ABSTRACT
Two-dimensional (2D) materials have gained tremendous interest in the scientific community because of their remarkable properties. Nevertheless, the atomic scale thickness of 2D materials make it challenging to characterize their geometry, evaluate their properties, and identify defects present in these materials. Traditional techniques for detecting 2D materials involved hundreds of hours of manual labor invested by researchers. Despite this assiduous investigation, the structure-property relationships of 2D materials are perplexing and inconclusive. Machine learning (ML) deploys a wide range of algorithms to tackle enormous datasets and extract meaningful interpretations. ML algorithms can take the voluminous datasets as input features and streamline them to predict the fingerprint features of 2D materials within seconds. This chapter provides the progress of ML tools in optical and Raman spectroscopy characterization of 2D materials. These two spectroscopy techniques are the most widely utilized tools for the initial screening of 2D materials. First, we discuss the ML-powered optical identification method to realize the intelligent identification of 2D materials from the color characteristics of their optical micrograph. Then, we explore three algorithms: random forest regression, kernel ridge regression, and the Gaussian mixture model utilized in Raman spectroscopy to extract valuable insights using the examples of graphene and molybdenum disulfide. Furthermore, we outline the challenges and future research prospects of ML-enhanced algorithms to translate the ML models into industry-specific applications. The fusion of ML and nanoscience provides an opportunity to accelerate the design and discovery of new 2D materials.
