MxDiffusion: A Physics-Aware Maxwells Law-Guided Diffusion Model Strategy for Inverse Photonic Metasurface Design

TL;DR

MxDiffusion is a novel physics-guided diffusion framework that combines Maxwell's law with data-driven models to enhance the accuracy and physical consistency of inverse photonic metasurface design.

Contribution

The paper introduces a two-stage diffusion-based inverse design method that incorporates Maxwell's equations, improving physical accuracy and out-of-distribution performance.

Findings

Outperforms conventional data-driven diffusion models

Achieves high fidelity in physically constrained scenarios

Effective for diverse photonic structure applications

Abstract

We introduce MxDiffusion, a hybrid physics- and data-driven diffusion-based framework that enables efficient and highly accurate generation of photonic structures from target optical properties. The improved accuracy is achieved through a two-stage generation strategy, in which the first diffusion model is explicitly trained with Maxwells equation-based loss to embed physical insight directly into the inverse design process, while the second model maps the physically consistent intermediate representation to the final structural geometry with significantly higher fidelity than solely data-driven approaches. The performance of MxDiffusion is validated on two representative applications: gold nanostructures patterned on a silica substrate and a highly tunable bandpass filter based on phase change material. In both cases, the proposed framework consistently outperforms a conventional data-driven diffusion model benchmark, particularly for out-of-training-distribution design targets and highly constrained resonance conditions. These results demonstrate the efficacy and superiority of MxDiffusion as a general physics-guided inverse design paradigm.