(545e) Comparison of a One Dimensional and Two Dimensional Population Balance Models for the Crystallization Process of a Needle-Shaped API
AIChE Annual Meeting
2020
2020 Virtual AIChE Annual Meeting
Separations Division
Modeling and Control of Crystallization
Tuesday, November 17, 2020 - 9:00am to 9:15am
Crystallization is the most widely used method for separating an active pharmaceutical ingredient (API) from a reaction product mixture. It functions to control purity, crystal form, particle morphology, and particle size distribution (PSD). The application of population balance modeling (PBM) to crystallization process development has been growing as a result of improvements in process analytical tools (PAT) and numerical software packages. One such improvement has been the ability to account for high aspect ratios particle morphologies, such needles or plates, by treating the major and minor particle axis as independent, and assigning them independent growth and nucleation kinetics. Many APIs crystallize with needle morphology, and modeling these particles as spheres, as is the case with one-dimensional PBM, fails to accurately predict PSD. The advantage of two-dimensional population balance modeling for a needle-shaped API is highlighted here by comparing the one-dimensional model that has been developed for an antisolvent crystallization with the two-dimensional model. While the one dimensional model is useful in optimizing the crystallization process to enhance desaturation kinetics, the two-dimensional model can be utilized to optimize process conditions for minimizing particle aspect ratio. The API utilized for this work had extremely slow desupersaturation, and even after aging the seed bed for more than twenty four hours, the solubility concentration was not achieved. The 1D PBM revealed that the crystallization had a high energy of activation, indicating that the growth kinetics were rate limited by the surface attachment of API molecules, and not diffusion limited. As such, agitation rate was not predicted to effect the desupersaturation, and this was corroborated by experimental work. The model directed the lab scale crystallization development to alter the composition of the mother liquor at the seed point, increasing the relative supersaturation at seeding, which indeed resulted in faster desupersaturation kinetics. However, the 1D model was not able to inform on process conditions for generating lower aspect ratio particles. Given that needles are a challenging morphology for many downstream processing operations, reducing the particle aspect ratio can be beneficial for enhancing proccessability. This capability requires two dimensional PBM.