Accurate and efficient prediction of the optical properties of two-dimensional (2D) materials is crucial for photonic applications, but this task remains challenging due to the differences between theoretical and experimental methods. Here, we propose a physics-based machine learning (ML) framework for accelerating the screening of 2D materials. It combines first-principles density functional theory (DFT) calculations and graph neural network models, along with experimental spectroscopy validation and Cauchy model integration. Within this framework, we have collected a database containing over 1000 transition metal dichalcogenide (TMD) monolayers and their optical properties. We also propose a general method for defining the thickness of 2D structures with physical significance, thereby correcting the optical properties obtained from PBE-based density functional theory. Using the collected database, we developed a machine learning model based on the Cauchy model for calculating the refractive index in the near-infrared (755–1064 nm) region. The developed method reflects the correlation between the atomic structure of a single layer and its optical properties, which has been confirmed through extensive testing of independent 2D material databases. Therefore, our machine learning-driven strategy provides a powerful tool for rapidly screening novel single-layer materials with customized optical functions, significantly accelerating the discovery and design of next-generation photonic materials. As an application, we also further demonstrated how high refractive index candidate materials such as Bi2Te2Se can achieve enhanced field confinement and long string interference length in single-layer waveguides, highlighting their potential in integrated photonics.
This research is published in ACS Nano under the title "Materials Informatics Framework for Accelerated Discovery of High-Refractive-Index 2D Materials". References:
DOI: 10.1021/acsnano.5c1064