Prediction of Bandgap in Lithium-Ion Battery Materials Based on Explainable Boosting Machine Learning Techniques.

Area of Science:

• Materials Science
• Computational Chemistry
• Energy Storage

Background:

• The bandgap is crucial for battery energy density and material semiconducting properties.
• Accurate bandgap prediction in silicon oxide materials is essential for advancing lithium-ion battery technology.

Purpose of the Study:

• To enhance the prediction accuracy of bandgaps in silicon oxide lithium-ion battery materials.
• To identify key material features influencing bandgap predictions using machine learning.

Main Methods:

• Development and evaluation of a boosting machine learning model, specifically AdaBoost, for bandgap prediction.
• Application of the SHapley Additive exPlanations (SHAP) method to analyze feature importance and model interpretability.

Main Results:

• AdaBoost demonstrated superior prediction accuracy compared to five other models.
• SHAP analysis revealed a positive correlation between conduction band minimum energy and bandgap.
• A negative correlation was found between the Fermi level and bandgap, with decreasing Fermi levels leading to wider bandgaps.

Conclusions:

• Boosting machine learning models, particularly AdaBoost, are highly effective for predicting silicon oxide bandgaps.
• Understanding the relationship between electronic structure (cbm, Fermi level) and bandgap is key for material design.
• This approach offers a pathway to optimize silicon oxide materials for improved lithium-ion battery performance.