Enhanced fluoride removal by modified water hyacinth: response surface methodology and machine learning approach

This study investigates the potential of surface-modified water hyacinth for efficient fluoride removal from aqueous solutions. Sodium dodecyl sulfate treatment was employed to modify the water hyacinth powder, leading to enhanced adsorption capacity. A central composite design (CCD) within the response surface methodology (RSM) framework was employed to optimize the key process variables impacting fluoride removal: adsorbent dosage, contact time, initial fluoride concentration, and solution pH. The optimal conditions for fluoride removal were determined to be a pH of 4, an adsorbent dosage of 1.25 g/L, and a contact time of 180 min. Under these parameters, a fluoride removal efficiency of 95.43% was achieved for an initial fluoride concentration of 32 ppm. Adsorption isotherms and kinetics were investigated to understand the underlying mechanisms. The adsorption kinetics and equilibrium data were best fitted by three models: the Freundlich isotherm (for equilibrium), pseudo-second-order kinetics (for reaction rate), and intraparticle diffusion (for mass transfer mechanisms). Further characterization through techniques like FTIR, SEM, and EDX provided insights into the changes in the water hyacinth properties after fluoride adsorption, solidifying the effectiveness of the modification. The extreme gradient boosting machine learning technique was implemented to enhance prediction accuracy, resulting in exceptional prediction accuracy for fluoride removal percentages, with maximum errors of 0.84% and 0.95% for training and testing datasets, respectively. Integrating computational methods helps in developing a deep understanding of environmental remediation technologies. This research envisages the potential of modified water hyacinth, an eco-friendly material with defluoridation efficiency, as a sustainable and viable solution for water defluoridation.