Adsorption of rhodamine B onto cotton straw-derived biochar: kinetic, equilibrium, thermodynamics, and predictive studies using artificial intelligence.

This study investigates the efficiency, mechanisms, and artificial intelligence (AI) modeling of rhodamine B (RhB) adsorption using biochar derived from cotton straw (CS@B). Characterization through SEM, FTIR, and pH_PZC revealed that CS@B possesses a porous structure, with RhB adsorption involving hydrogen bonding, electrostatic interactions, and π-π interactions, and a pH_PZC of 8.27. Maximum RhB removal (99.7%) was achieved at pH 2.0. Kinetic studies aligned with the pseudo-second-order model, while the Freundlich isotherm model accurately described the equilibrium data. The maximum adsorption capacity of 117.84 mg g^-1 surpasses many other adsorbents. Thermodynamic analysis confirmed a spontaneous and endothermic process. Artificial intelligence models, including artificial neural networks (ANN) and support vector regression (SVR), predicted adsorption capacity with high accuracy. The ANN models, particularly the MLP 5-7-1 architecture, achieved R² values up to 0.994 and low RMSE values for the testing dataset, while the SVR model attained an R² of 0.984. Reusability tests showed that CS@B remained effective over several cycles, with a slight decline in efficiency. These results underscore the potential of CS@B for effective RhB removal in water treatment. Furthermore, the integration of AI models provides a robust framework for enhancing the predictability and efficiency of adsorption systems.