🔬 optics

Inverse designed full-Stokes polarimetric metasurface with simultaneous wavefront sensing for visible light

This paper presents an inverse-designed metasurface that integrates full-Stokes polarimetry and Shack-Hartmann wavefront sensing within a single continuous aperture for visible light, achieving high-precision simultaneous measurements by breaking the traditional aperture-sharing paradigm and utilizing a neural network for automated data processing.

Original authors: Ondřej Červinka, Martin Hrtoň, Štěpán Venos, Jakub Lelek, Libor Úlehla, Tomáš Šikola, Filip Ligmajer

Published 2026-04-22

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Ondřej Červinka, Martin Hrtoň, Štěpán Venos, Jakub Lelek, Libor Úlehla, Tomáš Šikola, Filip Ligmajer

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a quality control inspector for a factory that makes high-precision camera lenses. To ensure these lenses are perfect, you need to check two things:

Are they perfectly flat? (If they are warped, the image will be blurry).
Are they stressed? (If the glass is under internal pressure, it changes the "polarization" of the light passing through it, which can ruin the image quality).

Traditionally, to check these two things, you need two big, clunky machines. One machine measures the flatness (like a topographical map), and another measures the light's polarization (like a special filter analyzer). You have to run the lens through both, which takes time, costs money, and requires a lot of space.

The "Magic Sheet" Solution
This paper introduces a revolutionary new tool: a metasurface. Think of this not as a thick glass lens, but as a flat, microscopic sheet of glass covered in billions of tiny, invisible pillars (nanostructures). It's so thin it's almost like a sticker.

The researchers didn't just design this sheet by hand; they used a super-smart computer program (called inverse design) to figure out the perfect shape for every single tiny pillar. It's like asking a chef, "I want a cake that tastes like chocolate, looks like a rose, and melts at exactly 37 degrees," and the computer instantly invents a new recipe that no human would ever think of.

How It Works: The "Traffic Light" Analogy
Usually, if you want to measure different things with a small sensor, you have to split the sensor into separate zones. Imagine a parking lot where you have to paint three different colored zones to park three different types of cars. This wastes space and makes the parking spots smaller.

This new device breaks that rule. Instead of splitting the parking lot, it uses the entire lot for every type of car simultaneously.

Here is the magic trick:

When light hits this sheet, the tiny pillars act like a super-advanced traffic director.
Depending on the "color" of the light's polarization (horizontal, vertical, circular, etc.), the sheet bends the light into six different tiny beams that land on six different spots on a camera sensor.
The Polarization Check: By measuring how bright each of those six spots is, a small computer brain (a neural network) can instantly calculate the exact polarization state of the light.
The Wavefront Check: If the lens being tested is warped or stressed, the light hits the sheet at a slightly weird angle. This causes the six tiny spots to shift position (slide left, right, up, or down). By measuring how much they slide, the device knows exactly how warped the lens is.

Why This is a Big Deal

It's Tiny: It replaces a room full of equipment with a chip the size of a fingernail.
It's Efficient: Because it doesn't waste space by dividing the sensor, it captures much more light, making the measurements sharper and more accurate.
It's Fast: It does both jobs at the exact same time, in a single snapshot.
It's Smart: The researchers used a "neural network" (a simple type of AI) to act as the interpreter. Since the real-world device isn't perfect (tiny manufacturing errors exist), the AI learns to correct those mistakes, turning raw data into a perfect reading.

The Bottom Line
The researchers built a "Swiss Army Knife" for light. Instead of needing a screwdriver and a hammer, you now have one tool that can measure the shape of a surface and the stress inside a material simultaneously. This could revolutionize how we make cameras, telescopes, and medical imaging devices, making them smaller, cheaper, and capable of real-time quality checks that were previously impossible.

1. Problem Statement

In precision optical metrology, characterizing optical components (e.g., lenses) requires measuring two critical parameters simultaneously:

Stress-induced birefringence: Which distorts the polarization state of light.
Surface flatness: Which distorts the wavefront (phase profile) of light.

Traditionally, these are measured using two separate instruments (a polarimeter and a Shack-Hartmann wavefront sensor), which increases system cost, footprint, and processing time. While metasurfaces have been used to miniaturize these sensors, existing solutions face significant limitations:

Aperture Partitioning: Most designs divide the aperture into discrete zones (e.g., 1/3 or 1/6 of the pixel area) to handle different polarization states. This drastically reduces light-gathering efficiency and the Numerical Aperture (NA), limiting sensitivity and spatial resolution.
Analytical Design Constraints: Traditional analytical methods struggle to independently map all six polarization bases within a single continuous aperture without complex "meta-molecules" that still suffer from efficiency losses.
Lack of Integration: No unified architecture existed that could simultaneously provide full-Stokes polarimetry and high-precision wavefront sensing in a single, compact, tile-able pixel without sacrificing performance.

2. Methodology

The authors propose a single-zone, inverse-designed metasurface that operates in the visible spectrum (target wavelength $\lambda = 532$ nm).

A. Inverse Design Strategy

Instead of using analytical formulas or partitioned zones, the team employed an adjoint optimization algorithm to search a vast parameter space of nanostructure layouts.

Parameterization: Each nanostructure (meta-atom) is defined by its lateral dimensions (width $W_m$ , length $L_m$ ) and rotation angle ( $\theta_m$ ).
Optimization Goal: The algorithm maximizes the intensity of the weakest focal point among six target polarization channels simultaneously. The cost function is defined as $\varepsilon_n = \min \{I_{\vec{p}_j}\}$ , ensuring balanced efficiency across all channels.
Gradient Calculation: The algorithm computes gradients of the cost function with respect to geometry parameters using the Green's function formalism to propagate near-field changes to far-field focal points.
Constraints: The design incorporates fabrication constraints (e.g., minimum feature size, aspect ratio) directly into the optimization loop, ensuring the final design is manufacturable via nanoimprint lithography.

B. Device Architecture

Material: High-aspect-ratio Titanium Dioxide ( $TiO_2$ ) nanostructures on a fused silica substrate.
Geometry: A square pixel of $100 \times 100$ µm with a focal length of $f = 1$ mm.
Operation: The metasurface splits incident light into six distinct focal spots based on the incident polarization state.
- Polarimetry: The intensities of the six focal spots are used to reconstruct the Stokes vector.
- Wavefront Sensing: The displacement (centroid shift) of the focal spots relative to their ideal positions is used to detect wavefront tilt (Shack-Hartmann principle).

C. Data Processing

To handle experimental non-idealities (crosstalk, sensor non-linearity, fabrication imperfections), the authors utilized a shallow neural network (6-3-3 architecture).

Input: Six measured focal spot intensities.
Output: Stokes parameters ( $S_1, S_2, S_3$ ).
Advantage: The network automates peak identification and corrects for hardware artifacts better than standard linear calibration matrices.

3. Key Contributions

Aperture-Sharing Breakthrough: The device utilizes the entire pixel area for all polarization channels, eliminating the efficiency and NA penalties associated with zoned designs. This results in an NA ( $\approx 0.05$ ) roughly 3 times larger than comparable zoned designs at the same focal length.
Simultaneous Dual-Functionality: It is the first demonstration of a single, continuous-aperture metasurface capable of performing full-Stokes polarimetry and Shack-Hartmann wavefront sensing simultaneously in the visible spectrum.
Inverse Design Validation: The work demonstrates that inverse design can generate "non-intuitive" nanostructure layouts that outperform traditional analytical designs in multi-functionality and precision.
Machine Learning Integration: The successful application of a shallow neural network to calibrate the system against real-world fabrication and sensor non-idealities, achieving high reconstruction accuracy.

4. Results

Polarimetry Performance:
- Tested against 100 polarization states uniformly distributed on the Poincaré sphere.
- Achieved a mean polarization reconstruction error of 0.046 (Euclidean distance on the Stokes sphere).
- This outperforms or matches state-of-the-art division-of-image and spatial multiplexing metasurfaces (which typically range from 0.075 to 0.130).
Wavefront Sensing Performance:
- Measured focal spot displacement sensitivity of 435 µm/deg at the camera plane (21.8 µm/deg at the metasurface focal plane).
- Focal spots are diffraction-limited (FWHM $\approx 4$ µm) and spaced 25 µm apart, allowing for clear separation.
- The effective angular measurement range is 1.4°, sufficient for detecting typical planar distortions in optical components.
Fabrication:
- Successfully fabricated using electron beam lithography and atomic layer deposition (ALD) of $TiO_2$ .
- The device features a square tiling capability, allowing for seamless coverage of large sensor areas without dead space.

5. Significance

This work represents a major step forward in compact optical metrology. By merging full-Stokes polarimetry and wavefront sensing into a single, tile-able metasurface pixel, the technology enables:

Real-time Diagnostics: Immediate detection of stress-induced birefringence and surface flatness errors in optical components during manufacturing or assembly.
System Miniaturization: Replacing bulky, separate optical benches with a single flat optical element.
Enhanced Performance: The elimination of aperture partitioning significantly improves light collection efficiency and spatial resolution, making the sensor more sensitive and robust.

The study validates that inverse design is a superior approach for creating multifunctional optical systems, capable of solving complex, competing optimization problems that traditional analytical methods cannot address.