Speeding up the development of solid state electrolyte by machine learning
Qianyu Hu, Kunfeng Chen, J. Y. Li, Tingting Zhao, Feng Liang, Dongfeng Xue
Abstract
Solid-state electrolytes have been demonstrated immense potential with their high density and safety for Li, Na batteries. The discovery of novel crystals is of fundamental scientific and technological interest in solid-state chemistry. The discovery, synthesis and application of energetically favourable solid-state electrolytes has been bottlenecked by expensive trial-and-error approaches. Machine learning has brought breakthroughs to solid-state electrolytes. Numerous solid-state electrolyte candidates have been screened by different models at multiscale, i.e., interatomic potentials, molecular dynamics, ionic conductivity. Machine learning method also accelerate the synthesis prediction, mechanism discovery and interface design. This review would answer the question what can be done for solid-state electrolytes by machine learning, including descriptor, model, algorithm etc. This paper will promote fast integration between scientists in materials, software, computing discipline.