Spectral Learning and Inverse Design of Broadband Infrared Absorbers

Abstract

Designing broadband metamaterial absorbers with high absorption over a wide spectral range remains a significant challenge. Here, we develop an integrated simulation and machine-learning framework for the rapid modeling and design of broadband metamaterial absorbers. Electromagnetic simulations are first performed and benchmarked against previous studies and experimental data. From these, we find that the dielectric function is crucial for accurate spectral prediction and select an optimal material system and construct a training dataset. Machine-learning models are then used for forward prediction, inverse design, and spectral correlation analysis with all tasks achieving accuracies above 93%. These results show that the proposed framework can reduce the computational effort required for parameter exploration and accelerate the design of broadband infrared absorbers.