Machine-Learning-Guided Discovery and Optimization of Additives in Preparing Cu Catalysts for CO₂ Reduction

Area of Science:

• Electrochemistry
• Materials Science
• Catalysis

Background:

• Catalyst discovery and optimization are crucial for chemical transformations.
• Machine learning (ML) offers powerful tools for accelerating materials development through data analysis.
• Copper (Cu) catalysts are widely studied for carbon dioxide reduction (CO2RR) into various products.

Purpose of the Study:

• To demonstrate the application of ML in the iterative search for additives in electrochemical copper catalyst preparation for CO2RR.
• To accelerate the discovery and optimization of Cu catalysts with enhanced selectivity for specific CO2RR products.

Main Methods:

• An iterative ML workflow involving experimental testing, ML analysis, and prediction/redesign was employed.
• Electrochemical deposition was used to prepare Cu catalysts with varying additives.
• ML models were trained and utilized to guide catalyst redesign based on experimental data.

Main Results:

• The ML-driven iterative process successfully identified catalyst formulations with improved CO2RR selectivity.
• Achieved catalysts demonstrated selectivity towards carbon monoxide (CO), formic acid (HCOOH), and C2+ products (e.g., ethylene, ethanol).
• Demonstrated that subtle changes in catalyst morphology and surface structure, influenced by additives, significantly impact CO2RR selectivity.

Conclusions:

• ML significantly accelerates the discovery and optimization of catalysts by efficiently extracting insights from limited experimental data.
• The developed ML approach provides a viable strategy for designing selective Cu catalysts for CO2RR.
• This work highlights the potential of integrating ML with experimental efforts in materials science for rapid catalyst development.