Machine learning-optimized hybrid graphene/polymer electrodes for high-performance and scalable supercapacitors

Abstract

The demand for carbon-based materials with superior electrochemical properties continues to grow, particularly for scalable supercapacitor applications. In this study, we present a hybrid synthesis route for graphene electrodes that combines chemical vapor deposition (CVD) and microwave-assisted reduction, guided by machine learning (ML) optimization. A predictive neural network trained on over 100 synthesis experiments enabled precise tuning of key parameters, resulting in high conductivity (~ 9.6 × 105 S m⁻¹). Raman analysis across ≥ 5 spots per sample showed hybrid graphene at ID/IG = 0.29 [mean ± SD to be reported], while the CVD control exhibited ID/IG ≈ 0.10 [mean ± SD]. These properties underpinned the electrodes’ high specific capacitance (up to 500 F g⁻¹), energy density (120 Wh kg⁻¹), and 95% retention over 10,000 cycles. The synthesized graphene was further hybridized with MnO₂, RuO₂, and conductive polymers (polyaniline, polypyrrole), leading to enhanced specific capacitance, energy density, and power density (70 kW kg⁻¹), while maintaining long-term stability. Structural, thermal, and electrochemical evaluations confirmed the durability and high performance of the optimized electrodes. This work demonstrates a scalable and cost-effective graphene synthesis strategy for advanced energy storage, enabled by machine learning. The integration of data-driven optimization with roll-to-roll-compatible processing provides a promising pathway toward industrial deployment of graphene-based supercapacitors in transportation, grid systems, and flexible electronics.

Graphical abstract