Machine learning-driven optimization of biochar-based supercapacitors for sustainable energy storage: mechanisms, trends, and perspectives

Global transition toward sustainable and efficient energy storage has intensified research into supercapacitors, valued for their rapid charge-discharge capability, high power density, and long operational lifespan. Among carbonaceous materials, biomass-derived biochar has emerged as a low-cost, renewable, and structurally tunable precursor for electrode design. This review provides a structured and critical synthesis of biochar-based supercapacitor strategies, systematically categorizing heteroatom doping, nanocomposite engineering, and hybrid architecture within an integrated electrochemical performance framework. Unlike existing literature that predominantly summarizes modification approaches, this review emphasizes cross-study comparability, mechanistic attribution, and device-level relevance to clarify transferable design principles. A consolidated analysis of recent machine learning (ML) applications reveals a paradigm shift from empirical optimization to data-driven predictive design. Benchmarking against reported datasets, descriptors, and model architectures, we confirm that tree-based ensemble and neural-network models (e.g., XGBoost, LightGBM, ANN) consistently achieve high predictive accuracy (R² > 0.9), while also identifying current limitations in data standardization and descriptor harmonization. The review critically evaluates how ML can move beyond correlation towards interpretable and transferable optimization under realistic device constraints.Finally, a converging research roadmap is proposed, prioritizing standardized reporting, device-level benchmarking, and constraint-aware ML integration to accelerate scalable implementation. By integrating materials engineering, data-driven modeling, and practical deployment considerations, this review establishes a comprehensive framework for advancing sustainable biochar-based supercapacitors toward real-world energy storage applications.

Subjects

Carbon electrode modification

Electrochemical energy storage

Heteroatom doping

Nanocomposite engineering

Predictive modeling

DDC Class

621.3: Electrical Engineering, Electronic Engineering

541.3: Physical Chemistry

Lizenz

https://creativecommons.org/licenses/by/4.0/

Publication version

publishedVersion

Name

1-s2.0-S2949822826003886-main.pdf

Type

Main Article

Size

2.48 MB

Format

Adobe PDF

Options

Machine learning-driven optimization of biochar-based supercapacitors for sustainable energy storage: mechanisms, trends, and perspectives