Доступ предоставлен для: Guest

ISSN Online: 2377-424X

ISBN Print: 978-1-56700-474-8

ISBN Online: 978-1-56700-473-1

International Heat Transfer Conference 16
August, 10-15, 2018, Beijing, China

HIGH TEMPERATURE THERMAL CONDUCTIVITY OF SILICON FROM MACHINE-LEARNING-BASED INTERATOMIC POTENTIAL

Get access (open in a dialog) DOI: 10.1615/IHTC16.mpe.022399
pages 6001-6011

Аннотация

The researches of phonon transport and thermal conduction in solid materials at high temperatures are of great importance in designing thermal barrier coatings, developing thermoelectric materials and finding solutions for heat dissipation in high power electronics. Molecular dynamics simulation is a powerful approach to investigate thermal transport at high-temperature conditions. However, its inputs, the interatomic potentials, are usually a bottleneck for predicting phonon and thermal properties of materials, as the prediction of most empirical interatomic potentials on vibrational properties of materials is far from satisfactory. In this study, we utilize machine learning techniques to develop potentials based on the data from first-principles calculations. The phonon dispersion and mode Gruneisen parameters of silicon calculated using the resulting machine-learning-based potential show great agreement with the first-principles data, indicating the interatomic potential derived in this work is able to capture both the harmonic and anharmonic vibrational properties of silicon. We also show that the calculated thermal conductivity from molecular dynamics simulations fairly agrees with the measured thermal conductivity. Our work provides a framework to predict thermal transport properties of solid materials at high temperatures from first-principles.