Machine Learning Interatomic Potentials Enable Molecular Dynamics Simulations of Doped MoS2

Abstract

Dopants can tune the performance of MoS2 in various applications, but use of molecular dynamics simulations for doped MoS2 materials discovery is limited by the lack of multi-dopant interatomic potentials. Universal machine learning interatomic potentials (MLIPs) could be a solution, but the accuracy of these potentials must first be evaluated. Here, we evaluate the accuracy of a recently developed MLIP, META's Universal Model for Atoms (UMA), for 25 different MoS2 dopants spanning metals, non-metals, and transition metals in Mo substitution, S substitution, and intercalated positions by benchmarking the MLIP-predicted formation energy and the dopant-induced structural change against density functional theory calculations. The computational framework for MLIP validation and simulations are described in detail and the source code is made open source. The MLIP is then demonstrated by performing heating-cooling simulations of MoS2 supercells with all 25 dopants. These simulations capture complex phenomena including dopant clustering, MoS2 layer fracturing, interlayer diffusion, and chemical compound formation at orders-of-magnitude reduced computational cost compared to density functional theory. This work provides a computational workflow for application-oriented design of doped-MoS2, enabling high-throughput screening of dopant candidates and optimization of compositions for targeted tribological, electronic, and optoelectronic performance. Github Repo Link: https://github.com/AbrarFaiyad/Machine-Learning-Interatomic-Potential-UMA-Enables-MD-Simulations-of-MoS2-Doped

Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.