A New Approach To AR Waveguides: Polarization-Based Zoning Boosts Prototype Brightness Uniformity To 70%

The waveguides in AR glasses require a certain margin for eye movement—a concept known as the "eye box." However, a larger eye box means that the projected light is distributed over a wider area, while the user’s pupil captures only a small fraction of it. Simply increasing the power of the light engine would lead to higher power consumption and greater thermal load. Is it possible to maintain a functional eye box range while illuminating only the specific area actually needed at any given moment?

A paper by Nilsen et al., published in Advanced Photonics Research in 2026, proposes a specific approach: dividing the exit-pupil extraction region into zones that respond to different circular polarization states, and using the input polarization to select which zone is active. Instead of illuminating the entire large eye box simultaneously, this method employs "switchable, smaller eye boxes" to cover a larger, usable range.

Where do the losses occur in traditional exit-pupil expansion?

In near-eye display (NED) systems, a waveguide typically receives an image from an optical engine via an input coupler. The light then propagates through the waveguide via total internal reflection (TIR) before being directed toward the user's eye by an output coupler. To ensure that the full field of view remains visible even when the eye shifts position, the output coupler must replicate the exit pupil across a larger area. The challenge is that each time a light beam interacts with the output coupler, a portion of its energy may be extracted. Consequently, as the light propagates, less energy remains available for subsequent sections of the output coupler, which can lead to a noticeable brightness gradient across the eye box.

This paper investigates polarization volume gratings (PVGs) used as out-couplers. PVGs exhibit selectivity based on the handedness of circularly polarized light. The authors divide what would otherwise be a single out-coupling region—covering the entire extent of the eye box—into two equal-length segments: one responsive to left-circularly polarized (LCP) light, and the other to right-circularly polarized (RCP) light. By switching the input polarization, the light is directed along specific in-coupling and out-coupling paths, illuminating only the selected half-region.

It is important to distinguish between two concepts that are easily confused: the total addressable range and the area illuminated at any given moment. While the two segments combined cover the original length of the eye box, only one segment is active at a time. For a practical product, enabling the illuminated region to track the pupil would require eye-tracking technology combined with electronically controlled polarization switching. The prototype presented in the paper does not incorporate such a closed-loop system; instead, it relies on a linear polarizer and a quarter-wave plate for manual polarization switching.

How to validate a research paper？

The authors first compared a full-length outcoupler design with a dual-zone design using OpticStudio. The initial simulation employed a 550 nm collimated rectangular beam, a 55 mm long by 1 mm thick waveguide (refractive index 1.8), and an outcoupler with a 400 nm period; the total area of the dual-zone outcoupler was identical to that of the full-length version. The eye-box uniformity for the central field-of-view beam improved from 35.8% to 69.4%. In this paper, uniformity is defined as the ratio of the minimum intensity to the maximum intensity within the sampled area, expressed as a percentage; it does not represent the overall optical efficiency of the system.

Subsequently, the authors fabricated a dual-zone prototype and compared it against a 15 mm long full-length outcoupler. The dual-zone design consisted of two 7.5 mm long sections, each responding to an opposite circular polarization. Testing with a 532 nm laser at normal incidence showed that, for the samples with lower diffraction efficiency, uniformity improved from 44.0% to 70.1%, with total powers of 9.11 μW and 9.85 μW, respectively, measured over an aperture of approximately 6 mm × 15 mm. While these results support the claim that "partitioning improves uniformity," the total power values for the two types of samples are merely comparable; therefore, it would be inaccurate to state that a doubling of brightness has been experimentally verified.

The authors also adjusted the liquid crystal material composition, increasing the diffraction efficiency of the outcoupling gratings from approximately 60.5% to 74%. For the corresponding full-length and dual-zone samples, the uniformity values were 15.9% and 38.8%, respectively, while the measured total power within the eye-box was 10.52 μW and 11.01 μW. Increasing the outcoupling ratio causes a larger portion of light to exit the waveguide earlier; consequently, an increase in total power can coincide with a decrease in spatial uniformity. The dual-zone approach mitigates this trade-off, though it does not eliminate it entirely. The paper also presents the design and simulation of a four-zone, two-dimensional exit-pupil expansion system; in a separate model utilizing 532 nm collimated light, uniformity improved from 19.2% to 51.2%. As the four-zone configuration has not been experimentally verified, this figure cannot be placed on the same evidentiary level as the measured results from the aforementioned two-zone prototype.

Brightness and Eye-box: Beyond a Single Magnification Factor

The author employs a simplified one-dimensional model to illustrate the potential: assuming a 20% out-coupling efficiency per interaction—with 10 interactions occurring in a full-length out-coupling region versus 5 in a half-length region—the calculated uniformity is approximately 41% for the half-length region and 13% for the full-length region. By further increasing the out-coupling efficiency of the half-length region, it is theoretically possible to extract a total power comparable to that of the full-length region from a smaller illuminated area, resulting in a local light intensity nearly twice as high.

However, this "twofold" increase represents an upper limit derived from an idealized model, rather than an empirically measured gain in brightness for a physical prototype. In reality, the efficiency of a PVG (Polarization Volume Grating) varies with wavelength, angle, and polarization state; furthermore, polarization drift during propagation and losses at the interfaces between layers can also alter the outcome. More importantly, if a smaller region is selected, the system must dynamically switch if the user’s eye shifts toward the boundary or into a different area; if both regions are illuminated simultaneously, the available power must be redistributed between them. Consequently, the claims of an "expanded" eye-box and "enhanced" brightness—as suggested in the article's title—should be understood as two distinct design possibilities offered by the architecture, rather than benefits that can be achieved simultaneously and unconditionally.

What is still needed to turn this into a pair of glasses?

For system designers, the value of this paper lies not merely in the use of a different type of grating, but in shifting the approach to exit-pupil expansion: moving from "passively illuminating the entire area" to "distributing light based on the pupil's position." This also introduces new engineering challenges:

1. Closed-loop eye tracking

It is necessary to verify tracking accuracy, the latency involved in polarization switching, and image continuity as the eye moves across different zones. The prototype presented in the paper has not yet validated this closed-loop system.

2. Boundaries and Crosstalk

The boundaries between zones may truncate the field of view (FOV), while unintended incoupling paths can lead to "ghost" images (stray light) appearing in unselected regions. Although the paper presents photographs of these ghost images, it does not provide a tolerance budget that could be directly applied to mass-production design.

3. Full-Color and Wide Field of View

The experiments are primarily based on a 532 nm monochromatic setup. Achieving full-color performance requires addressing the dispersion and angular response characteristics of the polarization volume gratings (PVGs) and geometric-phase deflectors used in the four-zone design. The authors also note that increasing the number of zones may result in FOV truncation or optical crosstalk.

4. Multi-layer Fabrication and Alignment

The dual-zone prototype consists of two waveguides with distinct polarization responses, bonded together with an air gap in between. As the number of zones increases, the spatial relationships between the incouplers, folding couplers, and outcouplers become significantly more complex; any misalignment can cause relative displacement between the images produced by different zones.

Therefore, a reliable conclusion from this work is that, under the specific monochromatic conditions, prototype configuration, and measurement parameters presented in the paper, polarization-selective zoning significantly improves the brightness uniformity of the eyebox; furthermore, a four-zone, two-dimensional scheme offers the potential for even greater expansion. If the objective is a wearable, full-color AR system, then additional factors—such as optical efficiency, field of view, motion-to-photon latency, crosstalk, and assembly tolerances—must also be evaluated within a comprehensive system-level specification framework.

What does this mean for lightweight AI-powered glasses?

The value of this "zoned out-coupling" approach lies primarily in offering new engineering options for distributing optical output in waveguides, rather than providing a ready-to-manufacture, full-color solution. At SOTECH, when evaluating near-eye display technologies, we also assess their ability to deliver tangible benefits in real-world usage scenarios—such as readability under bright outdoor light, power consumption and heat generation during extended wear, eye-box uniformity, and the delicate balance between overall device weight and wearer comfort.

For lightweight AI glasses like the SOTECH G55, the current focus is on the synergy between mature waveguide optics, Micro-LED displays, and integrated structural design. This ensures that information—such as subtitles, navigation prompts, or teleprompter text—remains clear and stable during daily use without obstructing the user’s natural field of view. As technologies like eye-tracking, polarization switching, and zoned out-coupling continue to mature, similar approaches could help such devices achieve a more refined balance between brightness, power efficiency, and the quality of the eye-box experience.