Identifying the ‘inorganic gene’ for high-temperature piezoelectric perovskites through statistical learning

Abstract

This paper develops a statistical learning approach to identify potentially new high-temperature ferroelectric piezoelectric perovskite compounds. Unlike most computational studies on crystal chemistry, where the starting point is some form of electronic structure calculation, we use a data-driven approach to initiate our search. This is accomplished by identifying patterns of behaviour between discrete scalar descriptors associated with crystal and electronic structure and the reported Curie temperature (TC) of known compounds; extracting design rules that govern critical structure–property relationships; and discovering in a quantitative fashion the exact role of these materials descriptors. Our approach applies linear manifold methods for data dimensionality reduction to discover the dominant descriptors governing structure–property correlations (the ‘genes’) and Shannon entropy metrics coupled to recursive partitioning methods to quantitatively assess the specific combination of descriptors that govern the link between crystal chemistry and TC (their ‘sequencing’). We use this information to develop predictive models that can suggest new structure/chemistries and/or properties. In this manner, BiTmO3–PbTiO3 and BiLuO3–PbTiO3 are predicted to have a TC of 730°C and 705°C, respectively. A quantitative structure–property relationship model similar to those used in biology and drug discovery not only predicts our new chemistries but also validates published reports.

Theory-Guided Machine Learning in Materials Science

“…For example, we used data from a recent study on predicting high-temperature piezoelectric perovskites (Balachandran et al, 2011), and indeed find that qualitatively different results may be obtained (Figure 1). In the first case where the gini impurity attribute (Rokach and Maimon, 2005) is used, the global instability index (GII) (Lufaso and Woodward, 2001) shows up as an important feature as does the calculated Goldschmidt tolerance factor (Calc t_IR).…”

Research Update: Towards designed functionalities in oxide-based electronic materials

“…Such functionality metrics are created in practice using a combination of detailed study of the underlying physics of the problem, and more recently facilitated by statistical learning tools. [25][26][27][28] Note that whereas a FM is an a priori quantity developed from a theoretical or empirical model from domain knowledge, the related but distinct term of "descriptors" is typically distilled ex post facto from mining data. Calculations such as those based on DFT then provide a theoretical value for the functionality metric-usually some composite parameters beyond total energies and standard single-particle band structures-that reflects the desired functionality.…”

Accelerated discovery of cathode materials with prolonged cycle life for lithium-ion battery

“…This can be used for high-efficiency material exploration. Another approach is based upon state-of-the-art machine-learning algorithms to search the optimum in the chemical compositional space 7 . A combined approach of both techniques has also been used 8,9 .…”