Data-driven design and controllable synthesis of Pt/carbon electrocatalysts for H₂ evolution

Area of Science:

• Materials Science
• Electrochemistry
• Data Science

Background:

• Achieving net-zero emissions necessitates efficient catalysts for hydrogen (H2) evolution.
• Designing optimal H2 evolution catalysts is complex and time-consuming.
• Data-driven approaches offer a promising strategy for catalyst development.

Purpose of the Study:

• To develop an optimized catalyst for hydrogen evolution using a data-driven strategy.
• To identify key features influencing catalyst performance through machine learning.
• To fabricate and evaluate novel platinum-based catalysts for enhanced activity.

Main Methods:

• Collected and analyzed data for platinum/carbon (Pt/carbon) catalysts.
• Applied machine learning algorithms to rank feature importance for catalyst overpotentials.
• Utilized a space-confined method to fabricate size-controllable platinum nanoclusters on nitrogen-doped mesoporous carbon nanosheet networks.

Main Results:

• Machine learning identified Pt content and Pt size as critical factors for catalyst overpotentials.
• Fabricated catalysts demonstrated superior catalytic activity in alkaline electrolytes compared to commercial standards.
• The developed catalysts utilize reduced amounts of platinum.

Conclusions:

• A data-driven strategy effectively accelerates the development of high-performance hydrogen evolution catalysts.
• Size-controlled Pt nanoclusters on N-doped carbon offer a promising alternative to current commercial catalysts.
• The generated data can iteratively improve predictive models for future catalyst design.