← Back to all notes
R

Roel Huis in ‘t Veld

Director, Kriya Materials · Week of November 2, 2026

Wolfspeed wants to sell you a waveguide substrate that reflects 20% per face

Wolfspeed put out a press release on 29 October saying its transparent silicon carbide, TSiC, is now commercially available for AR waveguides, with 200 mm optical-grade substrates on offer and 300 mm in development. My view: real material, real physics advantage, and the hard part starts at the surface, which the release does not mention once. The physics first. Silicon carbide has a refractive index around 2.6 in the visible, against 1.7 to 2.0 for the best waveguide glasses. A higher index lets a diffractive waveguide carry a wider range of angles by total internal reflection, so you get more field of view out of one thin plate instead of stacking two or three. Barry Silverstein of the University of Rochester, formerly at Meta Reality Labs, is quoted in the release saying the high index lets a single-layer waveguide do the work that previously took stacked optics. That is true, and it is why Meta went to SiC for Orion. The other side of the same number. Fresnel reflection at a surface is ((n minus 1) over (n plus 1)) squared. For n of 2.6 that is about 20% per face, and a plate has two. An uncoated SiC waveguide loses roughly a third of the see-through light to reflection before any grating does anything, and every reflection is a potential ghost image or a spot of eye glow for the person you are talking to. On glass of index 1.9 the same figure is under 10%. So the anti-reflection coating on a SiC waveguide is not a finishing touch. It is part of the optical design, it has to work across red, green and blue at the angles a waveguide uses, and it has to be applied over or next to nanostructured gratings without filling them in. The release claims low absorption, superior flatness and low birefringence, with no number for any of them. Semiconductor Today's report of 30 October carries the same claims and the same absence of numbers: no absorption spec, no index value, no price, no shipping volumes. Karl Guttag's year-in-review piece of 23 October lists transparency problems, rainbow effects, eye glow and slanted-grating artifacts as the things still being chased in waveguides, and his line that if you solve one, roughly 17 others remain, applies with full force here. A substrate vendor solves field of view. The coatings, the grating etch into a very hard crystal, colour uniformity and cost per wafer remain for someone else. What customers will ask for next is not a different substrate but a surface stack: a durable broadband AR on a high-index crystal, an index-matched layer where the grating meets air, and a cladding of defined low index to control which modes leak. Those are coating questions, and they get harder, not easier, as the index of what is underneath goes up. A higher index buys you angles on the inside and costs you reflections on the outside; the physics does not let you keep only the first half. Sources Business Wire (Wolfspeed press release), “Wolfspeed Commercially Introduces Transparent Silicon Carbide Technology for Augmented Reality and Quantum Computing”, 29 October 2026. https://www.businesswire.com/news/home/20261029500897/en/ Semiconductor Today, “Wolfspeed commercially introduces transparent silicon carbide technology for AR and quantum computing”, 30 October 2026. https://semiconductor-today.com/news_items/2026/oct/wolfspeed-301026.shtml KGOnTech (Karl Guttag), “AR in 2026 Year In Review Presentation (Part 1: AR Challenges)”, 23 October 2026. https://kguttag.com/2026/10/23/ar-in-2026-year-in-review-presentation-part-1-ar-challenges/