Low-Frequency In-Plane Phonons Dominate Thermal Transport in Montmorillonite: Machine-Learning Molecular Dynamics Insights

Montmorillonite (MMT), a foundational constituent of permafrost systems, plays a critical role in determining the thermophysical properties of frozen soils through its thermal transport behavior. The current understanding of these properties remains limited because of the experimental challenges associated with cryogenic conditions and the inherent limitations of classical molecular dynamics potentials in accurately capturing low-temperature phonon interactions. In this study, we integrate the neuroevolution potential (NEP) framework with active learning to develop a high-precision, data-driven model for evaluating the thermal conductivity of MMT. Nonequilibrium molecular dynamics, homogeneous nonequilibrium molecular dynamics, and spectral heat current are employed as complementary computational approaches to systematically simulate and rigorously compare the thermal conductivity of MMT. The results show that the thermal conductivity of MMT reaches 4.28 W m –1 K –1 at 300 K, with thermal transport predominantly governed by in-plane phonons (>85% contribution) and low-frequency phonon modes (