The separation of benzene and trace thiophene by extractive distillation was intensified in two aspects, that is, selection of a suitable entrainer and improvement of the process. The mixture of dimethylformamide (DMF) and an ionic liquid (IL) was chosen as the entrainer. Vapor–liquid equilibrium (VLE) experiments using pure DMF and a mixed entrainer were conducted, and UNIFAC model for ILs was extended to the benzene‐thiophene‐DMF‐IL system. The results demonstrated that volatilization loss of DMF in the vapor phase was significantly reduced by adding IL. Moreover, an improved process with only four columns using a mixed entrainer was proposed. The results indicated that the improved process is more promising for decreasing energy consumption and equipment investment compared with the conventional six‐column process. The total heat duties of reboilers and condensers was decreased by 6.47% and 6.41%, respectively. The process intensification strategy may be directly extended to separate trace components of other systems. © 2015 American Institute of Chemical Engineers AIChE J, 61: 4470–4480, 2015

A united chemical thermodynamic model, that is, the COSMO-UNIFAC model, was first proposed to predict the phase equilibrium of multicomponent systems in which the UNIFAC model parameters are missing. This model combines the advantages of the UNIFAC model (accurate prediction) and the COSMO-based models (a priori prediction). The predicted vapor−liquid equilibrium results by the COSMO-UNIFAC model were compared with experimental data from the literature and this work, confirming that it can provide a moderate quantitative prediction even if the UNIFAC model parameters are missing.

The united chemical thermodynamic model, that is, COSMO-UNIFAC model was first extended to the systems containing ionic liquids (ILs). This model for ILs combines the respective advantages of COSMO-based (priori prediction) and UNIFAC (relatively accurate prediction) models. The comparison of the predicted values by COSMO-UNIFAC model with experimental data indicates that this model can provide a moderate quantitative prediction for the systems containing ILs when the UNIFAC model parameters are vacant.

Vapor pressure data for the binary systems (water/methanol/ethanol + 1-ethyl-3-methylimidazolium acetate ([EMIM]+[Ac]−)) and the ternary systems (methanol + water + [EMIM]+[Ac]−, ethanol + water + [EMIM]+[Ac]−, and methanol + ethanol + [EMIM]+[Ac]−) were measured by a modified equilibrium still. For the above systems, the maximum average relative deviation between experimental data and the UNIFAC-Lei model predicted values was 7%, confirming the prediction accuracy of the UNIFAC-Lei model. Thus, this model was further used to predict the isobaric vapor–liquid equilibrium (VLE) data at 101.3 kPa for the methanol + water + [EMIM]+[Ac]−, ethanol + water + [EMIM]+[Ac]−, and methanol + ethanol + [EMIM]+[Ac]− systems at a fixed mole fraction of ionic liquid (IL) (x IL = 0.2). It was demonstrated that the ionic liquid [EMIM]+[Ac]− was an appropriate entrainer to separate the methanol–ethanol–water mixture. On this basis, the extractive distillation process was simulated using the rigorous equilibrium (EQ) stage model. The results showed that the entrainer consumption, the heat duty of total reboilers, and the heat duty of total condensers decrease by 25, 6, and 6%, respectively, when [EMIM]+[Ac]− replaces the conventional entrainer ethylene glycol (EG). Furthermore, the σ-profiles and excess enthalpies obtained by the COSMO-RS model provided theoretical insights into the separation mechanism at the molecular level.

No abstract