Zero-Shot Discovery of High-Performance, Low-Cost Organic Battery Materials Using Machine Learning.

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry

Background:

• Organic electrode materials (OEMs) offer sustainable alternatives to metal-based electrodes due to their abundance and structural diversity.
• Exploring the vast design space of OEMs via traditional trial-and-error is inefficient and costly.

Purpose of the Study:

• To develop a computational framework, SPARKLE, for accelerated discovery of high-performance, cost-effective, and stable OEMs.
• To enable zero-shot predictions of novel OEMs balancing specific energy, solubility, and synthesizability.

Main Methods:

• Integration of computational chemistry, molecular generation, and machine learning within the SPARKLE framework.
• Deployment of SPARKLE across a large design space (>670,000 organic compounds) for candidate identification.
• Experimental synthesis and battery performance testing of 27 novel OEM candidates.

Main Results:

• SPARKLE demonstrated superior performance over black-box ML algorithms in interpolation and extrapolation tasks.
• Identified approximately 5000 novel OEM candidates, with 62.9% exceeding benchmark performance metrics.
• Synthesized OEMs showed a 3-fold improvement over human-selected materials, with top performers surpassing state-of-the-art specific energy and stability at lower cost.

Conclusions:

• The SPARKLE framework enables efficient and accurate discovery of advanced organic electrode materials.
• AI-driven material discovery accelerates the development of sustainable and high-performance batteries.
• SPARKLE-identified OEMs represent a significant advancement in cost-effective and sustainable energy storage solutions.