Machine learning and DFT database for C-H dissociation on single-atom alloy surfaces in methane decomposition

Area of Science:

• Catalysis
• Materials Science
• Computational Chemistry

Background:

• Single-atom alloy (SAA) catalysts offer uniform active sites and high selectivity for efficient methane decomposition.
• Methane decomposition is a key process for hydrogen production, but optimizing catalysts is challenging.
• Reducing CO2 emissions during hydrogen production is a critical global energy goal.

Purpose of the Study:

• To develop a comprehensive database of C-H dissociation energy barriers on single-atom alloy (SAA) surfaces.
• To leverage machine learning (ML) and density functional theory (DFT) for predicting catalytic properties.
• To support the advancement of efficient and low-emission hydrogen production technologies.

Main Methods:

• Utilized first-principles density functional theory (DFT) calculations to determine energy barriers on various SAA surfaces.
• Trained machine learning (ML) models on DFT-derived data to predict energy barriers across a wide range of SAA compositions.
• Compiled a dataset of 10,950 entries, including descriptors and energy barriers, for SAA catalytic performance.

Main Results:

• Generated a large-scale, validated dataset of C-H dissociation energy barriers for SAAs.
• ML models demonstrated high reliability in predicting energy barriers, confirmed by comparison with existing DFT calculations.
• The dataset provides crucial insights into SAA catalytic mechanisms.

Conclusions:

• The developed database and computational tools significantly enhance understanding of SAA catalysts for methane decomposition.
• This resource accelerates the design of novel SAAs for efficient, sustainable hydrogen production.
• Public accessibility of the data and tools promotes further innovation in catalysis and clean energy.