Screening of potential candidates for solid electrolyte interphase materials for lithium-ion batteries through a...

Area of Science:

• Materials Science
• Computational Chemistry
• Electrochemistry

Background:

• Machine learning accelerates material property prediction, reducing challenges in designing optimal materials.
• Solid electrolyte interphase (SEI) materials are crucial for lithium-ion battery performance and longevity.
• Developing stable SEI layers is key to suppressing dendrite formation and enhancing battery safety.

Purpose of the Study:

• To refine a large dataset of SEI materials using machine learning.
• To identify promising SEI candidates based on chemical stability, solvation energy, and synthesis ease.
• To uncover key molecular features influencing SEI properties for future material discovery.

Main Methods:

• Utilized a machine learning approach to analyze 11,664 SEI materials.
• Predicted chemical reactivity and solvation energy with high accuracy (86.7-91.3%).
• Identified critical atomistic features like dipole moments, heteroatom counts, and various descriptors influencing SEI properties.

Main Results:

• Identified key molecular descriptors (e.g., PEOE_VSA, kappa index, functional groups) impacting solvation energy and chemical reactivity.
• Determined that molecules containing fluorine, nitrogen, and carbon promote stable SEI formation.
• Highlighted that sulfur, oxygen, nitrogen, and carbon-containing molecules can reduce stable SEI formation capability.

Conclusions:

• Developed a robust machine learning workflow for discovering and optimizing SEI materials.
• Strategic selection of SEI materials based on predicted properties can enhance lithium-ion battery performance and lifespan.
• The study provides critical guidance for screening materials to effectively suppress dendrite formation.