Recent Developments in Reactor Automation for Multistep Chemical Synthesis

Abstract

Reactor automation is revolutionising the way new chemical processes are discovered and developed. Assigning repetitive aspects of chemical synthesis to machines, such as experimental execution and data collection, provides more time for researchers to focus on critical interpretation and creative problem solving. The ability to autonomously prepare late‐stage intermediates and complex products, rather than just simple starting materials, will play a central role in applications such as the efficient exploration of chemical space and responsive manufacturing. However, translating automated technologies from specific single‐step tasks to more general multistep syntheses remains a significant challenge, owing to high structural diversity and chemical/physical interdependencies between the steps. Robotic batch and continuous flow platforms are gradually becoming more universal, providing access to a wider range of chemistries required to achieve autonomous multistep synthesis. Advances in process analytical technologies have enhanced our ability to monitor interconnected reactions in real‐time, thus accelerating data collection and giving greater process control for ensuring a high standard of safety and product quality. Integration of these tools with control software creates a feedback loop, which can be harnessed for adaptive and flexible multistep screening or holistic self‐optimisation. This review presents recent developments in the application of automated reactor technologies for multistep chemical synthesis, including batch and continuous flow platforms. Specifically, this review highlights how the integration of control software with advanced process analytical technologies and machine learning algorithms are accelerating the synthesis of complex molecules.

Adaptive mixed variable Bayesian self-optimisation of catalytic reactions

“…[6][7][8][9][10][11][12][13][14][15] Self-optimising platforms leverage automation and robotics to rapidly conduct experiments suggested by optimisation algorithms, thus accelerating process development. 16,17 For example, Taylor et al explored the use of multi-task Bayesian optimisation (MTBO) to efficiently optimise previously unseen C-H activation reactions. 18 Data from previous optimisation campaigns of similar reactions was used to inform and accelerate the optimisation of reactions with differing substrates.…”