Deep Learning‐Assisted Quantification of Atomic Dopants and Defects in 2D Materials

Abstract

Atomic dopants and defects play a crucial role in creating new functionalities in 2D transition metal dichalcogenides (2D TMDs). Therefore, atomic‐scale identification and their quantification warrant precise engineering that widens their application to many fields, ranging from development of optoelectronic devices to magnetic semiconductors. Scanning transmission electron microscopy with a sub‐Å probe has provided a facile way to observe local dopants and defects in 2D TMDs. However, manual data analytics of experimental images is a time‐consuming task, and often requires subjective decisions to interpret observed signals. Therefore, an approach is required to automate the detection and classification of dopants and defects. In this study, based on a deep learning algorithm, fully convolutional neural network that shows a superior ability of image segmentation, an efficient and automated method for reliable quantification of dopants and defects in TMDs is proposed with single‐atom precision. The approach demonstrates that atomic dopants and defects are precisely mapped with a detection limit of ≈1 × 1012 cm−2, and with a measurement accuracy of ≈98% for most atomic sites. Furthermore, this methodology is applicable to large volume of image data to extract atomic site‐specific information, thus providing insights into the formation mechanisms of various defects under stimuli.

Monitoring the Material Quality of Two-Dimensional Transition Metal Dichalcogenides

“…High-performance applications of TMDs are guaranteed by the production of high-quality material. In terms of sample preparation, 2D TMDs can be synthesized through top-down methods, including mechanical exfoliation (ME) − and liquid phase exfoliation, − and bottom-up methods mainly including molecular beam epitaxy (MBE) − and chemical vapor deposition (CVD). − However, due to the 2D nature and ultrahigh specific surface areas of TMDs, it is impossible to avoid introducing defects during sample synthesis with any method. , These lattice defects can be divided into two main types: point defects in individual island and grain boundaries (GBs) between two stitched islands. − Different from that of the single-element materials (such as graphene), the formation of point defects in TMDs is mainly due to the insufficient supply of the transition metal or the chalcogen precursor, as it is particularly difficult to precisely control these two compounds. GBs in polycrystalline TMDs result from the stitching of grains with random grain orientations during the growth.…”

“…58−63 However, due to the 2D nature and ultrahigh specific surface areas of TMDs, it is impossible to avoid introducing defects during sample synthesis with any method. 64,65 These lattice defects can be divided into two main types: point defects in individual island and grain boundaries (GBs) between two stitched islands. 66−71 Different from that of the single-element materials (such as graphene), the formation of point defects in TMDs is mainly due to the insufficient supply of the transition metal or the chalcogen precursor, as it is particularly difficult to precisely control these two compounds.…”

Recent Advances and Applications of Deep Learning Methods in Materials Science

“…Despite only using the first frame for training, the model successfully identified and tracked defects in the subsequent frames for each sequence, even when the lattice underwent significant deformation. Similarly, Yang et al [276] used U-net architecture (as shown in Fig. 4) to detect vacancies and dopants in WSe 2 in STEM images with model accuracy up to 98 %.…”

STEM Image Analysis Based on Deep Learning: Identification of Vacancy Defects and Polymorphs of MoS2

“…Recent advancements in deep learning, especially the advent of convolutional neural networks (CNNs), hold great promise for feature recognition and analysis from TEM data. − Indeed, CNN-based image analysis has recently been applied to identification of various structural features in TEM and STEM image analysis. ,− CNNs have been successfully adapted for identification of point defects and their evolution under e-beam irradiation, denoising of TEM/STEM images, subangstrom reconstruction around point defects, and structural phase evolution. ,− Although these previous studies have clearly demonstrated the promise of CNNs, there are still many challenges to overcome. One of the most pressing challenges is the major bottleneck in the range of TEM/STEM image quality that CNNs can reliably handle.…”

“…These structural features are important for determining the various properties of TMDCs, including MoS 2 . For example, the sulfur vacancy is a predominant point defect in MoS 2 and has a strong influence on the material’s doping level, charge transport behavior, and chemical reactivity. ,− Previous deep-learning-based analysis on chalcogen vacancies in TMDCs was mostly on Se or Te ,, and S vacancies were rarely analyzed. The lower Z-contrast of S atoms in STEM images compared to Se or Te atoms may pose some challenges in precise vacancy assignment from FCN.…”