A Deep-learning Approach for Modeling Phase-change Metasurface in the Mid-infrared

Abstract

Reconfigurable metasurface constitutes an important block for future adaptive and smart nanophotonic applications. In this work we introduce a new modeling approach for the fast design of tunable and reconfigurable metasurface structures using convolutional deep learning network. The metasurface structure is modeled as a multilayer image tensor to model the material properties as image maps. The dimensionality mismatch problem is avoided by using the operating wavelength as an input to the network, so the model is used as single-point solver. As a case study, we model the response of a reconfigurable absorber employing phase transition of vanadium dioxide in the mid-infrared. The results show that our model provides accurate prediction of the metasurface response using small training dataset.