Predicting Anion Redox in Secondary Battery Cathode Materials with a Data-Driven Model

Abstract

In this study, a new empirical model for predicting the likelihood that a material will exhibit anion redox under cation intercalation is developed with a machine learning approach, and many promising new materials are predicted by applying the model to thousands of candidates. This model is applied to a subset of the Inorganic Crystal Structure Database to determine trends in reported literature materials that can guide design and exploration of new materials that exhibit anion redox to obtain high energy storage capacities without the pitfalls, such as low cyclability, that plague known anion redox materials. Anion redox cathodes improve the energy density of current lithium and sodium secondary batteries owing to their ability to charge compensate mobile cation insertion/extraction through changing the oxidation state of the anion in addition to a transition-metal species. Although the true mechanism through which anion charge compensation occurs has not been fully elucidated, materials that exhibit this phenomenon have recently become the topic of intense interest given their potential to help improve the energy density of secondary batteries beyond current capabilities. Anion close-packed structures and high-valent transition metals are confirmed to be key attributes for enabling anion redox in these materials.