The selection of optimal solvent or a mixture of solvents is essentially useful to achieve pharmaceutical production with desirable product and production qualities as well as a minimum cost of production and a maximum space-time yield. Despite the availability of high-throughput technologies, experimental solvent screening is most often expensive and laborious and sometimes simply not possible because of lack of sufficient amount of API or relevant impurities. This is especially the case when many combinations of design needs to be tested for identifying an optimal mixture of binary or ternary solvents made up among several common solvents in pharmaceutical industry. To this end, application of predictive thermodynamic models for estimation of phase behaviors in pure and mixed solvent systems have been shown to be a supporting and complementary tool to design pharmaceutical processes. In this work, different predictive thermodynamic models such as UNIFAC, COSMO-SAC, COSMO-RS and NRTL-SAC were applied for solubility prediction of a steroid-like structure antibiotic. The predicted solubility data were compared with experimental solubility data in n-butanol, isobutanol, methyl tert-butyl ether, methyl isobutyl ketone, ethyl acetate, hexane, methanol, acetone and toluene by calculating the root mean square error (RMSE) in order to evaluate the models capability. Therefore, a model which gives more accurate solubility prediction can be selected. For this specific case study, NRTL-SAC is capable of predicting solubility in a good agreement with experimental solubility with RMSE of 4.5%. Moreover, the application of such predictive thermodynamic models for the design of separation/purification unit operations is sometimes limited by the lack of pure component thermophysical properties like melting point and enthalpy of fusion. This mostly happens in the design of separation/purification unit operations to separate analytically identified but not commercially available impurities. Therefore, in addition to the application of those predictive thermodynamic models for solubility prediction, a group-contribution+ (GC+) based model is applied for pure component thermosphysical property prediction. However, the GC+ based model has a limited capability for a large and complex organic molecules that pure component thermosphysical property prediction is prone to uncertainty. In such circumstance, a reference solvent approach can be applied to predict solubility[1]. The reference solvent approach requires no pure component thermosphysical properties but an experimental solubility data in a single solvent. The experimental solubility data of a noncommercially available impurity can be measured by concentrating a sample containing the impurity. This work concludes by critical examination of the performance and the applicability of the above mentioned four property models under different pure component data constraints especially for solubility calculations in life science product (API) separation applications.
Acknowledgment: We would like to thank the Danish Council for Independent Research (DFF) for financing the project with grant ID: DFF-6111600077B.
See more of this Group/Topical: Engineering Sciences and Fundamentals