Antisolvent selection method based on predictive thermodynamic models

초록

Drowning-out crystallization is an effective technique for separating solutes from solutions by introducing a mass separation agent called an antisolvent. While previous studies using COSMO-SAC (SLE) were limited to binary solute-solvent equilibria, this research introduces the first integrated protocol for selecting an antisolvent. This protocol combines COSMO-SAC (SLE) with an infinite-dilution selectivity descriptor and a precipitation mass balance to analyze full ternary systems (solute, solvent, and antisolvent). New solid-liquid equilibrium (SLE) data were collected for succinic, glutaric, and adipic acids in 1-butanol or DMF with nine different antisolvents at a temperature of 298.15 K. Four predictive models-COSMO-SAC (SLE), COSMO-SAC (2017), NIST-modified UNIFAC, and original UNIFAC-were benchmarked against these data. The COSMO-SAC (SLE) model achieved a median relative deviation in precipitated mass of 58%, representing a threefold improvement over the next best model, marking the first quantitative validation of COSMO-SAC (SLE) for predicting precipitation yields. High selectivity values accurately forecast complete drowning-out or liquid-liquid phase separation, while low values indicate full solubility. Analysis of the sigma-profile reveals a connection between the extent of polar and non-polar surface overlap and the observed phase behavior, providing a mechanistic understanding beyond simple empirical benchmarking. The resulting workflow-comprising model screening, selectivity ranking, sigma-profile interpretation, and experimental validation-offers a reliable approach for selecting solvents and antisolvents in pharmaceutical and fine chemical crystallization.

키워드