Predictive removal of nickel (II) from wastewater onto cellulose nanocrystal nanocomposite using deep machine learning

This study investigates the removal of nickel (II) from aqueous solutions using a biodegradable cellulose nanocrystal (CNC) nanocomposite. Fourier Transform Infrared (FTIR) analysis confirmed successful functionalization, with characteristic peaks observed at 1735 cm⁻¹ (C = O stretching of carboxyl groups) and 1050 cm⁻¹ (C-O-C stretching), indicating the effective incorporation of EDTA into the CNC structure. Scanning Electron Microscopy (SEM) revealed a rough and porous surface morphology, favorable for enhanced adsorption performance. Thermogravimetric Analysis (TGA) demonstrated the composite's thermal stability up to 320 °C, with a significant weight loss of 65% between 300-400 °C corresponding to cellulose decomposition. Batch adsorption experiments examined the effects of pH, contact time, adsorbent dosage, and initial nickel (II) concentration. The maximum removal efficiency of 98.3% was achieved at a pH of 6, a 120-min contact time, an 8 g/100 mL dosage, and an initial concentration of 150 mg/L. Film diffusion was identified as the rate-limiting step with an R² of 0.983. Machine learning models were also developed to predict adsorption performance. The Artificial Neural Network (ANN) model achieved R² of 0.987 and RMSE of 0.012, while the Adaptive Neuro-Fuzzy Inference System (ANFIS) demonstrated superior accuracy with R² of 0.995 and RMSE of 0.008. The nickel(II) adsorption is best represented by the Langmuir model, with an R² value of 0.996. The pseudo-second-order model governs the adsorption process; the Dubini-Radushkevich model confirms chemisorption with an energy of 9.375 kJ/mol. These findings confirm that the CNC nanocomposite is an efficient, thermally stable, and sustainable adsorbent for Ni(II) removal from aqueous media, with ANN and ANFIS models providing reliable predictive capability for process optimization.