Search for Lithium Ion Conducting Oxides Using the Predicted Ionic Conductivity by Machine Learning

A machine learning method was developed, which predicts ionic conductivity based on chemical composition alone, aiming to develop an efficient search for lithium conductive oxides. Under the obtained guideline, material focused Li2O-SiO2-MoO3 pseudo-ternary phase diagram, is predicted have high (>10−4 S • cm−1). We investigated formation range, conductivity, and crystal structure of superionic conductor (LISICON) solid solution Li4SiO4-Li2MoO4 tie line. The LISICON phases about 10−7 cm−1, higher than that end members; however, two orders magnitude lower analogous materials. In addition, experimental values were or three by learning. analysis revealed distance between sites occupancy each site in contributed decrease conductivity. This strong correlation one reasons discrepancy alone value.