Study of Interfacial Thermal Resistance of Heterostructures Using Machine Learning (ML)-Molecular Dynamics

Diamond, as an ultra-wide bandgap semiconductor, holds significant potential in many cutting-edge applications from electronics, nanotechnology, biosensors and quantum information processing, owing to its ultra-high thermal conductivity and mobility. The interfacial thermal conductance between diamond and metals is crucial for the thermal management of diamond-based electronic devices and determines the maximum working temperature. However, current knowledge about the metal/diamond thermal interfacial transport is still very limited. There is no reliable interatomic potential between metals and diamond. In this work, we train machine learning interatomic potentials at the interface based on density functional theory (DFT) calculations. The accuracy of the potential is validated against DFT calculations. Using the potentials, we conduct MD simulations and obtain the interfacial thermal conductance. Different surface terminations of diamond have been studied. The results from this research should have significant contribution to improved understanding of interfacial thermal properties, which is key to developing new diamond-based electronics devices.