Rules Describing CO₂ Activation on Single-Atom Alloys from DFT-Meta-GGA Calculations and Artificial Intelligence

Area of Science:

• Materials Science
• Catalysis
• Computational Chemistry

Background:

• Single-atom alloys (SAAs) show promise for CO2 hydrogenation catalysis.
• High-throughput screening of SAA compositions is currently inefficient.
• Effective CO2 activation is a critical first step in CO2 conversion.

Purpose of the Study:

• To develop AI-driven rules for identifying effective CO2 activation sites on SAAs.
• To overcome the limitations of traditional materials screening methods.
• To accelerate the discovery of novel SAAs for CO2 hydrogenation.

Main Methods:

• Density Functional Theory (DFT) with the BEEF (Bader-End-to-End Functional) force field was used to model CO2 interaction with 780 SAA surface sites.
• Subgroup discovery (a machine learning technique) was applied to identify patterns in site properties related to CO2 activation.
• Key physicochemical and geometric parameters were identified to describe active surface sites.

Main Results:

• AI successfully derived rules correlating surface site properties with effective CO2 activation, indicated by C-O bond elongation.
• Key descriptors identified include free-atom properties (electron affinity, electronegativity, d-orbital radii) and generalized coordination number.
• The derived rules predicted over 1500 promising surface sites in various single- and dual-atom alloys.
• Validation through DFT calculations confirmed the predictive power of the AI-derived rules.

Conclusions:

• AI-driven rule discovery provides an efficient method for screening SAAs for CO2 hydrogenation.
• The identified rules offer insights into the fundamental properties governing CO2 activation on alloy surfaces.
• This approach significantly accelerates the identification of advanced catalytic materials for CO2 conversion.