Design of Continuous Crystallizers for Production of Active Pharmaceutical Ingredients

The production of Active Pharmaceutical Ingredients (APIs) is conducted primarily in batch processes. This manufacturing approach is reinforced by a patent-driven business model and the need to minimize the process development times for newly patented drugs. However, the regulatory and business environments are now changing.

The increasing costs of drug development, combined with the strict regulations and the competition from generic manufacturers, have pushed pharmaceutical companies to seek cheaper and more sustainable production methods. Transition from batch to Continuous Pharmaceutical Manufacturing (CPM) could lead to significant reductions in the production costs and an improved consistency of the product quality.

As a result, development of such processes has received a significant interest in the past decade. To be able to compete in a patent-driven industry with relatively small annual production rates, CPM should be conducted in versatile units that offer short process development times and can be used for production of different compounds.

This PhD project deals with the development of novel crystallizer configurations and process design methods oriented to the crystallization of APIs with strict requirements for the control of crystal size and shape. The project includes the development of methods for the early assessment of crystal quality and the evaluation of techniques for improved control of crystallization kinetics in continuous systems.

In the first block of the PhD, a two-stage continuous Mixed Suspension Mixed Product Removal (MSMPR) crystallization setup was designed for the production of an API presenting elongated crystals. A step by step characterization was applied based on image analysis of the crystallization magma, from which the effects of process conditions on crystal size and shape were evaluated.

Crystal breakage was found to be highly selective for a single crystal plane, leading to a significant broadening of the crystal shape distribution. This behavior was consistent with the observations in full-scale batch production, where the crystallization product is subject to significant mechanical stress in downstream processing.

The attainable regions for the MSMPR cascade were obtained through a population balance model that is based on the real crystal dimensions obtained from image analysis. Finally, the crystallizer was optimized for a crystal dimension that is consistent through a moderate degree of crystal breakage during downstream processing.

The second block of the PhD involves a fundamental study of the effect of gas dispersion on crystal nucleation kinetics. It is frequently stated in the literature that the presence of an inert gas in a crystallizer can have an impact on crystallization kinetics, either via an improved mass transfer in the crystallizing suspension or by promoting heterogeneous nucleation.

These statements are supported by a variety of studies in batch mode. However, the mechanisms are not yet fully understood. In this thesis, the effect of injecting a saturated gas on batch crystallization kinetics has been evaluated from experimental induction times. Combining induction time statistics with a detection method based on sample turbidity, the average time for crystal formation is separated from a detection delay that is a function of the rates of secondary nucleation and crystal growth.

Results show a consistent 5-fold reduction in the detection delay for two model systems, and an effect on primary nucleation that is sensitive to the gas injection rate and the studied solute. These results indicate that the induction time reductions frequently reported in the literature could actually be a consequence of a faster crystallization rate after the first nuclei is formed.

The mechanism behind these observations is presumably related to a significant improvement in the mixing pattern and intensity. A novel continuous crystallizer design based on self-induced gas dispersion is presented and evaluated in the last block of the PhD. The objective was to evaluate if gas dispersion could be used to generate smaller crystals in an MSMPR crystallizer, as well as to further develop the understanding of the effect of a moving gas on secondary nucleation and crystal growth.

The effect of gas dispersion on crystallization yield and crystal size distribution has been evaluated for a configuration that would be expected in an implemented process, and for operating conditions that are already optimized for the generation of small crystals. Results show that, in contrast with the observations in batch crystallizers, the effect of gas dispersion in a well-mixed MSMPR crystallizer is very limited.

Further studies on the effect of impeller speed revealed that crystallization kinetics are not sensitive to variations in the mixing intensity for conditions that meet the requirements for homogeneous three-phase mixing. Results from this study further support the hypothesis that a moving gas phase is an alternative to promote different mixing conditions and demonstrate the limited applicability of this technique in a continuous MSMPR crystallizer.