Predicting Interfacial Thermal Conductance and Thermal Conductivity across Multilayer TiS₂/MoS₂ van der Waals Heterostructures Using Moment Tensor Potentials

Thermal conductivity calculations using classical molecular dynamics (MD) simulations are significantly influenced by the selection of accurate interatomic potentials. However, these interatomic potentials are not readily available for several novel two-dimensional (2D) materials and their heterostructures. Here, we propose a machine learning interatomic potential (MLIP) and D3-dispersion correction approach to determine the out-of-plane thermal conductivity and interfacial thermal conductance of multilayer TiS2/MoS2 heterostructures. We employ an MLIP-based moment tensor potential (MTP) to characterize the intralayer interactions within layers, while the interlayer interactions are characterized using the D3-dispersion correction. This approach demonstrates greater accuracy compared with Lennard-Jones-based potentials for interlayer interactions. Furthermore, this work uses a simplified supercell of the TiS2/MoS2 heterostructure and cost-effective ab initio molecular dynamics (AIMD) simulations with a duration of less than 1 ps. Additionally, this work investigates the effects of tuning the ratio of TiS2 to MoS2 layers on the interfacial thermal conductance and the overall out-of-plane thermal conductivity of the multilayer heterostructure. This MLIP-based methodology provides a reliable approach for estimating the thermal properties of multilayer heterostructures.