Deep eutectic solvent-modified polyvinyl alcohol/chitosan thin film membrane for dye adsorption: Machine learning modeling, experimental, and density functional theory calculations.

The polyvinyl alcohol/chitosan (PVA/CS) thin film membrane was modified using a deep eutectic solvent (DES) to enhance its adsorption capability and mechanical strength for the removal of brilliant green (BG) dye. Batch adsorption experiments, machine learning (ML) modeling, and density functional theory (DFT) analyses were performed to evaluate the adsorption of BG using PVA/CS and DES-modified PVA/CS (DES/PVA/CS) membranes. Incorporating DES (5 wt%) into the PVA/CS membrane increased its elongation at break from 8.176 % to 22.817 %. The random forest ML model exhibited superior predictive accuracy (R = 0.93) compared to the artificial neural network (R = 0.68) for modeling the adsorption process. The adsorption experiments were conducted under optimal operating conditions for PVA/CS (pH 7.5, adsorbent mass 0.06 g, and initial BG concentration 65 mg/L) and DES/PVA/CS (pH 8, adsorbent mass 0.06 g, and initial BG concentration 80 mg/L), achieving maximum adsorption capacities of 23.15 mg/g for PVA/CS and 124.63 mg/g for DES/PVA/CS. DFT calculations showed adsorption energies of -20.76 kcal/mol and -23.13 kcal/mol for BG/PVA/CS and BG/DES/PVA/CS complexes, respectively. DES, a green modifier, significantly enhanced the adsorption capacity, mechanical stability, and functional group diversity of PVA/CS membranes, thereby enabling more efficient dye removal.