1. Introduction
N-nitrosodimethylamine(NDMA) is a class of disinfection byproducts and a frequently detected nitrosamine with a cancer risk level of 10−6 based on a 0.7 ng/L lifetime exposure(USEPA, 1987). As a result, certain countries have recently conducted surveys to determine the presence of NDMA in various source waters and have set an acceptable limit range for NDMA in drinking water as 3 to 100 ng/L (CDPH, 2009;Van Huy et al., 2011).
The fate and concentration of NDMA are primarily influenced by the source waters (precursor availability) and the type of disinfectant used(i.e., chlorine, chlorine dioxide, ozone, and chloramine). Several studies have extensively studied amine-functionalized organic sources as potential NDMA precursors (Gan et al., 2015;Selbes et al., 2012;Shen and Andrews, 2011). For example, a pharmaceutical ranitidine(RNT) with a class of N, N-dimethyl-α-arylamine moieties is characterized by its high molar conversion rate(>90 %) from chloramination (Shen and Andrews, 2011). In contrast, the NDMA yield from ozonation, chlorination, and chlorine dioxide oxidation is usually less than 1% (Gan et al., 2015;Jeon et al., 2016). Typically, chloramine is blamed for higher NDMA formation potential(FP), which largely depends on the structural susceptibility of the precursor and dichloramine formation (Selbes et al., 2012, 2018).
Early studies on NDMA mitigation strategies revealed that, regardless of disinfectant type, precursor pretreatment before chloramination could reduce the NDMA-FP. The key structural groups of precursor can be transformed or deactivated by oxidants such as free chlorine and potassium permanganate (Selbes et al., 2014;Shen and Andrews, 2013;Wang et al., 2015). However, in some cases, both oxidants can promote NDMA formation (Selbes et al., 2014;Wang et al., 2015). Furthermore, Zhao et al. (2021) revealed that neither chlorination nor photoirradiation alone could completely remove NDMA-FP from such as RNT. Although pH is the most important factor for determining the efficacy of pretreatment to reduce NDMA formation (Selbes et al., 2014;Shen and Andrews, 2013), the integration of chemical oxidants with ultraviolet(UV), and a variety of chloramination methods, shows that other factors need to be considered as well (Orak et al., 2018;Szczuka et al., 2020). For example, the photolysis in the presence of free chlorine(UV/Cl2) and permanganate(UV/MnO4-) can provide alternative oxidants for water treatment. Different reactive chlorine radicals and manganese(Mn) species, in particular, have been shown to be effective oxidants for the removal of organic pollutants (Guo et al., 2018;Yang et al., 2018). To our best knowledge, no previous studies have addressed the UV/MnO4- system for the degradation of NDMA precursors. Therefore, the current study focused on the formation of NDMA during wastewater chloramination. RNT was selected as the target precursor and the effects of different chloramination methods (breakpoint chlorination, stepwise chlorination, or rechloramination) on NDMA-FP were studied. Additionally, the efficiency of UV/Cl2 and UV/MnO4- in the effluents of wastewater treatment plants (WWTPs) was investigated as a novel pretreatment method for NDMA-FP reduction.
2. Material and methods
All chemicals and reagents were obtained from Sigma Aldrich or Showa and Junsei Chemicals. Unless otherwise stated, all experiments were conducted using deionized(DI) water with a conductivity lower than 14.4 μS/cm(Millipore, Bedford, MA, USA). A phosphate buffer(pH 7) was prepared from potassium monobasic and potassium dibasic salts. The pH values were adjusted as needed using 5 N sodium hydroxide and sulfuric acid(95%). Both influent and effluent wastewater samples were collected from four different WWTPs located in Busan, Gwangju, Daegu, and Seoul, Republic of Korea. Water quality parameters of WWTPs are listed in Table 1. All experiments were performed in triplicate.
2.1 Chloramination of precursors and oxidative pretreatment
2.1.1 Chloramination of precursors
Chloramination experiments were conducted in 0.1 L amber bottles by adding 3.0 mg Cl2/L preformed NH2Cl and 125 ng/L RNT at pH 7.5(10 mM phosphate buffer) for 72 h contact time. Samples were periodically withdrawn, quenched with excess sodium thiosulfate, and filtered through a polytetrafluoroethylene Whatman membrane filter before further analysis. For comparison, chlorination(3.0 mg Cl2/L), breakpoint chlorination(ammonification of water prior to adding free chlorine; Cl2/N weight ratio = 1:4, pH = 7.5), prechlorination(adding 3.0 mg Cl2/L preformed NH2Cl after premixing with water and free chlorine), stepwise chloramination(water was first chlorinated for 1 h and then ammonium was injected into the premixed solution; Cl2/N weight ratio = 4:1, pH = 7.5), and rechloramination(first, water was chloraminated for 10 h, after the total Cl2 of the solution neared zero, then the solution was rechloraminated with 3.0 mg Cl2/L preformed NH2Cl) were performed as reported elsewhere (Oak et al., 2018; Park et al., 2015;Yang et al., 2007). Experiments with WWTP effluents were also conducted under similar conditions.
2.1.2. Oxidative pretreatment followed by chloramination
Oxidative pretreatment experiments were conducted using free chlorine(3 mg Cl2/L) and permanganate(2 mg/L) with or without UVC irradiation(UV intensity = 0.56 mW/cm2). Adopted oxidant doses were chosen to meet manganese(0.05 to 0.1 mg/L) and residual chlorine(up to 0.5 mg/L) guidelines in typical water treatment plants(Jeong et al., 2017;Sgroi et al., 2018). Precise doses of oxidants and 10 μM RNT(buffered with 2 mM phosphate solution) were reacted at pH 7.5 for 30 min. During the reaction, no pH adjustments were made, unless otherwise specified. UV photolysis oxidation was conducted for 30 min inside a black box equipped with two 4-W(l = 254 nm, Sankyo Denki, Japan) and a quartz reactor. Sample aliquots(1 mL) were periodically withdrawn from the reactor and filtered through a polyvinylidene fluoride(PVDF) filter(0.45 μm, Whatman) before further analysis. Post-chloramination experiments were conducted in a manner similar to that mentioned in section 2.1.1, the only difference was that the samples were buffered with 10 mM phosphate at pH 7.5 for 24 h.
2.2 Analysis
Total chlorine and total phosphorus were quantified using a UV-Vis spectrophotometer(Scinco 3100, Korea or Hach DR 2700, USA) according to the DPD/KI and ascorbic acid methods, respectively. The concentration of permanganate was determined by a UV−vis spectrophotometer. The residual Mn and total nitrogen and dissolved organic carbon concentrations were measured using an automatic inductively coupled plasma optical emission spectroscopy(ICP-OES; 730 series, Agilent, USA) and total organic carbon analyzer(TOC-5000, Shimadzu, Japan), respectively. RNT quantification was performed using an ultra-high performance liquid chromatography Vanquish system(Thermo Scientific, San Jose, USA). An Agilent gas chromatography((GC)/TOF MS 6890N GC system, Agilent Technologies, USA) system was used to determine the NDMA concentration. Other details are described in our previous study (Seid et al., 2022).
3. Results and discussion
3.1 NDMA occurrence and formation potential in wastewater
The concentrations of RNT and NDMA in the influent and effluent of the four WWTPs are shown in Fig. 1. RNT was found in all WWTP samples with concentrations ranging from 1.32–186.9 ng/L (Fig. 1a). The mean total RNT concentrations measured in this study were much lower than the 2.9–11,664 ng/L range measured for wastewater in the US (Batt et al., 2008) and UK (Kasprzyk-Hordern et al., 2009), while the values were consistent with the RNT concentrations from wastewater influents (188 ng/L) in Italy (Castiglioni et al., 2006), and higher than wastewater effluents across other parts of Europe (< 43.6 ng/L) (Loos et al., 2013). NDMA concentrations in the WWTP samples ranged from less than limit of detection (2.5 ng/L) to 72.4 ng/L (Fig. 1b). The highest detected NDMA concentration was 72.4 ng/L in the influent and 28.3 ng/L in the effluent. These values were significantly lower than the NDMA levels observed in the wastewaters of the USA and China, as reported by other studies (Zeng et al., 2016;Wang et al., 2014). However, the values agreed with the NDMA levels found in Swiss and Korean secondary effluents and surface waters (Kim et al., 2013;Krauss et al., 2009).
The NDMA-FP was further evaluated by chloraminating the RNT and WWTP samples. As shown in Fig. 2, the molar yield of NDMA from RNT was as high as 29% (~36.2 ng NDMA per 125 ng RNT), which was consistent with the range found in the other studies (14–37%) (Chen et al., 2021;Hinneh et al., 2019), confirming the reliability of our protocol for NDMA-FP. The NDMA in the RNT-spiked WWTP samples was higher than that in DI water after 72 h of chloramination (109–227.8 ng/L).
NDMA concentrations in chloraminated raw WWTP influents and effluents, were approximately 212.6 and 68.5 ng/L, respectively. The levels of NDMA-FPs were within the range reported by previous study (Yoon et al., 2011). In this study, we found that RNT accounted for 7% and 21.8% of the total NDMA-FP in WWTP influents and effluents, respectively, while 78.2 to 93% of the NDMA-FP content was not quantified. This means that the NDMA-FP in WWTPs can be attributed to additional contributions from specific or non-specific precursors (Krauss et al., 2009;Wang et al., 2014;Yoon et al., 2011). Given the diversity of precursors present in WWTP effluents, tracing and characterization of unknown NDMA precursors requires further research for a better understanding of the array of NDMA formation.
3.2 Importance of chloramine methods
The NDMA-FP levels varied greatly depending on the chloramine methodology employed (Chen et al., 2021;Li et al., 2018;Orak et al., 2019). To gain further insights, NDMA formation from RNT was investigated under various chloramination regimes, which showed significant differences, as depicted in Fig. 3. Chlorination alone produced an insignificant quantity of NDMA, while relatively low NDMA concentrations (~9.1–199 ng/L) were obtained when preformed NH2Cl was added after chlorination. In contrast, excess dichloramine, breakpoint chlorination procedures, or stepwise chloramination increased the NDMA-FPs by 7.5–25.7%.
The addition of preformed NH2Cl to chloraminated RNT, i.e., rechloramination, also significantly promoted the NDMA concentrations, from 36.2 to 41.8 ng/L and from 227.8 to 270 ng/L in the DI water and wastewater effluents, respectively. The formation of localized dichloramine, peroxynitrite, or chlorinated species could explain these results (Chen et al., 2021;Orak et al., 2019;Selbes et al., 2018). Previous study by Schreiber and Mitch (2007) also suggested that reactive radicals generated from stepwise chloramination enhanced the overall NDMA formation. These results suggest that chloramination methods influenced NDMA formation from precursors such as RNT. Particularly, in water samples containing high concentrations of ammonia, NDMA formation can be mitigated by proper selection of disinfection methods and pretreatment, such as prechlorination. A more detailed discussion of this issue is presented in the following sections.
3.3 NDMA formation potential reduction during pretreatment
3.3.1 Effects of prechlorination on NDMA-FP
RNT degradation in (photo)chlorination was investigated, and the results are shown in Fig. 4a. Approximately 56% of RNT was removed within 30 min of chlorination (with 3 mg Cl2/L), which is consistent with previous study on RNT chlorination (Jeon et al., 2016). In contrast, UV photolysis, with or without free chlorine, was effective in removing RNT (>98%). Latch et al. (2003) reported that pharmaceutics (e.g., RNT) containing electron-rich heterocyclic moieties are more susceptible to photolysis. In UV/Cl2, RNT’s pseudo first-kinetics rate constant (kobs) was 12.4 h-1 (Fig. 4b), which is 42.4% faster than photolysis alone (8.1 h-1).
After pretreatment, the chlorinated samples were exposed to chloramination to assess the variations in NDMA-FP. As shown in Fig. 4c, NDMA-FP decreased by only 25% after 30 min of prechlorination (with free chlorine), while UV photolysis and UV/Cl2 eliminated 38 to 70% of NDMA-FP. However, UV/Cl2 was more effective for WWTP effluent water samples (Fig. 4d), displaying a higher NDMA-FP reduction (52.4%) than prechlorination and photolysis alone (12% to 22.3%). The reduction in NDMA-FP during prechlorination is associated with the removal or deactivation of core moieties (e.g., amine and furan rings) of RNT (Jeon et al., 2016;Wang et al., 2015). Photolysis of chlorine can produce various radicals such as hydroxyl radical (•OH) and reactive chlorine radicals (Eqs. 1–5). These reactive radicals have been shown to react with amine and furan moieties via hydroxylation, demethylation, chlorine addition, and ring opening (Guo et al., 2018;Jasper et al., 2016;Jeon et al., 2016), which could influence the stability of leaving group and NDMA-FP during subsequent chloramination.
3.3.2 Effects of permanganate on NDMA-FP
The effects of permanganate treatment on RNT removal and NDMA-FP were assessed. As shown in Fig. 5a, more than 50% of RNT was removed after oxidation with 2 mg/L MnO4-, which is consistent with previous study (Wang et al., 2015). Remarkably, RNT was completely removed within 20 min by (Fig. 5a and b) using UV/MnO4- (kobs = 10.4 h-1). The increase in RNT oxidation can be attributed to the activation of MnO4- by UV irradiation (Guo et al., 2018).
However, after exposure to permanganate, the NDMA-FP in post-chloramination significantly increased (Fig. 5c). Increasing the contact time by up to 60 min led to comparatively marginal decreases in the final NDMA-FP, as reported by Wang et al. (2015). By UV photolysis alone, the NDMA-FP of RNT was reduced by approximately 50%, which is in agreement with a previous finding (Farré et al., 2012). Analogous irradiation with MnO4- further reduced the NDMA-FP by up to 79.0%. Permanganate pretreatment and subsequent chloramination were also performed in WWTP effluent (Fig. 5d). For instance, permanganate exposure increased the NDMA-FP levels, whereas UV/MnO4- pretreatment effectively reduced NDMA-FPs (i.e., up to 40.3%) in WWTP effluent.
The poor efficacy of NDMA-FP reduction may be due to the lower reaction selectivity of Mn(IV) towards furan and tertiary amine than the thioether moiety. Previous study reported that oxygen transfer products (i.e., RNT sulfone or sulfoxide) were mainly formed from permanganate oxidation, while amine-N and furan rings (the core structures for NDMA-FP) were stable (Wang et al., 2015).
As given by Eqs. 6–15, the photodecomposition of permanganate can produce reactive •OH and Mn species (Ataee et al., 2020;Guo et al., 2018;Wei et al., 2021). A trapped charge transfers between reactive Mn species, on the other hand, facilitates the generation of Mn (V) peroxide (MnO2(η2-O2)-) and photoinduced reactive oxygen species (e.g., superoxide ions (O2∙-)). RNT is easily degraded by •OH and O2•– mediated oxidation (Radjenovic et al., 2010). Direct hole oxidation and permanganate species can also enhance the degradation of sulphur-containing compounds (Guo et al., 2018;Konstantinou et al., 2001;Yang et al., 2018). Thus, reactive oxygen species, including Mn(V) peroxide, can quickly attack the photo-labile moieties of RNT, leading to N-dealkylated products. Oxygen transfer to the amine moiety and N-dealkylation was reported to inhibit nucleophilic initiation and reduce NDMA-FP during subsequent chloramination (Wang et al., 2015).
4. Conclusions
This study investigated the presence of NDMA and the effects of pretreatments on NDMA formation potential during RNT chloramination in wastewater. The average concertation of RNT and NDMA in the wastewater influent and effluent samples was 94.1 and 37.5 ng/L, respectively, while the average NDMA-FP in the influents and effluents of the WWPTs was 132 ng/L. The observed NDMA-FP variations were primarily resulted from the type of pretreatments. It has been found that the NDMA-FP of RNT can be effectively reduced by UV/Cl2 and UV/MnO4- with 3 mg Cl2/L and 2 mg MnO4-/L, respectively. Furthermore, the results revealed that both pretreatment processes were more efficient than the sole prechlorination, UV photolysis, and permanganate oxidation under similar reaction conditions. These results collectively suggest that UV/Cl2 and UV/MnO4- processes could be viable options for the reduction of nitrosamines during water/wastewater treatment.