(141a) A Streamlined Approach to Avoiding Oiling in Crystallization Processes of Api's and Intermediates

Crystallizations involving the presence of two liquid phases are sometimes carried out despite their inherent shortcomings in controlling mixing, particle attributes and impurity profile. The occurrence of a Liquid-Liquid phase separation, also refered to as demixion or oiling depends on thermodynamic considerations through the relative position of the solubility and oiling curves. In addition, a kinetic component needs to be considered mainly through the position of the metastable curve relative to the oiling curve. Demixion has also a kinetic component to it and can be quantified by the spinodal curve. The crystallization process of a pharmaceutical intermediate is obtained by the neutralization of its sodium salt using acetic acid in the presence of ethanol and water. The substrate is crystallized out by cooling and further addition of water to improve yield. Oiling is observed in the cooling stage despite the full miscibility of the solvents involved in the process. The most comprehensive way of tackling an oiling issue lies in generating solubility, supersaturation, demixion and spinodal curves. Such undertaking becomes a daunting task when three solvents are involved. Besides, material availability and time constraints do not allow such extensive thermodynamic and kinetic measurements. The objective of this presentation is to present a simple, yet effective method of avoiding oiling through a reverse addition process while keeping the same relative final solvent composition. The method relies on mapping areas where oiling and self seeding take place while generating solubility and self seeding curves expressed as a function of the total amount of transferred solution. These sets of data are obtained in the course of running several reverse addition experiments with a substrate solution of a fixed composition while the composition of the quench solution is varied. These experiments serve the purpose of collecting key thermodynamic data tied to the operating procedure as well as information relative to crystal attributes and desupersaturation kinetics. We will show how the new approach to solubility and supersaturation curves generation can be used to estimate the optimal operating conditions with minimal effort. Overall, only a set of four solubility and self seeding curves have been needed for the process optimization effort. The lab development described herein has led to straightforward implementation at the commercial scale.