STEM image analysis based on deep learning: identification of vacancy defects and polymorphs of ${MoS_2}$

TL;DR

This paper demonstrates how a fully convolutional network can efficiently identify defects and polymorphs in MoS2 STEM images, matching expert analysis accuracy and enabling high-throughput structural analysis.

Contribution

It introduces a ResUNet-based deep learning approach trained on simulated data for automated STEM image analysis of MoS2, improving efficiency and scalability.

Findings

FCN achieves accuracy comparable to expert analysis

Models trained on simulated data generalize well to experimental images

Provides guidelines for training deep learning models for STEM analysis

Abstract

Scanning transmission electron microscopy (STEM) is an indispensable tool for atomic-resolution structural analysis for a wide range of materials. The conventional analysis of STEM images is an extensive hands-on process, which limits efficient handling of high-throughput data. Here we apply a fully convolutional network (FCN) for identification of important structural features of two-dimensional crystals. ResUNet, a type of FCN, is utilized in identifying sulfur vacancies and polymorph types of ${MoS_2}$ from atomic resolution STEM images. Efficient models are achieved based on training with simulated images in the presence of different levels of noise, aberrations, and carbon contamination. The accuracy of the FCN models toward extensive experimental STEM images is comparable to that of careful hands-on analysis. Our work provides a guideline on best practices to train a deep learning model for STEM image analysis and demonstrates FCN's application for efficient processing of a large volume of STEM data.