Machine Learning-Guided Design of a Flexible Highly Conductive Additive-Free Polymer Cathode.

Area of Science:

• Materials Science
• Electrochemistry
• Sustainable Energy

Background:

• Organic cathode materials (OCMs) offer sustainable alternatives for batteries but suffer from low conductivity and dissolution.
• Discovering optimal OCMs is challenging due to the vast chemical space.
• Existing OCMs often require conductive additives and binders, complicating fabrication and performance.

Purpose of the Study:

• To address the limitations of OCMs by developing a high-performance, intrinsically conductive material.
• To utilize machine learning for accelerated discovery of novel OCM candidates.
• To demonstrate a flexible, additive-free organic cathode for advanced battery applications.

Main Methods:

• Employed a machine learning (ML) discovery process using a pretrained transformer model on the ZINC organic molecules database.
• Screened for potential high-performance OCM candidates, identifying isoindigo-type redox units.
• Designed and synthesized poly-benzodifurandione (PBFO) based on ML predictions for flexible, free-standing cathode applications.

Main Results:

• Achieved a breakthrough electrical conductivity of 5.9×10^2 S cm^-1 in flexible PBFO films, a new benchmark for additive-free organic cathodes.
• Demonstrated high reversible capacity (262 mAh g^-1) at 2.5 V vs. Li+/Li with a high electrode-level energy density of 655 Wh kg^-1.
• Successfully fabricated neat PBFO cathodes without conductive additives or binders, showcasing their potential for high-performance Li-ion and Na-ion storage.

Conclusions:

• The developed PBFO represents the first flexible, high-conductivity organic cathode material without additives.
• The ML-driven discovery process efficiently identified promising OCM candidates.
• This work paves the way for practical and sustainable organic batteries with enhanced performance.