Simultaneous prediction of vapor-liquid and liquid-liquid phase equilibria in systems of ionic liquids belonging to [Cnmim][BF4] and [Cnmim][PF6] families by CP-PC-SAFT and SAFT-VR-Mie with universal kij values

This study examines the capabilities of CP-PC-SAFT and SAFT-VR-Mie without association, polar and elec-trostatic contributions to predict phase equilibria in the systems of ionic liquids (ILs) [Cnmim][BF4] and [Cnmim][PF6] (2 < n < 12). The molecular parameters of both models for these ILs were generalized by the 1st order polynomials as functions of molecular weights. The models were fitted to VLE in the system CO2 - [C4mim][BF4]. k(12) = 0.04 was obtained for CP-PC-SAFT and k(12 )= 0.03 - for SAFT-VR-Mie. These val-ues of the binary parameters were applied for predicting phase equilibria in other systems of CO2 , N2O, CH4 , O-2 , CO, C(3)H(8 )and vinyl chloride with the considered ILs. It was found that both models exhibit robust and far-going predictive capacities. SAFT-VR-Mie was found superior in predicting the available data on LLVE compositions in the systems of CO2 and N2O. CP-PC-SAFT was more accurate in estimating the densities of saturated phases. Some VLE data were predicted more accurately by SAFT-VR-Mie and others - by CP-PC-SAFT. Both models correctly estimate that the ranges of LLE phase splits increase in the sequence of thiophene -> benzene -> toluene -> xylenes -> n-alkanes and yielded reasonably good results for phase equilibria and densities of saturated phases in ternary systems of the considered ILs with ali-phatic and aromatic hydrocarbons. At the same time, the overall accuracy of CP-PC-SAFT in predicting the available binary LLE data was better. With the universal value of k(12) = 0.04 this model precisely predicts the UCST in the system aniline-[C4mim][BF4]. (C) 2022 Elsevier B.V. All rights reserved.

Automated summary

This study examines the capabilities of CP-PC-SAFT and SAFT-VR-Mie models to predict phase equilibria in different ionic liquid systems. The results show that both models exhibit robust predictive capacities, although their performance varies in different systems.