An Artificial Intelligence-enabled 2D Materials-based Color and Spectral Recognition System

Color-detection&spectroscopy,two distinct yet inter-related applications of semiconductors,suffer from the same limitations:necessarily large number of identical photoreactors,sophisticated engineering,intricate beam path,and lack of flexibility to adjust for the damaged pixels.2D materials are desirable candidates for optoelectronic applications,but face great challenges due to inherent variabilities in their properties.We report on a novel technique in simultaneously estimating the color&wavelength spectrum of any broadband visible light.The small number of optical thin-film filters of transition metal dichalcogenides(TMDs),fabricated via vapor phase chalcogenization(VPC),are designed to have enough variabilities with respect to each other.We apply artificial intelligence(AI)algorithms such as machine learning(ML)and pattern recognition on emergent optoelectronic responses of the filters upon shining various spectra,and obtain the underlying nonlinear ruling function for the collective of filters.Later,we estimate the color&spectrum of any unknown incident light using the learned function on the optoelectronic data of the light that is collected from the filters.