Jump to content

Diffractive optics

From VR & AR Wiki

Diffractive optics is the branch of optics that controls light through diffraction rather than through refraction by curved lenses or reflection by mirrors. Its components are called diffractive optical elements (DOEs); the simplest examples are diffraction gratings, and the term is most often used for thin micro-optic plates that impose a designed pattern of phase changes on an incoming beam.[1] In their 2021 review of AR and VR displays, Xiong, Hsiang, He, Zhan and Wu note that "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", and they divide the field into two major types: holographic optical elements (HOEs) made by recording a wavefront, and surface relief gratings (SRGs) written by lithography.[2]

Diffractive optics is central to augmented reality hardware. Gratings are the in-couplers and out-couplers of diffractive waveguide combiners, and in a 2022 review written from Microsoft's HoloLens team, Bernard Kress and Maria Pace describe how surface relief gratings were applied to the waveguide combiners of the Microsoft HoloLens and the Magic Leap One.[3] DOEs also split an infrared laser beam into the dot patterns used for structured-light depth sensing, as in the first Kinect and Apple's Face ID.[4][5] A key limitation of diffractive elements is that their behavior depends strongly on wavelength, which shows up in AR glasses as color non-uniformity and "rainbow" artifacts.[1][2]

Reviewed 27 September 2026. Checked every citation against its source (Xiong 2021 and Kress and Pace 2022 full texts, RP Photonics, Veldkamp 1989 and Levola 2006 abstracts, Nobel summary, Meta Orion and research blogs, Butler 2012, Hsu 2025, Maimone 2020, Xia 2022, KGOnTech) for all figures, dates, quotes and attributions About review dates.

How it works

A diffractive element works by the phase differences it introduces across a beam, and because those phase differences depend on the optical wavelength, the performance of a DOE is typically wavelength dependent. That is often not a problem with narrowband laser light, but a more general difficulty is that a substantial part of the optical power can remain in the undiffracted zero order, where it is not usable.[1] A periodic diffraction grating sends light into discrete orders; a Fresnel zone plate is a grating with circular instead of straight lines and can focus light, so it can act as a diffractive lens.[1]

Micro-optic DOEs are usually thin plates. Binary elements have only two phase levels across the surface, while others approximate an "analog" phase profile with a series of discrete height steps, each step typically equal to the vacuum wavelength divided by the refractive index difference between the element and the surrounding medium.[1] The number of levels sets the achievable efficiency. In the scalar-theory figures given by the RP Photonics Encyclopedia for an ideal quantized blaze:[1]

Phase levels Diffraction efficiency into the design order
2 (binary, 50% duty cycle) About 40.5% into each first order (about 81% if both symmetric first orders are used)
4 About 81%
8 About 95%
16 About 98.7%
Continuous profile Approaches 100%

Kress and Pace point out two costs of replacing curved optics with flat diffractive ones: flat diffractive lenses often have much worse aberrations than their curved refractive counterparts, and their limited efficiency can sharply reduce the wall-plug efficiency of a head-mounted display.[3] Xiong et al. make the same point about color: replacing the refractive elements of a VR eyepiece with a single thin diffractive lens is "tempting", but "the diffractive nature of such a lens will result in serious color aberrations".[2]

Surface relief and volume elements

Surface relief gratings carry corrugated microstructures on their surface, which can be binary, blazed, slanted or analog in shape; the geometry is chosen according to the target diffraction order and the required spectral and angular bandwidth.[2] Volume elements instead record a hologram as a modulation of refractive index inside a layer, either in photopolymer or in liquid crystal. Photopolymer HOEs have a small refractive index modulation and are therefore strongly selective in wavelength and angle, and several holograms can be recorded into one film. Liquid-crystal HOEs, including the polarization volume gratings used as waveguide couplers, have a much larger index modulation (equal to the liquid crystal's birefringence, 0.04 to 0.5) and respond differently to left- and right-handed circular polarization.[2] Xiong et al. give simulated bandwidths for these recorded elements at 550 nm:[2]

Element Angular bandwidth Spectral bandwidth
Transmissive photopolymer grating About 5 degrees (efficiency above 80%) About 175 nm (7 degree incidence)
Reflective photopolymer grating Not stated About 10 nm
Transmissive liquid-crystal grating About 20 degrees About 300 nm (7 degree incidence)
Reflective liquid-crystal grating 15 to 50 degrees, depending on wavelength About 80 nm

Optical metasurfaces extend the same flat-optics approach with arrays of engineered "meta-atoms". Xiong et al. describe them as reaching "beyond diffractive and refractive optics", with functions such as achromatic focusing and beam steering, but note that a geometric-phase metalens "is intrinsically a diffractive lens that also suffers from strong chromatic aberrations", and that metasurfaces in existing AR and VR prototypes still trade off numerical aperture, aberrations, efficiency, aperture size and fabrication complexity.[2]

Fabrication

Surface relief gratings for waveguide couplers in the visible spectrum have periods below 500 nm, and multilevel versions need features of only a few tens of nanometers. Masters are made by direct electron-beam writing, i-line or deep-UV lithography, or interference lithography, and the gratings are then replicated in volume by nanoimprint lithography, a wafer process originally developed for the integrated-circuit industry.[3] Kress and Pace write that slanted-grating nanoimprint with slants up to 50 degrees "has been mastered by many foundries around the world"; Xiong et al. cite a report that SRGs 300 nm high with slant angles up to 50 degrees can be replicated with high yield.[3][2] Volume holograms are instead produced by optical exposure: thin photopolymer holograms by contact printing roll to roll, and thick multiplexed photopolymer holograms by multiple exposures.[3]

History

Kress and Pace note that spectroscopy "was the first and still it is the main application pool for gratings and holograms".[3] Holographic optical elements are built on holography; Dennis Gabor received the 1971 Nobel Prize in Physics "for his invention and development of the holographic method".[6]

By the late 1980s, computer-designed, lithographically etched DOEs were being developed under the name "binary optics". At the 1989 Conference on Lasers and Electro-Optics, Wilfrid Veldkamp and Gary Swanson of MIT Lincoln Laboratory described binary optics as a diffractive technology that uses computer-generated designs of microscopic surface relief patterns and VLSI etching; a single etching step produces a two-level relief, which gave the technique its name. They listed lenses, prisms, beam splitters, beam multiplexers, filters and microlenses among the efficient elements already made.[7] According to Kress and Pace, surface relief gratings became a commodity technology once mastering and mass-replication techniques were established in the early 1990s.[3]

For near-eye displays, Tapani Levola's 2006 paper "Diffractive optics for virtual reality displays" studied how DOEs on planar waveguides could miniaturize the optics of portable near-to-eye displays, where low power consumption, low weight and small size are the main requirements.[8] Diffractive waveguides later reached mixed reality headsets: Kress and Pace illustrate the grating architectures of the Microsoft HoloLens 1, which they date to 2015, and the Magic Leap One of 2018. They also note that WaveOptics, which used quasi-analog surface relief computer-generated holograms, is now part of Snap Inc.[3] In 2025 Meta described the silicon carbide diffractive waveguides of its Orion AR glasses prototype, whose gratings are slant-etched directly into the high-index material.[9]

Applications in AR

Waveguide couplers

In a diffractive waveguide, an input grating couples light from a small projector into a glass plate at an angle steep enough for total internal reflection. An output grating, usually with the same period as the input grating, couples it out toward the eye a little at a time; the output coupler can also be an off-axis lens with slight curvature so that the image appears at a finite distance.[2] Xiong et al. identify three main diffractive coupler types: SRGs, photopolymer gratings, and liquid-crystal polarization volume gratings. The in-coupler should be efficient, while the out-coupler is usually made weak and locally modulated so that light leaves the plate evenly across the eye box.[2] Replicating the output across a grid of positions is called exit pupil expansion; it can be done with a separate turning grating followed by an out-coupler, or with a single two-dimensional grating such as a hexagonal lattice, which gives the designer more freedom because local grating parameters can vary in two directions.[2]

Color is the recurring problem. Xiong et al. explain that when in-coupler and out-coupler apply opposite grating vectors, their dispersions cancel, so the output image generally shows no color dispersion even with a broadband LED source. The difficulty is instead field of view and uniformity: red light propagates at a larger angle than blue light in the same plate, so two or three layers of waveguide with different grating pitches are usually used.[2] Kress and Pace add that spectral spread matters most with broadband LED illumination, as in the HoloLens 1, Vuzix, Magic Leap, DigiLens and Nokia devices, and name the laser MEMS display engine of the HoloLens 2 as a notable exception.[3]

Xiong et al. summarize typical performance of diffractive waveguide combiners as a field of view of about 50 degrees, an eye box of around 10 mm, and a combiner efficiency of about 50 to 200 nit per lumen. The theoretical upper limit on field of view is set by the waveguide's refractive index.[2] Meta applied this principle in Orion: it states that silicon carbide has a refractive index of 2.7 and that its waveguides give Orion a field of view of approximately 70 degrees with a single plate per lens instead of a multi-plate glass stack. Meta research manager Nihar Mohanty said that the team was the first to "do slant etch directly on the devices", because "the whole industry used to rely on nano imprint, which doesn't work for substrates with such a high refractive index".[9]

Kress and Pace compare the diffractive coupler technologies used in commercial waveguides:[3]

Coupler technology Spectral dispersion Mass production method Companies or products named by Kress and Pace
Slanted surface relief grating Strong Nanoimprint (wafer or plate) Microsoft HoloLens, Vuzix, Nokia
Blazed surface relief grating Strong Nanoimprint (wafer or plate) Magic Leap One (in-couplers)
Binary surface relief grating Strong Nanoimprint (wafer or plate) Magic Leap One (out-couplers)
Multilevel surface relief grating Strong Nanoimprint (wafer or plate) WaveOptics, BAE, Dispelix
Thin photopolymer hologram Strong Contact print, roll to roll Sony, TrueLife Optics
H-PDLC volume hologram Strong Exposure DigiLens
Thick photopolymer hologram Minimal Multiple exposure Akonia (now part of Apple)
Resonant waveguide grating Can be mitigated Roll-to-roll nanoimprint CSEM / Resonant Screens
Metasurface coupler Can be mitigated Nanoimprint (wafer or plate) Metalenz

The authors add that most AR, mixed reality and smart glasses products of the time were based on only a few coupler technologies: thin volume holograms, slanted surface relief gratings, and embedded half-tone mirrors.[3] Waveguides with embedded partial mirrors, such as those of Lumus, are reflective rather than diffractive;[3] Xiong et al. note that such geometric waveguides usually give better image sharpness (MTF) and color uniformity than diffractive ones.[2]

Artifacts: leakage, ghosts and rainbows

Because a grating in front of the eye also diffracts light from the outside world, diffractive waveguides have characteristic see-through artifacts. Xiong et al. list three: light leakage, where guided light diffracts outward toward the world; see-through ghosts, formed by consecutive in-coupling and out-coupling at the out-coupler grating; and the rainbow, "caused by the diffraction of environment light into user's eye". A higher-index substrate or an optimized grating structure reduces the rainbow.[2] Kress and Pace call the world-side leakage "eye glow" and rank waveguide types by it:[3]

Waveguide type World-side leakage (as a share of the eye-side leakage)
Surface relief gratings with binary profiles About 50%
Surface relief gratings with slanted profiles About 40%
Holographic grating couplers About 8-10%
Reflective waveguides 5% without anti-reflection coating, under 1% with it

Xia, Guan, Cai and Magnenat Thalmann call chromatic aberration, or the rainbow effect, "one major drawback" of diffractive waveguide displays, and note that using HOEs instead of DOEs as the in-couplers and out-couplers can eliminate it because of their narrower spectral bandwidth.[10] In a 2018 analysis of the Magic Leap One on his KGOnTech blog, Karl Guttag wrote that on both it and the HoloLens "the exit gratings run horizontally and thus make them susceptible to 'capturing' light from above, such as overhead lights", and estimated that the Magic Leap One blocked about 85% of real-world light compared with about 60% for the HoloLens.[11] Barry Silverstein, Meta's director of research science, gave an everyday example of the rainbow problem: when "driving at night with moving car lights around you", the wearer will "have rainbows that move, too". He said that silicon carbide "gets rid of those rainbows".[9]

Holographic free-space combiners

Diffractive elements are also used as reflective combiners without a waveguide. Xiong et al. note that HOEs are generally preferred over half mirrors in these designs because they allow an off-axis layout that places the combiner where an eyeglass lens would sit and reflect less environment light; their strong chromatic dispersion, however, usually requires a laser source, which leads to a Maxwellian-view (retinal projection) system.[2] Kress and Pace list North Focals, Intel Vaunt and Sony smart glasses as examples of conventional volume holographic optics in head-worn displays.[3]

Applications in VR

Most VR headsets use refractive magnifier lenses, but diffraction still affects them. Xiong et al. note that the grooves of Fresnel lenses cause diffraction artifacts and stray light that degrade image quality, whereas pancake lenses have smooth surfaces and much fewer diffraction artifacts and stray light.[2]

Research groups have built VR optics from diffractive and holographic films. In 2020 the Facebook researchers Andrew Maimone and Junren Wang presented display designs that combine holographic optics, directional backlighting, laser illumination and polarization-based optical folding; using only thin, flat films as optical components, their prototypes were less than 9 mm thick with fields of view over 90 degrees horizontally.[12] Meta described the work as "purely research" and the device as a "proof-of-concept research device".[13] Switchable liquid-crystal diffractive lenses, based on the Pancharatnam-Berry (geometric) phase, can change focus electronically for varifocal designs, but because their diffraction efficiency is usually optimized for a single wavelength they have limited spectral bandwidth and a focal length that varies with color.[2]

Depth sensing

DOEs are a standard way to create the projected patterns used for structured light. In the first Kinect, "light from an IR laser diode passes through a diffractive optical element (DOE) to project a pseudo-random pattern of IR dots into the scene", and the offset between the known pattern and the dots seen by an infrared camera gives depth.[4] Apple's Face ID dot projector, as described by Hsu and colleagues in a 2025 paper, generates more than 32,000 dots with a module that integrates a VCSEL array of about 366 emitters, two lenses, a folding waveguide and a DOE; the lenses magnify the laser emission and the DOE diffracts it into a dense dot pattern.[5] The same paper compares these DOE-based projectors with metasurface-based designs, including Metalenz's Orion metasurface-VCSEL projector, and reports a research chip that integrates a metasurface directly on a single photonic-crystal surface-emitting laser in a volume of 0.025 cubic millimeters.[5]

See also

References

  1. ↑ 1.0 1.1 1.2 1.3 1.4 1.5 Rüdiger Paschotta. "Diffractive optics". RP Photonics Encyclopedia. RP Photonics. doi:10.61835/hpf. https://www.rp-photonics.com/diffractive_optics.html. Retrieved 2026-09-27.
  2. ↑ 2.00 2.01 2.02 2.03 2.04 2.05 2.06 2.07 2.08 2.09 2.10 2.11 2.12 2.13 2.14 2.15 2.16 2.17 Xiong J, Hsiang E-L, He Z, Zhan T, Wu S-T (2021). "Augmented reality and virtual reality displays: emerging technologies and future perspectives". Light: Science & Applications, vol. 10, article 216. doi:10.1038/s41377-021-00658-8. https://www.nature.com/articles/s41377-021-00658-8. Retrieved 2026-09-27.
  3. ↑ 3.00 3.01 3.02 3.03 3.04 3.05 3.06 3.07 3.08 3.09 3.10 3.11 3.12 3.13 Kress B C, Pace M (2022). "Holographic optics in planar optical systems for next generation small form factor mixed reality headsets". Light: Advanced Manufacturing, vol. 3, no. 4, pp. 771-801. doi:10.37188/lam.2022.042. https://www.light-am.com/article/doi/10.37188/lam.2022.042. Retrieved 2026-09-27.
  4. ↑ 4.0 4.1 Butler A, Izadi S, Hilliges O, Molyneaux D, Hodges S, Kim D (2012). "Shake'n'Sense: Reducing Interference for Overlapping Structured Light Depth Cameras". Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '12). ACM. https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/shake27n27Sense.pdf. Retrieved 2026-09-27.
  5. ↑ 5.0 5.1 5.2 Hsu W-C, Miao W-C, Hong Y-H, Kuo H-C, Huang Y-W (2025). "Monolithically Integrated Metasurface on a PCSEL for Depth Perception". Nano Letters, vol. 25, no. 29, pp. 11382-11390. doi:10.1021/acs.nanolett.5c02540. https://pmc.ncbi.nlm.nih.gov/articles/PMC12291584/. Retrieved 2026-09-27.
  6. ↑ "The Nobel Prize in Physics 1971". NobelPrize.org. Nobel Prize Outreach. https://www.nobelprize.org/prizes/physics/1971/summary/. Retrieved 2026-09-27.
  7. ↑ Veldkamp W B, Swanson G (1989). "Binary optics". Conference on Lasers and Electro-Optics (CLEO) 1989. Optica Publishing Group. https://opg.optica.org/abstract.cfm?URI=CLEO-1989-MA4. Retrieved 2026-09-27.
  8. ↑ Levola T (2006). "Diffractive optics for virtual reality displays". Journal of the Society for Information Display, vol. 14, no. 5, pp. 467-475. doi:10.1889/1.2206112. https://doi.org/10.1889/1.2206112. Retrieved 2026-09-27.
  9. ↑ 9.0 9.1 9.2 "Crystal Clear: Our Silicon Carbide Waveguides and the Path to Orion's Large FoV". Meta. Meta Platforms. 2025-03-06. https://www.meta.com/blog/orion-silicon-carbide-waveguides-ar-glasses-large-field-of-view/. Retrieved 2026-09-27.
  10. ↑ Xia X, Guan F Y, Cai Y, Magnenat Thalmann N (2022). "Challenges and Advancements for AR Optical See-Through Near-Eye Displays: A Review". Frontiers in Virtual Reality, vol. 3, article 838237. doi:10.3389/frvir.2022.838237. https://www.frontiersin.org/articles/10.3389/frvir.2022.838237/full. Retrieved 2026-09-27.
  11. ↑ Karl Guttag (2018-09-26). "Magic Leap Review Part 1 - The Terrible View Through Diffraction Gratings". KGOnTech. https://kguttag.com/2018/09/26/magic-leap-review-part-1-the-terrible-view-through-diffraction-gratings/. Retrieved 2026-09-27.
  12. ↑ Maimone A, Wang J (2020). "Holographic optics for thin and lightweight virtual reality". ACM Transactions on Graphics, vol. 39, no. 4. doi:10.1145/3386569.3392416. https://doi.org/10.1145/3386569.3392416. Retrieved 2026-09-27.
  13. ↑ Andrew Maimone, Junren Wang (2020-06-29). "Holographic optics for thin and lightweight virtual reality". Meta Research. Meta Platforms. https://research.facebook.com/blog/2020/06/holographic-optics-for-thin-and-lightweight-virtual-reality/. Retrieved 2026-09-27.