Automated optimization under dynamic flow conditions

Abstract

Automated optimization in flow reactors is a technology that continues to gain interest in academic and industrial research. For drug substance applications, where limited material is available for extensive studies, it is imperative that the automated optimization procedure identify ideal conditions for manufacturing in a resource sparing manner. It is equally as important that these investigations provide data-rich results so that the information can be used for process understanding. Achieving these two objectives in parallel is challenging with traditional automated optimization systems that rely on steady-state data. Dynamic flow systems, which adjust process inputs in a controlled manner to collect transient reaction results, maximize reaction information content. In this work, the gains in reaction knowledge by performing the automated optimization in a dynamic flow system are demonstrated using a nucleophilic aromatic substitution as a case study. A gradient-based search algorithm is used to optimize a multi-faceted objective function that accounts for yield, material input, and productivity. The immense dataset from the automated dynamic optimization was used to establish a reaction model to provide greater insight to the reaction kinetics and selectivity.