Automated identification and characterization of two-dimensional materials via machine learning-based processing of optical microscope images

Abstract

Mechanical characterization of two-dimensional (2D) materials has always been a challenging task due to their extremely small thickness. The current prevailing methods to measure the strength of 2D materials normally involve sophisticated testing facilities and complicated procedures of sample preparation, which are usually costly and time-consuming. In this paper, we propose a cost-effective and rapid approach to characterizing the strength of 2D materials by processing optical microscope images of the mechanically exfoliated 2D materials. Specifically, a machine learning-based model is developed to automate the identification of 2D material flakes of different layers from the optical microscope images, followed by the determination of their lateral size. The statistical distribution of the flakes’ size is obtained and used to estimate the strength of the associated 2D material based on a distribution-property relationship we developed before. A case study with graphene indicates that the present machine learning-based method, as compared to the previous manual one, enhances the efficiency of characterization by more than one order of magnitude with no sacrifice of the accuracy.