Association Modeling of Ethanol + Chloroform + Dioxane

Recently Killian et al. (https://dx.doi.org/10.1016/j.fluid.2021.113299, https://dx.doi.org/10.1016/j.saa.2022.121837, https://dx.doi.org/10.1021/acs.iecr.3c00968) have demonstrated that infrared spectroscopic data of alcohols in hydrocarbon exhibit cooperative hydrogen bonding that is inconsistent with the widely used Thermodynamic Perturbation Theory 1 (TPT1) association model implemented in the SAFT equation of state. Experimental spectroscopic data are more accurately modeled with Resummed Thermodynamic Perturbation Theory (RTPT) of Marshall and Chapman (https://dx.doi.org/10.1063/1.4834637). RTPT uses a smaller association strength for the dimer and a single larger association strength for all subsequent oligomers. Application of RTPT to phase equilibria was also demonstrated by Killian et al. The model has been published only for a single associating species (alcohol) in an inert solvent. This work extends both TPT1 and RTPT to an inadequately studied ternary mixture with both self- and cross-associating species (ethanol + chloroform + dioxane) and compares vapor-liquid equilibrium (VLE) modeling results from both to experimental VLE data from Gonzalez et al. (https://dx.doi.org/10.1021/je00034a018).

Vapor-liquid equilibria are calculated using y_iP = x_iγ_iP^SAT_i, where γ_i is the activity coefficient calculated by contributions as ln(γ_i) = ln(γ^association_i) + ln(γ^{combinatorial}_i) + ln(γ^attractive_i). ln(γ^association_i) is the deviation caused by molecular association of molecules, ln(γ^{combinatorial}_i) is the deviation due to sizes of molecules, and ln(γ^attractive_i) is the empirically adjusted residual term for dispersion and polarity. This work focuses on modeling the association contribution to the activity coefficient for each component and the impact of the RTPT or TPT1 assumption when the same cross-association strengths are used to model the experimental data. The association contribution differs significantly for the two models. Spectroscopic data for mixtures are needed to determine which model is most appropriate.