BioMCM-41 mesoporous silica from agricultural waste for methylene blue removal and ANN-based prediction
(2026) Next Materials — Vol. 13, p. 103346 (2026)
Files
Pub307.pdf
Open Access - Adobe PDF
- 19.78 MB
Details
- Authors
Khelif, Meriem 
Bouchenafa-Saib, Naima Ibrir, Abdellah
- Abstract
- The development of sustainable mesoporous silica adsorbents from agricultural residues represents an important strategy for improving dye removal while reducing dependence on conventional silica precursors. However, the conversion of low-cost biomasses into ordered MCM-41-type materials and the prediction of their adsorption performance under variable conditions remain insufficiently explored. In this work, bio-silica-derived MCM-41 materials were synthesized from two abundant agricultural wastes, Festuca arundinacea and barley bran, using a cost-effective and environmentally friendly route. The extracted silicas were highly pure and amorphous, and were converted into sodium silicate precursors for mesoporous material synthesis and compared with commercial MCM-41. The materials were characterized using X-ray diffraction, Fourier-transform infrared spectroscopy, X-ray fluorescence, N₂ adsorption–desorption, transmission electron microscopy, energy-dispersive X-ray spectroscopy, and thermogravimetric–differential scanning calorimetry. The prepared materials exhibited ordered hexagonal mesoporosity, high specific surface areas of 1090, 1080, and 1030 m² g⁻¹ for BioMCM-41FA, BioMCM-41BB, and MCM-41C, respectively, with pore diameters ranging from 2.9 to 3.9 nm. Their adsorption performance was evaluated using methylene blue as a model cationic dye. BioMCM-41FA and BioMCM-41BB achieved complete dye removal, while MCM-41C reached 95% under the same conditions. The corresponding adsorption capacities were 80.00, 80.00, and 77.29 mg g⁻¹. In addition, an artificial neural network model was developed to predict adsorption yield using eight input variables. After hyperparameter optimization, the ANN showed strong predictive performance, with R² = 0.941 for training, R² = 0.900 for external testing, Q² = 0.900, RMSE = 0.0526, MAE = 0.0354, and MAPE = 4.79%. These results confirm the potential of agricultural wastes for producing efficient mesoporous adsorbents and demonstrate the usefulness of ANN modeling for adsorption prediction.
- Affiliations
Citations
Khelif, M., Bouchenafa-Saib, N., Ibrir, A., Chennouf, Z., Boumessaidia, S., & Gaigneaux, E. (2026). BioMCM-41 mesoporous silica from agricultural waste for methylene blue removal and ANN-based prediction. Next Materials, 13, 103346. https://doi.org/10.1016/j.nxmate.2026.103346 (Original work published 2026)