Anti-reflection coating

An anti-reflection coating (also called an anti-reflective, antireflection or AR coating) is a surface treatment on a lens, window or display that reduces the share of light reflected at the boundary between two media, so that more light is transmitted. Conventional anti-reflection coatings are thin films whose thickness and refractive index are chosen so that light reflected from the top and bottom of each layer interferes destructively. A second family replaces the film with a surface pattern smaller than the wavelength of light, the so-called moth-eye structure, which gives a gradual change in refractive index between air and the substrate.

In virtual reality (VR) and augmented reality (AR) hardware, unwanted reflections between closely spaced optical surfaces produce ghost images and stray light that lower image contrast. Anti-reflection coatings and nanostructured anti-reflective surfaces are used or proposed on the polarization optics of pancake lenses, on AR waveguides, on prescription lens inserts for headsets and on the lenses of smart glasses.

Reviewed 13 October 2026. Checked every cited source: the Hedayati and Elbahri review, HyperPhysics, Zeiss, Fraunhofer IOF and Tydex pages, the Luo, Ding, Peng, Tan, Strong and Clapham and Hutley paper metadata and abstracts, the Meta, Corning and Google patents, Guttag's G1 article and Meta's Zenni and Ray-Ban Meta pages. About review dates.

How it works

Whenever light crosses a boundary between materials with different refractive indices, part of it is reflected; the Fresnel equations give how much. For ordinary optical glass the loss is about 4 percent at each uncoated surface, according to Zeiss, and it rises above 6 percent for high-index glass. The Fraunhofer Institute for Applied Optics and Precision Engineering (IOF) put the figure for plastic lenses, with a refractive index of about 1.5, at roughly 8 percent of incoming light. In a system with many surfaces these losses add up, and the reflected light does not simply vanish: it can travel back through the optics and reach the eye as a ghost or a veil of stray light.

Single-layer coatings

A single thin film creates two reflected waves, one from the air-to-coating boundary and one from the coating-to-substrate boundary. If the film has a lower index than the substrate and an optical thickness of a quarter wavelength, the two reflections are 180 degrees out of phase and partly or fully cancel. Complete cancellation at one wavelength requires a coating index equal to the square root of the product of the substrate index and the surrounding medium's index, and a physical thickness of the wavelength divided by four times the coating index.

These conditions explain the limits of a single layer. It is non-reflective at only one wavelength, usually chosen in the middle of the visible spectrum near 550 nm, and its performance falls off at oblique angles of incidence because the optical path through the film changes. For glass with an index of 1.5 the ideal coating index is about 1.22, and Hedayati and Elbahri note that no natural material has that value. Magnesium fluoride is one such compromise material: the optics manufacturer Tydex states that a quarter-wave magnesium fluoride film lowers the reflection of K8 glass (index 1.52) from 4.1 percent to 1.2 percent. Porous or composite single layers can reach lower effective indices; the 2016 review cites a nanoimprinted sol-gel layer of silica nano-cylinders with an effective index of 1.20 that cut the reflection of a glass sheet coated on both sides from 7.35 percent to 0.57 percent.

Multilayer coatings

Stacks of two or more dielectric layers remove reflection over a wider band. In a two-layer (V-shaped) design the layer facing air has the lowest index, and layer thicknesses are usually a quarter or half of the design wavelength. Broadband multilayer coatings alternate high-index and low-index materials; titanium dioxide and silicon dioxide are the most used pair. Their reflection depends strongly on every layer's thickness and index, so designs are found by numerical optimization, and more layers bring adhesion, mechanical stability and cost problems. Tydex states that broadband visible and near-infrared coatings use three to six layers and that its standard V-coat leaves less than 0.25 percent residual reflection per surface. A Google patent application on AR displays describes traditional anti-reflective coatings as alternating high-index and low-index metal oxide layers, some with up to seven layers, and calls them expensive or time consuming to apply.

Graded-index and moth-eye surfaces

A surface textured with features smaller than the wavelength of the incoming light behaves like a medium whose refractive index changes gradually with depth, instead of a sharp step. Features equal to or larger than the wavelength mostly scatter light. The best-known example comes from biology. Using electron microscopy, Bernhard found that the corneal lenses of moths are covered with an ordered array of conical bumps, typically about 200 nm high and 200 nm apart. P. B. Clapham and M. C. Hutley reproduced the structure on glass in 1973, reducing its reflection from 5.5 percent to 0.2 percent, although their surface was too delicate for practical use.

Moth-eye surfaces can also be made by imprinting. In 2017 Guanjun Tan, Shin-Tson Wu and colleagues described a broadband moth-eye-like anti-reflection film imprinted onto a flexible substrate for sunlight-readable displays; it tolerated a bending radius of 8 mm, and a fluoroalkyl top coating gave it a water contact angle above 100 degrees to resist dust and fingerprints. In October 2016 the Fraunhofer IOF in Jena announced an antireflection system developed by its researchers that combines several nanostructured layers with conventional homogeneous oxide layers and lowers the refractive index at the surface of plastic optics to almost 1.1. The announcement quoted Ulrike Schulz of the institute's plastic optics department and named virtual reality among the fields that could use the coating.

History

According to the 2016 review by Hedayati and Elbahri, Lord Rayleigh proposed in 1879 that reflection from a surface could be reduced if the change in refractive index between the object and its surroundings was small. The English lens designer H. Dennis Taylor noticed in 1892 that tarnished camera lenses allowed photographs with shorter exposures and linked the effect to the tarnish layer, and F. E. Wright tarnished glass deliberately with chemicals and explained the anti-glare effect as a gradual index transition. The same review credits Fraunhofer with an early observation that etching a surface in acid vapors decreased its reflection.

Zeiss states that its engineer Alexander Smakula developed an anti-reflective coating in 1935 that raised the light transmission of binoculars by 50 percent, with the patent allowed the same year. This single-layer coating was known as the "T" coating; Zeiss replaced it with the multilayer T* coating at the end of the 1970s. Zeiss dates the commercial launch of its first anti-reflective coating for spectacle lenses to 1959. John Strong published "On a Method of Decreasing the Reflection from Nonmetallic Substances" in the Journal of the Optical Society of America in January 1936. The moth-eye approach followed Bernhard's biological work and the 1973 experiments of Clapham and Hutley described above.

Applications in VR and AR

Pancake lenses

Pancake optics fold the light path between a half mirror, a reflective polarizer and quarter-wave plates, which makes headsets thinner but limits optical efficiency to a theoretical maximum of 25 percent. Because the design relies on several closely spaced surfaces, it also suffers from ghost images caused by stray light from multiple surface reflections and imperfect polarization control. A 2024 study in Optics Express by Zhenyi Luo, Yuqiang Ding, Qian Yang and Shin-Tson Wu traced the light paths of these ghosts experimentally and by simulation and measured how the intensity of each ghost changes with the angle of incidence. In a related 2024 paper on a pancake design that replaces the half mirror with a commercial Faraday rotator and reflective polarizers, Ding, Luo and colleagues evaluated ghost images with a micro-OLED panel and reported that they "can be suppressed to undetectable level if the optics are with antireflection coating."

Patent filings also treat coating reflection as a source of ghosts. A Meta (Facebook Technologies) patent titled "Pancake lens ghosting mitigation", filed in February 2019 and granted in January 2021, analyzes three ghost paths in a pancake assembly and states that all three are formed by reflection from the anti-reflection coating on the quarter-wave plate of the circular polarizer between the display and the lens. In its example the first two ghosts carry 0.5 times that coating's reflection coefficient relative to the main image. Instead of relying only on the coating, the patent positions the waveplate in the gap between the display and the lens, closer to the lens than to the display, so that the ghosts are defocused, and one claim requires the coating's reflectivities in the red, green and blue channels to differ by less than 50 percent of its reflectivity in green. A 2020 paper by Xu Peng and colleagues in Virtual Reality & Intelligent Hardware took a similar optical-design route for polarized catadioptric VR optics, adjusting the focal power of the three main ghost paths so that the ghost spot area grew by 40 percent and the stray-light energy ratio fell by 40 percent.

Headset optics and displays

A patent application filed by Corning in April 2018 describes VR and AR headsets in which stray light reflects from the surfaces of multiple air-spaced optical components and reaches the viewer's eyes. It proposes a stray-light-reducing nanostructured surface, with features wider than 1 nm and narrower than 1 micron, on the lens, the display or both. The application was published in November 2018 and is listed as abandoned.

AR waveguides

According to a Google patent application published in October 2024, the structures that couple and expand light within an AR display waveguide typically have highly reflective surfaces made of high-index polymer, and this can cause "rear view or cosmetic artifacts." The application proposes moth-eye nanopillars imprinted in the non-grating regions of the waveguide in the same nanoimprint step that transfers the other diffractive optical elements to the glass substrate. A separate Google patent, filed in June 2022 and granted in April 2025 as US 12,276,792 B2, notes that, because of problems with designing gratings and coatings together, either the see-through performance of the display or the coating performance is often sacrificed. It places a four-layer anti-reflection coating in its own area of the waveguide, formed during the same fabrication used for the incoupler, outcoupler and exit pupil expander.

Air gaps between a waveguide and a vision-correction lens are another source of reflections. In his 2024 analysis of the Even Realities G1, display analyst Karl Guttag wrote that bonding the prescription lens optically to the waveguide "eliminates reflection from air gaps between the vision correction lens and the waveguide," while snap-in inserts "will be more prone to reflections between the waveguide and inserts."

Prescription inserts and smart glasses

Prescription lens inserts for headsets can include anti-reflection coatings. Meta's support page for the Zenni inserts made for Meta Quest 3, Meta Quest 3S and Meta Quest 2 states that the lenses come "with anti-reflective and anti-scratch coatings included," with oil-resistant and anti-fog coatings as options. For Ray-Ban Meta smart glasses, Meta said at launch that prescription lenses would be available in a range of materials and treatments, including Crizal anti-reflective coating.

Research

Hedayati and Elbahri write that devices such as displays, touchscreens and very compact cameras have led the optics market to require three properties from anti-reflection coatings: ultrathin construction, insensitivity to the angle of incidence, and broadband performance. Conventional stacks are strongly angle-sensitive because they depend on light circulating many times inside the cavities formed by the films. Their review describes plasmonic and metasurface designs as a way to make broadband, angle-insensitive coatings about an order of magnitude thinner than the operating wavelength. In the VR studies and patents cited above, the reflectance of coated surfaces in folded optics is treated as one of the factors that set the brightness of ghost images.

See also

• Pancake lenses
• Ghost image
• Stray light
• Polarizer
• Optical aberration
• Waveguide
• Prescription lens inserts
• Metasurface