Machine Learning-Guided Optimization of Bimetallic Oxide Nanostructures for Enhanced OER Electrocatalytic Activity

Abstract

Transition metals are increasingly replacing noble-metal-based OER electrocatalysts due to their low cost and availability. However, they often face challenges of low electrocatalytic activity and stability. To address these limitations, bimetallic or composite materials have been extensively developed, leveraging synergistic effects to enhance catalytic efficiency. Unfortunately, optimizing these systems remains challenging, as the individual contributions of each component to the overall electrocatalytic performance are not yet fully understood, and limited attention has been given to this issue. To address this issue, herein we employed machine learning (ML) algorithms to optimize and identify the most influential component governing OER electrocatalytic efficacy. We synthesized a bimetallic OER catalyst by coating electrospun nanofibers of polyaniline (PA) and cellulose acetate (CA) on nickel foam (NF), followed by drop-casting of CuO-NiO (CNO) onto the nanofiber surface. ML was applied to optimize and construct the best-fit combination of the designed bimetallic OER catalyst. Results reveal that ML-optimized CNO/CA-PA@NF shows high electrocatalytic activity, showing a low overpotential of 326 mV at 10 mA cm^–2, Tafel slope of 52 mV dec^–1, and an onset potential of 1.48 V vs RHE. Additionally, it shows high stability, which could be ascribed to cohesive interfacial interactions between CNO and CA-PA nanofibers. To the best of our knowledge, this is the first report to highlight the transformative role of ML optimization in advancing bimetallic transition-metal-based electrocatalysts, thus paving the way for durable and efficient OER systems for sustainable energy applications.