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

During chemical reactions, especially for electrocatalysis and electrosynthesis, the electric field is the most central driving force to regulate the reaction process. However, due to the difficulty of quantitatively measuring the electric field effects caused at the microscopic level, the regulation of electrocatalytic reactions by electric fields has not been well digitally understood yet. Herein, we took the infrared/Raman spectral signals of CO molecules as descriptors to quantitatively predict the effects of different electric fields on the catalytic properties. Taking the metal-doped graphitic CN (-CN) catalyst as an example, we theoretically investigated the adsorption mode and energy of CO molecules adsorbed on 27 distinct metal single-atom catalysts under different directions and intensities of electric field. Through a machine learning approach, a spectroscopy-property model between infrared/Raman spectral descriptors and adsorption energy/charge transfer was established, which quantified the facilitation of electric field effects on the CO catalytic conversion. Meanwhile, based on the attention mechanism, the catalytic insight of the relationship between spectra and adsorption modes was mined, and the inverse prediction of electric field strength from spectra was realized. This work opens a new quantitative pathway for monitoring and regulating electrocatalytic reactions using machine learning spectroscopy.