Using Morphological Population Balance to Develop a Model-Driven Quality-by-Design Approach for Crystallization Processes

Quality by design-based crystallization of active pharmaceutical ingredient (API) particles with optimal properties for lean drug-product manufacturing remains a long-standing goal of the biopharmaceutical industry; however, its practical application is difficult. Particle and powder properties, such as size, shape, and flowability, along with the key performance indicators of crystallization, such as purity, yield, and robustness, must be considered for the process design. However, considering these parameters for a dynamical system, such as crystallization, is particularly challenging as the involved process conditions and product attributes are strongly interdependent. This challenge is addressed herein by applying morphological population balance (MPB) modeling to the crystallization of an API with a needle-like morphology and poor powder properties, which pose challenges during downstream processing. The potential of seeded batch cooling crystallization is explored for simultaneously improving the particle size and aspect ratio of the API by manipulating seeding conditions, such as seed loading and morphology. Moreover, a novel calibration strategy was developed for the obtained MPB model by performing sequential thermal cycling–wet-milling experiments. The desired kinetic parameters were obtained via these experiments with minimal material consumption and time. The calibrated model was then used to construct probabilistic-design spaces for different seeding conditions and assess the feasibility and reproducibility of the crystallization yields. Results revealed that for persistent needles, relatively small uncertainty in the estimated kinetic parameters cascades to significant shrinkage of the design space. Owing to this feature, batch seeded cooling crystallization is not a suitable platform to control the particle size and morphology of the API of interest; even if the seeding strategy includes high, by industrial standards, seed loadings (5–10%) and low aspect ratio seeds (<2.5). Thus, alternative approaches, such as crystal nucleation control and/or particle engineering, are required to overcome the limitations of APIs having persistent needle-like morphologies. Overall, this study demonstrates a novel paradigm for the calibration and interrogation of the MPB model, which are essential for their industrial application and regulatory compliance.