Quantitative Insight into the Electric Field Effect on CO₂ Electrocatalysis via Machine Learning Spectroscopy

Area of Science:

• Computational chemistry
• Materials science
• Spectroscopy

Background:

• Electric fields are crucial for electrocatalysis and electrosynthesis but are difficult to quantify at the micro-level.
• Understanding electric field regulation of electrocatalytic reactions is essential for optimizing chemical processes.

Purpose of the Study:

• To develop a quantitative method for predicting electric field effects on catalytic properties using spectroscopy.
• To establish a machine learning model linking spectral signals to catalytic performance.

Main Methods:

• Investigated CO2 adsorption on metal-doped graphitic carbon nitride (g-C3N4) catalysts under varying electric fields.
• Utilized infrared/Raman spectral signals as descriptors for machine learning models.
• Employed attention mechanisms to analyze spectral-property relationships and enable inverse prediction.

Main Results:

• Developed a spectroscopy-property model correlating spectral descriptors with adsorption energy and charge transfer.
• Quantified the enhancement of electric field effects on CO2 catalytic conversion.
• Achieved inverse prediction of electric field strength from spectral data.

Conclusions:

• This work presents a novel quantitative pathway for monitoring and regulating electrocatalytic reactions via machine learning spectroscopy.
• The findings offer new insights into the interplay between electric fields, spectral properties, and catalytic activity.
• The developed model facilitates precise control over electrocatalytic processes.