Functionalised mesoporous silica nanoparticles for dye removal: Experimental insights and predictive modelling
Tajudeen A. Oyehan, Christian Pfrang, Eugenia Valsami‐Jones
Abstract
• Mesoporous silica nanoparticles (MSNs) were synthesised and shown to be effective in dye removal • Functionalisation at 12 mol% (functional group to MSN) was optimal for dye removal • MSN-COOH is particularly efficient in removing both cationic and anionic dyes • Random forest (RF) models are excellent at predicting dye removal efficiency • RF permutation-importance identified functional group and pH as key parameters The removal of synthetic dyes from wastewater remains a major environmental challenge, requiring efficient and sustainable adsorbents. In this study, mesoporous silica nanoparticles (MSNs) were synthesised via a low-temperature route and functionalised with carboxyl (COOH) and thiol (SH) groups through co-condensation to enhance adsorption of cationic and anionic dyes. Structural, thermal, and surface analyses confirmed successful synthesis and retention of functional groups. Batch adsorption experiments under varying pH, adsorbent dosage, dye concentration and contact time showed that MSN-COOH achieved the highest adsorption capacity (149.3 mg g -1 for crystal violet; 10.7 mg g -1 for methyl orange) and the fastest kinetics (two-fold faster than bare MSNs). Kinetic, isotherm, and zeta potential analyses indicated that electrostatic interactions and surface chemistry governed removal performance. A machine learning approach using the Random Forest (RF) model was applied to predict adsorption efficiency and evaluate the relative influence of experimental parameters. The RF model achieved high accuracy (r ≥ 0.97), effectively capturing complex adsorption trends and enabling rapid performance estimation without extensive laboratory trials. Permutation importance analysis identified functionalisation and pH as the dominant factors. Overall, this study optimised functional group loading for efficient dye removal, advancing both the scientific understanding and practical engineering of MSN-based adsorbents. It also highlights the combined benefits of low-temperature synthesis and ML-driven insights, positioning MSN-COOH as a promising and sustainable adsorbent for wastewater treatment. Future scale-up studies and integration with continuous treatment systems may accelerate the translation of this approach to real-world applications.