Discovery of novel Li SSE and anode coatings using interpretable machine learning and high-throughput multi-property...

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry
• Machine Learning

Background:

• All-solid-state batteries (ASSBs) with lithium (Li) metal anodes offer enhanced safety and energy density compared to traditional Li-ion batteries.
• Developing stable solid electrolytes and protective anode coatings with high ionic conductivity remains a significant challenge for ASSB technology.
• The exploration of novel Li-containing compounds is crucial for advancing ASSB performance.

Purpose of the Study:

• To employ an informatics approach combining high-throughput screening and interpretable machine learning to identify promising materials for solid electrolytes and Li metal anode coatings.
• To accelerate the discovery and design of novel materials for safer and higher-energy-density all-solid-state batteries.

Main Methods:

• Generated a database of over 15,000 Li-containing compounds from Materials Project, computing migration barriers and stability windows.
• Screened compounds for thermodynamic/electrochemical stability and low Li-ion migration barriers.
• Developed ensemble machine learning models (gradient boosting regression) to predict migration barriers and redox potentials, validated using Shapley additive explanations and permutation feature importance.

Main Results:

• Identified several promising new candidate materials, including Li3N, Li2O, Li2S, LiF, and Li3P, exhibiting good stability and low Li-ion migration barriers.
• Machine learning models achieved high accuracy (R² values of 0.95, 0.92, and 0.86) in predicting oxidation potentials, reduction potentials, and migration barriers, respectively.
• Interpretable ML analyses revealed key material properties influencing performance, aiding in understanding structure-property relationships.

Conclusions:

• The informatics and machine learning approach effectively accelerates the discovery of novel solid electrolytes and anode coatings for all-solid-state batteries.
• Identified promising Li-containing compounds and key material descriptors that can guide future material design.
• This strategy holds significant potential for the rapid development of next-generation battery technologies.