Computational Investigations of Features for Predicting Ionic Conductivity in Multivalent Solid Electrolytes

Solid-state batteries based on the redox of multivalent ions (Mg²⁺, Ca²⁺, Zn²⁺, etc.) may offer improved safety and performance relative to today’s Li-ion batteries. A significant challenge hindering their development is the sluggish mobility of multivalent ions in most solids. Computational methods for efficiently predicting conductivity can accelerate the discovery of faster ion conductors. Direct first-principles calculations of conductivity are expensive and difficult to automate, which has prompted a search for other properties related to conductivity that are easier to calculate or measure. Previous studies have identified features related to the electronic charge density and phonon spectrum that are correlated with energy barriers for ion migration in monovalent conductors. Results from our first-principles simulations demonstrate that these features are not well correlated with energy barriers for multivalent ion migration. I will discuss potential reasons for this lack of correlation and propose modifications that are found to improve correlations. These findings quantify the promise of using such features to efficiently screen for better multivalent ion conductors.