Kinetics of diatrizoate degradation by ozone and the formation of disinfection by-products in the sequential chlorination

In this study, we studied the degradation kinetics of a common iodine contrast agent, diatrizoate, by ozone and the formation of disinfection by-products (DBPs) in the sequential chlorination. Effects of ozone concentration, solution pH, and bromide concentration on diatrizoate degradation were evaluated. The results indicate that diatrizoate can be effectively degraded (over 80% within 1 h) by ozone, and the degradation kinetics can be well described using the pseudo- ﬁ rst-order kinetic model. The pseudo- ﬁ rst-order rate constant ( k obs ) of diatrizoate degradation signi ﬁ cantly increased with increasing ozone concentration and decreasing bromide concentration. The k obs kept increasing with the increase of pH value and reached a maximum of 6.5 ( + 0.05) (cid:1) 10 (cid:3) 2 min (cid:3) 1 at pH 9. As the ozone concentration gradually increased from 0.342 to 1.316 mg/L, the corresponding k obs of diatrizoate degradation increased from 1.76 ( + 0.20) (cid:1) 10 (cid:3) 3 to 4.22 ( + 0.3) (cid:1) 10 (cid:3) 2 min (cid:3) 1 . The bromide concentration exhibited a strong inhibitory effect on diatrizoate degradation because of the competition for ozone with diatrizoate. Trichloromethane was the only detected DBP in the subsequent chlorination in the absence of bromide. However, in the presence of bromide, six other DBPs were detected, and bromochloroiodomethane and tribromomethane became the major products with concentrations 1 – 2 orders higher than those of the other DBPs. In order to provide safe drinking water to the public, water should be maintained at circumneutral pH values and low bromine concentrations ( , 5 μ M) before reaching the chlorine disinfection process to effectively control the formation of DBPs.


INTRODUCTION
Iodinated X-ray contrast media (ICM) are a class of pharmaceutical products widely used to increase the contrast of tissue, structures, and details of organs and blood vessels during medical imaging at hospitals and medical centers (Hu et al. 2019a). The annual global consumption of ICM is approximately 3.5 Â 10 6 kg (Perez & Barcelo 2007;Wang et al. 2016), and the annual estimated production of ICM is more than 5 Â 10 6 kg (Gharekhanloo & Torabian 2012;Hu et al. 2019a). Typically, 95% of non-metabolized ICM can be eliminated through urine and feces within 24 h after adoption (Sandra Perez et al. 2006). The molecular structures of ICM consist of 2,4,6-triiodinated benzoic derivatives, and their molecular weights vary between 600 and 900 Da (Jeong et al. 2017). The structures of ICM are very stable and cannot be effectively removed by traditional water treatment processes (Hu et al. 2020a). Therefore, trace amounts of ICM are commonly detected in surface waters (Duirk et al. 2011;Kormos et al. 2011). In a previous study, many kinds of ICM were detected in the rivers (Duirk et al. 2011) of drinking water sources in the US at the levels of 10-2,700 ng/L, including iopamidol, iomeprol, iopromide, iohexol, and diatrizoate.
ICM can be divided into ionic and nonionic groups (Hu et al. 2019b), which are all commonly used ICM nowadays . Diatrizoate is a nonionic ICM that is especially resistant to conventional wastewater and drinking water treatment processes (Sugihara et al. 2013). Therefore, many researchers have focused on the degradation of diatrizoate using advanced oxidation techniques (Hennebel et al. 2010;Velo-Gala et al. 2014;Azerrad et al. 2016;Polo et al. 2016;Meng et al. 2017). For example, Hennebel et al. (2010) removed diatrizoate using a catalytically active membrane, and the removal efficiency reached 77% after 48 h. Velo-Gala et al. (2014) compared the degradation of diatrizoate in a solution with various advanced oxidation processes (AOPs) using iron salts and ultraviolet light (UV) radiation. Experimental results showed that UV/K 2 S 2 O 8 is more effective than UV/H 2 O 2 with a higher reaction rate constant for diatrizoate degradation. Polo et al. used solar radiation to degrade diatrizoate (Polo et al. 2016) and reported that the formation of free radical promoting substances (including complexes of peroxygen and iron hydroperoxide) can absorb radiation in the solar radiation area, resulting in 100% sunlight degradation of diatrizoate. A German municipal sewage treatment plant (STP) using ozonation only achieved 14% removal of diatrizoate (Ternes et al. 2003).
Halogenated disinfection by-products (DBPs) are produced by natural organic matter (NOM) reacting with chlorine during drinking water disinfection ( Jiang et al. 2020). The presence of ICM in the source water may also react with chlorine to produce highly toxic iodinated DBPs (I-DBPs), which have aroused high concerns for public health (Jeong et al. 2017). DBPs have chronic cytotoxicity and genomic induction. Research results indicated that 13 kinds of haloacetamides can damage the DNA of Chinese hamster ovary cells (Michael et al. 2008). Moreover, Jiang et al. (2017) reported that the presence of bromide ions in the raw water may participate in the reactions between NOM and chlorine to generate brominated-DBPs (Br-DBPs), which are more carcinogenic and mutagenic than their chlorinated analogues. Epidemiological studies have consistently observed a correlation between chlorine consumption in drinking water and increasing risk of bladder cancer (Li & Mitch 2018). DBPs and their toxic effects will put pressure on the living creatures in the environment and organisms (Shang et al. 2019;Ranjan et al. 2021). ICM is a potential source of toxic I-DBPs during water disinfection, such as iopamidol (Mao et al. 2020). It has been reported that I-DBPs are generated during chlorine disinfection of iodide-containing water in sewage treatment systems (Allard et al. 2013;Xia et al. 2017;Liu et al. 2018;Zhang et al. 2019). Allard et al. (2013) reported that in iodide-containing waters, I-DBPs can be produced during chlorination or especially chloramination; a pre-ozonation step to oxidize iodide to iodate is an efficient process to mitigate I-DBP formation. I-DBPs have recently become an emerging group of DBPs with public health concerns (Yolanta et al. 2014). To the authors best knowledge, no literature regarding the degradation kinetics of diatrizoate ozonation and DBP formation after post-chlorination have been reported, which are worth research attention in order to provide safe drinking water to the public.
Therefore, the purposes of this study were (1) to investigate the kinetics of diatrizoate degradation during ozone oxidation, (2) to evaluate the effects of bromide concentration and solution pH on diatrizoate degradation, and (3) to investigate the formation of regulated and emerging DBPs of ozonated diatrizoate in the subsequent chlorination to simulate water distribution in the pipelines.

Ozone reaction system
The schematic diagram of the ozone reaction system is displayed in Figure 1. Ozone gas was produced by a COM-AD-01 ozone generator (Anseros, Germany). The desired ozone concentration was controlled by adjusting the flow rate of a stream of pure ozone gas into the reactor. Before each experiment, the concentration of dissolved ozone in the reactor was monitored by measuring the amount of I 2 formed in the reaction between O 3 and potassium iodide solution according to the sodium thiosulfate titration method based on the following equations (Giuseppe et al. 2007). (1) An internal micro pump (MP-10R, Keyuan Pumps Co., Ltd, Shanghai, China) and an external peristaltic pump (DP-130, Shanghai Magnetic Pump Co., Ltd, Shanghai, China) were adopted to circulate the water inside the double-layer glass reactor and the water flowed in the outer layer, respectively, to maintain the system temperature at 25 + 1°C.

Experimental procedures
The ozonation experiments were carried out in a ventilated hood. Diatrizoate stock solution (1 mM) was prepared to make each reacting solution of 20 μM and 200 mL using ultrapure water. The pH value of the reaction solution was adjusted to a desired value (pH 5-9) by adding a small amount (,5 mL) of H 2 SO 4 (0.18M) or NaOH (1M) in the presence of 10 mM phosphate buffer. In order to ensure the purity of the produced ozone, the ozone gas produced by the generator was pre-ventilated for 30 min into a saturated potassium iodide solution to adsorb ozone before releasing into the atmosphere. After pre-aeration, the pure ozone gas was passed into the double-layer reactor to start the reaction. For the kinetic experiments, the ozone concentrations were set at 0.342-1.316 mg/L, while for the degradation experiments, the O 3 concentration was fixed at 0.713 mg/L. At a certain time interval, 1 mL of the reaction solution was taken from the sampling port and transferred into an HPLC vial containing Na 2 S 2 O 3 (concentration more than 1.2 times the ozone concentration) to quench the reaction, and the sample was stored at 4°C in the dark for analyzing the residual diatrizoate concentration using an HPLC as soon as possible. Parallel experiments were performed simultaneously. On the other hand, TBA with concentration more than 1.2 times the ozone concentration was dosed in the solution to verify the contribution of hydroxyl radicals (·OH) on diatrizoate degradation during ozonation.
To study the effect of bromide concentration on diatrizoate degradation, the experiments were performed by dosing bromide concentrations of 0-150 μM at pH 7 and ozone concentration of 0.713 mg/L. For the evaluation of DBP formation after diatrizoate ozonation in the sequential chlorination to simulate drinking water distribution in the pipelines, the diatrizoate solution was prepared in a 45 mL vial dosing 100 μM free chlorine, sealed and preserved in a thermostatic incubator at 25°C for 3 d. Then, all samples were quenched with Na 2 S 2 O 3 . Each DBP formation experiment was performed in duplicate with averaged values being reported in this study.

Analytical methods
An Agilent 1200 infinity series HPLC system (Agilent, USA) was used to analyze the concentration of diatrizoate with an XTerra MS C18 column (5 μm, 250 mm Â 4.6 mm, Waters, USA), a UV detector, and an autosampler with an injection volume of 5 μL. The mobile phase was composed of 10%/90% (v/v) methanol and 1% phosphoric acid solution (v/v) at a flow rate of 1 mL min À1 , and the volume ratio is 10%/90%. The detection wavelength of diatrizoate was 237 nm.
A GC (GC-2010 Plus, Shimadzu, Japan) equipped with an RTX-5 fused-silica capillary column (30 m Â 0.25 mm id, 0.25 μm film thickness) and an electron capture detector was used to analyze the concentrations of the formed DBPs. The analysis was conducted according to the US EPA method 551.1 (Richardson et al. 2007).

RESULTS AND DISCUSSION
3.1. Kinetics of diatrizoate degradation during ozone oxidation Figure 2 displays the pseudo-first-order kinetics plot of diatrizoate degradation during ozonation with different initial ozone concentrations. Diatrizoate can be degraded by more than 91% after 60 min ozonation. The fitted line in Figure 2 exhibited a high linearity (R 2 . 0.99), indicating that diatrizoate degradation follows pseudo-first-order kinetics during ozonation. The degradation rate of diatrizoate can be expressed as follows: where [Diatrizoate] represents the diatrizoate concentration at time t, and k obs (min -1 ) represents the observed pseudo-firstorder rate constant of diatrizoate degradation during ozonation. The slope in Figure 2 is k obs . As the initial ozone concentration increased from 0.342 to 1.316 mg/L, k obs increased linearly from 1.76 (+0.20) Â 10 À3 to 4.22 (+0.3) Â 10 À2 min À1 (R 2 . 0.923), which can be explained by the increasing concentrations of dissolved ozone and the resulted hydroxyl radicals in the solution (Huang et al. 2020a). Therefore, the ozonation degradation of diatrizoate can be further expressed as follows: where k app represents the apparent second-order reaction rate constant of the entire reaction. In sum, the ozonation degradation of diatrizoate fitted the second-order kinetics well, one order in ozone concentration and one order in diatrizoate concentration. The relationship between k obs and k app can be expressed (Equation (5)) and calculated as 0.03768 L (min mg) À1 , which is higher than that of iohexol degradation during ozonation (Hu et al. 2020a).

Effect of tert-butanol (TBA) on diatrizoate degradation during ozonation
In order to verify the contribution of hydroxyl radicals (·OH) on diatrizoate degradation during ozonation, the commonly used quencher, TBA, was dosed during the reaction. As displayed in Figure 3(a) and 3(b), the degradation rate of diatrizoate decreased sharply in the absence of TBA, while the degradation rate of diatrizoate only slightly decreased in the presence of TBA. Therefore, it can be concluded that ·OH played the major role in diatrizoate degradation during ozonation along with partial contribution of ozone and other radicals.

Effect of solution pH on the degradation of diatrizoate during ozonation
The effect of solution pH on diatrizoate degradation during ozonation was investigated at pH 5.5-9, and the results are presented in Figure 4. The degradation rate of diatrizoate gradually increased from 4.17 (+0.04) Â 10 À3 min À1 to the maximum Uncorrected Proof of 6.5 (+0.05) Â 10 À2 min À1 as pH increased from 5.5 to 9.0. The dosed diatrizoate can be completely decomposed at pH 7.5-9.0 within an hour. The rate constants of diatrizoate degradation during ozonation in alkaline conditions were significantly higher than those in acidic conditions. This phenomenon can be explained by the major contribution of hydroxyl radicals   (·OH) to diatrizoate degradation, which will be further discussed in Section 3.5. At pH . 7, increasing OH À concentration in the solution can promote the formation of hydroxyl radicals according to Equations (6) and (7). A series of reactions can happen with the generation of other active radicals at different pH values, such as ·OH, O 2 À ·, O 3 À ·, HO 3 ·, O À ·, and HO 2 ·(Equations (8)- (13)). Therefore, the pH of the solution played an important role in diatrizoate degradation during ozonation (Gunten 2003;Virmani et al. 2020).

Effects of bromide concentration on diatrizoate degradation during ozonation
Figure 5(a) and 5(b) displays the effect of bromide concentration on the degradation of diatrizoate. It is obvious that the rate of diatrizoate degradation significantly decreased as the bromide concentration increased in the solution. The reason is that Br À can compete with diatrizoate for ozone consumption and cause a series of chain reactions (Equations (14)-(20); Song et al. 1996). Solutions containing Br À are prone to form BrO 3 À during ozonation (Equation (18)), which has weak oxidizing power for organic compounds (Gunten 2003;Nie et al. 2013). Although in Equation (16), a strong oxidant HOBr and several bromine radicals can be formed, a part of them can further react with ·OH to form BrO À and BrO 2 À (Equations (17)-(20)), which are also weak oxidants (Huang et al. 2020b). Therefore, the presence of bromide exhibited an inhibitory effect on diatrizoate degradation during ozonation.
Br Á þBr À ! Br 2 À Á, k ¼ 6 Â 10 12 M À1 min À1 Br 2 À Á þ Br 2 À Á ! Br À þ Br 3 À Á, k ¼ 1:2 Â 10 11 M À1 min À1 3.5. Effect of solution pH on DBP formation of ozonated diatrizoate in the sequential chlorination In previous literature, pH can have significant effects on DBP formation during disinfection (especially using chlorine). Therefore, DBP formation potential of diatrizoate after ozonation was also evaluated after 3 d chlorination to simulate water distribution in pipelines. As shown in Figure 6 in the absence of bromide, only trichloromethane (CHCl 3 ) was detected in the sequential chlorine process. Due to the strong oxidizing ability of ozone and ·OH, diatrizoate can be decomposed into small molecular structures. ICMs can be degraded to a different extent during AOPs to form inorganic iodine and organic iodide at pH .7 (Ina et al. 2009), and ozone can oxidize iodide ions to HOI (Equations (21) and (22)) that can further react with O 3 to generate IO 2 À and IO 3 À (Equations (23) and (24)). Therefore, iodate will be the major product during  Uncorrected Proof the ozone disinfection process without forming a considerable amount of I-DBPs (Von 2003).
H þ þ OI À ! HOI, k ¼ 2:16 Â 10 6 M À1 min À1 (22) 3.6. Effects of bromide concentration on the formation of DBPs after post-chlorination Figure 7 displays the effect of bromide concentration on the formation of DBPs after chlorination of ozone oxidized diatrizoate. It can be seen that six more kinds of DBPs besides CHCl 3 were detected, including CHClBr 2 , CHCl 2 Br, CHClBrI, CHBr 2 I, C 2 HBr 2 N, and CHBr 2 Cl, and the dominant one was CHClBrI. As the bromide concentration increased from 5 to 200 μM, C 2 HCl 3 , C 2 HBrN, and CHClBr 2 formation remained relatively low and stable. However, the concentrations of CHBr 3 increased with increasing bromine concentration, while the concentrations of two detected I-DBPs, CHBr 2 I, and CHClBrI exhibited an increasing and then decreasing trend. As the bromide concentration reached 100 μM, CHBr 2 I was detected. Accordingly, the presence of Br À has a considerable impact on the production of DBPs. In order to analyze the contribution of bromine to the formation of THMs, the bromine incorporation factor (BIF) was calculated according to the following equation: The range of BIF is 0-3 (Rathbun 1996). The larger the BIF value, the greater the contribution of bromine to the formation of THMs, indicating the greater proportion of Br-THMs in the total concentration of THMs. As the concentration of bromide increased from 10 to 200 μM, the BIF increased from 1.15 to 2.01, implying that Br-DBPs were preferentially produced, and chlorinated-DBPs were inhibited in the presence of bromide in the subsequent chlorination after diatrizoate ozonation (Ates et al. 2007). The presence of Br À in the solution can react with HOCl to form HOBr (Cowman & Singer 1996;Brix et al. 2017), which is a strong oxidizing agent with oxidation power stronger than that of HOCl (Symons et al. 1993;Maryam et al. 2020) and can react with diatrizoate or its degradation intermediates at higher reaction rates (Chang et al. 2001), causing more formation of Br-DBPs (and even I-DBPs) with higher toxicity. Therefore, although ozone is efficient to decompose diatrizoate, it is essential to control the potential risks of forming toxic Br-DBPs and I-DBPs in the presence of bromide in the sequential chlorination.

CONCLUSION
This study shows the successful application of ozone to completely decompose diatrizoate in alkaline conditions. Diatrizoate degradation during ozonation followed pseudo-first-order kinetics well. ·OH played a major role in the diatrizoate degradation process, which was confirmed by the quenching experiments using TBA. At 25°C, the degradation rate of diatrizoate by ozonation increased with increasing solution pH and ozone concentration. In the sequential chlorination simulating for water distribution in the pipelines, only one DBP, CHCl 3 , was detected, and its formation was not pH-dependent. On the other hand, as Br À concentration increased, the diatrizoate degradation efficiency decreased significantly due to the competition of ozone with diatrizoate, and six more kinds of DBPs were detected in the water after post-chlorination. Moreover, as Br À concentration increased to 100 μM, Br-DBPs were preferentially produced, and CHBr 2 I was detected. Therefore, although ozone is efficient to decompose diatrizoate, it is essential to control the potential risks of forming toxic Br-DBPs and I-DBPs in the presence of bromide in the sequential chlorination.