Predictive Design of Filtration Processes in the Pharmaceutical Industry – the Impact of Crystal Size and Shape

Abstract

The ubiquitous use of crystallization in pharmaceutical manufacturing requires extensive use of filtration processes to separate the crystals. Filtration is a critical step in the production chain, but this operation can also become one of the bottlenecks of the process due to the intrinsically long processing times. Moreover, filtration can also have a detrimental impact on the properties of the final product. It is for these reasons that the predictive design of this unit operation via computer modeling is an ever-increasing requirement in order to optimize and accelerate process development. Filtration performance is influenced by compound and system-specific parameters, but as well by the size and shape of the crystals, along with their distribution. Unlike the substance-specific properties determined by the synthetic routes of the active ingredients, the particle size and shape have the potential to be controlled. Understanding the relationship between crystal attributes and filterability is paramount to predict filtration and, ultimately, to design the overall process. This work presents a series of research investigations that combine experiments and simulation strategies to bridge this gap. A challenging category of crystals, namely needle-like crystals, i.e., particles exhibiting an elongated morphology, are the main focus of this research project due to their omnipresence in pharmaceutical manufacturing. In the first part of this work, a series of populations of two compounds exhibiting needle-like morphologies are used to study the change in filtration performance with respect to the size and shape of the crystals obtained during the crystallization. The attributes of the crystals are characterized by a standard industrial technique, laser diffraction, and a state-of-the-art stereoscopic imaging tool. The data resulting from both strategies are used to execute the prediction of filterability via statistical regression analysis. It is shown that this modeling approach can predict the best filtering population out of a pool of possible systems and that it can be extrapolated to a different compound exhibiting similar morphologies. The filter cake structure, which determines the performance of the process, is simulated using a Monte Carlo sampling strategy. Here needle-like crystals are represented by spherocylinders, namely cylinders with hemispherical caps at each end. This geometrical model is suitable to study needle-like crystals as it is possible to modify the aspect ratio of the particles. Populations with different aspect ratios and distributions of characteristic dimensions are generated and used to simulate the packing inside the cake. The work confirms previous studies that showed that the porosity of these structures increases with aspect ratio, but also provides evidence that the polydispersity of the populations has a considerable effect on the cake structure as well. Experimental observations have also been used to validate the outcome of the simulations. Finally, a complementary investigation of experimental cake structures is proposed exploiting X-ray tomography. The cake structure and its building blocks are characterized. This analysis enables the extraction of multiple characteristic dimensions of the particles and the appreciation of detailed features. The results are validated with particle size and shape distributions obtained via optical imaging of the crystals prior to filtration. Attention is drawn to the potential use of this approach to characterize how crystal size and shape change during filtration.

Date of Award	1 Aug 2021
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Roger Davey (Supervisor) & Carlos Avendano (Supervisor)