Autonomous model-based experimental design for rapid reaction development

Abstract

Currently, the development of flow processes is seen as a resource intensive and specialized activity, when compared to batch processes. We describe an optimization workflow, based on newly-developed open-source software...

A field guide to flow chemistry for synthetic organic chemists

“…Besides the use of human intervention to rene the automated continuous-ow platforms, algorithm-driven optimization has gained signicant attention in both academia and industry as a way to explore high-dimensional chemical space and achieve optimal conditions with fewer experiments. 107,[130][131][132][133][134][135][136] There are three main types of algorithms applied for selfoptimization experimentation: local optimization algorithms such as design of experiments (DoE) 137,138 and Nelder-Mead simplex, 139,140 global optimization algorithms like SNOBFIT 141,142 and Bayesian Optimizations, 114,130,[143][144][145] and machine learning algorithms like deep reinforcement learning. 146 Jensen et al are pioneers in this growing eld, having developed various versions of automated continuous-ow platforms, including a fridge-size recongurable platform, 147 a 'plug-and-play' platform, 142 and a robotic platform.…”

Will the next generation of chemical plants be in miniaturized flow reactors?

“…78,79 Combined to intelligent algorithms, it would ultimately lead to autonomous flow platforms able to self-plan and -execute its own synthetic pathway to designated target molecules. [79][80][81] Whereas this road might take some times, efficient black-box algorithms have proven their effectiveness. These algorithms do not require information on the reaction studied and focus on identifying an optimum (yield) rather than generating data to develop reaction models.…”

“…77 The merging of flow reactors with advanced automation and process analytic technology (PAT) significantly strengthens the robustness and versatility of flow processes, either at the R&D or production scales, and provide invaluable tools for fast optimization. 78,79 Combined to intelligent algorithms, it would ultimately lead to autonomous flow platforms able to self-plan and -execute its own synthetic pathway to designated target molecules. [79][80][81] Whereas this road might take some times, efficient black-box algorithms have proven their effectiveness.…”

Design and characterization of a flow reaction calorimeter based on FlowPlate® Lab and Peltier elements

“…In this context, the FlowPlate® Lab modules represent well characterized and established systems. [31][32][33][34][35][36][37][38][39][40] The developed design allows for both isoperibolic and isothermal operation mode. The focus in this contribution is on design of the calorimeter.…”

“…In this context, the FlowPlate® Lab modules represent well characterized and established systems. 31–40…”