Machine learning-assisted optimization of multi-metal hydroxide electrocatalysts for overall water splitting

Area of Science:

• Electrochemistry
• Materials Science
• Catalysis

Background:

• Green hydrogen production via electrochemical water splitting is key to replacing carbon-intensive fuels.
• Widespread adoption requires affordable, earth-abundant catalysts.
• Current catalyst development is often time-consuming and labor-intensive.

Purpose of the Study:

• To optimize precursor ratios of hydroxide-based electrocatalysts using machine learning.
• To enhance electrocatalytic performance for overall water splitting.
• To accelerate the discovery of efficient electrocatalysts for green hydrogen.

Main Methods:

• Utilized machine learning models, specifically Neural Networks, trained on experimental data.
• Optimized precursor ratios for hydroxide-based electrocatalysts.
• Characterized the optimized catalyst (molybdate-intercalated CoFe LDH) and tested its performance.

Main Results:

• Neural Network models effectively predicted and minimized catalyst overpotentials within two iterations.
• The optimized CoFe LDH catalyst showed low overpotentials (266 mV OER, 272 mV HER at 10 mA cm⁻²).
• Demonstrated excellent stability for overall water splitting over 50 hours in alkaline electrolyte.

Conclusions:

• Machine learning significantly accelerates electrocatalyst optimization compared to traditional methods.
• The developed synthesis procedure is scalable and suitable for industrial implementation.
• Optimized catalysts show promise for cost-effective green hydrogen production.