Correlation and prediction of small to large sized pharmaceuticals solubility, and crystallization in binary and ternary mixed solvents using the UNIQUAC-SAC model

Abstract

The recently reported UNIversal QUAsiChemical Segment Activity coefficient (UNIQUAC-SAC) model [developed by Haghtalab and Yousefi Seyf Ind. Eng. Chem. Res. 2015, 54, 8611] provides a practical thermodynamic framework to be used in VLE, LLE, and SLE calculations. The UNIQUAC-SAC model has the advantage of being independent of area (q) and volume (r) structural parameters used in the combinatorial part. While the UNIFAC or UNIFAC-DMD could not apply to the 47% (44 of 94) of the studied molecules because of the undefined groups. Here, the numbers of solvents with identified segment numbers were extended from 82 to 130 with a slight refinement to the previous values. The model parameters obtained via a large consistent set of VLE (isobaric and isothermal), and LLE experimental data. Also, the model was tested with pharmaceutical solubility experimental data in pure (94 solutes with 6210 solubility data), binary, and ternary solvents. The model provides a robust correlation of pharmaceuticals solubility in a mono-solvent and robust prediction of pharmaceuticals solubility in mixed solvents. The UNIQUAC-SAC model was successfully evaluated in solvent screening in cooling crystallization of ibuprofen and valsartan in both pure and binary solvents mixture (combined cooling and anti-solvent crystallization). The model with 130 solvents as a robust database provides a useful and practical thermodynamic tool in the conceptual design of pharmaceutical processes. A MATLAB based graphical user interface (GUI), finally, was developed to be used in the conceptual segment numbers regression procedure.