Experimental optimization and machine learning modeling for sustainable Congo red dye removal from wastewater using activated goat bone biochar

Abstract

This study explores the use of goat bone-based activated biochar (GB_PAC) synthesized from animal waste as an efficient and sustainable adsorbent for removing Congo Red (CR) dye from aqueous solutions. GB_PAC, prepared through chemical activation with phosphoric acid, was tested in batch adsorption experiments. FTIR analysis revealed key functional groups such as hydroxyl (O–H), carboxyl (CO), and phosphate groups, which play a crucial role in the adsorption of CR dye through interactions like hydrogen bonding and electrostatic attraction. BET surface area analysis showed that GB_PAC exhibited a surface area of 91.27 m²/g, with a mesoporous structure that enhances its adsorption capacity. The study systematically analyzed factors such as dye concentration (10–50 mg/L), adsorbent dosage (0.15–0.75 g/100 mL), pH (7.5), and contact time (30–180 min). The maximum adsorption capacity of GB_PAC for CR dye was 83.33 mg/g, and the adsorption process followed the Langmuir isotherm model (R² = 0.9907) and pseudo-second-order kinetics. Process Optimization was performed using Response Surface Methodology (RSM), which enabled statistically guided experimental design and optimization of influential variables. Optimal conditions were identified as 48.596 mg/L dye concentration, 0.398 g adsorbent dose, and 88.23 min contact time, achieving a predicted removal efficiency of 94.34 %. To enhance prediction capabilities, machine learning (ML) models, specifically Decision Tree and Random Forest, were trained using experimental data. These models demonstrated strong predictive accuracy, with R² values of 0.91 and 0.87, respectively. This dual-framework approach, combining RSM for optimization and ML for predictive modeling, underscores the novelty of using waste-derived GB_PAC for wastewater treatment applications. The findings support GB_PAC as a cost-effective, sustainable, and data-driven solution for CR dye removal from contaminated water.