Machine Learning Optimization of Energy Materials Design and Development

Abstract

Machine learning (ML) is quickly becoming a core tool for speeding up the discovery as well as the design and development of various materials used in energy conversion and storage. The joint use of data-driven models, density functional theory (DFT), high-throughput experimentation and material design approaches have enabled scientists to explore and screen the chemical space that were computationally or experimentally inaccessible.This article describes most commonly used ML techniques for the study of energy materials, presents major applications in areas of electrolytes and electrodes batteries, photoelectric materials, electrocatalysts and thermoelectrics and discusses the main issues such as small quantity of data at disposal, lack of understanding and low performance prediction that have limited the implementation of these materials in real life.