Evidence for multi-Li+ clustering in lithium-ion battery electrolytes.

In lithium-ion batteries, the bulk solvation structure of the electrolyte is a major determinant of the physical and electrochemical properties of the cell, for example, conductivity, viscosity, salt dissociation and subsequent reactivity, the electrochemical stability window (and thus solid electrolyte interphase formation), and the composition of the electrochemical double layer1-3. However, elucidation of this structure at an atomic level of detail has been challenging to accomplish via a purely experimental approach. Here we combine machine-learning-enabled atomistic simulations4-16, nuclear magnetic resonance spectroscopy, electrochemical data3,17-20 and quantum-chemical calculations to reveal previously unstudied multi-Li+ complexes without bound anions that are uniquely facile for reduction. This insight explains the success of empirically chosen ethylene-carbonate-containing solvent mixtures and provides a foundation for the rational design of improved electrolytes. Furthermore, our machine-learning force-field-based lithium-ion-battery electrolyte simulations accurately reproduce a wide range of condensed-phase experiments3,21-24, despite being parameterized to only quantum-chemical gas-phase cluster calculations, and hence provide a compelling demonstration that machine-learning force-field-based models developed from first principles can robustly represent a complex chemical system to atomic resolution.