Boric acid-activated carbon derived from Scots pine sawdust for phenol removal: adsorption mechanism, linear and nonlinear kinetic and isotherm modeling, and artificial neural network prediction

In this study, a sustainable activated carbon was successfully prepared from Scots pine sawdust via boric acid activation (H3BO3-AC-SPS) and evaluated for phenol adsorption from aqueous solutions. The physicochemical characteristics of the adsorbent were investigated using FTIR, SEM-EDX, XRD, and BET analyses, while the effects of solution pH, contact time, adsorbent dosage, initial phenol concentration, temperature, ionic strength, and reusability without regeneration on the adsorption performance were systematically examined. H3BO3-AC-SPS exhibited excellent adsorption performance over a wide pH range (5–8), showed good tolerance to coexisting inorganic ions and organic contaminants, and retained appreciable adsorption efficiency over five consecutive adsorption cycles without regeneration. Phenol uptake increased with temperature, and NaOH enabled partial desorption of phenol. The adsorption kinetics and equilibrium data were analyzed using both linear and nonlinear regression approaches and comparatively evaluated based on the coefficient of determination (R2), sum of squared errors (SSE), and root mean square error (RMSE). The pseudo-second order kinetic model and the Langmuir isotherm exhibited the best predictive performance, with the Langmuir model estimating a maximum adsorption capacity of 65.4 mg g−1. The adsorption mechanism may involve π–π interactions, hydrogen bonding, pore filling, and hydrophobic interactions. In addition, an artificial neural network (ANN) model accurately predicted the adsorption behavior, yielding a high correlation coefficient (R = 0.9789). The applicability of H3BO3-AC-SPS was further demonstrated in stream water and mining wastewater. These findings demonstrate that boric acid activation is an effective strategy for converting Scots pine sawdust into a high-performance adsorbent and that integrating linear and nonlinear adsorption modeling with statistical error analysis provides a reliable framework for interpreting adsorption behavior. This journal is