Machine Learning-Enhanced Screening of Single-Atom Alloy Clusters for Nitrogen Reduction Reaction

Area of Science:

• Catalysis
• Materials Science
• Computational Chemistry

Background:

• Electrochemical nitrogen reduction reaction (eNRR) offers a sustainable route to ammonia (NH3) production.
• Single-atom alloy catalysts (SAACs) are highly efficient due to atomic precision and optimal atom utilization.

Purpose of the Study:

• To investigate structure-activity relationships in dodecahedral nanocluster-based SAACs for eNRR.
• To accelerate the discovery of novel SAACs for efficient and selective ammonia synthesis.

Main Methods:

• Density functional theory (DFT) calculations combined with machine learning (ML) for catalyst screening.
• Screening over 300 nanocluster-based SAACs with various transition metal dopants.
• Utilizing DFT+ML to predict catalyst stability and eNRR performance.

Main Results:

• Facet sites identified as optimal doping positions for SAACs.
• Late group transition metals demonstrated superior stability and activity for eNRR.
• Eight highly suitable SAACs were identified, with valence d-electron count being a key factor for activity.
• Identified catalysts showed minimal hydrogen evolution reaction (HER) activity.

Conclusions:

• Machine learning significantly accelerates the design and discovery of SAACs for eNRR.
• The developed DFT+ML approach provides a powerful tool for identifying efficient catalysts for sustainable ammonia production.
• This study highlights the potential of SAACs in advancing green chemistry and fertilizer production.