Combating Deviant Crystallisation Behaviours in Commercial API Manufacturing

This thesis aims at investigating the root causes of aggregate formation during the industrial production of Ticagrelor, an anticoagulant active pharmaceutical ingredient patented by AstraZeneca. The study focuses on the crystallisation stage, where the addition of an antisolvent generates supersaturation, inducing crystal formation. Antisolvent crystallisation is widely used for isolating and purifying pharmaceutical compounds, however, it presents challenges such as the formation of unwanted aggregates that affect particle size distribution (PSD), a critical parameter for product processability, quality and bioavailability. For technical and economic reasons, experiments were conducted using a laboratory-scale reactor designed to replicate as closely as possible the industrial process. DynoChem simulations facilitated the scale-down by translating key parameters, keeping constant the power per unit mass, trying to guarantee comparable mixing behavior. A preliminary phase aimed at validating the lab-scale model, verifying that results were comparable across both scales. Then, the study proceeded with a systematic variation of the most influential operative parameters. The primary variable analyzed was the agitator speed, chosen for its significant impact on mixing efficiency and its adjustability on the industrial scale. The exploratory phase was divided into two parts. The first one followed a more traditional approach, changing "one-variable-at-a-time" to evaluate the direct relationship between agitator speed and resulting variations in particle size and aggregate formation. The working hypothesis was that a higher agitation speed would promote better mixing, leading to smaller average particle sizes and a reduced number and size of aggregates. On the other hand, slower agitation was expected to result in insufficient mixing, increasing both number and size of aggregates. While this relationship may seem intuitive, the numerous competing mechanisms in crystallisation processes can lead to unexpected outcomes. The second exploratory phase used a Design of Experiments (DoE) methodology via Modde software to explore the combined effects of agitator speed values and the timing of speed increase during the antisolvent addition stage, identifying an optimal "safe operative area" where the deviation is minimized. To characterize the product, two primary analytical methods were used: microscopy and laser diffraction. Microscopy analysis conducted on the slurry sampled after the end of the addition phase, confirmed that aggregates form during the early stages of crystallization, while laser diffraction provided quantitative PSD measurements on dried samples. Additionally, X-ray powder diffraction verified that the desired polymorphic form was obtained through the laboratory crystallisation. The scope of the research could be expanded by increasing the number of experiments and exploring a wider range of operational variables. Despite the difficulties and technical limitations inherent in scale-down studies, this work provides valuable insights offering a practical reference for future process optimization efforts, helping to guide more targeted modifications to the industrial process, saving time and resources.