🔬 optics

Moth's eye-inspired perfectly vertical subwavelength grating coupler for silicon photonics

This paper proposes a bio-inspired, single-etch anisotropy design for a perfectly vertical subwavelength grating coupler on the SOI platform that achieves high in-coupling efficiency (41%) and strong unidirectionality (over 10 dB) for applications like heterogeneous light source integration.

Original authors: Ivan A. Kazakov, Ilona Popova, Arkady Shipulin

Published 2026-04-28

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Ivan A. Kazakov, Ilona Popova, Arkady Shipulin

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

The "One-Way Mirror" for Light: A Simple Guide to Moth-Eye Photonics

Imagine you are trying to direct a massive crowd of people through a single, narrow doorway. If the crowd comes from all directions, they’ll bump into each other, cause a jam, and most people will end up walking in the wrong direction.

In the world of high-tech microchips (specifically "silicon photonics"), we have a similar problem. We need to get light from a laser or a fiber-optic cable into a tiny, microscopic circuit. Usually, we have to tilt the light at an angle to get it in, which is like making everyone walk through the door sideways. This takes up a lot of space and is hard to do perfectly.

Researchers at the Skolkovo Institute of Science and Technology have come up with a clever way to solve this by looking at nature—specifically, the eyes of a moth.

1. The Problem: The "Two-Way Street" Headache

Most current ways to get light into a chip act like a two-way street. Light goes in, but some of it bounces right back out, or worse, it leaks into the wrong part of the circuit. This "backflow" causes interference and wastes energy. If we want to build things like LiDAR (the "eyes" for self-driving cars) or ultra-fast quantum computers, we need a "one-way street" for light.

2. The Inspiration: The Moth’s Secret

Have you ever wondered why moths don't get blinded by headlights at night? It’s because their eyes have a special microscopic texture called a "Moth-Eye" structure. Instead of being perfectly smooth, their surface is covered in tiny, cone-shaped bumps.

These bumps act like a "gradient" for light. Instead of hitting a hard wall and bouncing back (like a mirror), the light "slides" into the eye smoothly. It’s like walking down a gentle ramp instead of running into a brick wall.

3. The Invention: The "Asymmetric Ramp"

The scientists took this "Moth-Eye" idea and applied it to a tiny grating (a series of microscopic ridges) on a silicon chip.

But they added a brilliant twist: they made the ridges asymmetrical.

Imagine a slide at a playground. On one side, the slide is smooth and steep, allowing you to zip down easily. On the other side, the slide is covered in bumpy, jagged "teeth."

The Smooth Side: Acts like a mirror, reflecting light toward the circuit.
The Bumpy Side: Acts like an "anti-reflective" sponge, catching the light and pulling it into the chip.

Because one side is smooth and the other is "toothy," the light is forced to go in one direction only. It’s like a specialized turnstile at a stadium: you can go in, but you can't go back out the same way.

4. Why This Matters (The "So What?")

This design is a "triple threat":

It’s Vertical: You don't have to tilt the light; you can shine it straight down, which saves massive amounts of space on the chip.
It’s One-Way: It directs light with incredible precision (over 10 decibels of "unidirectionality"), meaning almost no light leaks into the wrong place.
It’s Easy to Build: Most high-tech light guides require multiple complex "etching" steps (like carving a statue in layers). This design only requires one single step, making it much cheaper and easier to mass-produce.

The Bottom Line

By mimicking the way a moth avoids reflections, these scientists have created a microscopic "one-way valve" for light. This could pave the way for smaller, faster, and more efficient technology—from the sensors that help cars "see" the road to the ultra-secure communication networks of the future.

Technical Summary: Moth’s Eye-Inspired Perfectly Vertical Subwavelength Grating Coupler

Problem Statement

The integration of light sources into Photonic Integrated Circuits (PICs) is a critical bottleneck in silicon photonics. While edge couplers (ECs) offer high bandwidth and low loss, they are unsuitable for applications requiring wafer-level testing, phased-array antennas, or the heterogeneous integration of Vertical-Cavity Surface-Emitting Lasers (VCSELs). Grating couplers (GCs) are the preferred alternative for these applications, but standard GCs often suffer from bidirectional light propagation (causing interference issues) and complex fabrication requirements. There is a need for a coupler that is unidirectional, highly efficient, and easy to manufacture (requiring only a single etching step).

Methodology

The authors propose a bio-inspired design that breaks the symmetry of the grating stripe to achieve unidirectional light propagation.

Design Principle (Moth’s Eye Metasurface): The grating is designed with an asymmetric transverse profile. One side of the grating stripe is flat and reflective, while the other side is patterned with "teeth" that act as an anti-reflective surface, mimicking the "moth’s eye" structure. This asymmetry directs light to one side of the grating.
Subwavelength Grating (SWG) Engineering: To avoid complex multi-level etching, the authors use SWG structures to engineer the effective refractive index ( $n_{eff}$ ) of the grating stripes. By varying the fill-factor ( $f$ ), they can achieve a wide range of effective indices between $SiO_2$ and $Si$.
Numerical Simulation: 3D Finite-Difference Time-Domain (FDTD) simulations were performed using the MEEP software. The optimization involved five parameters: fill-factor ( $f$ ), two subwavelength periods ( $\Lambda_x, \Lambda_{x \cdot f_x}$ ), and two subwavelength lengths ( $f_x, f_s$ ). The efficiency was calculated by decomposing the electric field into forward- and backward-traveling eigenmodes to ensure reflected light did not skew the results.
Experimental Validation: The design was fabricated on a 220 nm Silicon-on-Insulator (SOI) platform via a CUMEC MPW run. Two layouts were tested: a 2-port structure to measure single-coupler insertion loss and a 3-port structure to quantify unidirectionality.

Key Contributions

Novel Bio-inspired Design: Introduction of a single-etch, asymmetric grating geometry that achieves unidirectionality through effective index engineering rather than complex multi-step lithography.
Effective Medium Theory (EMT) Application: Utilization of second-order EMT approximations to accurately model the refractive index of the SWG regions for 2D-FDTD optimization.
Proof-of-Concept for VCSEL Integration: Providing a viable pathway for the heterogeneous integration of energy-efficient VCSELs onto silicon PICs.

Results

Simulated Performance:
- In-coupling Efficiency: 41% at $\lambda = 1550$ nm.
- Unidirectionality: $>20$ dB at 1550 nm (less than 1% of light couples into the opposite direction).
- Bandwidth: 50 nm (at the -1 dB level).
Experimental Performance:
- Unidirectionality: $12.80 \pm 0.02$ dB near 1528 nm.
- Insertion Loss: $-8.35 \pm 0.02$ dB for a single coupler.
- Note: While experimental unidirectionality and loss were lower than simulated values, they successfully validated the physical principle of the design.

Significance

This work demonstrates a highly manufacturable solution for vertical light coupling in silicon photonics. By combining the simplicity of a single-etch process with the advanced light-steering capabilities of moth-eye metasurfaces, the design is highly relevant for next-generation technologies, including LiDAR, quantum communications, sensing systems, and high-density classical optical interconnects. The approach is also platform-agnostic and could be extended to silicon nitride (SiN) photonics.