Thermodegradation kinetics of hemicelluloses, cellulose, and lignin: A review of mechanistic interpretation and hybrid predictive frameworks

Abstract

Understanding the thermodegradation kinetics of individual lignocellulosic components is essential for resolving inconsistencies in biomass decomposition modeling and improving predictive accuracy. This review aims to: (1) compare the kinetic models of different lignocellulosic components, (2) interpret the kinetic parameters at the molecular level, and (3) assess a hybrid mechanistic-kinetic-machine learning framework. Hemicelluloses generally exhibit lower activation energies during torrefaction (from ∼48 to ∼55 kJ·mol −1 ), whereas pyrolysis activation energies span a much broader range (from ∼30.09- to ∼247.87 kJ·mol −1 ) due to their heterogeneous structure and overlapping decomposition reactions. Cellulose typically exhibits higher activation energies, ranging from ∼74.2 to ∼287.56 kJ·mol −1 during pyrolysis, reflecting the greater thermal stability of its crystalline β-(1 → 4)-linked glucan structure. Lignin demonstrates the broadest distribution (from ∼21 to ∼673 kJ·mol −1 ) owing to its heterogeneous aromatic network and multiple bond types. Model comparisons indicate that single-step approaches are suitable for cellulose, whereas distributed activation energy and multi-pseudo-reaction models better describe hemicelluloses and lignin. For lignocellulosic biomass, model-free isoconversional methods are more suitable because they capture complex, distributed activation energies without assuming a predefined reaction mechanism. Machine learning, particularly random forest models (R 2 up to nearly 0.9), significantly improves predictive performance by capturing nonlinear relationships between feedstock properties and activation energy. This review proposes a solution that integrates component-specific mechanistic interpretation with model-free kinetics and machine-learning predictions, enabling more reliable model selection, activation energy estimation, and process optimization for thermochemical conversion.