The application of Central composite design and Convolutional Neural Network to optimise the removal of copper (II) from wastewater using a biodegradable composite

This study developed a novel biodegradable composite adsorbent based on carboxyl- and amide-grafted cellulose nanocrystals (CNCs) to efficiently remove copper (VI) from aqueous media. Structural and physicochemical characterisation confirmed the successful preparation of the composite adsorbent. The FTIR spectra revealed overlapping O–H and N–H stretching between 3500–3800 cm⁻¹ and a carbonyl band at 1640 cm⁻¹, indicating amide and hydroxyl group grafting. SEM images revealed a porous, interconnected morphology formed through physical crosslinking between the CNCs and chitosan chains. The BET results revealed an increased on the pore volume of 0.057 cm³/g and specific surface area of 3.452 m²/g compared with those of CNCs (2.968 m²/g). Adsorption experiments, optimised via Central Composite Design (CCD), identified a pH of 6.0, a dosage of 10 g/200 mL, an initial copper (II) concentration of 50 mg/L, and a contact time of 120 min. as the optimal conditions, achieving a maximum adsorption capacity of 350.62 mg/g. Comparative modelling via convolutional neural network (CNN) and central composite design (CCD) revealed superior CNN performance, with an R² of 0.996 and an MSE of 0.0013. Quantum chemical (HOMO–LUMO) analysis further confirmed that the amide-modified CNCs exhibited the strongest binding affinity toward copper (II), highlighting the potential of this eco-friendly nanocomposite for sustainable wastewater treatment applications