429529 Prediction of Liquid-Liquid Equilibria in Electrolyte Systems with COSMO-RS

Tuesday, November 10, 2015: 3:53 PM
255C (Salt Palace Convention Center)
Thomas Gerlach and Irina Smirnova, Institute of Thermal Separation Processes, Hamburg University of Technology, Hamburg, Germany

Due to the strong influence of salts on the solubility of organic components in aqueous solutions, electrolyte systems are of great interest for the development of separation processes. Considering liquid-liquid extraction processes, the partition behavior of polar organic components can be strongly influenced by the addition of salts, allowing the extraction of polar organic substances from aqueous solution. For such applications a large number of different salt and solvent combinations are possible. Furthermore, especially in biotechnological processes various minor components are present in the mixtures and knowledge about their partition behavior is also relevant for the development of an extraction process.

To reduce the experimental effort, predictive models for electrolyte systems are necessary. Previously, we developed an extension of the model COSMO-RS to predict the influence of alkali halide salts on liquid-liquid equilibria [1]. For this purpose, COSMO-RS was combined with the Pitzer‑Debye‑Hückel term to account for long-range ion-ion interactions and element specific parameters for the interaction energies of the ions were introduced. The model was parameterized based on mean ionic activity coefficients in aqueous solutions.

In this work a re-implementation of COSMO-RS was developed to further refine the electrolyte extension. Based on this, the model was enhanced to describe a higher number of ion species. Besides the alkali halide salts, divalent monoatomic ions as well as polyatomic ions have been introduced into the parameterization data set. The number of element specific parameters was reduced by introducing general interaction energy terms into the model. Using these parameters it can be shown that mean ionic activity coefficients can also be predicted for ions that were not part of the training set used for the parametrization. Furthermore, the model was applied to predict mean ionic activity coefficient in mixed-solvent systems. While a qualitative prediction of liquid-liquid equilibria with the resulting parameter set can already be shown, the parameterization is further refined based on liquid-liquid equilibria of salt containing systems to increase the quantitative agreement with experimental results.

Extended Abstract: File Not Uploaded
See more of this Session: Thermophysical Properties and Phase Behavior II
See more of this Group/Topical: Engineering Sciences and Fundamentals