Automated High-Throughput Partition Coefficient Determination with Image Analysis for Rapid Reaction Workup Process Development and Modeling
Sophie Duffield, Luigi Da Vià, Amelia Celeste Bellman, Fabio Chiti
Abstract
With this work, we explore the application of a novel image analysis algorithm in combination with a high-throughput automated workflow to extract partition coefficient measurements and full mass balance from small-scale samples. An image analysis algorithm was developed in MATLAB R2018b to determine the volume of the aqueous and organic phases of the biphasic samples with 95% accuracy. The automated workflow used less than 1% of the typical reagent amounts and provided up to 94% time savings when compared with the conventional partition coefficient determination studies. This approach also proves that it is possible to build thermodynamic models for liquid–liquid equilibrium process steps using small-scale vessels (8 mL) and identify the impact of varying process parameters in silico. The model could predict the system behavior at a kilo scale and resulted in an optimized set of process conditions that increased the product recovery from 88 to 94% theoretical. The good agreement between the model and the experimental data also enabled the impact of process parameters on a critical impurity to be determined, supporting risk assessment and quality by design activities for the case study highlighted.