Rapid identification of two-dimensional materials via machine learning assisted optic microscopy

Abstract

A combination of Fresnel law and machine learning method is proposed to identify the layer counts of 2D materials. Three indexes, which are optical contrast, red-green-blue, total color difference, are presented to illustrate and simulate the visibility of 2D materials on Si/SiO2 substrate, and the machine learning algorithms, which are k-mean clustering and k-nearest neighbors, are employed to obtain thickness database of 2D material and test the optical images of 2D materials via red-green-blue index. The results show that this method can provide fast, accurate and large-area property of 2D material. With the combination of artificial intelligence and nanoscience, this machine learning assisted method eases the workload and promotes fundamental research of 2D materials. Keywords: Two-dimensional materials, Optical contrast, Total color difference, Red-green-blue, K-means clustering, K-nearest neighbors (k-NN)