Electrolytes for lithium ion batteries are complex formulations of solvents, salts, and additives. The formulation components affect properties of both the bulk solution and the solid electrolyte interphases on the cathode and the anode. Bulk properties, such as ionic conductivity and viscosity, are typically dominated by solvents and salts and can be predicted by models, such as the Advanced Electrolyte Model developed by Idaho National Lab. However, contributions to electrolyte performance due to additives are much more difficult to predict.

Systematic studies of electrolyte additives require evaluation of those additives in combination with a variety of solvents, salts, and other additives. The additive performance will also depend upon the active materials in the cell as well as the potentials at the anode and cathode. This presentation summarizes a practical approach to effective data gathering in such systems for use in establishing machine learning algorithms to predict performance of new formulations.