(71a) Automated Systems for Screening, Kinetics, and Optimization of Chemical Synthesis and On Demand Production

Jensen, K. F., Massachusetts Institute of Technology
Over the two past decades, small-scale continuous synthesis has matured from simple demonstration examples to applications in pharmaceuticals and fine chemicals. The field has moved beyond single transformations to continuous multistep synthesis of active pharmaceutical ingredients (APIs) by incorporating in-line workup techniques. Integration of on-line measurements of reactant flows, reactor temperature, and outlet concentrations with feedback control systems has enabled automated optimization of reaction yields as well as determining kinetic information. The present contribution starts with automated screening and optimization of chemical reaction in microsystems and continues with integration of small-scale reactors and separators in on-demand continuous synthesis of APIs. Automated continuous/oscillating droplet flow systems are demonstrated for simultaneous optimization of discrete process variables (e.g., catalyst, ligands, and solvents) and continuous conditions (temperature and reagent concentrations). Coupling the droplet reactor with a light source converts the system into an automated microfluidic platform for exploring and optimizing visible-light photoredox catalysis.

As an example of process integration for on demand manufacturing, we present a plug-and-play, reconfigurable, refrigerator-sized manufacturing platform for on-demand synthesis of common pharmaceuticals, e.g. ciprofloxacin.This flexible system is capable of complex multi-step synthesis, in-line purification, post-synthesis work-up and formulation. Multistep synthesis occurs at elevated temperatures and pressures to enhance reaction rates, and the resulting residence times are a few minutes, in contrast to the multiple hour-long synthesis typically needed for batch. Typical production rates are grams/hour sufficient to produce thousands of doses per day of common pharmaceuticals. Finally, we discuss opportunities for integration of machine learning for synthesis planning with fully automated chemical synthesis systems.