Universal Catalyst Design Framework for Electrochemical Hydrogen Peroxide Synthesis Facilitated by Local Atomic...

Area of Science:

• Catalysis
• Materials Science
• Computational Chemistry

Background:

• Designing high-performance catalysts is complex due to diverse material structures and active sites.
• A universal framework for catalyst design is needed to overcome these challenges.

Purpose of the Study:

• To develop a universal design framework for high-performance catalysts using machine learning and descriptor-based modeling.
• To accurately predict catalyst performance for the two-electron water oxidation reaction (2e- WOR).

Main Methods:

• Integration of weighted atom-centered symmetry function (wACSF) descriptors with machine learning (ML).
• Utilizing microkinetic modeling and high-throughput screening.
• Developing ML models to predict adsorption free energies (ΔGOH* and ΔGO*) for various catalyst types.

Main Results:

• ML models accurately predicted adsorption free energies (R² = 0.84 for ΔGOH*, R² = 0.91 for ΔGO*) across diverse catalyst families.
• A universal volcano model for 2e- WOR was established, showing good agreement with experimental data.
• High-throughput screening identified LiScO2, achieving 90% H2O2 Faradaic efficiency and 168-hour stability.

Conclusions:

• The developed universal framework enables rational catalyst design for electrochemical reactions.
• This approach facilitates efficient discovery of novel catalysts across different material classes.
• The framework is implemented in the Digital Catalysis Platform (DigCat) for broader application.