Electron transfer rules of minerals under pressure informed by machine learning

Area of Science:

• Geochemistry
• Computational Chemistry
• Materials Science

Background:

• Electron transfer is fundamental but poorly understood under high pressure.
• Deep Earth conditions present unique challenges for existing electron transfer theories.

Purpose of the Study:

• To develop a deep learning model for predicting element electronegativity under arbitrary pressure.
• To establish a unified formula quantifying the relationship between electronegativity, pressure, and electronic configuration.
• To investigate pressure-induced changes in mineral work functions and their impact on electron transfer.

Main Methods:

• A deep learning model was trained to predict electronegativity for 96 elements.
• A unified formula was regressed to describe electronegativity's dependence on pressure and electronic configuration.
• Relative work functions of minerals were calculated using predicted electronegativity.

Main Results:

• Electronegativity under pressure was accurately predicted for 96 elements.
• Mineral work functions decrease with increasing pressure due to electron delocalization.
• Enlarged work function differences drive directional electron transfer under pressure.

Conclusions:

• The model explains deep high-conductivity anomalies and predicts redox reactivity between Fe(II)-bearing minerals and water.
• This work provides fundamental insights into the physicochemical properties of elements and compounds under extreme pressure.
• The findings have implications for understanding deep Earth processes and geochemistry.