchaut and Asveld 2021 ). Other models include “participative design,” where the views and values of end-users are sought during the design phase, so designers can incorporate knowledge of the consequences of new technologies on those impacted by them (Mumford 1993 ). The acronym ETHICS was used in the formulation of this approach when considering the impact of new computing technologies on workers’ experiences, referring to Effective Technical and Human Implementation of Computer-based Systems (Mumford 1993 ). At its core, the purpose of participative design is to make new technologies fit-for-purpose and people-friendly.

While intuitively a system that progressively learns about risks as they manifest and adapts accordingly may seem superior to one that might ban an emerging technology from the outset due to unknown risks, one problem with this iterative approach is what has been dubbed the Collingridge dilemma. Mittelstadt, Stahl, and Fairweather ( 2015 ) explain this dilemma as follows:

it is impossible to know with certainty the consequences of an emerging technology at an early stage when it would be comparatively simple to change the technology’s trajectory. Once the technology is more established and it becomes clear what its social and ethical consequences are going to be, it becomes increasingly difficult to affect its outcomes and social context. (1028)

So, while a “Safe-by-Design” approach might be able to pivot easily if issues are discovered early in the process, once the technology has progressed to a certain stage, it is too late to intervene. To prevent the creation and release of potentially dangerous technologies requires a more speculative approach, as is present in the next three models to be discussed.

Technology Assessment (TA) to Ethical Technology Assessment (eTA)

Alongside the PP, technology assessment (TA) is one of the best-known methods of dealing with uncertainty (Mittelstadt, Stahl, and Fairweather 2015 ). Grunwald ( 2020 ) notes that because TA is not focused on technologies that actually exist yet, it is a method that “creates and assesses prospective knowledge about the future consequences of technology,” through evaluating and scrutinizing “ideas, designs, plans, or visions for future technology” (Grunwald 2020 , 97; Grunwald 2019 ). He further notes that participatory versions of this speculative evaluative process are less about imagining technologies per se, and more about envisaging future technologies as situated in a specific “societal environment” (Grunwald 2019 ). The inputs for analysis are thus “models, narratives, roadmaps, visions, scenarios, prototypes” etc. (Grunwald 2020 , 99). Mittelstadt, Stahl and Fairweather ( 2015 ) note traditional TA arose in response to “undesirable or unintentional side effects of emerging technologies” with a primary focus on considering the impact of technology on “the environment, industry and society” (1035). Its goal is to foster responsible regulation to maximize benefit and prevent harm caused by advances in technology.

While TA has been highly influential, particularly for establishing environmental impact assessments and other forms of risk analysis, Palm and Hansson ( 2006 ) claim ethical and social dimensions of emerging technologies have often been neglected (546). They propose the ethical technology assessment (eTA) approach that adjusts development in line with ethical concerns and guides decision-making (551). Their ethical “checklist” contains nine items that, if implicated in an emerging technology, indicate an eTA should be conducted. Examples include if the proposed technology might be expected to affect concepts of “privacy” or “gender minorities and justice” (551). Brey ( 2012 ) states the purpose of eTA is to “provide indicators of negative ethical implications at an early stage of technological development … by confronting projected features of the technology or projected social consequences with ethical concepts and principles” (3–4). Palm and Hansson ( 2006 ) note that current obstacles to this process include fear of the unknown, assumptions about the self-regulation of technological development, and citizens feeling ill-equipped to engage in discussions of technologies that are becoming increasingly complex (547). They describe the result in terms of W.F. Ogburn’s concept of “cultural lag,” where technology, as an instance of material culture, is now released into society before “non-material culture has stabilized its response to it” (547). In other words, social, ethical, legal, religious, and cultural systems have not yet grappled with the implications of technologies before they are unleased on society. Some of these challenges are met by scenario-based approaches to emerging technologies, as demonstrated below.

Scenario Approaches

While many scenario-based approaches to emerging technology ethics overlap methodologically with eTA, there are some features that are worth discussing separately. Brey’s ( 2012 ) account of the techno-ethical scenario approach describes it as ethical assessment that helps policymakers “anticipate ethical controversies regarding emerging technologies” through analyzing hypothetical scenarios (4). He notes a unique feature of the method is that it not only tries to predict what moral issues will arise with the advent of new technologies but also how those very technologies will impact morality and “the way we interpret moral values” (4). Boenink, Swierstra, and Stemerding’s ( 2010 ) framework breaks this process up into three distinct steps: 1) “Sketching the moral landscape,” which provides a baseline narrative from which the introduction of the new technology can be compared; 2) “Generating potential moral controversies” using “New and Emerging Science and Technology” (NEST) ethics, with the aim of predicting realistic ethical arguments and issues regarding emerging technologies; and 3) “Constructing closure by judging plausibility of resolutions,” where arguments and counterarguments are considered in the light of the most likely resolution to the issues raised in step 2 (11–13). The process can draw analogies to existing or historical examples of technological change, and the ethical consequences involved, or construct specific controversies and “alternative futures” (14). The most important step to consider here is the second, of which Brey ( 2012 ) states:

The NEST-ethics approach performs three tasks. First, it identifies promises and expectations concerning a new technology. Second, it identifies critical objections that may be raised against these promises, for example regarding efficiency and effectiveness, as well as many conventionally ethical objections, regarding rights, harms and obligations, just distribution, the good life, and others. Third, it identifies chains of arguments and counter-arguments regarding the positive and negative aspects of the technology, which can be used to anticipate how the moral debate on the new technology may develop. (4–5)

In this way, scenario analysis can consider how technology and ethics change in tandem when new technologies emerge.

Socio-technical scenario approaches are similar to the techno-ethical approach outlined above; however, according to Schick ( 2019 ), they owe their origins to utopian studies and traditional philosophical thought experimentation (261). Claiming they are now used “as a form of moral foresight; an attempt to keep the ethical discourse ahead of the technological curve,” Schick ( 2019 ) suggests the goal of socio-technical speculation is to “guide society toward morally sound decisions regarding emerging technologies” (261). Thus, the scenarios being discussed are deeply embedded in hypothetical future societies. The Collingridge dilemma is also noted as a potential pitfall for this method, which the final technique covered here tries to avoid through engaging anticipatory models of ethics and governance.

Anticipatory Technology Ethics/Governance

It is well recognized that governing emerging technologies is difficult due to uncertainty regarding their impact on human health, the environment, and society. Hester et al. ( 2015 ) suggest one method of addressing this is to develop regulatory systems that rely on “anticipatory ethics and governance, future-oriented responsibility, upstream public engagement and theories of justice” (124). These would be forward-looking and flexible, allowing cautious development of technology instead of enforcing bans or merely being used to impute responsibility for harm after the fact, as is often seen in current legal systems. Noting that existing ethico-legal approaches “tend to be reactive and static,” these authors promote a “future-care oriented responsible innovation” that protects public trust in science and technology (125, 131). Brey ( 2012 ) notes that most anticipatory ethics frameworks apply one of two approaches: restricting discussion to “generic qualities” of technology and their likely ethical ramifications or speculating on possible future devices and their impact on society (2–3). The latter relies on future studies and forecasting techniques to allow ethical reflection on technologies that are yet to materialize. When discussing the European Commission’s Ethical Issues of Emerging ICT Applications (ETICA) approach, Brey ( 2012 ) claims multiple such techniques were used in the aggregate in an attempt to circumvent any individual weaknesses in methodology (5). However, his own theory of “anticipatory technology ethics” (ATE) tries to overcome the limited capability of forecasting by separating ethical evaluation into three levels: “the technology, artifact and application level” (7). Technologies are considered collections of techniques with a common function, and thus the technology level of ATE just focuses on what the technology is, and the general ethical concerns arising from this. At the artifact level, the “functional artifacts, systems and procedures” developed from the technology of interest are ethically evaluated (8). Brey ( 2012 ) provides the example of nuclear technology yielding such artifacts as nuclear reactors , x-ray imaging , and bombs . The artifact level of ATE thus considers what a technology is likely to bring into being and the relevant consequences of this. The application level then focuses on the use and purpose of artifacts in practice. The latter two levels of ATE are included in Brey’s “ responsibility assignment stage ,” where moral actors are assigned responsibility for the impact of emerging technologies (12).

Other variations on ATE can be found in Nestor and Wilson’s ( 2020 ) anticipatory practical ethics methodology incorporating stakeholder analysis and intuitionism, which allows for ethical consideration of not just future technologies but also future stakeholders, for example, children produced using CRISPR technology (134). These authors distinguish between anticipatory ethics, where ethical theories are applied to novel situations impacting various stakeholders with the goal of providing policy recommendations, and anticipatory governance, which develops policies in line with predictions regarding human behaviour. They claim the two can be combined to produce “future-oriented legal analysis based on theories of justice for rapidly emerging technologies” (134). They suggest such an analysis should include 1) specific ethical principles, including common sense intuitions; 2) “intermediate” principles, such as harm minimisation, utility, justice, etc.; 3) normative ethical theories, such as consequentialism, deontology, social contract theory, etc.; 4) relevant professional ethics codes, e.g. medical ethics; and 5) “the possibility of emergent ethical principles arising due to the uniqueness and rapid pace of development of new technologies” (137). For Nestor and Wilson ( 2020 ), these are all considered legitimate sources for ethical decision-making and can be used in conjunction with stakeholder analysis to produce ethical guidance and policy recommendations (139).

Anticipatory ethical systems are also subject to criticisms, including that because they speculate on future technologies they might waste time conducting analyses on things that never come to pass. Schick ( 2019 ) also claims it is often unclear what constitutes success in anticipatory ethics, as the goal of settling all ethical concerns and establishing appropriate regulatory systems before a technology is released may be unrealistic (265). Further, in their attempt to pre-empt future applications of new technologies, Schick ( 2019 ) claims speculative ethical models may miss crucial stages in the process, as demonstrated by the example of genetic engineering:

the mainstream bioethics discourse on human genetic engineering (i.e. primarily in the US and the UK) was not indexed to the current state of science or slightly ahead of it, but instead took up questions entangled with more distant anticipated future developments. Keeping the discourse well ahead of the curve of emerging biomedical technologies probably generated interesting discussions, but it may also have contributed to the weakness of the consensus-based norms that were thought to be keeping human germline genetic engineering in check. In effect, the forward-looking discourse subjected them to what might be called “anticipatory obsolescence” by asking whether to maintain a distinction between somatic and germline therapies—long before there was a technique up to the task of altering the genome of a human embryo with sufficient efficacy to begin considering preclinical human embryonic interventions. (264)

Once human embryonic gene editing became possible, Schick ( 2019 ) claims “the newly urgent question of whether germline interventions were ethically permissible was no longer where the discussion was centered” as speculations regarding human enhancement had started to dominant bioethical debate on the subject (264). Schick ( 2019 ) continues: “[i]n retrospect, it seems almost inevitable that once germline engineering was accomplished, the ‘old’ question of whether it should be undertaken at all would suddenly become obsolete” (264). Thus, there is a risk that by focusing too much on future applications, ethicists will miss the opportunity to intervene in foundational stages of technological revolution.

While anticipatory ethics and governance systems are becoming a popular way of dealing with the uncertain risks of emerging technologies, Mittelstadt, Stahl, and Fairweather ( 2015 ) claim such prophetic decision-making aids “cannot be given the same epistemic status as facts and norms concerning existing phenomena” (1044). They note some technologies are so novel even the most basic risk data is unavailable when decisions need to be made about their development. This applies to several of the emerging technologies under discussion in this symposium issue.

The Ethical, Legal and Social Implications of Emerging Technologies (ELSIET) Symposium

The Ethical, Legal and Social Implications of Emerging Technologies (ELSIET) research group was established with support from Deakin University’s Science and Society Network in 2018. Over the next two years the group recruited forty members from eighteen academic institutions in six different countries and hosted three seminars focused on the ethics of emerging technologies. This special issue highlights some of the work arising from these meetings. The purpose of the group is to foster collaborations among specialists working in emerging technologies, including ethicists, scientists, lawyers, and artists. The group went on hiatus at the beginning of the COVID-19 pandemic but has resumed regular activities in 2022 under the auspices of the Iverson Health Innovation Research Institute, Swinburne University of Technology. In 2019, ELSIET was awarded a Brocher Foundation symposium grant in conjunction with members of the University of Melbourne’s School of Population and Global Health, Western Australia’s Department of Health, and the Gen(e)quality Network. Originally planned for 2020, the symposium was rescheduled to May 2022, with an online version occurring in May 2021. Footnote 1

The papers included in this symposium issue address emerging technologies and situations that would trigger Palm and Hansson’s ( 2006 ) ethical checklist, as they pertain to “dissemination and use of information” and “privacy,” particularly for genetic information, “human reproduction” in the form of artificial womb technology, and “impact on human values,” with particular focus on the potential commodification of human DNA. Each paper also engages with one or more of the practices outlined above for ethically evaluating emerging technologies.

The collection begins with Wise and Borry’s ( 2022 ) discussion of the ethical issues surrounding the use of CRISPR-based technologies for eliminating Anopheles gambiae mosquitoes, the dominant vector for malaria throughout sub-Saharan Africa. These authors consider ethical debates regarding whether the species possesses any intrinsic worth, moral status, or instrumental value in terms of increasing biodiversity. The significance of the CRISPR-based technologies under debate relate to the new-found ability to modify the genes and eventually eradicate this entire species of mosquitoes, rather than just eliminating some of them. The competing demands of minimizing human suffering and avoiding unintended side effects to natural ecosystems are recognized throughout. This paper considers the utility of the PP in addressing these ethical issues, as well as the environmental and risk assessment elements intrinsic to TA.

The second paper, by Ferreira ( 2022 ), considers the ethical implications of artificial womb technologies through the lens of utopian fiction, namely Helen Sedgwick’s The Growing Season (2017) and Rebecca Ann Smith’s Baby X (2016). Viewed as feminist rewritings of Aldous Huxley’s dystopian classic Brave New World (1932), these texts consider the emancipatory potential of ectogenesis for women. For Palm and Hansson ( 2006 ), advances in reproductive technologies represent the site of some of “the most blatant clashes” between “social norms and moral values” in society, influencing perceptions of family and human reproduction (553). The use of utopian fiction to guide ethical evaluation aligns with various elements of the socio-technical scenario approach to emerging technologies.

The third paper, by Koplin, Skeggs, and Gyngell ( 2022 ), similarly falls into one of Palm and Hansson’s ( 2006 ) key criteria for eTA, as these authors propose allowing a commercial market for the sale and purchase of human DNA. For Palm and Hansson ( 2006 ), such a proposal would require ethical evaluation to prevent the “negative consequences of commodification” leading to “reduced respect for human personhood” (554–555). Koplin, Skeggs, and Gyngell ( 2022 ) anticipate these objections when outlining how an ethical market in human DNA might be created, considering related concerns regarding exploitation and undue inducement. This analysis includes various stages of the techno-ethical scenario approach, particularly the sketching of the current moral landscape of gene banking, and exploration of arguments and counterarguments to the hypotheticals presented.

The fourth paper, by Delgado et al. ( 2022 ), provides a scoping review of academic literature focused on biases in artificial intelligence algorithms for predicting COVID-19 risk, triaging, and contact tracing. These authors identify issues with data collection, management, and privacy, as well as a lack of regulation for the use of these programmes as key practical and ethical concerns. With their focus on the impacts of these biases and the social determinants of health on various reported health disparities, these authors highlight a role for Brey’s ( 2012 ) ATE framework, which considers the social application of emerging technologies, and Hester et al.’s ( 2015 ) anticipatory ethics and governance.

The final paper in the collection is Benston’s ( 2022 ) protocol for developing policy recommendations regarding heritable gene editing. In this, potential benefits and harms are identified and evaluated in a way that guides the proposed study design. The focus on anticipatory ethics and governance incorporates several elements present in Nestor and Wilson’s ( 2020 ) anticipatory practical ethics methodology, particularly Benston’s focus on detailed stakeholder analysis.

Technological developments involve uncertainty and carry with them the potential for both significant benefit and harm. While we cannot know the future, various methods for ethically evaluating and regulating emerging technologies have arisen that aim to promote discovery while protecting safety. The more revolutionary a new technology is, the greater its potential impact on society and thus the ethical issues it might generate. The articles in this symposium issue all take a proactive, rather than reactive, approach to discussing such issues in advance of these technologies being fully realized in society.

Talks are available on demand here: https://mspgh.unimelb.edu.au/centres-institutes/centre-for-health-policy/research-group/evaluation-implementation-science/elsi-genomics-symposium .

Beauchamp, T.L., and J.F. Childress, 2001. Principles of biomedical ethics , 5th ed. Oxford: Oxford University Press.

Google Scholar  

Benston, S. 2022. Walking a fine germline: Synthesizing public opinion and legal precedent to develop policy recommendations for heritable gene-editing. Journal of Bioethical Inquiry 19(3): doi: https://doi.org/10.1007/s11673-022-10186-8

Boenink, M., T. Swierstra, and D. Stemerding. 2010. Anticipating the interaction between technology and morality: A scenario study of experimenting with humans in bionanotechnology. Studies in Ethics, Law, and Technology 4(2): 1–41.

Article   Google Scholar  

Bouchaut, B., and L. Asveld. 2021. Responsible learning about risks arising from emerging biotechnologies. Science and Engineering Ethics 27(2): 22.

Brey, P.A.E. 2012. Anticipatory ethics for emerging technologies. NanoEthics 6(1): 1-13.

Delgado et al. 2022. Bias in algorithms of AI systems developed for COVID-19: A scoping review. Journal of Bioethical Inquiry 19(3)

Edwards, J. 1991. New conceptions: Biosocial innovations and the family. Journal of Marriage and Family 53(2): 349–360.

Ferreira, A. 2022. The (un)ethical womb: The promises and perils of artificial gestation. Journal of Bioethical Inquiry 19(3): doi: https://doi.org/10.1007/s11673-022-10184-w

Friedman, B., M. Harbers, D. Hendry, J. van den Hoven, C. Jonker and N. Logler 2021. Eight grand challenges for value sensitive design from the 2016 Lorentz workshop. Ethics and Information Technology 23(1): 5–16.

Gouvea, R., J. Linton, M. Montoya, and S. Walsh. 2012. Emerging technologies and ethics: A race-to-the-bottom or the top? Journal of Business Ethics 109(4): 553–567.

Grunwald, A. 2020. The objects of technology assessment. Hermeneutic extension of consequentialist reasoning. Journal of Responsible Innovation 7(1): 96–112.

Grunwald, A. 2019. Technology assessment in practice and theory . Abingdon, Oxon: Routledge.

Guida, A. 2021. The precautionary principle and genetically modified organisms: A bone of contention between European institutions and member states. Journal of Law and the Biosciences 8(1): lsab012.

Hansson, S.O. 2020. How extreme is the precautionary principle? NanoEthics 14(3): 245–257.

Hester, K., M. Mullins, F. Murphy, and S. Tofail. 2015. Anticipatory ethics and governance (AEG): Towards a future care orientation around nanotechnology. NanoEthics 9(2): 123–136.

Koplin, J., J. Skeggs, and C. Gyngell. 2022. Ethics of buying DNA. Journal of Bioethical Inquiry 19(3): https://doi.org/10.xxx

Mittelstadt, B. D., Stahl, B. C., and Fairweather, N. B. 2015. How to shape a better future? Epistemic difficulties for ethical assessment and anticipatory governance of emerging technologies. Ethical Theory and Moral Practice 18: 1027–47. https://doi.org/10.1007/s10677-015-9582-8

Moor, J.H. 2005. Why we need better ethics for emerging technologies. Ethics and Information Technology 7(3):111–119.

Mumford, E. 1993. The participation of users in systems design: An account of the origin, evolution, and use of the ETHICS method. Florida: CRC Press.

Nestor, M.W., and R.L. Wilson. 2020. Beyond Mendelian genetics: Anticipatory biomedical ethics and policy implications for the use of CRISPR together with gene drive in humans. Journal of Bioethical Inquiry 17(1): 133–144.

Palm, E., and S.O. Hansson. 2006. The case for ethical technology assessment (eTA). Technological Forecasting and Social Change 73(5): 543–558.

Schick, A. 2019. What counts as “success” in speculative and anticipatory ethics? Lessons from the advent of germline gene editing. NanoEthics 13(3): 261–267.

Umbrello, S., and I. van de Poel. 2021. Mapping value sensitive design onto AI for social good principles. AI and Ethics 1(3): 283–296.

Wareham, C., and C. Nardini. 2015. Policy on synthetic biology: Deliberation, probability, and the precautionary paradox. Bioethics 29(2): 118-25.

Wise, I.J. and P. Borry. 2022. An ethical overview of the CRISPR-based elimination of Anopheles gambiae to combat malaria. Journal of Bioethical Inquiry 19(3): doi: https://doi.org/10.1007/s11673-022-10172-0

Wolff, J. 2014. The precautionary attitude: Asking preliminary questions. Hastings Center Report 44(S5): S27–S28.

Download references

Funding source

Deakin University Science and Society Network seed grant

Author information

Authors and affiliations.

School of Health Sciences and Biostatistics, Swinburne University of Technology, John St, Hawthorn, Victoria, Australia

Evie Kendal

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Evie Kendal .

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Kendal, E. Ethical, Legal and Social Implications of Emerging Technology (ELSIET) Symposium. Bioethical Inquiry 19 , 363–370 (2022). https://doi.org/10.1007/s11673-022-10197-5

Download citation

Received : 02 May 2022

Accepted : 17 May 2022

Published : 24 June 2022

Issue Date : September 2022

DOI : https://doi.org/10.1007/s11673-022-10197-5

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Assisted reproductive technology
• Biotechnology
• Emerging technology
• Anticipatory ethics
• Precautionary Principle
• Find a journal
• Publish with us
• Track your research