Machine Learning Interatomic Potentials Enable Molecular Dynamics Simulations of Doped MoS2

Abstract: We present the first computational framework for molecular dynamics simulation of MoS2 doped with 25 elements spanning metals, non-metals, and transition metals using Meta's Universal Model for Atoms machine learning interatomic potential (MLIP). Benchmarking against density functional theory calculations demonstrates the accuracy of the MLIP for simulating doped-MoS2 systems and highlights opportunities for improvement. Using the MLIP, we perform heating-cooling simulations of doped-MoS2 supercells. The 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 an open-source 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. The MLIP bridges the accuracy-efficiency gap between first-principles methods and empirical potentials, and the framework offers unprecedented opportunities for large-scale materials discovery in two-dimensional doped material systems.