Predicting Gas Adsorption on Dual-Atom Catalysts Embedded in Graphene via Integrated DFT and Machine Learning

Area of Science:

• Materials Science
• Computational Chemistry
• Catalysis

Background:

• Rational design of high-performance dual-atom catalysts is hindered by underdeveloped predictive frameworks for gas adsorption properties.
• Understanding atomic-scale interactions is essential for advancing catalyst performance.

Purpose of the Study:

• To systematically investigate gas adsorption on bimetallic-doped graphene systems.
• To establish a descriptor-driven framework for predicting adsorption energies and guiding catalyst design.

Main Methods:

• Density Functional Theory (DFT) calculations were employed to study CO, CO2, H2, and NO adsorption on ten bimetallic-doped graphene systems.
• Key descriptors such as d-band center, adsorption distance, and interfacial charge transfer were identified.
• Machine learning models, specifically gradient boosting decision trees, were trained using these descriptors.

Main Results:

• DFT analysis revealed d-band center, adsorption distance, and charge transfer as critical descriptors for adsorption strength.
• Machine learning models achieved high accuracy in predicting adsorption energies.
• A descriptor-driven framework correlating atomic-scale features with adsorption energies was successfully established.

Conclusions:

• The identified descriptors and machine learning model provide a foundation for the rational design of dual-atom catalysts.
• This framework enables efficient screening and optimization of catalyst materials.
• Further validation is needed for broader applicability to diverse catalyst and adsorbate systems.