Silicon Carbide Waveguides Pros and Cons

Source: KGOnTech (kguttag.com), Karl Guttag · Published: 2026-09-08

Note: lightly condensed capture — image embeds dropped and some passages abbreviated; conference-promotion, comment and navigation sections removed.

---

Introduction

At Meta Connect in September 2024, Meta demonstrated its Orion prototype, which uses Silicon Carbide (SiC) diffractive waveguides to support a 70-degree FOV. On several occasions, Meta has stated that it hoped Orion’s demonstration of SiC waveguides would encourage further development of the necessary infrastructure and optical-grade SiC, helping to drive down costs. Meta also made the technical case for SiC waveguides at SPIE’s AR/VR/MR conference in January 2025 in their paper “The evolution of display materials and processes for full augmented reality”.

Several China-based companies, including SEEV, Moldnano, LLVision, and Goeroptics, were likely already developing SiC waveguides before the Orion demonstration and are moving toward production. In the US, Coherent has been developing SiC substrates and promoting their use in AR for some time, and Magic Leap has publicly said it is developing SiC waveguides.

As I often say, proponents of a given technology tend to be very vocal about the pros but often leave it to others to speak to the cons. People are often enticed by the pros without considering the cons, i.e., “Too few factors analysis.” In this article, I will try to give a balanced analysis based on what I know so far about SiC waveguides.

At the SEMI Core-Display Conference held on October 29, 2025, Dr. Shi Rui, CTO & Co-founder of SEEV, delivered a keynote speech titled “Mass Production Technology for Silicon Carbide Diffractive Waveguide Chips.” On August 28, 2026, the Nimbo X1 Kickstarter campaign launched. The Nimbo X1 uses a SEEV SiC waveguide (openly admitted by both companies). It was the Nimbo X1 campaign and related YouTube video that inspired me to take a deeper dive into the pros and cons of SiC waveguides.

Moldnano and LLVision Reference Papers

Both Moldnano and LLVision have published informative papers (free to access) on their SiC waveguide work. Moldnano and Westlake University published “SiC diffractive waveguides for augmented reality: single-layer, full-color, rainbow-artifact-free display with vision correction” (SpringerOpen eLight); Moldnano also presented their SiC work at SPIE AR/VR/MR 2026. LLVision’s paper is titled “Single-Layer Full-Color SiC Diffractive Waveguide AR Glasses with Large FOV and Rainbow Effect Suppression” (Photonics, September 2025). I will be citing from these two sources in the rest of this article.

SiC Waveguides Pros (with some Yes-Buts)

The top-cited advantages of SiC are that it has a high refractive index, high thermal conductivity, and toughness against breaking and scratching.

• Optically transparent SiC uses a crystal stacking pattern that repeats every four SiC bilayers with hexagonal symmetry (4H).
• The main advantage cited for using SiC in waveguides is its high index of refraction. Typical waveguide glass has an index of refraction of 1.7 to 2.0, whereas SiC has an index of refraction of about 2.6. In a diffractive waveguide, this high index supports a wider field of view in a thinner waveguide.
• It has been cited by multiple companies, including Meta, Moldnano, and LLVision, that using SiC better supports full color in a single waveguide. Additionally all three companies have stated that the combination of SiC with their grating design greatly reduces the rainbow capture commonly associated with glass diffractive waveguides.
• SiC has very high thermal conductivity — nearly as conductive as copper, or about 200–400 times as thermally conductive as glass. This could spread and radiate heat via the waveguide surface, though the waveguide will likely be surrounded by thermally insulating coatings, films, lenses and dimming structures. SiC is also brittle and must be contained for eye safety, and whatever contains it will likely be thermally insulating.
• SiC is very tough: it will neither scratch, bend, nor break easily — nearly as scratch-resistant as diamond. Moldnano showed that even a thin SiC waveguide can withstand a typical eye safety ball drop test. But SiC is also brittle, which means that if it breaks, it will have very sharp edges. For eye safety, it should be laminated or otherwise contained (much of which will likely happen due to the other layers of material on either side of the waveguide, including lenses).
• SiC is very thermally stable, so very thin waveguides won’t warp or break with thermal cycling — though this might be a problem if it is glued to a less thermally stable material.

SiC Waveguide Drawbacks

The most obvious drawbacks being discussed: the expense of manufacturing raw SiC wafers; manufacturable SiC wafers are typically small, which limits processing throughput (and adds cost); being crystalline as opposed to amorphous glass, it can have defects and thus lower yield; and SiC’s toughness makes it very hard to polish smooth and to etch diffraction gratings. These drawbacks, one way or another, come back to the cost of manufacturing. Meta and others have stated that with volume production costs should come down; while this is true, SiC waveguides will inherently be significantly more expensive than their glass-based counterparts.

While the optical advantages are widely discussed by proponents, they typically fail to mention serious optical drawbacks. The most serious in terms of image quality is color dispersion — the various wavelengths see significantly different indices of refraction — and the crystal structure of SiC makes it highly birefringent. These two factors make diffractive waveguide design highly challenging. I have seen through two SiC waveguides (Moldnano and Goertek) and seen pictures taken through three others, and the color uniformity is extremely poor compared to other waveguides of similar FoVs.

The high refractive index of SiC means inherently worse Fresnel reflections and makes anti-reflective coatings more difficult — there is more likelihood of visible reflections from the front and ghost images due to reflections.

Trend Toward 2-Grating (WaveOptics-Like) Waveguides

Historically, most diffractive waveguides have used three diffraction gratings: the small entrance grating, a (typically triangular) expansion grating and an exit grating. WaveOptics (bought by Snap in 2021) invented a two-grating design — the entrance grating and a second grating that expands the image and redirects the light to exit.

In the last few years the 2-grating architecture (what Meta calls the “butterfly architecture”) seems to be gaining popularity. Perhaps the most popular use of the 2-grating approach is in the Even Realities G1 and G2 glasses, which use SEEV glass-based 2-grating waveguides; the 2-grating waveguide supports the smaller form factor of the G1 and G2. Some newer Vuzix waveguides have also gone to the two-grating approach (seen at CES 2026).

The 2-grating design is much more compact, resulting in a much smaller waveguide, and it eliminates the separate (usually triangular) expansion grating, which is one of the major sources of “rainbow” light capture. However, in my random sampling of various two- and three-grating waveguides over the years, the color uniformity seems worse with 2-grating designs (not necessarily inherent).

Most 2-grating designs also have significantly worse forward-light projection (eye glow) than some of the newer 3-grating designs; the Nimbo X1 with the SEEV waveguide has adopted a pantoscopic (top-to-bottom tilt) that directs the forward-projected light downward. This trick works most of the time, but it can be noticed if the wearer tilts their head back.

Meta’s Orion used a two-sided, two-grating version; they had three input gratings to support the combining of monochromatic Red, Green, and Blue MicroLEDs.

Poor Color Uniformity from SiC Waveguides Thus Far

The color uniformity of the five diffractive SiC waveguides I have seen, either firsthand or in pictures, is poor — generally much worse than any glass waveguide I have seen, even from many years ago. Issues like the color dispersion and perhaps the birefringence of SiC are major issues. The five SiC waveguides have different FoVs from 70° to 30°; two use 3-grating designs and three use 2-grating designs; two of the designs use monochrome MicroLEDs (one combines the color with the waveguide and one with an X-Cube), two use a DLP, and one uses an LCOS display. Regardless of the microdisplay type, number of gratings in the waveguide, or FoV, the color uniformity is abysmal to say the least.

I often discuss what I refer to as the Consumer vs. Technophile curve: appreciation for image quality is often tied to understanding the effort involved in creating that image. Consumers typically do not care about these production details; they are primarily concerned with how a display compares to the TVs and smartphones they already own.

I will freely admit that the camera being “objective” will tend to make color variations seem a bit worse than what the subjective human visual system will see, but not so different that people will not see these problems. The human visual system can tolerate large variations in brightness of a monochrome image (which helps monochrome green glasses), but the eye will notice substantial changes in color uniformity across the FOV.

Meta Orion

In October 2024, Meta first publicly demonstrated Orion lab prototype glasses with SiC waveguides, drawing considerable attention to the use of SiC. While Orion has a 70° FOV, it uses individual red, green, and blue 640 x 480-pixel MicroLED displays (by Jade Bird Display). The waveguide had three input gratings to combine the three monochrome displays. With a 70° FoV and only 640×480 resolution, Orion had a very low pixel density of 13 pixels per degree — more an issue of the lack of availability of higher-resolution MicroLEDs than of the waveguide, and Meta has said they have high-resolution prototypes. Meta did not allow people to take pictures, but they did show some images they took at conferences, including at SPIE AR/VR/MR 2025. In what is supposed to be a white image, the color uniformity is abysmal — various colors die out at different rates as they cross the waveguide. To be fair, Orion had a much bigger FOV and it becomes increasingly harder to maintain uniformity as the FoV increases.

Moldnano (and Westlake University)

Moldnano’s waveguide used 3 gratings (entrance, expansion, and exit). Their design supports a 30.82° FOV and uses a 1280×720 DLP (many companies use DLP engines for prototyping, then switch to LCOS for products, as LCOS AR designs are usually smaller and less expensive). The article includes several through-the-waveguide pictures against black and against ambient light; it is obvious that the color blue is severely lacking.

SEEV (Nimbo X1)

The Nimbo X1 (from Nimbopearl) using the SEEV SiC waveguide has a 30° FOV with X-cube MicroLED 640×480. SEEV uses a 2-grating waveguide. While Nimbopearl won’t say which company’s MicroLEDs they are using, they have confirmed that they are NOT from Jade Bird Display — the first confirmation I have seen of another company being ready to manufacture all three colors of MicroLEDs. In video footage, only a small part in the middle of the FoV renders white as intended.

Goeroptics (Subsidiary of Goertek) SiC Waveguide

At SPIE AR/VR/MR 2026, Goeroptics was demonstrating a 50° FOV SiC waveguide with a 1200 x 1560 LCOS panel, using a more conventional 3-grating waveguide. A picture through the optics of what should have been a mostly white image once again showed a lack of color uniformity.

LLVision

LLVision’s paper includes images demonstrating a 50° FOV with a 1280 × 720-pixel DLP. They use a 2-grating waveguide structure. Of particular interest was their use of my 1280 × 720 white-on-black test pattern from this blog’s test pattern page (edited to remove two pictures and black out my label). Overall, while still poor, their image uniformity seems better than the others — it seems red-deficient with a cyan (blue-green) cast, but at least not as radically varying as the others.

Conclusion

I want to reiterate the issue: “Too Few Factor Analysis.” It’s easy to see the advantages of SiC, including its high refractive index, high strength, and high thermal conductivity. But these advantages must be weighed against what mitigates them and the issues with SiC waveguides.

What does it matter if the wide FoV has poor image quality? How can you support the wide FoV with enough pixels to make it useful? What about the power consumption associated with supporting a wide FoV with more light, more data, and more processing?

I don’t expect many consumers to be happy with the image quality of the current SiC waveguides in any application, let alone when viewing pictures or videos. It’s unclear how hard it will be to improve color uniformity, but it is likely to be difficult because of the inherent problems caused by dispersion and birefringence.

Given the high cost of making SiC waveguides and the limited potential volume, it’s hard to see how manufacturing issues will be solved and image quality improved enough to satisfy a broad consumer market in the near future. There may be specialized applications where the unique properties of SiC will outweigh its drawbacks.

Curiously, the Nimbo X1 starts with SiC waveguides with only a 30-degree FoV for its first product. It does suggest that SiC manufacturing costs are coming down, but why lead with a more expensive technology they don’t really need?

Next Time Nimbo X1

When I started this article, I was planning on writing about the Nimbo X1 and the things I had noticed about it. But as I got into studying and writing about SiC, I thought the information on the Nimbo X1 got buried. I’m going to try and get the Nimbo X1 article out before I leave for Eindhoven, but it might have to wait until after MicroLED and AR/VR Connect.