Enhanced groundwater defluoridation using aluminium loaded Sesbania grandiflora activated carbon

BACKGROUND AND OBJECTIVES: Excess fluoride in groundwater threatens public health, especially in rural areas with limited treatment options. Biomass-derived activated carbon provides a sustainable, low-cost solution. This study synthesized aluminium-impregnated Sesbania grandiflora activated carbon to assess fluoride removal effectiveness and adsorption capacity in addition to structural characteristics, co-ion removal, regeneration, and predictive modeling. METHODS: Aluminium-impregnated Sesbania grandiflora activated carbon was synthesized from locally sourced biomass through chemical activation and aluminium loading. Characterization included Fourier Transform Infrared Spectroscopy (functional groups), Scanning Electron Microscopy (morphology), Energy Dispersive X-ray Spectroscopy (elemental composition), X-ray Diffraction (crystallinity), and Brunauer–Emmett–Teller analysis (surface area and porosity). Regeneration used diluted hydrochloric acid, batch adsorption optimized particle size, dosage, agitation speed, and contact duration, and an Artificial Neural Network in Matrix Laboratory software was used to model predictive performance.FINDINGS: Fourier Transform Infrared Spectroscopy confirmed hydroxyl, carboxyl, and aluminium-associated groups involved in fluoride surface complexation. Scanning Electron Microscopy showed a porous, heterogeneous surface partially filled after adsorption, while Energy Dispersive X-ray Spectroscopy verified fluoride incorporation and elemental changes. X-ray Diffraction indicated amorphous content increased from ۷۴.۸ to ۸۱.۱ percent, and Brunauer–Emmett–Teller analysis showed surface area decreased from ۷۸۰ to ۶۱۵ square metres per gram, confirming pore occupation. Maximum fluoride removal of ۹۲.۶ percent was achieved under optimized conditions corresponding to an adsorption capacity of ۰.۲۹ milligrams per gram at ۲.۵ milligrams per litre initial concentration and ۸ grams per litre dosage. The adsorbent also removed specific coexisting ions (calcium, magnesium, iron, nitrate, and total dissolved solids) while maintaining near-neutral water acidity. Regeneration efficiency remained ۸۴.۲ percent after five cycles. Predictive performance increased from the linear regression model (R² = ۰.۹۴۵) to the nonlinear regression model (R² = ۰.۹۷۸), with the Artificial Neural Network model achieving the highest accuracy (R² = ۰.۹۹۱) with a training regression coefficient of ۰.۹۹۴۸, demonstrating strong capability to capture nonlinear relationships and reliably estimate fluoride removal efficiency.CONCLUSION: Spectroscopic and microscopic evidence indicate surface complexation–driven adsorption, structural stability, and reusability, and the material provides quantified fluoride and co-ion reduction with reproducible predictive modeling for decentralized groundwater treatment.

کلیدواژه ها:

Bio-adsorbent ، Heavy metal removal ، Machine learning modeling ، Process optimization ، Sustainable treatment.