Designing a machine learning-optimized nitrogen and sulfur co-doped CoNi@SiO2 electrocatalyst for high performance oxygen evolution reaction

Abstract

Although various heteroatom-doped bimetallic composites have been explored for the oxygen evolution reaction (OER), they often suffer from aggregation and low electrical conductivity, which hinder their electrocatalytic efficacy. Thus, to overcome these challenges, herein, we have employed machine learning (ML) optimization to precisely control the growth of bimetallic Co and Ni nanoparticles on silica (SiO₂) nanospheres (CoNi@SiO₂), thus mitigating the aggregation effect. Additionally, the amount of thiourea, which serves as the source for S and N doping, was optimized using ML. The results reveal that NS-doped CoNi@SiO₂ exhibits enhanced electrocatalytic performance by providing more exposed active sites. The results reveal that the designed ML optimized NS-doped CoNi@SiO₂-based electrode has shown promising OER activity by exhibiting low overpotential (220 mV) and onset potential (1.28 V vs. RHE) compared to CoNi@SiO₂ (300 mV, 1.29 V), SiO₂ (340 mV, 1.34 V vs. RHE) and NF (341.7 mV, 1.51 V vs. RHE). The enhanced performance of NS-doped CoNi@SiO₂ can be attributed to the synergistic effects of NS doping and the bimetallic system, which leads to an increase in the exposition of active sites and surface area. This work highlights the potential of ML optimization in fine-tuning electrocatalyst composition to enhance electrocatalytic performance.