Session: K9-11 Phonon Modeling and Machine Learning for Thermal Transport

Paper Number: 117206

117206 - Fast and Accurate Machine Learning Prediction of Phonon Scattering Rates and Lattice Thermal Conductivity 

Lattice thermal conductivity is important for many applications of high scientific and societal impact including thermal insulation for energy savings, thermal management of semiconductor devices, thermoelectrics, and thermal barrier coatings. But experimental measurements or first principles calculations including three-phonon and four-phonon scattering are expensive or even unaffordable. Machine learning approaches that can achieve similar accuracy with first principles or experiments have been a long-standing open question. Despite recent progress in end-to-end machine learning models that use structural information of materials as descriptors, their accuracy falls far short of that of experiments or first principles. In this work, we provide the first machine learning approach that can predict phonon scattering rates and thermal conductivity at the experimental and first principles accuracy level, for a wide range of materials represented by Si, MgO and LiCoO2. The success of our approach is enabled by mitigating computational challenges associated with the high skewness of phonon scattering rates and their complex contributions to the total thermal resistance. Furthermore, transfer learning between different orders of phonon scattering is used to improve the model performance. Our surrogates offer up to two orders of magnitude acceleration compared to first principles calculations and would enable large-scale thermal transport informatics.

Presenting Author: Ziqi Guo Purdue University