High-throughput prediction of finite-temperature properties using the quasi-harmonic approximation

In order to calculate thermal properties in automatic fashion, the Quasi-Harmonic Approximation (QHA) has been combined with the Automatic Phonon Library (APL) and implemented within the AFLOW framework for high-throughput computational materials science. As a benchmark test to address the accuracy of the method and implementation, the specific heat capacities, thermal expansion coefficients, Grüneisen parameters and bulk moduli have been calculated for 130 compounds. It is found that QHA-APL can reliably predict such values for several different classes of solids with root mean square relative deviation smaller than 28% with respect to experimental values. The automation, robustness, accuracy and precision of QHA-APL enable the computation of large material data sets, the implementation of repositories containing thermal properties, and finally can serve the community for data mining and machine learning studies.

This article originally appeared in Computational Materials Science 125, 2016, Pages 82–91.