Population Balance Modeling To Predict Particle Size Distribution upon Scale-Up of a Combined Antisolvent and Cooling Crystallization of an Active Pharmaceutical Ingredient

Abstract

Herein, a population balance model (PBM) for a combined cooling and antisolvent crystallization process for an active pharmaceutical ingredient (API) has been developed and utilized to predict the product particle size distribution (PSD) for two sets of four ∼35 kg scale plant batches, with good agreement to data. The PBM was constructed from lab-scale (∼10 g) crystallization runs using seed and product PSD measurements along with concentration measurements of the API during batch desupersaturation experiments. The PBM was then used to predict the product PSD for two sets of four plant batches, run using different reactors equipped with different agitator types operated at different agitation rates. Analysis of the crystallization kinetics reveals that secondary nucleation due to attrition has a strong influence on the PSD in the crystallization process of the API, and thus mixing conditions (agitator type, agitator speed, pumping, and power numbers) have a strong effect on PSD. The model provides a more robust particle size control strategy than design of experiment (DOE) studies alone by incorporating fundamental crystallization kinetics, with data from a small set of lab experiments in lieu of extensive DOE studies. This firstprinciple-based approach was useful for enhancing the robustness of the technical transfer process by accounting for impacts on product PSD stemming from process scale-up and parameter changes.

Polymorphic Control and Scale-Up Strategy for Antisolvent Crystallization Using Direct Nucleation Control

“…To overcome this difficulty, a model-based approach can be used. 30 The basic idea of the model-based approach is to collect data from a limited number of well-designed small-scale experiments and use these data to estimate the kinetics of the different crystallization mechanisms and the associated uncertainty regions. 32 In addition, computational fluid dynamics (CFD) models are often coupled to the crystallization model to account for the hydrodynamics of different types of crystallization reactors, volumes or operating modes.…”

“…This makes the crystallization process particularly difficult to scale-up. A successful scale-up is accomplished by identifying mixing requirements and critical mechanisms to obtain the desired product quality. − Traditionally, the influence of various operating parameters is investigated empirically, using design of experiment (DOE), to establish the design space at each scale. In the early stages of the API development, only small quantities of the compound are available; this may enhance the difficulty of performing a successful scale-up due to constraints in the number of experiments that can be performed.…”

“…The DOE approach is resource- and labor-intensive. To overcome this difficulty, a model-based approach can be used . The basic idea of the model-based approach is to collect data from a limited number of well-designed small-scale experiments and use these data to estimate the kinetics of the different crystallization mechanisms and the associated uncertainty regions .…”

Comparison of One-Dimensional and Two-Dimensional Population Balance Models for Optimization of a Crystallization Process for a Needle-Shaped Active Pharmaceutical Ingredient

“…Despite this, the multidimensional PBM of the crystallization process is still underutilized, especially within the pharmaceutical industry. While one-dimensional PBMs are now widely utilized to optimize the API crystallization process and are prevalently published, − there are relatively fewer published multidimensional crystallization PBMs in comparison, and many of the multidimensional models originate from the same few academic groups. ,− …”

Cross-Pharma Collaboration for the Development of a Simulation Tool for the Model-Based Digital Design of Pharmaceutical Crystallization Processes (CrySiV)

“…The PBM concept was introduced in the mid-1960s and gained significant ground especially in academic research . Considerable attention has been focused on PBM-based crystallization process modeling, control, and design. − There are several software tools for crystallization process modeling, including gPROMS, Aspen Plus, and DynoChem. Nevertheless, an early consensus among the ETC crystallization working group members was that there was a need to improve the capabilities, robustness, and user friendliness of current crystallization process simulation software for pharmaceutical applications.…”