Probing Thermal Conductivity of Crystalline Materials with Universal Potential Driven Anharmonic Lattice Dynamics and Molecular Dynamic Simulations

Phonons, the quantized vibrations of atoms, are the major heat carriers in semiconductors and insulators. Due to the inherent inclusion of quantum effects of atomic vibrations and direct access to detailed microscopic insights into transport processes, anharmonic lattice dynamics (ALD) are widely regarded as the method of choice for computing and interpreting thermal transport in crystalline materials. In this talk, I will introduce a framework that bridges anharmonic lattice dynamics with MatterSim, a universal machine learning interatomic potential that delivers accurate predictions of energies, forces, and virials for materials across the periodic table and under diverse thermodynamic conditions. This framework enables robust high-throughput screening of candidate materials with intrinsically high lattice thermal conductivity.

To move beyond the perturbative regime of ALD, I will further demonstrate a complementary approach based on a novel heat-flux formulation and equilibrium molecular dynamics (MD) simulations powered by MatterSim. This allows us to accurately capture lattice thermal conductivity in strongly anharmonic materials and under extreme conditions. Ultimately, these ALD and MD approaches hold great promise for the design and exploration of next-generation materials for microelectronics, energy storage, and energy conversion applications.