Abstract

Automation, Machine Learning, and Robotics for Flow Chemistry Optimization

Klavs Jensen, Professor, Massachusetts Institute Of Technology

Advances in flow chemistry enabled by automation, machine learning, robotics, and on-line analytics  are highlighted through case studies, including an automated droplet microfluidic electrochemistry platform for redox neutral electrochemistry, reaction optimization and characterization of kinetics, (2) an automated cascade of miniaturized continuous stirred tank reactors (CSTRs) for optimization of flow chemistry and photo-redox catalysis involving suspensions of solids, and (3) an automatic, robot assembled, reconfigurable modular micro/mini-fluidic system for execution and optimization of multistep reactions. Machine learning models for retrosynthesis and forward prediction combined with reaction context identification provide computer aided synthesis pathway planning for this system.  We describe and compare different optimization strategies and discuss challenges and opportunities in further integration of machine learning and synthesis platforms.