Decoding Adsorption Mechanisms in Biomass-Derived Adsorbents: From Experimental Evidence and Molecular Simulation to Explainable and Transferable Prediction
- 1Departamento de Química y Producción, Facultad de Ciencias Exactas y Naturales, Universidad Técnica Particular de Loja, San Cayetano Alto, Loja 110107, Ecuador
- 2Carrera de Ingeniería Química, Facultad de Ciencias Exactas y Naturales, Universidad Técnica Particular de Loja, San Cayetano Alto, Loja 110107, Ecuador
Abstract
Biomass-derived adsorbents are increasingly used for aqueous pollutant removal, yet adsorption mechanisms are often inferred from model fitting or isolated characterization rather than convergent molecular evidence. This scoping evidence review integrates experimental surface chemistry, density functional theory (DFT), molecular dynamics (MD), and machine learning (ML) to examine the transition from mechanistic interpretation to transferable prediction. Using a PRISMA-ScR-informed systematic evidence mapping approach, a relevance-prioritized analytical corpus of 125 primary studies was constructed, comprising 70 mechanistic and 55 predictive studies. Evidence charting was closed after the mechanistic and predictive categories stabilized during relevance-prioritized synthesis. Electrostatic interactions, hydrogen bonding, surface complexation/coordination, π interactions, and pore filling were the most recurrent mechanisms and frequently coexisted. Static adsorbate–surface DFT predominated, whereas solvent/speciation-aware calculations and dynamic interfacial models remained comparatively limited. The synthesis shows that kinetic or isotherm fitting cannot uniquely establish molecular mechanisms and that ML feature importance or SHapley Additive exPlanations (SHAP) attribution should not be interpreted as causal evidence. Predictive transferability was more convincingly demonstrated through grouped, external, or experimental validation than through random data partitioning. Based on these findings, we propose the Molecular Mechanistic Evidence Framework (MMEF), integrating molecular depth, evidence convergence, environmental realism, interpretive discipline, and predictive transferability to guide adsorption research toward molecularly supported, experimentally testable, and prospectively predictive adsorbent design.
Related
- Engineered Biochar for Gaseous Air Pollutant Control: Modification Strategies, Synergistic Adsorption Mechanisms, and Comparative Evaluation with Conventional Adsorbents
- Adsorptive Removal of 1,2-Dichloroethane Using Novel Oleogel Adsorbents: Synthesis Strategy and Adsorption Mechanism Insight
- Explainable machine learning reveals orbital design rules for reversible H2 adsorption on single-atom-doped TiO2 nanoparticles
- From Publications to Physical Knowledge Models: Transferable physical understanding for AI-native experimental science
- Interfacial Mechanisms of Reactive Dye Adsorption on a Waste-Cotton Derived PVA/TiO2/Cellulose Biofilm: A Molecular Dynamics and Physics-Informed Predictive Framework
BibTeX
@article{jaramillofierro2026decoding,
title = {{Decoding Adsorption Mechanisms in Biomass-Derived Adsorbents: From Experimental Evidence and Molecular Simulation to Explainable and Transferable Prediction}},
author = {Ximena Jaramillo-Fierro and Julián Hidalgo and Eduardo Valarezo},
journal = {Molecules},
year = {2026},
month = nov,
doi = {10.3390/molecules31214018},
url = {https://www.mdpi.com/1420-3049/31/21/4018},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Decoding Adsorption Mechanisms in Biomass-Derived Adsorbents: From Experimental Evidence and Molecular Simulation to Explainable and Transferable Prediction
AU - Ximena Jaramillo-Fierro
AU - Julián Hidalgo
AU - Eduardo Valarezo
PY - 2026
DA - 2026/11/07
JO - Molecules
DO - 10.3390/molecules31214018
UR - https://www.mdpi.com/1420-3049/31/21/4018
AB - Biomass-derived adsorbents are increasingly used for aqueous pollutant removal, yet adsorption mechanisms are often inferred from model fitting or isolated characterization rather than convergent molecular evidence. This scoping evidence review integrates experimental surface chemistry, density functional theory (DFT), molecular dynamics (MD), and machine learning (ML) to examine the transition from mechanistic interpretation to transferable prediction. Using a PRISMA-ScR-informed systematic evidence mapping approach, a relevance-prioritized analytical corpus of 125 primary studies was constructed, comprising 70 mechanistic and 55 predictive studies. Evidence charting was closed after the mechanistic and predictive categories stabilized during relevance-prioritized synthesis. Electrostatic interactions, hydrogen bonding, surface complexation/coordination, π interactions, and pore filling were the most recurrent mechanisms and frequently coexisted. Static adsorbate–surface DFT predominated, whereas solvent/speciation-aware calculations and dynamic interfacial models remained comparatively limited. The synthesis shows that kinetic or isotherm fitting cannot uniquely establish molecular mechanisms and that ML feature importance or SHapley Additive exPlanations (SHAP) attribution should not be interpreted as causal evidence. Predictive transferability was more convincingly demonstrated through grouped, external, or experimental validation than through random data partitioning. Based on these findings, we propose the Molecular Mechanistic Evidence Framework (MMEF), integrating molecular depth, evidence convergence, environmental realism, interpretive discipline, and predictive transferability to guide adsorption research toward molecularly supported, experimentally testable, and prospectively predictive adsorbent design.
SN - 1420-3049
ER -