A Smart Workflow for Performing the Electrochemical Evaluation of Fuel Cell Catalysts in a Precise Way with Ultra-High Reproducibility

摘要

The reproducibility of the rotating disk electrode (RDE) tests has been decisively limited by environmental and process variability in conventional manual sample preparation. To address this challenge, we present autoRDE-film, an automated platform that integrates force-controlled robotic polishing and process-controlled drop-coating to facilitate highly consistent RDE sample preparation. This mechanism reduces the standard deviation of the oxygen reduction reaction (ORR) limiting current from 0.7 mA cm–2 (traditional scheme) to 0.07 mA cm–2 and the standard deviation of the ORR half-wave potential (E1/2) from 0.038 to 0.004 V, ensuring exceptional batch-to-batch consistency. With the help of 480 RDE sample datasets generated by autoRDE-film, we have developed a ResNet-18-based deep learning model that correlates the film morphology with their electrochemical performance, enabling rapid quality classification with an accuracy of 96.3%. The combined workflow achieves a success rate of 90% for uniform coatings in batch preparations, far surpassing those of traditional approaches. By quantitatively controlling all preparation parameters, autoRDE-film establishes a universal and high-efficiency framework for evaluating ORR catalysts, which can be extended to the evaluation of fuel cell and water electrolysis catalysts and offers a transformative tool for AI-aided electrocatalyst optimization.