Mechanical Properties of Boron Nitride Nanosheets Using Machine-Learned Interatomic Potential

Abstract

Over the past decade, the study of new materials has become more attainable through computational methods and has transformed into a rapidly expanding field of research. At present, several computational techniques, such as density functional theory (DFT) and classical molecular dynamics (MD) simulations, are accessible through which one can study the interactions of atoms and physical properties at nanoscale. Although the DFT technique is capable of accurately predicting attributes, its scope is confined to a few atoms or molecules. On the other hand, classical MD simulation can handle large structures but is deficit in operational effectiveness due to the use of semi-empirical interatomic potentials. To address this challenge, we are exploring an innovative and efficient approach of machine-learned interatomic potential (MLIP) technique trained using a set of configurations obtained from DFT-MD simulations to accurately investigate the physical properties of nanomaterials. This work focuses on the development of an MLIP for boron nitride (BN)-based nanostructures. Both energy and forces were compared to validate and assess the reliability and accuracy of newly generated potential. Furthermore, we conducted MD simulations using newly formed MLIP to examine the structural attributes and failure analysis of BN sheets with varying structural parameters and temperatures. In our findings, we obtained the critical stress, critical strain, and modulus of elasticity of 110 GPa, 0.18, and 994 GPa, respectively. Our study offers a novel approach that highlights the importance of advanced computational tools in the exploration of material properties at the atomic scale, paving the way for future investigations into the applications of BN materials in the fields, such as actuators, sensors, and energy storage devices. © 2025 Elsevier B.V., All rights reserved.