(672a) Control of Crystal Morphology through Process Parameter Optimization Guided By 2d PBM | AIChE

(672a) Control of Crystal Morphology through Process Parameter Optimization Guided By 2d PBM


Seifried, B. - Presenter, Massachusetts Institute of Technolgy
Rosenbaum, T., Bristol-Myers Squibb
Gamble, J., Bristol-Myers Squibb
Engstrom, J., Bristol-Myers Squibb
Schacht, U., EPSRC Centre for Innovative Manufacturing in Continuous Manufacturing and Crystallisation University of Strathclyde
Melkeri, Y., GSK R&D
Mitchell, N., Process Systems Enterprise
The primary method of controlling API powder properties is through the design of a robust API crystallization process. The crystallization process has historically been developed through a combination of screening work and design of experiments (DoE), which can be both materials and labor intensive. To rapidly address evolving API powder properties control needs, we have developed a hybrid modeling and experimental workflow for crystallization process development. The workflow involves creation of a population balance model (PBM) via gPROMS FormulatedProducts (gFP, PSE Siemens), based on desupersaturation and PSD data gathered from initial crystallization experiments utilizing EasySampler equipped with a novel filter frit (Mettler Toledo), and Morphologi G4 (Malvern Panalytical Ltd). We demonstrate the effectiveness of the workflow by applying it to a needle-shaped API (Figure 1,b) with a cooling crystallization process. Separate growth kinetics were used to describe the major and minor axis growth, and it was found necessary to utilize a 2D seed model to accurately describe the seed PSD. The model provided a good fit to the desupersaturation data and particle aspect ratio (AR), accurately capturing that the relative supersaturation (RSS) at seed point was a main factor in influencing product AR (Figure 1, a). The model was then used to probe product AR when the seed type (PSD and AR) was varied. The model predicted that use of lower AR seeds would yield lower product AR that in turn led to efforts to generate more equant seeds both via micronization (jet milling) and temperature cycling. Experiments with the different seed types validated the model predictions and yielded lower AR products (Figure 1, c).