Bridging Theory with Experiment: Digital Twins and Deep Learning Segmentation of Defects in Monolayer MX2 Phases

TL;DR

This paper presents a digital twin approach using first principles calculations to train deep learning models for defect segmentation in monolayer MX2 materials, improving defect identification in microscopy images.

Contribution

It introduces a method to generate digital twins from density functional theory for training and benchmarking deep learning models on defect identification in nanomaterials.

Findings

Deep learning models trained on digital twins outperform traditional methods.

The approach enables defect classification under various experimental conditions.

Benchmarking reveals physical factors affecting model performance.

Abstract

Developing methods to understand and control defect formation in nanomaterials offers a promising route for materials discovery. Monolayer MX2 phases represent a particularly compelling case for defect engineering of nanomaterials due to the large variability in their physical properties as different defects are introduced into their structure. However, effective identification and quantification of defects remains a challenge even as high-throughput scanning tunneling electron microscopy (STEM) methods improve. This study highlights the benefits of employing first principles calculations to produce digital twins for training deep learning segmentation models for defect identification in monolayer MX2 phases. Around 600 defect structures were obtained using density functional theory calculations, with each monolayer MX2 structure being subjected to multislice simulations for the purpose of generating the digital twins. Several deep learning segmentation architectures were trained on this dataset, and their performances evaluated under a variety of conditions such as recognizing defects in the presence of unidentified impurities, beam damage, grain boundaries, and with reduced image quality from low electron doses. This digital twin approach allows benchmarking different deep learning architectures on a theory dataset, which enables the study of defect classification under a broad array of finely controlled conditions. It thus opens the door to resolving the underpinning physical reasons for model shortcomings, and potentially chart paths forward for automated discovery of materials defect phases in experiments.