(330e) Optimization of Cooling Crystallization of an Active Pharmaceutical Ingredient Undergoing Degradation

Authors: 
Pal, K., Purdue University
Nagy, Z. K., Purdue University
Crystallization is one of the most important unit operations in pharmaceutical manufacturing. Crystallization is the “missing link” between Drug substance and Drug product design and thereby is very important to the successful operation of downstream unit operations. The most important performance indicators of any crystallization process are the process yield and the end-of-batch mean particle size. Frequently Active Pharmaceutical Ingredients(APIs) degrade at higher temperatures due to various reasons viz. solvolysis, etc. This leads to loss of process yield since the crystallization process needs to be carried out from a lower starting temperature. Sometimes degradation of APIs can be so problematic that cooling crystallization has to be discarded as a crystallization strategy and alternate crystallization strategies viz. Anti-solvent crystallization, etc need to be considered.

Unseeded Cooling Crystallization processes typically follow a parabolic cooling profile where initially the system is nucleated very fast and then allowed to grow until the temperature is decreased sharply towards the end of the batch. In this work, the crystallization process has been modeled mathematically using population balance equations for an API undergoing degradation at high temperatures. The optimal cooling profile for the crystallization process has been obtained for different optimization objectives viz. Minimization of concentration of degraded product and Maximization of Mean Crystal Size at end of batch. It has been seen that the typical parabolic cooling profile is not always the optimal cooling profile when the API degrades and a completely counterintuitive optimal cooling profile results when the objective is to minimize the degraded product concentration. An online Ultra High Performance Liquid Chromatography was also used to monitor the concentrations of the API and the degraded impurities. It was seen that adsorption of the impurity on the crystal faces occurs for a particular class of cooling profiles. The process model was modified to account for the simultaneous degradation of API and adsorption of impurity and the parameters of the rate processes were estimated using the on-line measured concentrations.