Abstract
Experimental search for high-temperature ferroelectric perovskites is a challenging task due to the vast chemical space and lack of predictive guidelines. Here, we demonstrate a two-step machine learning approach to guide experiments in search of xBi[ {{mathrm{Me}}_y' {mathrm{Me}}_{(1 - y)}'' } ]O3–(1 − x)PbTiO3-based perovskites with high ferroelectric Curie temperature. These involve classification learning to screen for compositions in the perovskite structures, and regression coupled to active learning to identify promising perovskites for synthesis and feedback. The problem is challenging because the search space is vast, spanning ~61,500 compositions and only 167 are experimentally studied. Furthermore, not every composition can be synthesized in the perovskite phase. In this work, we predict x, y, Me′, and Me″ such that the resulting compositions have both high Curie temperature and form in the perovskite structure. Outcomes from both successful and failed experiments then iteratively refine the machine learning models via an active learning loop. Our approach finds six perovskites out of ten compositions synthesized, including three previously unexplored {Me′Me″} pairs, with 0.2Bi(Fe0.12Co0.88)O3–0.8PbTiO3 showing the highest measured Curie temperature of 898 K among them.
Highlights
Experimental search for high-temperature ferroelectric perovskites is a challenging task due to the vast chemical space and lack of predictive guidelines
We demonstrate a materials design approach driven by machine learning (ML) and active learning methods to simultaneously predict x, y, Me′, and Me′′ such that the new candidate solid solutions are expected to (i) form in a perovskite structure and (ii) have high ferroelectric Curie temperature (TC)
The novelty of our ML approach lies in the integration of classification learning with regression methods to constrain the search space of possible perovskites so that only promising compositions are recommended for experimental synthesis, characterization, and feedback
Summary
Experimental search for high-temperature ferroelectric perovskites is a challenging task due to the vast chemical space and lack of predictive guidelines. One of the emerging areas in the nascent field of materials informatics is the active learning or adaptive design approach, where the ML models are combined with algorithms that recommend informative experiments (from a vast pool of possible experiments) such that the new data are expected to improve the performance of the ML models in the iteration[17,18,19] Recent demonstrations of these methods to experimentally discover complex organic−inorganic molecules, alloys, and functional oxides are worth mentioning[12,13,14,15,16]. We demonstrate a materials design approach driven by ML and active learning methods to simultaneously predict x, y, Me′, and Me′′ such that the new candidate solid solutions are expected to (i) form in a perovskite structure (with at least 95% phase purity) and (ii) have high ferroelectric Curie temperature (TC). While the classification learning models allow us to screen for candidate chemical compositions that can have perovskite structure, the Perovskite crystal structure (ABO3)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
