ABO3 Perovskites' Formability Prediction and Crystal Structure Classification using Machine Learning

TL;DR

This paper presents a machine learning framework that accurately predicts the formability and classifies the crystal structure of ABO3 perovskites, enabling faster material screening for photovoltaic applications.

Contribution

It introduces a high-accuracy machine learning approach for predicting perovskite formability and structure, streamlining the material development process.

Findings

Formability prediction accuracy of 98.57%

Crystal structure classification accuracy of 90.53%

Framework applicable to other material properties

Abstract

Renewable energy sources are of great interest to combat global warming, yet promising sources like photovoltaic (PV) cells are not efficient and cheap enough to act as an alternative to traditional energy sources. Perovskite has high potential as a PV material but engineering the right material for a specific application is often a lengthy process. In this paper, ABO3 type perovskites' formability is predicted and its crystal structure is classified using machine learning with high accuracy, which provides a fast screening process. Although the study was done with solar-cell application in mind, the prediction framework is generic enough to be used for other purposes. Formability of perovskite is predicted and its crystal structure is classified with an accuracy of 98.57% and 90.53% respectively using Random Forest after 5-fold cross-validation. Our machine learning model may aid in the accelerated development of a desired perovskite structure by providing a quick mechanism to get insight into the material's properties in advance.