Parametrization of PC-SAFT EoS for Solvents Reviewed for Use in Pharmaceutical Process Design: VLE, LLE, VLLE, and SLE Study

Abstract

The perturbed chain-statistical associating fluid theory equation of state (PC-SAFT EoS) is one of the state-of-the-art thermodynamic models used in the phase equilibrium calculation of associating mixtures, in particular, in the pharmaceutical industry. Accordingly, parametrization of the PC-SAFT EoS for approved solvents reviewed for use in pharmaceutical process design by the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) was done in the present study. First, the PeC-SAFT EoS model parameters for 41 pure solvents were regressed (of 62 solvents). The available high-quality binary vapor–liquid equilibrium (VLE) data (944 out of 1891 binary systems) and also binary liquid–liquid equilibrium (LLE) data were used to tune the interaction parameters of the PC-SAFT model between the 62 solvents. The interaction parameters of 944 binary systems from 1891 (50%) solvent–solvent systems were regressed. To best fit the PC-SAFT model to the experimental data of 254 binary systems, in addition to the interaction parameters of the dispersion contribution (k ij ), the temperature-dependent effective cross-association volume of the association contribution (k A i B j ) was also fitted. The latter term plays a very important role in the fitting of the experimental VLE (systems that contain at least one associating component) and also LLE data (254 out of 944 binary systems). Using the obtained binary LLE parameters, the LLE phase behavior of 11 ternary LLE systems was successfully predicted by the PC-SAFT model. In addition, the heterogeneous residue curve map, azeotrope compositions, and temperatures of the water–ethanol–benzene system were successfully predicted by the PC-SAFT model. It was also shown that the performance of the PC-SAFT model was superior to those of UNIFAC, UNIFAC-LLE, UNIFAC-DMD, PSRK, VTPR, and UMR-PR models. Finally, the solubility of aripiprazole, cilostazol, imatinib, linagliptin, and naftopidil was regressed in pure solvents, and their solubility was successfully predicted in binary solvents. Hence, the PC-SAFT EoS in the current state together with the solvents confirmed by the ICH is a useful thermodynamic package to be used in the synthesis of active pharmaceutical ingredients, finished product formulations, and also purification operations.

Phase Equilibrium Calculation of Bio-Oil-Related Molecules Using Predictive Thermodynamic Models

“…So, this model can predict the VLE data well. , Although both VLE and LLE data have been used in the development of the UNIFAC-DMD and NIST-UNIFAC models, the number of available LLE data sets is very low compared to the number of available VLE data sets. So, the weight of the VLE data in the objective function is higher than the LLE, and the obtained interaction parameters inherit the VLE properties . The results of the COSMO-based models are poor even for VLE, and this is considerable in the case of highly nonideal positive LLE systems.…”

“…So, the weight of the VLE data in the objective function is higher than the LLE, and the obtained interaction parameters inherit the VLE properties. 19 The results of the COSMO-based models are poor even for VLE, and this is considerable in the case of highly nonideal positive LLE systems.…”

“…Because of the high cost of the experimental work, the large numbers of compounds present in FPBO, and the nature of bio-oil-related molecules (molecules with high melting points for vapor–liquid equilibrium studies), the availability of experimental phase equilibria data is limited. Therefore, thermodynamic models are an alternative to experimental research to reduce costs and save time. , Various Gibbs free energy models have been developed for phase equilibria calculations of polar and complex mixtures including correlative, predictive, and pure predictive models. Notably, the Wilson, NRTL, and UNIQUAC models are well-established correlative local composition models.…”

“…Therefore, thermodynamic models are an alternative to experimental research to reduce costs and save time. 18,19 Various Gibbs free energy models have been developed for phase equilibria calculations of polar and complex mixtures including correlative, predictive, and pure predictive models. Notably, the Wilson, 20 NRTL, 21 and UNIQUAC 22 models are well-established correlative local composition models.…”

Virtual Issue: A Multiscale View on Molecular and Process Design

“…Thermodynamic modeling was discussed by Specht et al . for the case of poorly specified mixtures, while Yousefi Seyf and Asgari focused on parametrizing the PC-SAFT equation of state for the solvents used in the pharmaceutical industry.…”