Inverse Design of Diffractive Metasurfaces Using Diffusion Models

2025-06-26

AI Summary

Inverse design of diffractive metasurfaces faces challenges including strong nonlinearity in the structure–optical response mapping, susceptibility to local optima, and high computational cost. Method: This work introduces, for the first time, conditional diffusion models into metasurface inverse design, enabling continuous wavelength-band generation and integrating dual initialization paradigms—direct sampling and gradient-based optimization. A physics-consistent dataset is constructed using Rigorous Coupled-Wave Analysis (RCWA), and RCWA-guided posterior sampling is jointly employed with gradient optimization for efficient co-solution. Contribution/Results: The method designs uniform beam splitters and polarization beam splitters within 30 minutes, achieving low reconstruction error and strong generalization across wavelengths and target specifications. All code and datasets are publicly released.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Computer Vision: Diffusion Models for Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Algorithmic accountability and transparency on the web
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Metasurfaces are ultra-thin optical elements composed of engineered sub-wavelength structures that enable precise control of light. Their inverse design - determining a geometry that yields a desired optical response - is challenging due to the complex, nonlinear relationship between structure and optical properties. This often requires expert tuning, is prone to local minima, and involves significant computational overhead. In this work, we address these challenges by integrating the generative capabilities of diffusion models into computational design workflows. Using an RCWA simulator, we generate training data consisting of metasurface geometries and their corresponding far-field scattering patterns. We then train a conditional diffusion model to predict meta-atom geometry and height from a target spatial power distribution at a specified wavelength, sampled from a continuous supported band. Once trained, the model can generate metasurfaces with low error, either directly using RCWA-guided posterior sampling or by serving as an initializer for traditional optimization methods. We demonstrate our approach on the design of a spatially uniform intensity splitter and a polarization beam splitter, both produced with low error in under 30 minutes. To support further research in data-driven metasurface design, we publicly release our code and datasets.

Problem

• Inverse design of metasurfaces for desired optical responses
• Overcoming complex nonlinear structure-property relationships
• Reducing computational overhead and expert tuning requirements

Innovation

• Diffusion models for metasurface inverse design
• RCWA-guided posterior sampling for low error
• Public release of code and datasets