Prediction of Spectral Absorption of Anisotropic Α-Moo3 Nanostructure Using Deep Neural Networks

Deep neural networks (DNN) are currently one of the most effective tools in nanophotonics research owing to their ability to automate predictions and decisions based on complex data by establishing a nonlinear mapping process between the inputs and outputs. This paper presents an application of DNN to predict the spectral absorption of anisotropic α-MoO3 nanostructure, while accelerating its design process. The model validation results demonstrate the effectiveness of the algorithm, and the computational accuracy is controlled within the acceptable range. The spectrum of broadband absorption can be laterally shifted by changing the optical axis rotation angle of α-MoO3, and the broadband-narrowband absorption transition can be achieved by changing the width of the conical structure. The effect of the incident angle on the absorption spectrum is considered, and the absorber is found to be angle-insensitive over a wide angle range. This study provides guidance for the design of artificial intelligence-based nanophotonic devices and supports the development of solar thermal utilization, radiation cooling, etc