Investigation of an electrolysis system with boron-doped diamond anode and gas diffusion cathode to remove water micropollutants

A diverse group of dangerous substances can be discharged during human activities. This group contains pharmaceuticals, drugs, personal care products (PCPs), steroids, hormones, surfactants, perfluorinated compounds (PFCs), flame retardants, industrial additives, as well as their transformation products (TPs) in the aquatic environment (García et al. 2021), whose share to damage health was published by EU to be 74% (Eurostat 2021). These substances are known as micropollutants (MPs), since their concentrations range from ng/L to μg/L, and are distinguished by chronic toxicity, endocrine disruption, and the development of pathogen resistance through its persistence and accumulative effect in water (Rosal et al. 2010). Accordingly, different biological, physical, and chemical processes were investigated and used to remove the micropollutants from water, especially from treated wastewater discharged from wastewater treatment plants (WWTPs). The efficiency of the treatment is determined by tracking the decrease in the concentration of certain ‘guide substances’ chosen from over 300 MPs discovered in WWTPs effluents (Schymanski et al. 2014). Considering the guide substances depends on their frequency in the in- and effluent of the WWTPs in adequate concentrations for testing, insufficient biological treatment, treatability via other processes (ozonation, adsorption on active carbon, etc.), and the testability without limitations (Götz & Otto 2015).

MPs are recalcitrant to biodegradation and sometimes toxic to microorganisms. It is also not easy to adsorb them onto active surfaces or retain them by NF/RO membranes (Stott 2003). For those reasons, MPs cannot be sufficiently removed via biological and physical processes, although the cost of such processes can be lower than chemical ones. The advanced oxidation processes (AOPs) are the most investigated ones among the chemical methods applied to remove MPs. Despite their ability to remove MPs effectively, there are some drawbacks to be considered for AOPs like high energy requirements, high operational and maintenance costs as well as the formation of toxic or persistent by-products during the oxidation process (Taoufik et al. 2020). Ozonation as one of the AOPs is widely investigated and even integrated into WWTPs to remove MPs from effluents, especially in Europe (Antakyali et al. 2015). According to different studies, municipal effluent needs 0.4–0.9 mg O₃ for 1 g of dissolved organic carbon (DOC) to remove standardized guide substances up to 80–85% from effluent (Schachtler et al. 2020). As the production of 1 g O₃ demands up to 12.5 Wh (Bui et al. 2016), the energy input of ozonation can dependently be calculated to be ∼62 Wh for each cubic meter of wastewater containing a DOC of ∼5 mg/L.

Ozone has high reactivity with pollutants, and it is accordingly used to remove MPs from municipal wastewater types on a wide range. However, some industrial wastewaters need a higher O₃ dose than usual for municipal wastewater because of their complex water matrix, which consumes additional ozone fraction for constituents other than organics. This increases the treatment costs and produces more toxic by-products (Wunderlin & Grelot 2021). For instance, increasing the ozone dose contributes to increasing the formation of bromate from bromide in the bromide-containing effluent (Soltermann et al. 2016), where bromate is the most critical byproduct of ozonation. Ozonation also leads to the generation of cancerogenic nitrosamines (Soltermann & Götz 2016), whose maximal allowed concentration limit in waters is 100 ng/L (WHO 2008). Additionally, ozonation oxidizes the trivalent chromium Cr (III) to the toxic form Cr (VI) in the form of chromate CrO₄²⁻ and dichromate Cr₂O₇²⁻, which must be only considered at a high O₃ dose like 1.25 g O₃/gDOC (Katsoyiannis et al. 2018). On the other side, ozone has a short half-life time (15–20 min in water at 20–25 °C) and is incredibly affected by pH, temperature, DOC, and the presence of scavengers. The high reactivity of O₃ and its short half-life time make it difficult to get a homogenously dissolved ozone concentration which demands a strong mixing power to support its reaction with pollutants in a short time, particularly at the trace concentrations of MPs (Bui et al. 2016).

Looking at other alternatives that can remove MPs from water even better than ozonation, the electrolysis systems show a good example through their intended ability to completely degrade organics in water into CO₂ and H₂O. The high generation rate of hydroxyl radicals at BDD ensures the completion of oxidation without the formation of bio-toxic by-products as reported by He et al. (2019). This distinction is very decisive in MPs oxidation which is mostly associated with the formation of bio-toxic intermediate products.

All the reaction pathways (Eqs. (6)–(12)) consume the radicals generated in the BDD-GDE-cell which eliminates its contribution to the MPs oxidation reducing the removal efficiency. However, the controlling of kinetics to promote the oxidation pathways can be investigated by optimizing the operating parameters to remove MPs as high as possible.

As mentioned, MPs are supposed to be completely oxidized into CO₂ and H₂O in the BDD-GDE-cell without the formation of any dangerous by-products. Additionally, there is another worthy distinction, which is the long-term disinfection effect of H₂O₂ remaining in water after treatment. A considerable amount of H₂O₂ remains in the effluent after treatment, which only hardly decomposes in water (activation energy E_a = 200 kJ/mol) at ambient temperature compared to ozone decomposing much easier with E_a = 76 kJ/mol (Ershov & Morozov 2009). This property enables H₂O₂ to better disinfect microbially polluted waters as it can longer stay in the water. It was stated that a concentration of about 100 mgH₂O₂/L can lead to a sufficient sterilizing level after 1-h contact time (Chung et al. 2020). Higher concentration up to ∼184 mg/L was easily reachable at 1 mA/cm² in the BDD-GDE-system after removing phenol from wastewater (Muddemann et al. 2021). Although the BDD-GDE-system has all those distinctive properties, its effectiveness in removing MPs and the related demanded energy input are still not sufficiently investigated and will somehow be addressed in the present work for the first time.

Electrical conductivity, pH, and temperature were monitored in water storage by portable measurement devices (Windaus, Germany). TOC was determined by TOC-analyzer-Primacs (Skalar, Netherlands). The concentration of 10 guide substances was quantified using an Agilent 1290 Infinity II UHPLC system coupled to an Agilent 6470 Triple Quadrupole MS system via an Agilent Jet Stream electrospray ionization source. Separation was performed on an Agilent ZORBAX Eclipse Plus C-18 column (50 × 2.1 mm, 1.8 μm particle size) with a gradient flow of two eluents – ultra pure water (A) and acetonitrile (B)— both acidified with 0.05% formic acid and set at a flow rate of 0.3 mL min⁻¹. All samples were filtered through 0.45 μm polyethersulfone syringe filters (PES, Macherey-Nagel, Germany) before the analysis (Chabilan et al. 2022). The treatment effectivity of the BDD-GDE-cell was determined as a mean to the percentages of all removed guide substances. The 10 approved substances to be observed in the investigated water had different initial concentrations as flow: 5-Methyl-1H-benzotriazole (868.3 ng/L), Acetyl-sulfamethoxazole (126.1 ng/L), Benzotriazole (5,396.9 ng/L), Carbamazepine (1,188.4 ng/L), Clarithromycin (471.7 ng/L), Diclofenac (3,013.2 ng/L), Hydrochlorothiazide (1,839.7 ng/L), Irbesartan (471.7 ng/L), Metoprolol (3,888.7 ng/L), and Sulfamethoxazole (267.7 ng/L). The information on the chemical structure and related registry number are listed in Table S3.

To determine the current density with the best MPs removal, four different current densities (5, 4, 3, 2 mA/cm²) in the current BDD-GDE-cell were first investigated at a recirculation flow rate of 4 L/min. The time for each run was adjusted to meet the maximal energy input reported for ozonation to remove ∼80% MPs in a WWTPs effluent with 5 g_DOC/m³ and can be calculated from the published studies as ∼62 Wh/m³ (Hollender et al. 2009). This way allows comparing the recent results (MPs removal and energy input) with that achieved by ozonation.

Then, the influence of turbulence of electrolyte flow on the organic removal was investigated at Re number of 1500 (laminar flow) and 27,700 (extremely turbulent flow) in two individual 3-h experiments by adjusting the flow rate through the BDD-GDE-cell at 0.22 and 4 L/min, respectively. Upon defining the optimal parameters, the BDD-GDE-cell was further operated for 180 min observing how the treatment efficiency of the cell would be improved without considering the extent of the demanded energy input.

Applying four current densities (5, 4, 3, 2 mA/cm²) at different treatment times (3.5–15 min) to meet the same energy input, MPs could be removed under all conditions but to different extents. However, MPs removal after the treatment showed different values as listed in Table 1.

At ∼62 Wh/m³, the highest removal of MPs was 10.2% at 2 mA/cm² (61 Wh/m³; 15 min) versus 7.8% at 5 mA/cm² (62 Wh/m³; 4 min) despite the higher generation rate of ·OH radicals at BDD (Saad et al. 2016) and the higher efficiency of H₂O₂ production at GDE (Muddemann et al. 2020) expected at 5 mA/cm². The higher removal at 2 mA/cm² is attributed to the longer treatment time of 15 min against 4 min at 5 mA/cm². This finding agrees with what Lan et al. (2017) stated that lowering the treatment time reduces the reaction probability between the radicals and organics, especially under low organic concentrations as that of MPs. This statement was also confirmed as the treatment time was increased from 3.5 to 7 min keeping the current density at 5 mA/cm² whereby the MPs removal increased to 12.7%; however, it duplicated the energy input to 123 Wh/m³. Dependently, it can be specified that increasing the current density, e.g. from 2 to 5 mA/cm², can improve the MPs removal very slightly. This result suggests operating the electrolysis cell at the lower current density to be able to increase the treatment time as long as possible and without demanding more energy. The concentration of H₂O₂ after treatment was low at all current densities and ranged between 5 and 10 ppm (Table 1). This low amount of H₂O₂ diminished its contribution as ·O₂H in removing MPs.

Working at lower turbulence was another factor that helped extend the treatment time at a certain energy input. The reason was that the lower turbulence in the space between the two electrodes in the BDD-GDE-cell improves the ions concentration gradient near each electrode surface and accordingly increases the electric conductivity there, which reduced the needed voltage for a specific current density as the result. This offered the possibility of a further increase in the treatment period without exceeding the allowed energy input limit and accordingly improved the removal value. The turbulence effect on the treatment performance in this work was investigated at two different Re numbers (1520 and 27,700) and the results are shown in Table S1. Conducting both 3-h treatment experiments at the equal current density (2 mA/cm²), the removed organics indicated as TOC removal (%) was higher for the laminar flow than that for the turbulent one (Table 2), which indirectly indicates that MPs as a part of the organics was also better removed at laminar flow. The last result can be confirmed considering the low concentration of H₂O₂ (5 mg/L) measured at Re = 27,700 compared to 30 mg/L at Re = 1520. As the generation rate of ·OH at BDD and H₂O₂ at GDE is expected to be the same in both experiments because of the equal adjusted current density at 2 mA/cm² (Schmidt 2003), the turbulence increase seems to support the side reactions of radical HO₂⁻ derived from H₂O₂ before it reacts with the organic molecules (radicals-organics) in the medium. Consequently, H₂O₂ is supposed to be extensively degraded by side reactions to produce H₂O as a final product through Equations (6) and (7) according to (Muddemann et al. 2021). On the other side, energy input became higher during the turbulent flow experiment than that during the laminar one because the voltage at Re = 27,700 increased from 3.24 to 5.4 V after 180 min to keep the current density constant at 2 mA/cm². Conversely, the voltage 2.97 became 4 V after 180 min at Re ≈ 1500. According to this statement, the turbulent flow showed an increase in the specific energy input by ∼50% as explained in Table 2.

At ∼62 Wh/m³, ozonation removes 77% of MPs (Schachtler et al. 2020) which is better than the removal and the energy demand achieved by the present BDD-GDE-system (∼46% at 571 Wh/m³). Indeed, the distinctively lower energy demand of ozonation is attributed to two reasons: (1) in slightly alkaline solution (up to pH = 10) and the presence of scavengers (carbonate, bicarbonate, etc.), the direct ozonation reaction dominates and works beyond the inhabitation effect of scavengers (von Sonntag & von Gunten 2012). This condition existed in the experiment, as pH increased from ∼7 to ∼8. Additionally, the inorganic carbon (carbonate and bicarbonate) in the treated water was ∼8 mg/L (∼40% of the total carbon (TC) in this water) as defined in Table S1; (2) the dominating direct reaction of ozone has high selectivity to unsaturated bonds as stated by (Gottschalk et al. 2010). As the chemical structures of all investigated guide substances listed in Table S3 are rich with unsaturated bonds (PubChem 2022), it is accordingly expected that ozone will oxidize these substances selectively reducing their concentrations. The last fact explains why more MPs can disappear by ozonation in comparison to that by the BDD-GDE-system, which generated nonselective radicals (Sopaj et al. 2015). Those radicals can be highly consumed through undesired reaction pathways as in Equations (6) and (12) and disappear before reacting with MPs existing at very low concentrations. The total organic carbon (TOC) concentration of the 10-guide-substances in this work was calculated in the electrolyte to be TOC_{10-guide-substances} = 0.014 mg/L. This low concentration of those substances brings its oxidation in hard competition with other organics found in the electrolyte (TOC_water = 12 mg/L) which consume most of the ·OH radicals before they oxidize MPs. Further side reaction pathways are presented in Equations (4)–(7) that consume an additional amount of ·OH and H₂O₂ and reduce the oxidation rate of MPs.

On the other side, the considerable concentration of total inorganic carbon (TIC_water) in the electrolyte by 8 mg/L promotes the reactions of Equations (8)–(10) consuming H₂O₂ and the reactions described in Equations (11) and (12) which consume ·OH radicals through HCO₃⁻. All these conditions support the side reaction pathways producing O₂, H₂O, and HCO₃⁻ as final products instead to oxidize MPs decreasing the oxidation efficiency.

Despite the moderate removal value (∼46% at 571 Wh/m³) reached in this work with the BDD-GDE-cell, the possibility to improve the MPs oxidation performance suggests the BDD-GDE-system as another alternative besides the ozonation, if the removal can be more increased up to ∼77% at lower energy by ∼62% and observing the possible formation of bio-toxic intermediate products. Working at lower current density (<< 2 mA/cm²) and longer treatment time in future experiments makes this intended result possible. In this context, it must be taken into consideration that during the treatment of industrial wastewater using ozone, a higher dose is to be expected than that published for municipal wastewater which increases the energy, as it needs more oxygen for producing ozone (Böhler et al. 2020). It is also highly necessary to refer to the H₂O₂ (up to 30 mg/L) remaining in the treated water, which can be disinfected for a long time after treatment and longer than disinfection with ozone.

The electrolysis systems are innovative techniques to remove MPs from wastewater. In this work, an electrolysis BDD-GDE-system showed that MPs can be effectively achieved at different settings of the operating parameters. Decreasing the current density to 2 mA/cm² and Re number to the laminar range increased the treatment time 3-fold at the same energy demand, which increased the removal of the different investigated micropollutants. Oxidizing MPs at a longer treatment time (180 min) brought the removal value of MPs to a high value so that diclofenac for example was removed up to 84%, exceeding the obligatory level. Also, reducing Reynolds number to the laminar range (∼1500) promoted the MPs removal, as it let – besides the longer treatment time – the radical-organic reactions outweigh the radical-consuming side reactions increasing the MPs (removal) oxidation probability. Indeed, these results indicate a high potential of possible application in the future for such an electrolysis system to degrade MPs in aquatic solutions.

This study was supported by the German Federal Ministry of Education and Research ‘Bundesministerium für Bildung und Forschung (BMBF)’ (grant No. 03XP0107E). Prof. Dr Harald Horn and his group in Water Chemistry and Water Technology of Engler-Bunte-Institut at Karlsruhe Institute of Technology are acknowledged for the micropollutants analysis. Many thanks also to Covestro AG for providing the efficient GDEs.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.