UC Berkeley

Reliable prediction of freezing point depression in liquid electrolytes will accelerate the development of improved Li-ion batteries which can operate in low temperature environments. In this work we establish a computational methodology to calculate activity coefficients and liquidus lines for battery-relevant liquid electrolytes. Electronic structure methods are used in conjuction with classical molecular dynamics simulations and theoretical expressions for Born solvation energy, ion-atmosphere effects from Debye-Hückel theory and solvent entropic effects. The framework uses no a priori knowledge beyond neat solvent properties and the concentration of salt. LiPF6 in propylene carbonate (PC), LiPF6 in dimethyl carbonate (DMC) and LiClO4 in DMC are investigated up to 1 molal with accuracy better than 3 C when compared to experimental freezing point measurements. We find that the difference in freezing point depression between the propylene carbonate-based electrolyte and the dimethyl carbonate electrolytes originates from the difference in the solvent dielectric constant.