Deep Learning-Driven Design of Plasmonic Optical Coloring Filters

Abstract

Plasmonic Coloring Filters can be dynamically tuned by manipulating the geometric parameters of meta-materials, such as unit cell size, shape, and arrangement. In the presented work, the impact of the unit cell dimensions of the plasmonic color tunable filter on the transmitted color is investigated. A deep neural network model is implemented to efficiently predict transmission coefficients across various dimensional configurations and to predict the optimal one, enabling the rapid optimization of design parameters for desired wavelengths. The model's high accuracy significantly reduces both the simulation time and the computational resources.