Progressive Learning-Guided Discovery of Single-Atom Metal Oxide Catalysts for Acidic Oxygen Evolution Reaction

Area of Science:

• Catalysis
• Materials Science
• Computational Chemistry

Background:

• The oxygen evolution reaction (OER) is crucial for clean energy but is limited by slow kinetics and high energy requirements.
• Transition metal single-atom catalysts show potential for OER, but conventional methods struggle with intrinsic scaling relationships and identifying high-activity candidates.

Purpose of the Study:

• To overcome limitations in OER catalyst discovery by integrating density functional theory (DFT) with an active learning framework.
• To accelerate the identification of efficient and durable OER catalysts using a novel machine learning approach.

Main Methods:

• Employed a progressive learning strategy within an active learning framework, using DFT to predict adsorption energies as auxiliary features.
• Performed high-throughput screening of 261 transition metal single-atom-doped metal oxides (M_SA-MO_x).
• Utilized electronic structure analysis and constant-potential DFT calculations for mechanistic insights and validation.

Main Results:

• Identified nine top-performing OER catalysts with theoretical overpotentials < 0.5 V, including Mn_SA-RuO₂ and Fe_SA-TiO₂ (< 0.3 V).
• Revealed intermediate binding strength as a key descriptor for OER activity.
• Experimental validation of Mn_SA-RuO₂ confirmed its low overpotential and excellent durability in acidic conditions.

Conclusions:

• The integrated framework effectively accelerates the discovery of efficient OER catalysts.
• Provides mechanistic insights for the rational design of advanced materials for sustainable energy technologies.