Evolution of kinetic motifs through rate-based experimental design in flow reactors

Continuous processing is revolutionising drug discovery and manufacture. Currently there is no established robust scale-up workflow for the development of flow processes in the pharmaceutical manufacturing sector. This presents a significant development risk for the implementation of continuous processing within the pharmaceutical industry in comparison to traditional batch processing, which limits the uptake of flow processes within this industry. The primary goal of this thesis is to devise and demonstrate a development workflow that maximises both understanding and the degree of optimisation of the process operating conditions for a reaction. In particular, this thesis develops an experimental design and fitting strategy, termed rate-based design of experiments (rDoE), which can be applied within the development constraints. The concept of a kinetic motif is applied as a way to represent the time-dependent behaviour in a single-step or multistep reaction system. Initially a workflow for the evolution of kinetic motifs in flow through a rate-based experimental design methodology is established using a Paal‒Knorr reaction. The experimental design investigates process-relevant conditions, and focuses on the regions where most change is observed to improve confidence. The established kinetic model is successfully validated in batch and at a larger scale in an Alfa Laval ART® Plate Reactor. A three-level definitive screening design is combined with reaction profiling (rDSD) to measure the reaction kinetics for a multistep Friedel‒Crafts heteroarylation reaction. The responses are measured by UPLC analysis. The kinetic motif acts as a simplified representation of the system to reveal a sweet spot operating region which is not predicted by statistical models. An automated kinetic platform is developed with at-line HPLC analysis at the reactor outlet, and aromatic nucleophilic reactions are demonstrated using the approach. The reactions are multistep networks which makes prediction of the kinetic parameters for individual pathways difficult. This problem is overcome by exploring a wide design space, from mild to very aggressive conditions using a new linear flow ramp gradient approach. The approach gave high confidence in the established kinetic model. The workflow is applied to a telescoped multiphase process. The example shows that a rDoE approach is fit-for-purpose for establishing rate-based behaviour, even for a multiphase system. The model developed is used to successfully identify the experiments that give the best results at a larger scale.