The transformation mechanism of ofloxacin (OFL) by chlorine dioxide disinfection in water distribution system: kinetics, byproducts and toxicity

Chlorine disinfection is currently the most cost-effective and widely used drinking water disinfection process for water supply systems (Gallard & von Gunten 2002; Gao et al. 2018). However, this inevitably produces halogenated disinfection byproducts that are harmful to humans, such as trihalomethanes (THMs) and haloacetic acids (HAAs) (Sharma et al. 2014). The US Environmental Protection Agency (EPA) has defined the maximum content of THMs and HAAs in drinking water as 80 μg/L and 60 μg/L, respectively (Ruecker et al. 2017). In recent years, following increasing research on the hazards of traditional chlorination disinfection byproducts and to meet the drinking water quality standards for THMs and HAAs, ClO₂ disinfection technology has received increasing attention (Volk et al. 2002; Lv et al. 2021). ClO₂ results in rapid and efficient sterilization and reduces the generation of halogenated disinfection byproducts (Navalon et al. 2008; Sorlini et al. 2014).

Fluoroquinolone antibiotics (FQs) are a class of broad-spectrum antibacterials that are widely used to treat diseases in humans and animals (Xu et al. 2015; Guo et al. 2017; Castrignanò et al. 2020). However, after ingestion, they cannot be completely absorbed and some FQs can be excreted as technical components or active metabolites and enter into the sewage system (Tong & Xiang 2007; Lastre-Acosta et al. 2019). Traditional wastewater treatment systems are not able to remove FQs efficiently, resulting in the migration of FQs among environmental media such as sewage plants, surface water, and groundwater, ultimately affecting systems of drinking water treatment (Yahya et al. 2014; Annabi et al. 2016; Y. Jia et al. 2018). An investigation conducted by Ye et al. (2007) found that FQ antibiotics in finished water in North Carolina were present at a concentration of 1.2–4.0 ng/L. In addition, the detection rates of FQ antibiotics in the drinking water of Guangzhou and Macau in China have been determined to be as high as 77.5% and 100%, respectively. The contamination of FQs in drinking water mainly comes from the incomplete treatment of FQs in drinking water plants. The long-term consumption of water containing FQs can lead to an increased risk of drug resistance in humans and other organisms (Clark et al. 2001; Janecko et al. 2016). Therefore, when ClO₂ is used for disinfection, the transformation mechanism of residual FQs in water supply systems remains to be further studied, so as to ensure the safety of drinking water.

Ofloxacin (OFL) is a third-generation FQ antibiotic, mainly used for urinary infections, upper respiratory tract infections, and gastrointestinal infections caused by sensitive bacteria (Yassine et al. 2017; Wang et al. 2019). OFL contains a quinolone ring in its chemical structure, which can be removed by oxidation with a variety of oxidizing agents including manganese dioxide, ozonation, ferrates, and chlorine (Sharma 2008; Carbajo et al. 2015; Yassine et al. 2017). However, most current studies on OFL have focused on simple reaction environments, such as beakers of deionized water or organic solvents. The study of the migration and transformation of residual OFL in the water supply network under the action of ClO₂ has not yet been reported. Different pipes, flow rates, temperatures, and other factors in the actual water supply network can affect the decay of ClO₂ and the reaction rate between ClO₂ and pollutants (He et al. 2019; Li et al. 2019). For example, in the study of the degradation of fleroxacin (FLE) and chloramphenicol (CAP) by ClO₂ in different pipes, results have shown that the degradation of FLE or CAP has an obvious difference in various pipes (He et al. 2019; Li et al. 2019). The differences in results obtained in a simple reaction environment would be more notable than those in an actual water supply system. Therefore, this study systematically investigated the transformation mechanism of residual OFL in a water supply system under the action of ClO₂. According to the previous study of degradation of OFL by other oxidants (Liu et al. 2020)_, both OFL and its intermediate products have acute toxicity, and so the acute toxicity during OFL degradation by ClO₂ in this study was also studied.

The details of this study include the following: (i) examining the effects of ClO₂ concentration, pH, and pipe materials on OFL degradation by ClO₂; (ii) identifying the production of OFL intermediates under the action of ClO₂ and elucidating the transformation mechanism; (iii) analyzing the generation of disinfection byproducts; (iv) assessing the actual toxicity during OFL degradation by ClO₂.

All chemicals were obtained from various manufacturers and were used directly without further purification. OFL with a purity of 98%, methanol, acetonitrile and MTBE were purchased from Aladdin Reagent Company (Shanghai, China). Sodium hypochlorite, sodium hydroxide, phosphoric acid, sodium thiosulfate and ascorbic acid were provided by Sinopharm Reagent Company (Shanghai, China). THMs standard and HAAs standard with gas-chromatographically pure mixture were obtained from Cansyn (Toronto, Canada).

The OFL degradation experiments were carried out both in deionized water and in the pilot-scale water supply system, respectively. The pilot-scale pipe network consists of three loops, which are ductile iron pipe, PE pipe and stainless steel pipe (Figure S1). Before the start of the experiment, the pipe and the dosing pump were cleaned with fresh tap water for about 30 minutes. When the cleaning was completed, the cleaning water was drained, and fresh tap water was injected again to start the experiment. The characteristics of fresh tap water were measured each day, with results given in Table S1. The experimental parameters such as flow rate and water temperature were adjusted to the set values by the automatic adjustment system set on the water supply system. Appropriate amounts of phosphoric acid or NaOH solution were added to the water supply system through the dosing pump to adjust the pH, and then the ClO₂ solution and OFL solution were added to the pipe network in turn to start the experiment. The active ClO₂ stock solutions were prepared based on the hydrochloric acid method by mixing hydrochloric acid with sodium chlorite (NaClO₂) to the desired concentrations, and the purity was more than 99%. After the experiment, the concentration of OFL at different times was measured by high performance liquid chromatography (HPLC). In order to reduce accidental errors and ensure the reliability of experimental results, each group of experiments was repeated three times.

The OFL concentration changes during the degradation process were detected by HPLC (Agilent 1200, Santa Clara, CA, USA). The detector was a UV detector with a wavelength of 293 nm. The column temperature was 30 °C and the ratio of mobile phase A (0.5% phosphoric acid) and mobile phase B (methanol) was 75:25 with a flow rate of 1 mL/min. Table S2 lists the limit of detection (LOD) and limit of quantification (LOQ) for OFL. Before the detection of intermediates of OFL oxidation by ClO₂, the water samples were enriched and concentrated by solid-phase extraction. Addition in turn of 6 mL of methanol and 6 mL of deionized water activated the CNWBOND HC-18 SPE (500 mg/3 mL) extraction column. A 1,000 mL water sample flowed through the extraction cartridge at a flow rate of 5 mL/min. After the sample passed, 6 mL of methanol was added to elute the extraction cartridge. The eluate was then blown dry with nitrogen at 45 °C, and then 1 mL of methanol was added to collect oxidation products. The samples were detected by LC-MS (Agilent 6460, Santa Clara, CA, USA). The LC-MS detection conditions were as follows: the chromatographic column model was a Zorbax SB-18 column (2.4 mm × 150 mm, 5 μm). The chromatographic conditions were mobile phase A (0.2% formic acid) and phase B (acetonitrile), using gradient elution: 0–5 min 20% B, 15 min 35% B, 20 min 80% B, 20 –23 min 100% B at a flow rate of 0.3 mL/min. The ionization mode was positive ion electrospray ionization (Electrospray Ionization ESI+), with carrier gas temperature 350 °C, flow rate of 10 mL/min, scanning range of 50–500, pressure of 25 psig, fragmentor voltage of 110 eV, capillary voltage of 4,000 V.

THMs were detected by GC-ECD (GC-450, Varian, Palo Alto, CA, USA). Column type was HP-5 (30 m × 0.32 mm × 0.25 mm), and the injection mode was split with an inlet temperature of 210 °C and injection volume of 3 μL. The detector temperature was 290 °C, while the heating program was as follows: 35 °C for 13 min, then keep at 30 °C/min to 200 °C for 3 min. Makeup gas flow rate was 30 mL/min, and the carrier gas was high-purity nitrogen with a flow rate of 1 mL/min. The method detection limits and recovery for THMs are provided in Table S3(a).

Before the analysis of HAAs, MTBE liquid–liquid extraction was used to pretreat the water samples, and then GC (Varian, Palo Alto, CA, USA) was used for detection. The column model was an HP-5 capillary column (15 m × 0.25 mm × 0.25 μm, Agilent). The injection method was headspace injection/splitless injection, and the detector temperature was 300 °C. The injection port temperature was 175 °C, with an injection volume of 1 μL. The heating program was as follows: 40 °C for 5 min, then 10 °C/min to 140 °C, followed by 25 °C/min to 190 °C, and hold for 0.5 min. The makeup gas flow rate was 30 mL/min. Carrier gas was high-purity nitrogen, with a flow rate of 1 mL/min. The method detection limits and recovery for HAAs are provided in Table S3(b).

The concentrations of ClO₂ were measured by spectrophotometry, using a DR-2800 spectrophotometer (HACH, Loveland, CO, USA). The chlorite in the water sample was detected by ion chromatograph (Dionex ICS-2000, Sunnyvale, CA, USA). The total organic carbon (TOC) of tap water in the pilot-scale pipe was detected by a TOC analyzer (Shimadzu, Kyoto, Japan) (Dong et al. 2017). The acute toxicity of water samples collected at different reaction times during OFL oxidation by ClO₂ were evaluated based on the luminescence inhibition of Vibrio fischeri bacteria. The luminescence inhibition was detected by a Microtox^® Model 500 comprehensive toxicity detector (Strategic Diagnostics Inc., Newark, DE, USA).

As seen from Figure 1, the degradation rate of OFL in both deionized water and PE pipes accelerated with an increasing ClO₂ concentration. When the initial ClO₂ concentration was 0.3 mg/L, the degradation efficiencies of OFL in deionized water and PE pipe, after 90 s of reaction, were 73% and 68%, respectively. When the initial chlorine concentration was increased to 1.3 mg/L and the reaction was performed for 90 s, the degradation rate was 100% in both deionized water and PE pipes. In addition, the degradation rate of OFL in deionized water was higher than that in PE pipe. This was mainly owing to the organic matter, pipe scale, and biofilm of the pipe-wall consuming part of the ClO₂, which made the proportion of ClO₂ used for OFL degradation in PE pipe decrease. In addition, within the selected concentration of ClO₂ in this study, OFL can be effectively removed by 600 s reaction, indicating that OFL degradation mainly occurs at the beginning of the system.

As can be seen from the visible pipe section of the pipe system (Figure S3), the pipe scale in the three pipe networks was as follows: stainless steel pipe ≈ PE pipe > ductile iron pipe. The pipe scale contains corroded metals, solid metal oxides, metal ions, biofilms, and natural organic matter, and all of these factors can contribute to the consumption of oxidants. In this case, the ClO₂ consumed by the pipe wall in the three pipe networks was in the following order: stainless steel pipe ≈ PE pipe > ductile iron pipe. In addition, it was found that the organic concentrations of the ductile iron pipe, PE pipe, and stainless steel pipe were 2.37 mg/L, 3.38 mg/L, and 3.09 mg/L, respectively. Therefore, the ClO₂ content available for OFL degradation in ductile iron pipes was higher than that of PE pipe and stainless steel pipe. Furthermore, the concentrations of iron in the water samples extracted from stainless steel pipe and ductile iron pipe were 0.27 mg/L and 0.14 mg/L, respectively, whereas the PE pipe water samples were found to contain no iron ions. Fe²⁺ can be oxidized into Fe³⁺, and the free radicals generated during the iron oxidation can promote OFL degradation in stainless steel pipe and ductile iron pipe (Dong et al. 2017). Comprehensively considering pipe scale, the organic matter in the main water bodies, and the effect of iron on OFL degradation, it should be noted that OFL degradation was fastest in ductile iron pipe, followed by that in stainless steel pipe, and it was slowest in PE pipe.

The OFL degradation in PE pipe was examined at an initial ClO₂ concentration of 1.0 mg/L and an initial OFL concentration of 250 μg/L. Eight intermediates were identified by LC-MS analysis under the action of ClO₂.

In pathway I, the piperazine group of OP-335 first undergoes the fracture and removal of ammonia nitrogen and alkyl (−C₃H₇N) to produce the intermediate OP-278. Subsequently, the intermediate OP-278 undergoes decarboxylation and dealkylation to produce the product OP-220. In addition to that in pathway I, the intermediate product OP-335 can undergo decarboxylation, degradation of the piperazine group, defluorination, and chlorine substitution reactions. The intermediate P-291 is the product of the decarboxylation of P-325 catalyzed by the N4 of the piperazine group. Subsequently, the hydrogen atom on the quinolone group of the intermediate OP-291 is replaced by a chlorine atom via halogenated decarboxylation to form OP-325. Then, the piperazine group in the intermediate P-325 is further oxidized to produce P-268. In addition, OP-291 can generate OP-260 via the removal of ammonia nitrogen and alkyl on the piperazine group. Finally, the product P-268 undergoes sequential dealkylation to produce the compound P-220.

The acute toxicity of the water during the degradation of OFL was examined using experimental conditions consistent with those of the analysis of the intermediates. Meanwhile, two groups of blank control tests were set up as follows: (1) only OFL solution was added without ClO₂ solution (blank experiment A); (2) only ClO₂ was added without OFL solution (blank experiment B). The conditions of the blank experiment were consistent with those of the OFL oxidation experiment, except that no ClO₂ or OFL was added.

The degradation rates of OFL under the action of ClO₂ in deionized water and PE pipes were positively correlated with the ClO₂ concentration and pH, and the degradation rates in different pipes were as follows: ductile iron pipe > stainless steel pipe > PE pipe. The detection of OFL degradation intermediates showed that the product of piperazine-group cleavage was the initial and the main intermediate. Furthermore, the intermediate of quinolone group decarboxylation was also detected. The contributions of OFL degradation under the action of ClO₂ to the generation of halogenated DBPs and chlorite were limited. OFL degradation by ClO₂ leads to an increase of acute toxicity, indicating that a decrease in OFL concentration might not mean a decrease of toxicity. In the actual water supply, while removing organic pollutants, necessary measures should be taken to reduce the generation of intermediate products during ClO₂ disinfection so as to reduce the toxicity of drinking water.

This work has been financially supported by the China Postdoctoral Science Foundation (No. 2018M632465) and the Doctoral Research Fund of Shandong Jianzhu University (X20038Z0101).

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

The authors declare there is no conflict.