Accurate Modeling of Interfacial Thermal Transport in van der Waals Heterostructures via Hybrid Machine Learning and Registry-Dependent Potentials.

Two-dimensional transition metal dichalcogenides (TMDs) exhibit remarkable thermal anisotropy due to their strong intralayer covalent bonding and weak interlayer van der Waals (vdW) interactions. Accurately modeling their thermal transport properties remains a significant challenge, as density functional theory (DFT) is computationally prohibitive for large systems while traditional classical force fields often fail to capture the long-range, registry-dependent vdW energetics critical for twisted interfaces. To address this, we employ a modular hybrid computational framework that integrates a single-layer machine-learned potential (sMLP) for intralayer dynamics with a registry-dependent interlayer potential (ILP) for anisotropic vdW coupling. This modular framework achieves near quantum-mechanical accuracy across the group-VI TMD family, demonstrating good agreement with both DFT calculations and experimental measurements for key properties including lattice constants, bulk modulus, moiré reconstruction, phonon spectra, and anisotropic thermal transport conductivities. The scalability of this framework enables accurate simulations of TMD heterostructures featuring large-scale moiré superlattices, making it a transformative tool for the design of TMD-based thermal metamaterials and devices by bridging the gap between accuracy and computational efficiency.