ZnO-modified activated carbon derived from rambutan (Nephelium lappaceum) peel and seeds for efficient methylene blue removal: adsorption mechanism and artificial neural network modeling

Abstract

A ZnO-carbon hybrid (ZnO-ACRPS) was constructed through a dual-biomass route combining rambutan peel and seeds as structurally complementary carbon sources. Hydrothermal carbonization, subsequent pyrolysis, and controlled ZnO deposition converted the lignin-dominated peel framework and heteroatom-rich seed fractions into a turbostratic carbon network with regulated disorder and hierarchical mesoporosity. The composite reached a specific surface area of 575.81 m² g⁻¹ with an average pore diameter of 3.23 nm. Wurtzite ZnO nanocrystals (10–40 nm) were uniformly dispersed throughout the matrix, introducing accessible Lewis acidic centers within the pore architecture. Interfacial compositional tuning influenced adsorption energetics. Methylene blue uptake was better described by the Langmuir model, yielding a maximum capacity of 82.95 mg g⁻¹ at pH 7, and followed pseudo-second-order kinetics. Mass transfer was governed by coupled film diffusion and intraparticle diffusion steps. Negative Gibbs free energy values indicate spontaneous uptake, whereas an enthalpy change near 68 kJ mol⁻¹ suggests an endothermic contribution consistent with strong surface interactions rather than purely weak physisorption. Dye binding can be attributed to π–π interactions between graphitic domains and aromatic rings, electrostatic attraction governed by pH_pzc, and may involve ZnO-centered Lewis acid–base interactions. A feedforward artificial neural network (5–13–1) demonstrated good predictive performance and agreement with experimental results (R_test² = 0.921; RMSE = 5.86). Sensitivity analysis identified temperature and adsorbent dosage as the most influential factors, consistent with thermally activated adsorption and active-site accessibility inferred from surface and thermodynamic analyses. Regeneration trials indicated maintained performance during initial cycles, followed by a noticeable decline upon repeated use, while structural features remained preserved. Dual-precursor compositional control and ANN-assisted analysis establish a framework for understanding structure–performance relationships in ZnO-carbon systems, while providing a basis for future evaluation under more complex wastewater conditions.