Materials informatics for discovery of ion conductive ceramics for batteries

Abstract

In this review, we introduce our work in the field of materials informatics for the prediction of ionic conduction properties in inorganic crystalline solids. Rational material development based on information-derived prediction of the ionic conductivity for the materials listed in the crystal structure database is attractive to reduce processing time and labor costs. For this purpose, the development of general descriptors and a sufficient volume of ionic conductivity datasets are required. As an example, herein we describe machine learning regression and Bayes optimization schemes and their results by using histogram descriptors and a bond valence-based force field approach.