Synergizing Machine Learning with High-Throughput DFT to Design Efficient Single-Atom Catalysts for Hydrogen...

Area of Science:

• Materials Science
• Catalysis
• Computational Chemistry

Background:

• Efficient electrocatalytic hydrogen evolution reaction (HER) is crucial for clean energy technologies.
• Developing novel single-atom catalysts (SACs) offers high efficiency but faces challenges in material discovery due to high costs and long research cycles.

Purpose of the Study:

• To systematically investigate the HER activity of 90 types of single transition metal (TM) and/or nonmetal (NM) atoms bonded in graphyne (TM-NM-GY) SACs.
• To leverage machine learning and high-throughput DFT computations to accelerate the discovery of efficient HER SACs.
• To identify key factors governing HER activity in these SACs.

Main Methods:

• High-throughput density functional theory (DFT) computations were employed to screen 90 graphyne-based SACs.
• Machine learning algorithms, specifically stacking models, were utilized for predicting and designing catalysts.
• Analysis of structure-property relationships, including bond length, d-band center, binding height, charge transfer, and ICOHP.

Main Results:

• Several SACs, including Fe-GY, Fe-B-GY, Ni-B-GY, Pd-B-GY, Sc-N-GY, Co-N-GY, Y-N-GY, and Pd-N-GY, demonstrated superior HER catalytic activity compared to commercial Pt/C.
• Doping with non-metallic B or N atoms was found to effectively modulate the HER performance of SACs.
• HER activity was correlated with specific characteristic factors, providing insights into catalyst design principles.

Conclusions:

• The study successfully identified highly active SACs for HER, surpassing commercial benchmarks.
• Machine learning models proved effective in predicting and designing novel HER SACs, significantly reducing research time and cost.
• The findings provide a pathway for accelerating the rational design and discovery of advanced electrocatalysts for the hydrogen evolution reaction.