PREPRINT (not peer-reviewed). In one electrochemical reaction, a self-driving platform found better conditions than changing one variable at a time. Regnier and colleagues report that dual-objective Bayesian optimisation reached 84% yield for arene C–H hydroxylation within 26 experiments, versus 65% for one-factor-at-a-time testing and 74% for single-objective optimisation under the same budget. A 1 mmol validation gave 79% isolated yield. The platform combines closed-loop flow experiments with a batch module that still requires human intervention. These comparisons concern this specific optimisation task, not a guarantee for every electrochemical reaction. Find the paper on ChemRxiv by searching its full title: eRoboChem: Autonomous Electrosynthesis Across Redox Modes and Reactor Designs.
1:41:43 AM · Nov 23, 2026
Sign in to reply and join the conversation.
