Precise yet Fast High-Throughput Search for Thermal Insulators

We present a systematic and numerically precise computational search for thermal insulators in material space performed with the FHI-vibes high-throughput framework [1]. FHI-vibes employes a robust metric that quantifies the degree of anharmonicity in the nuclear dynamics via the statistical comparison of first-principles forces with those forces that would act in the harmonic approximation.
This enables us to efficiently scan over thousands of materials, including complex oxides and chalcogenides as well as ternary structures like perovskites, and single out strongly anharmonic ones. For these systems, we perform ab initio Green-Kubo simulations to model their thermal-transport properties, thereby naturally including all anharmonic effects [2]. Our strategy allows to avoid redundant calculations and achieve a much higher quality of information than achieved in traditional high-thoughput studies. Besides validating the performed search and analyzing its results, we discuss how big-data analytics techniques can be utilized to further accelerate and guide this search.

[1] https://vibes.fhi-berlin.mpg.de/
[2] C. Carbogno, R. Ramprasad, and M. Scheffler, Phys. Rev. Lett. 118, 175901 (2017)