Extension of COSMO-SAC Solvation Model for Electrolytes Shu Wang, Yuhua Song, and Chau-Chyun Chen* Aspen Technology, Inc., Burlington, Massachusetts 01803, United States COSMO-based activity coefficient models such as COSMO-SAC (conductor-like screening model-segment activity coefficient) have been shown to be relatively successful predictive models for molecular systems. [As cited in Lin, S. T.; Sandler, S. I. Ind. Eng. Chem. Res. 2002, 41, 899-913 and Mullins, E. et al. Ind. Eng. Chem. Res. 2006, 45, 4389-4415.] In this study, we present an extension of COSMO-SAC to electrolytes (eCOSMO-SAC) that combines the COSMO-SAC term for short-range molecule-molecule, molecule-ion, and ion-ion interactions with the extended symmetric Pitzer-Debye-Hu ¨ckel term for long-range ion-ion interactions. [As cited in Song, Y.; Chen, C.-C. Ind. Eng. Chem. Res. 2009, 48, 7788-7797.] The extension recognizes that like-ion repulsion and local electroneutrality govern the surface segment contacts, and it introduces a dual sigma profile concept for electrolyte systems. [Chen, C.-C. et al. AIChE J. 1982, 28, 588-596.] While the model formulation and parameters remain to be optimized with a greater selection of electrolytes, the eCOSMO-SAC predictions for a few representative electrolyte systems show trends that are in qualitative agreement with experimental data and those generated from the eNRTL model (as mentioned in the Song et al. work previously noted) and demonstrate essential characteristics that are consistent with the general behavior of electrolyte systems. Introduction Electrolyte solutions are ubiquitous in chemical process industries. Current efforts in the development of activity coefficient-based electrolyte thermodynamic models largely follow two main tracks: (1) virial expansion empirical expres- sions represented by the Pitzer equation 1 and (2) local composi- tion semiempirical expressions represented by the electrolyte NRTL model. 2-4 These models provide sound thermodynamic frameworks to quantitatively correlate available thermodynamic data for interpolation and limited extrapolation. More recently, a segment-based electrolyte activity coefficient model has been proposed 5,6 as a correlative and predictive thermodynamic framework. The model requires component-specific “conceptual segment” parameters that can be determined from correlating experimental data in a few representative systems. The model can then be used to qualitatively predict phase behavior of any electrolyte systems, as long as the conceptual segment param- eters are known for the molecules and electrolytes. While validation of the segment-based predictive model is far from adequate and its utility is uncertain, there is a clear need for predictive electrolyte thermodynamic models. COSMO-based activity coefficient models such as COSMO- SAC (conductor-like screening model-segment activity coeffi- cient) 7,8 and COSMO-RS 9 have been shown to be relatively successful predictive models for molecular systems. The COSMO-SAC solvation model uses the “screening charge density” or “sigma profile” of the molecular surface calculated from quantum chemistry as a descriptor to compute the activity coefficient of each component in mixtures. These models are capable of reasonably robust predictions for thermodynamic properties of thousands of components and their mixtures. 10 Although COSMO-based models were originally developed for molecular systems, they were later successfully applied without modification to molecular species in ionic liquids. 9,11 The success suggests that the COSMO-SAC formulation provides adequate representation for the short-range molecule-molecule interactions and, to a certain extent, the short-range molecule-ion interactions. Except in the dilute electrolyte concentration region, short- range interactions are known to play the dominant role in the phase behavior of electrolyte solutions. 3,12 In this study, we present an extension of COSMO-SAC to electrolytes by combining the COSMO-SAC term for short-range molecule- molecule, molecule-ion, and ion-ion interactions with the extended symmetric Pitzer-Debye-Hu ¨ckel term for long-range ion-ion interactions. 2 The extension recognizes that like-ion repulsion and local electroneutrality 3 govern the surface segment contacts, and it introduces a dual sigma profile concept for electrolyte systems. As we explore ways to extend COSMO- SAC to describe the complex behaviors of electrolytes, existing successful correlative models are used as useful references. Specifically, we compare the eCOSMO-SAC predictions for a few representative electrolyte systems with those calculated from eNRTL, 2 which is a widely practiced correlative model for electrolyte activity coefficients. In addition, we examine the sensitivities of the model predictions to the various model parameters. Moreover, the general characteristics of eCOSMO- SAC model predictions are compared with experimental data for selected electrolyte systems, including aqueous single electrolytes, aqueous multielectrolytes, and single electrolytes in mixed solvents. COSMO-SAC. There are many references for the COSMO- SAC model. 7,8 A brief summary of COSMO-SAC is given here as a basis for extension to electrolytes. For molecular systems, the activity coefficient of species i in a solution S can be computed from eq 1: The restoring free energy of the solute i in solution S is * To whom correspondence should be addressed. Phone: 781-221- 6420. Fax: 781-221-6410. E-mail: chauchyun.chen@aspentech.com. ln γ i COSMO-SAC ) ΔG i/S *res - ΔG i/i *res RT + ln γ i SG (1) Ind. Eng. Chem. Res. 2011, 50, 176–187 176 10.1021/ie100689g  2011 American Chemical Society Published on Web 11/24/2010