Antibiotic Pollution Meets Its Match: Predictive Model Cracks Electrocoagulation Chemistry

Metronidazole is one of those pharmaceuticals that refuses to disappear. The nitroimidazole antibiotic, used worldwide against anaerobic bacterial and parasitic infections, routinely slips through conventional wastewater treatment plants and shows up in hospital effluents and surface waters, where it can drive the emergence of antibiotic resistance and threaten aquatic ecosystems. Now a team of Algerian researchers has delivered what may be the clearest quantitative picture yet of how an electrochemical water-treatment technique dismantles this stubborn molecule, wrapping the entire process in a single predictive model that could reshape how engineers design treatment systems for pharmaceutical-contaminated water.

The study, published in Environmental Science and Pollution Research by Ahlem Aicha Fakir, Nadjib Drouiche and Hakim Lounici, focuses on aluminum electrocoagulation, a process in which sacrificial aluminum electrodes dissolve under an applied current to generate aluminum hydroxide flocs in situ. These gelatinous precipitates sweep contaminants out of solution by adsorption, while parallel electrochemical reactions at the electrode surface generate oxidative species, including hydroxyl radicals and reactive chlorine species such as hypochlorous acid and hypochlorite, that can chemically degrade the antibiotic. The trouble, historically, has been that these two removal routes operate simultaneously and are devilishly hard to separate, which has left much of the electrocoagulation literature relying on empirical fits rather than genuine mechanistic understanding.

To break that impasse, the researchers built what they call a unified adsorption–reaction model, a dynamic framework that tracks four interconnected reservoirs over time: metronidazole dissolved in the aqueous phase, metronidazole adsorbed onto the flocs, the concentration of active chlorine species, and the accumulating mass of aluminum hydroxide. Rather than treating adsorption and oxidation as competing black boxes, the model couples them explicitly, with rate constants for oxidation, adsorption, desorption and surface reaction all estimated simultaneously. The team calibrated the model against sixty concentration–time observations spanning four current intensities and fifteen sampling times, in batch experiments covering initial antibiotic concentrations from 5 to 50 milligrams per liter and temperatures from 15 to 40 degrees Celsius.

The results are striking. The unified model achieved a coefficient of determination of 0.996 with a root mean square error of just 3.2 percent, and it decisively outperformed the classical alternatives. Using the corrected Akaike Information Criterion, a statistical measure that penalizes model complexity while rewarding predictive accuracy, the adsorption–reaction framework beat the Langmuir–Hinshelwood model by 34.5 points, the pseudo-second-order model by 51.3 points and the pseudo-first-order model by 58.3 points. In model-selection terms, those margins are not marginal; they represent overwhelming evidence that the coupled mechanism, not a simple empirical decay curve, is the right description of what happens inside the reactor.

Crucially, the model did not merely fit the data it was trained on. In independent hold-out validation across the current, concentration and temperature domains, the combined root mean square error was 3.7 percent, essentially matching the calibration performance. That kind of transferability is the holy grail of process engineering, because it means the model can be trusted to predict behavior under operating conditions it has never seen, which is precisely what is needed when scaling a laboratory reactor up to a real treatment train.

To dissect the underlying chemistry, the team ran probe experiments with tert-butanol, a scavenger that selectively quenches hydroxyl radicals, and combined this with a dissolved organic carbon mass balance. The analysis suggested an apparent oxidative contribution of about 42 percent to metronidazole elimination, split between 35 percent complete mineralization and 7 percent partial oxidation, with the remaining 58 percent attributed to adsorptive capture. But the authors are refreshingly candid about the limits of that interpretation: because tert-butanol does not efficiently quench reactive chlorine species, those percentages should be read as operationally estimated apparent contributions that bundle together hydroxyl-radical-mediated and chlorine-mediated pathways rather than as cleanly separated mechanistic channels. It is a nuance that matters for anyone hoping to fine-tune the process chemistry.

The thermodynamic side of the study is equally rigorous, and it addresses a well-known pitfall in the adsorption literature. Instead of the frequently misused linear Van’t Hoff approach, the researchers applied an electrochemical Van’t Hoff formalism with nonlinear regression, yielding a standard adsorption enthalpy of minus 28.5 kilojoules per mole, consistent with a spontaneous, exothermic adsorption process. They also extracted a current-dependent enthalpy coefficient of 1.85 kilojoules per ampere per mole, though they caution that this parameter is best regarded as an empirical fitting quantity whose molecular interpretation remains phenomenological until direct surface characterization can confirm what it physically represents. That kind of restraint is rare and welcome in a field where thermodynamic parameters are sometimes over-interpreted.

On the practical engineering front, the study maps out an energy and cost landscape that plant designers will recognize immediately. The electrical energy per order of removal, a standard figure of merit for water treatment, came to 17.4 kilowatt-hours per cubic meter per order at 1.50 amperes, achieving 86 percent removal at the lowest total operating cost of 2.7 euros per cubic meter. Pushing the current to 1.98 amperes lifted removal above 90 percent at a modest energy penalty of 19.2 kilowatt-hours per cubic meter per order. The lowest energy per order, 13.5 kilowatt-hours, appeared at the highest tested current of 2.30 amperes, but that condition demanded the greatest total energy input and generated the most sludge, illustrating the classic trade-off between efficiency metrics, throughput and waste production. Operating costs across the tested range spanned 2.7 to 3.5 euros per cubic meter.

Perhaps the most forward-looking element is the uncertainty analysis. Through Monte Carlo simulation, the team estimated a 94 percent probability of achieving better than 85 percent removal even under realistic operational uncertainties, a robustness figure that speaks directly to regulators and utilities weighing whether the technology can be trusted outside pristine laboratory conditions. Sensitivity analysis identified the adsorption rate constant as the dominant control parameter, with a sensitivity index of 0.89, meaning that anything operators can do to enhance floc adsorption kinetics, from pH management to electrode configuration, will pay the largest dividends in overall performance.

The authors are honest about the boundaries of their work: mineralization was only partial, transformation products were not characterized, and validation was confined to synthetic solutions in a single laboratory reactor. Real wastewater, with its competing organics and variable salinity, will pose sterner tests. Even so, the framework represents a meaningful shift in how electrocoagulation can be practiced, moving the technology from trial-and-error optimization toward genuine predictive process engineering. As concerns over pharmaceutical micropollutants and antibiotic resistance intensify worldwide, tools that let engineers forecast removal performance, energy demand and cost before a single electrode is installed could prove indispensable in the effort to build sustainable treatment systems for antibiotic-contaminated waters.

Subject of Research: Mechanistic modeling of metronidazole removal from water by aluminum electrocoagulation

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

Article References: Mechanistic adsorption–reaction modeling and thermodynamic analysis of metronidazole removal by aluminum electrocoagulation: a unified framework for predictive process engineering. (n.d.). https://doi.org/10.1007/s11356-026-38251-4

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38251-4

Keywords: electrocoagulation, metronidazole, antibiotic removal, water treatment, adsorption–reaction model, electrochemical oxidation, reactive chlorine species, Van't Hoff thermodynamics, energy efficiency, pharmaceutical micropollutants, process engineering, Monte Carlo simulation