Machine-Learning-Assisted Discovery of High-Efficient Oxygen Evolution Electrocatalysts.

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry

Background:

• Iridium oxide (IrO2) is a key electrocatalyst for the oxygen evolution reaction (OER).
• Its widespread application is hindered by low efficiency and high cost.
• Novel IrO2-based materials are needed to overcome these limitations.

Purpose of the Study:

• To develop a computational strategy for discovering enhanced IrO2-based electrocatalysts.
• To screen for novel binary and ternary metal oxides with improved OER activity.
• To accelerate the identification of superior electrocatalyst candidates.

Main Methods:

• Employed high-throughput density functional theory (DFT) calculations.
• Integrated machine learning (ML), specifically a neural network language model (NNLM).
• Evaluated 36 metal dopants across 4648 stable surface structures for OER activity.

Main Results:

• Successfully associated atomic environments with crystal formation energies and OER intermediate free energies.
• Screened a series of potential candidates with superior OER catalytic activity.
• Demonstrated an efficient method for exploring complex multi-metallic compounds.

Conclusions:

• The combined DFT and ML approach effectively identifies promising electrocatalyst materials.
• This strategy significantly accelerates the discovery of advanced IrO2-based OER catalysts.
• The findings pave the way for more efficient and cost-effective electrochemical applications.