Systematic evaluation of parameterization approaches for the ePPC-SAFT model for aqueous alkali halide solutions

This work presents an analysis of various electrolyte SAFT model approaches through a rigorous benchmarking on extensively collected and critically evaluated databases. The primitive mean spherical approximation (MSA) and the Born equation are used respectively for the long-range ion-ion and ion-solvent interactions. For the short range interactions either dispersion, or association, or both (full) are investigated. Doing so, state-of-the-art parameter sets are obtained for the ePPC-SAFT model. Physical consistency is enforced for the parameters in the regression. Efforts are made to reduce the number of adjustable parameters with minimum loss of accuracy. This is done by analyzing the physical indication of the parameters, parameter sensitivity analysis, parameter trends, and trial-and-error. The model and parameter sets accurately represent the mean ionic activity coefficient (MIAC), vapor-liquid equilibria, and density, and accurately predict the osmotic coefficient extrapolated to temperature and salt composition ranges beyond the range of the MIAC data used in the regression. The ion-specific association strategies are found to be approximately as accurate as the salt-specific strategies, and are more accurate than the ion-specific dispersion and full strategies. Contributions of the model terms to the MIAC are analyzed. Temperature-dependence of the MIAC is discussed. The ion-specific association strategies successfully predicts the opposite relative magnitudes of the cation and anion individual ion activity coefficient of the aqueous NaCl and KCl solutions. The information is not included in model parameterization, while all the salt-specific strategies and ion-specific dispersion and full strategies fail. We recommend including the Wertheim association for the short-range ion-ion and ion-solvent interactions, and parameterizing SAFT models in an ion-specific manner using physically consistent parameters.