Crystallisation Scale-Up and Optimisation

Small scale high throughput screening can be used to rapidly screen process solvents and crystallisation conditions. However, it provides a limited range of crystallisation conditions (often with limited control) and data types and quality. Scaling up to larger crystallisation vessels (100 mL plus) provides a wider range of process conditions and allows for more data type (e.g. particle size, concentration, etc.) to be collected.

However, working at these larger scales consumes significant quantities of material, which is often limited, and generates waste.

This project will develop an integrated workflow for the modelling and optimisation of process conditions to allow translation of crystallisation behaviours observed in high throughput screening equipment to larger crystallisation processes. This will allow the design of the larger scale processes to be targeted, thereby reducing the number of experiments. Resulting in reduced material usage and waste production.

To achieve this, the project will first perform characterisation of the crystallisation environments in the high throughput screening equipment and larger process equipment. Throught the use of computational fluid dynamic tools, common hydrodynamic parameters such as shear rate, kinetic energy, dissipation rate, etc. can be quantified and compared between the scales.

Secondly, a collection of datasets will be generated, covering a range of active pharmaceutical ingredients with a range of behaviours, in both the high throughput screening equipment and larger scales (the high throughput screening data will be generated in an aligned project within the PhD cohort). This data will include process conditions as inputs and particle size, shape, polymorphic form, yield, etc. as outputs.

Finally, correlations between the scale's characterisation and crystallisation outcomes will be investigated. Based on these correlations the experimental design for the both the high throughput screening and larger scale vessels will be optimised to allow for the best translation of conditions between scales.