Application of Computer-Aided Molecular Design to a Pharmaceutical Separation

Solvent selection is a key challenge in designing separation processes for pharmaceuticals. In this work, we seek to find suitable solvents and anti-solvents for the removal of 2-phenylacetamide (PA) from N,N-dimethylformamide (DMF). Solvents are sought which are immiscible in DMF, for use in liquid-liquid extraction processes, but we also consider anti-solvents which are miscible in DMF, but change the solubility to allow direct crystallization of the solute. The software ICAS (PSE for Speed, Inc.) was applied to perform the molecular design. Property targets were set for boiling point, octanol/water partition coefficient, Hansen solubility parameter, and molecular weight. The software starts with a molecular space spanning thousands of potential solvents, and provides a list of the five best solvents in terms of matching the property targets. Some of these solvents have been previously synthesized, and have properties measured in literature, while others are truly novel. The software also provides property estimations for each solvent. A liquid-liquid extraction process was then simulated in ASPEN, with properties of the novel solvents estimated from structure. This process was then optimized to minimize energy use while maintaining a given purity. Results show an efficient process which recovers a high percentage of the solute. Future work will apply CAMD to the anti-solvent crystallization process as well.