Mechanistic adsorption-reaction modeling and thermodynamic analysis of metronidazole removal by aluminum electrocoagulation: a unified framework for predictive process engineering.

Metronidazole (MNZ) is a nitroimidazole antibiotic widely detected in hospital effluents and surface waters, yet conventional treatment plants are ineffective at removing such micropollutants. This study presents a comprehensive mechanistic framework for MNZ removal by aluminium electrocoagulation (EC), integrating electrochemical oxidation, adsorption onto in situ generated Al(OH)₃ flocs and reactive chlorine chemistry. Batch experiments were conducted at current intensities of from 1.00 to 2.30 A, initial concentrations of 5-50 mg/L and temperatures of 15-40 °C. A unified adsorption-reaction (A-R) model coupling four dynamic reservoirs (aqueous MNZ, adsorbed MNZ, active chlorine and floc mass) was developed and validated against 60 concentration-time observations from four current intensities and fifteen sampling times. Tert-butanol probe experiments combined with DOC mass balance revealed that MNZ elimination proceeds via an apparent oxidative contribution of 42% (35% complete mineralisation, 7% partial oxidation) and 58% adsorptive capture. However, because TBA selectively scavenges •OH radicals and does not efficiently quench reactive chlorine species (HOCl/ClO⁻), these values should be interpreted as operationally estimated apparent contributions that include both •OH- and RCS-mediated pathways. The A-R model achieved good predictive capacity (R2 = 0.996, RMSE = 3.2%) with the lowest corrected Akaike Information Criterion (AICc = 186.9; ΔAICc =  + 34.5 vs. the Langmuir-Hinshelwood model, + 51.3 vs. the pseudo-second-order model and + 58.3 vs. the pseudo-first-order model). Independent hold-out validation across current, concentration and temperature domains yielded a combined RMSE of 3.7%, indicating predictive capability beyond the calibration domain. Thermodynamic analysis using an electrochemical Van't Hoff formalism with nonlinear regression yielded ΔH°ads =  - 28.5 kJ/mol and β = 1.85 kJ·A⁻1·mol⁻1. β is best regarded as an empirical current-dependent fitting parameter rather than a fully validated physical thermodynamic quantity; its molecular interpretation remains phenomenological pending direct surface characterisation. Energy optimisation revealed a trade-off between cost-efficiency (EE/O = 17.4 kWh/m3/order at 1.50 A, 86% removal) and maximum removal (EE/O = 19.2 kWh/m3/order at 1.98 A, > 90% removal). The lowest EE/O (13.5 kWh/m3/order) was observed at 2.30 A, but this condition required the highest total energy per volume and produced the most sludge. The total operating cost ranged from 2.7 to 3.5 €/m3, with 1.50 A providing the lowest cost. Monte Carlo simulations showed 94% probability of achieving > 85% removal under operational uncertainties, with sensitivity analysis identifying adsorption kinetics (kads, S = 0.89) as the dominant control parameter. While partial mineralisation (35%) and uncharacterised transformation products remain limitations and validation was limited to synthetic solutions in a single laboratory reactor, the A-R model provides a reliable, transferable tool for EC process design and scale-up. The findings contribute to international efforts to develop sustainable treatment technologies for antibiotic-contaminated waters, positioning EC as a technically feasible and predictable option for pharmaceutical removal.