Rational Designing of Bimetallic/Trimetallic Hydrogen Evolution Reaction Catalysts Using Supervised Machine Learning

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry

Background:

• Precious platinum group metal (PGM) catalysts are expensive for hydrogen evolution reaction (HER).
• Machine learning (ML) offers a promising approach for discovering new catalysts.
• Earth-abundant alternatives are crucial for sustainable energy.

Purpose of the Study:

• To discover cost-efficient, earth-abundant heterogeneous catalysts for HER.
• To utilize a combined density functional theory (DFT) and ML approach.
• To screen NiCoCu alloy-based catalysts for enhanced HER activity.

Main Methods:

• Developed a generalized microstructure model to study adsorbate surface coverage.
• Generated input features for ML models based on surface properties.
• Employed optimized eXtreme Gradient Boost Regression (XGBR) models for catalyst screening.

Main Results:

• Successfully screened active NiCoCu alloy-based HER catalysts.
• Demonstrated the efficacy of the combined DFT and ML approach.
• Identified promising candidates from an extensive catalyst set.

Conclusions:

• The integrated DFT and ML strategy efficiently discovers novel heterogeneous catalysts.
• This approach accelerates the identification of active materials for electrochemical reactions.
• Enables the development of sustainable energy technologies by replacing costly PGMs.