An all-in-one multipurpose robotic platform for the self-optimization, intensification and scale-up of photocatalysis in flow

08 June 2023, Version 1
This content is a preprint and has not undergone peer review at the time of posting.


The optimization, intensification, and scaling up of chemical processes are essential and time-consuming aspects of contemporary chemical manufacturing, necessitating expertise and precision due to their intricate and sensitive nature. However, these process development problems are often carried out independently and consecutively, which can exacerbate the already significant consumption of time and resources involved in the process. In this work, we present a versatile, all-in-one robotic platform for the autonomous optimization, intensification, and scaling up of photocatalytic reactions in flow. This platform overcomes associated challenges through the integration of readily available hardware and custom software, offering a hands-off solution. Our open source platform combines a liquid-handler, syringe pumps, a tunable high-powered photoreactor, cheap IoT devices and an in-line NMR to enable automated, data-rich optimization using a Closed-Loop Bayesian Optimization strategy. The use of a high-power continuous-flow capillary photoreactor enables highly reproducible data to be obtained, as it mitigates issues related to mass, heat, and photon transport that are often the main sources of irreproducibility in photocatalytic transformations. A user-friendly graphical interface allows chemists without programming or machine learning expertise to easily optimize, monitor, and analyze photocatalytic reactions for chemical spaces of both continuous and discrete variables. The system's effectiveness was demonstrated by testing it on challenging photocatalytic transformations, which resulted in increased overall reaction yields and an impressive up to 550-fold improvement in space-time yields compared to batch processes. Additional tests on literature-reported reactions previously optimized in flow yielded substantial increases in both yield and space-time yield. Overall, our studies demonstrate that combining flow-based reactor technology with Bayesian optimization yields superior and unbiased results compared to human effort and intuition in terms of pace, precision, and outcomes for the optimization of photocatalytic reactions. Finally, due to its ability to autonomously generate datasets that include both optimal and suboptimal conditions, our RoboChem platform also contributes to advancing the field towards a digitally-driven era in synthetic chemistry.


self optimization
machine learning
bayesian optimization
robotic platform

Supplementary materials

Supplementary Information
Details of the platform, experimental data, procedures and spectroscopical analysis of compounds


Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.