(321bg) Electrolytes at Equilibrium: Accurate Prediction of Key Thermodynamic, Transport, and Ion Solvation Quantities | AIChE

(321bg) Electrolytes at Equilibrium: Accurate Prediction of Key Thermodynamic, Transport, and Ion Solvation Quantities

Authors 

Gering, K. L. - Presenter, Idaho National Laboratory


The demand for high-performance electrochemical systems that deliver safe, portable power has seen a sharp increase in recent years. This has led to a corresponding need to gain further understanding of molecular-scale interactions that are the basis for a broad spectrum of useful electrolyte properties, such as physical, transport, and thermodynamic quantities. This work has produced an accurate model wherein the treatment of chemical physics is based on the associative form of the Mean Spherical Approximation (AMSA) description of molecular properties and interactions. Effects from ion solvation are explicitly considered as well as ion-ion interactions at equilibrium (CIP and SSIP ion pairs, triple ions, solid solvates). Complementing the AMSA framework are molecular-based governing equations that express the effect of the molecular environment on various ionic transport properties. Properties predicted by the model include electrolyte conductivity, viscosity, density, ionic diffusivity, solvation numbers, activity and osmotic coefficients, dielectric depression, Gibbs free energy of ion solvation, solvent-ion binding energies, etc., from infinite dilution to near-saturation of salt concentration. The model applies equally to aqueous and nonaqueous systems, multi-solvent systems, symmetric and asymmetric salts, and has shown impressive accuracy over wide ranges of solvent composition, solute type, and temperature. Although the immediate application of this work is modeling electrolytes for lithium-ion batteries, other potential applications abound. Modeling results will be shown for an assortment of properties, and compared to experimental data where feasible. In many cases the overall average accuracy of the model is within 5-10 percent of experimental values per dataset, with many model predictions well below five percent deviation from the measured values.