Abstract
Molecular electrocatalysis campaigns often require tuning multiple experimental parameters to obtain kinetically insightful electrochemical measurements, a prohibitively time-consuming task when performing comprehensive studies across multiple catalysts and substrates. In this work, we present an autonomous workflow that combines Bayesian optimization and automated electrochemistry to perform fully unsupervised cyclic voltammetry (CV) studies of molecular electrocatalysis. We developed CV descriptors that leveraged the conceptual framework of the EC’ kinetic zone diagram to enable efficient Bayesian optimization. The CV descriptor’s effect on optimization performance was evaluated using a digital twin of our autonomous experimental platform, quantifying the accuracy of obtained kinetic values against the known ground truth. We demonstrated our platform experimentally by performing autonomous studies of TEMPO-catalyzed ethanol and isopropanol electro-oxidation, demonstrating rapid identification of kinetically insightful conditions in 10 or less iterations through the closed-loop workflow. Overall, this work highlights the application of autonomous electrochemical platforms to accelerate mechanistic studies in molecular electrocatalysis and beyond.
Supplementary materials
Title
Supporting Information
Description
Additional characterization and data.
Actions
Title
COMSOL Report
Description
Report for the CV model generated by COMSOL software
Actions
Supplementary weblinks
Title
Zeonodo Repository
Description
All data and code for the manuscript. Included are a large training dataset of simulated CVs, as well as experimental data for TEMPO-catalyzed alcohol oxidation. Jupyter notebooks and Python scripts for generating, processing, and analyzing the data are included.
Actions
View