Machine learning modeling of acid yellow 17 dye removal from solution using a biogenic hydroxyapatite-supported chitosan-graphene oxide adsorbent

Abstract

The focus of this study was the synthesis of a composite adsorbent (GO@CS-HAp) by integrating hydroxyapatite (HAp) derived from waste biomass with graphene oxide (GO) and chitosan (CS). The material was synthesized via a systematic procedure and applied for the treatment of acid yellow 17 dye. Characterization tests revealed that GO@CS-HAp integrated precursor features, shown by FTIR peaks at 3200, 2992.91, 2725, 2530, 1630, and 1450 cm −1 . XRD analysis confirmed HAp planes, semi-crystalline CS, and GO's disrupted hexagonal lattice. N 2 adsorption showed a surface area, pore size, and pore volume of 66.13 m 2 /g, 1.453 nm, and 0.0268 cm 3 /g, respectively. The influence of process parameters on dye uptake was modeled using artificial neural networks (ANN). The ANN model showed excellent predictive performance, as reflected by its high coefficient of determination (R 2 = 0.9980), along with low mean square error (0.1169) and root mean square error (0.3419). Optimization of the dye removal process was achieved using genetic algorithm, which identified the best operating conditions for maximum efficiency (87.33%). The optimal parameters included a contact time of 52.5 min, a pH of 6.16, an adsorbent dosage of 2.0 g, and a dye concentration of 117.5 mg/L. Analysis of adsorption isotherms indicated the suitability of the Langmuir model for the equilibrium data, while the kinetic behavior was accurately represented by the pseudo-second-order model. The thermodynamic evaluation indicated a spontaneous and feasible process. Overall, these results contribute to advancing dye wastewater treatment by demonstrating the potential of novel adsorbents combined with intelligent optimization strategies. • GO@CS-HAp composite was synthesized for dye adsorption. • Composite showed enhanced removal of acid yellow dye. • BET, XRD, and XRF confirmed structural and surface properties. • Adsorption followed favorable isotherm and kinetic models. • ANN modeling accurately predicted dye removal efficiencies.