WEB Exploring ultrahigh lattice thermal conductivity crystals via feature-based transfer learningThursday (24.09.2020) 09:30 - 09:45 M: Modelling and Simulation 2 Part of:
Ultrahigh lattice thermal conductivity materials hold great importance since they play a critical role in the thermal management of electronic and optical devices. Models using machine learning can search for materials with outstanding higher-order properties like thermal conductivity. However, the lack of sufficient data to train a model is a serious hurdle. Herein we show that big data can complement small data for accurate predictions when lower-order feature properties available in big data are selected properly and applied to transfer learning. The connection between the crystal information and thermal conductivity is directly built with a neural network by transferring descriptors acquired through a pre-trained model for the feature property. Transfer learning is employed to screen over 60,000 compounds to identify novel crystals that can serve as alternatives to diamond. The materials in the top-14 list obtained by feature-based and transfer learning screening all show high thermal conductivities, including boron arsenides, carbon, boron nitride, and heterodiamond. They have thermal conductivities on the order of 1,000 W/mK, validating the accuracy and high-efficiency of the developed screening method. What is more, the average or maximum dipole polarizability and the van der Waals radius are revealed to be the leading descriptors among those that can also be qualitatively related to anharmonicity. The successful prediction of high thermal conductivity crystals demonstrates the advantage of extrapolative prediction via transfer learning, and reveals the descriptors that are dominantly correlated with the anharmonic phonon properties.