Assessment of Degradation Byproducts and NDMA Formation

Article pubs.acs.org/est

Assessment of Degradation Byproducts and NDMA Formation Potential during UV and UV/H2O2 Treatment of Doxylamine in the Presence of Monochloramine Maria José Farré,*,† Jelena Radjenovic,*,† and Wolfgang Gernjak The University of Queensland, Advanced Water Management Centre, Queensland 4072, Australia S Supporting Information *

ABSTRACT: UV−C radiation is the U.S. EPA recommended technology to remove N-nitrosodimethylamine (NDMA) during drinking and recycled water production. Frequently, H2O2 is added to the treatment to remove other recalcitrant compounds and to prevent NDMA reformation. However, the transformation of NDMA precursors during the UV and UV/H2O2 process and the consequences for NDMA formation potential are currently not well understood, in particular in the presence of monochloramine. In this study, doxylamine has been chosen as a model compound to elucidate its degradation byproducts in the UV and UV/H2O2 process and correlate those with changes to the NDMA formation potential. This study shows that during UV treatment in the presence and absence of monochloramine, NDMA formation potential can be halved. However, an increase of more than 30% was observed when hydrogen peroxide was added. Ultrafast liquid chromatography coupled to quadrupole-linear ion trap mass spectrometer was used for screening and structural elucidation of degradation byproducts identifying 21 chemical structures from the original parent compound. This work shows that further oxidation of NDMA precursors does not necessarily lead to a decrease in NDMA formation potential. Degradation byproducts with increased electron density in the vicinity of the dimethylamino moiety, for example induced by hydroxylation, may have a higher yield of nucleophilic substitution and subsequent NDMA formation compared to the parent compound during chloramination. This work demonstrates the need to consider the formation of oxidation byproducts and associated implications for the control and management of NDMA formation in downstream processes and distribution when integrating oxidative treatments into a treatment train generating either drinking water or recycled water for potable reuse.

■

INTRODUCTION Received: Revised: Accepted: Published:

The U.S. Environmental Protection Agency (U.S. EPA) classifies N-nitrosodimethylamine (NDMA) as “B2 carcinogen - reasonably anticipated to be a human carcinogen”,1 and a 10−6 © 2012 American Chemical Society

12904

July 17, 2012 November 6, 2012 November 7, 2012 November 7, 2012 dx.doi.org/10.1021/es302883n | Environ. Sci. Technol. 2012, 46, 12904−12912

Environmental Science & Technology

Article

cancer risk level in drinking water at 0.7 ng/L has been determined.2 The U.S. Office of Environmental Health Hazard Assessment (OEHHA) has issued a public health goal of 3 ng/ L for NDMA3 and this compound is also included in the third Contaminant Candidate List for further evaluation in the U.S. EPA’s regulatory determination process. NDMA is included in the Australian Drinking Water Guidelines and the World Health Organization Guidelines for drinking water quality at 100 ng/L.4,5 It is accepted that the main mechanism of NDMA formation involves a nucleophilic substitution reaction between an organic nitrogen compound containing the N,N-dimethylamino group and chloramines.6 Significant efforts have been made to investigate the chemistry of NDMA as well as to understand the key parameters that enhance NDMA formation. Several critical parameters influencing the NDMA formation in water treatments have already been identified, including operational conditions for chloramination7,8 and contact time between water and disinfectant.9 NDMA exhibits two absorption bands, with one maximum at 228 nm (molar absorption coefficient ε228 nm = 7378 M−1 cm−1) and one at 332 nm (ε332 nm = 109 M−1 cm−1).10 As a consequence, the most common method to treat NDMA in water is UV photolysis by low-pressure UV lamps emitting mainly monochromatic light at 254 nm, medium-pressure lamps emitting polychromatic light, and pulsed UV systems.1 Hydrogen peroxide is commonly added to an UV reactor to remove other recalcitrant compounds and to prevent NDMA reformation.11 Upon addition of hydrogen peroxide, its direct photolysis generates reactive hydroxyl radicals (•OH) and a so-called advanced oxidation process (AOP) takes place. Hydroxyl radicals react electrophilically by either abstracting hydrogen atoms from C−H, N−H, or O−H bonds, or by addition to double bonds and aromatic rings leading to hydroxylation.12 Whereas the degradation of NDMA itself in UV and UV/H2O2 processes has been well studied, information about the fate of NDMA precursors, their degradation byproducts, and in particular potential changes to their reactivity with chloramines and NDMA formation potential is still very limited. Although several studies relate the presence of strong oxidants with an increase in NDMA formation potential,13,14 they focus on the relationship between the concentration of parent molecules or sum parameters such as total organic carbon removal and NDMA formation potential and do not investigate the molecular changes of the precursors that caused its increase. Recently, Shen and Andrews15 reported the NDMA formation potential of 20 pharmaceuticals and, by comparing their yield of NDMA, suggested that the presence of electron donating groups increases the electron density in the vicinity of nitrogen atom, which can favor the reaction with chloramines leading to NDMA formation. In analogy, it could be anticipated that UV photolysis and AOPs can generate byproducts with higher electron density in comparison to their parent compound as a result of oxidation and incorporation of hydroxyl groups, which may in turn lead to an increased NDMA formation potential compared to the parent compound. In contrast, Lee and coauthors16 observed a decrease in NDMA formation potential when several NDMA suspected precursors (organic compounds with dimethylamine or trimethylamine groups) were oxidized with relatively high doses of oxidants (ozone, chlorine dioxide, ferrate, and •OH). Consequently, due to scarcity of data concerning individual processes, a generalization is not possible.17

In a recent study we investigated the effect of UV and UV/ H2O2 in the presence of chloramines on the NDMA formation potential of tramadol,18 which due to its complex structure and resulting steric hindrance has a very low NDMA formation potential (0.4% molar yield) as reported by Shen and Andrews.15 In this previous study, it was found that in all oxidation processes investigated, the majority of the identified byproducts were likely to have a higher NDMA formation potential than the parent compound due to a reduced steric hindrance in the vicinity of the dimethylamine group. The aim of the present study is to investigate the effect of UV and UV/ H2O2 on the NDMA formation potential of a precursor with a higher initial NDMA yield to understand the oxidation effects on a molecule that does not have a sterically hindered tertiary amine moiety. Doxylamine (DOX) (CAS 469-21-6) (see Figure 3 for chemical structure) with a 7.5 ± 2.0% NDMA formation yield15,19 and containing a dimethylamino group on an aliphatic side chain without strong electron donating or electron withdrawing substituents in its vicinity has been selected to this aim. The reactions and behavior of DOX in oxidation processes could be anticipated to be representative of many other reported NDMA precursors. Concentration of DOX in secondary effluents has been previously reported around 300 ng/L,20 but the NDMA formation potential of oxidation byproducts of DOX in water/wastewater treatment has not been yet studied. Ultrafast liquid chromatography (UFLC) coupled to quadrupole-linear ion trap mass spectrometer (QqLIT-MS) has been employed in this work to determine the structures of major oxidation byproducts of DOX and to relate them to the changes in NDMA formation potential observed during the oxidative treatments. This work intends to generate new and significant insights concerning the fate of NDMA precursors in oxidative processes, which will assist the water industry to better understand the formation of NDMA during water treatment.

■

MATERIALS AND METHODS Materials. NDMA (>99%, 5000 μg/mL in methanol) was obtained from Supelco. Ammonium chloride (TraceSELECT, 99.9% purity), sodium hydroxide (SigmaUltra, 98%, pellets) and sodium hypochlorite solution (reagent grade, available chlorine 4%) were used to generate preformed monochloramine (NH2Cl). NH2Cl was prepared freshly before each experiment because of its ability to autodecompose at high concentrations. Prior to the preparation of the NH2Cl solution, the free chlorine concentration in the hypochlorite stock solution was determined. Based on the free chlorine concentration in the hypochlorite solution, the volume of hypochlorite stock solution to be added was calculated to achieve a molar ratio of ammonia to free chlorine of 1.2:1. The respective volume of hypochlorite stock solution was added dropwise to the ammonium chloride solution at pH = 8. Chloramine speciation was spectrophotometrically tested before initiating the experiment21 to avoid adding dichloramine. Potassium dihydrogenphosphate (Fluka, puriss. p.a., 99.5%) and disodiumhydrogenphosphate (Fluka, puriss. p.a., 99.5%) were used to prepare pH buffer solutions. To quench the chloramines solution, sodium sulphite (Fluka, puriss. p.a., 98.0%) was employed. Hydrogen peroxide (30% w/w, Merck) was purchased from Sigma Aldrich. Doxylamine (DOX) succinate salt (>98%) was purchased from Sigma Aldrich and dissolved in Milli-Q water at 20 mg/L DOX for further 12905

dx.doi.org/10.1021/es302883n | Environ. Sci. Technol. 2012, 46, 12904−12912


Article

Figure 1. Mean values with range of duplicate experiments of NDMA formation potential of doxylamine (DOX) during UV, UV/NH2Cl, and UV/ NH2Cl/H2O2 treatment. [NH2Cl]0 = 20 mg/L, [H2O2]0 = 50 mg/L, [DOX]0 = 20 mg/L. For the NDMA formation potential (FP): contact time = 7days, pH = 7, T = 23±2 °C, NH2Cl dose = 140 mg/L.

of the NDMA analysis are summarized in the Supporting Information (Text S1) and can be found elsewhere.18 Identification of DOX Oxidation Byproducts. Liquid chromatography−mass spectrometry (LC-MS) analyses were performed using a Shimadzu Prominence UFLC system (Shimadzu, Japan) coupled with a 4000 QTRAP QqLIT-MS equipped with a Turbo Ion Spray source (Applied BiosystemsSciex, U.S.A.). Chromatographic separation was achieved with an Alltima C18 Column (250 × 4.6 mm, particle size 5 μm) operated at 40 °C, supplied by Alltech Associates Inc. (U.S.). While the analyses were performed in both negative electrospray ionization ((−)ESI)) and positive electrospray ionization ((+)ESI) mode, oxidation byproducts of DOX were detected only in the latter case. They were tentatively identified by isolating the protonated molecular ions, collision induced dissociation (CID) MS2 and MS3 experiments in (+)ESI mode and mass spectral comparison with the parent compound. Additional confirmation of identity was obtained by the retention time (tR) and isotope abundance and distribution that enabled confirmation of halogenated intermediates. A more detailed description of the chromatographic conditions and MS settings is summarized in SI Text S2.

experiments. The initial concentration of DOX was high in order to be able to identify the unknown oxidation byproducts formed in the full-scan mode of analysis. UVC and AOP Experiments. UV and AOP experiments were performed by circulating 1 L of DOX solution at a flow rate of 1.6 L/min through a sealed reactor past a UV lamp emitting light at 254 nm (Ultraviolet Technology of Australasia LC 20, 60 W electrical, arc length 36 cm, the photoreactor consists of two fluoropolymer tubes with 25 mm inner diameter and an irradiated length of 36 cm placed in parallel to the arc resulting in an irradiated volume of 0.35 L and a nonirradiated volume of 0.65 L). The reactor geometry suggests a relatively broad UV dose distribution for a single pass of fluid through the reactor. However, due to the high recirculation rate and the multiple passes of each differential volume through the reactor the UV dose distribution will be narrowed considerably and volume averaged UV dose can be determined by chemical actinometry. Uridine actinometry was performed with a 0.009 mM uridine solution in a 1 mM phosphate buffer. Samples were taken every 2 min during 20 min, absorbance at 262 nm was measured, first order decay of uridine was determined, and UV dose was calculated using equation two in Jin et al.22,23 Three series of experiments were done in duplicate each for the following treatments: (i) UV radiation, (ii) UV radiation in the presence of NH2Cl (20 mg/L), and (iii) UV radiation in the presence of hydrogen peroxide (50 mg/L) and NH2Cl (20 mg/L). The experiments were performed at high initial concentrations of DOX (20 mg/L) in order to elucidate the structures of DOX byproducts. A detailed description of the experiments and further analyses are included in the Supporting Information (SI) (Text S1). To avoid scavenging of hydroxyl radicals, no buffer solution was used and, as seen in SI Figure S2, the pH did not change more than one unit throughout the experiments. NDMA Formation Potential Test and NDMA Analysis. The NDMA formation potential test follows closely the procedure by Mitch et al.24 and has been employed elsewhere.25 NDMA was analyzed by reversed-phase highperformance liquid chromatography equipped with a photodiode array detector (HPLC-DAD consisting of Shimadzu LC20 AT Prominence LC, SIL-20A HT Prominence auto sampler and SPD-M20 A Prominence diode array detector).The details

■

RESULTS AND DISCUSSION DOX Degradation. The study of DOX degradation during the three different oxidative treatments (UV, UV/NH2Cl, and UV/NH2Cl/H2O2) showed that the presence of NH2Cl did not strongly affect the direct photolysis rate and >90% removal was measured after 2250 mJ/cm2 of UV dose in the presence and absence of NH2Cl (SI Figure S1). Complete removal of DOX was observed after 1750 mJ/cm2, when hydrogen peroxide was added to the system, which confirms a stronger oxidation capacity of the UV/H2O2 in comparison to UV photolysis, resulting from the enhanced generation of hydroxyl radicals.26 DOX degradation kinetics could be fitted to pseudo-first order in all cases (data not shown). Degradation rate constants were calculated as 9.6 × 10−2, 9.3 × 10−2, and 2.5 × 10−1 1/min for UV, UV/NH2Cl, and UV/NH2Cl/H2O2, respectively (R2 ≥ 0.95 for the fits). Photolytic NH2Cl decay was not affected by the presence of hydrogen peroxide and it was completely removed after 1250 mJ/cm2 UV dose in the presence and absence of H2O2, to generate nitrate, nitrous oxide, and 12906



Article

Figure 2. (A) (+)ESI-QqLIT-MS2 spectrum of P89, m/z 90.1, (B) (+)ESI-QqLIT-MS2 spectrum of P105, m/z 106.1, insert (+)ESI-QqLIT-MS3 spectrum: m/z 106.1→m/z 88.1, (C) (+)ESI-QqLIT-MS2 spectrum of P135, m/z 136.1, insert (+)ESI-QqLIT-MS3 spectrum: m/z 136.1→m/z 106.1, and (D) (+)ESI-QqLIT-MS2 spectrum of P135, m/z 136.1, insert: (+)ESI- QqLIT-MS3 spectrum: m/z 136.1→m/z 106.1. 12907



Article

independently of the type of oxidative treatment.18 However, in the case of DOX, the NDMA formation potential increase was only observed during the AOP treatment. Recently, Chen and coauthors14 observed a similar NDMA formation potential increase as a result of medium pressure UV/H2O2 treatment of diltiazem. However, this study did not identify any oxidation byproducts other than dimethylamine (DMA) to explain the observed increase in NDMA formation potential. Kim and Clevenger28 previously reported that a good linear correlation could be found between NH2Cl/DMA and NDMA formation potential. However, Chen and coauthors14 insisted on the fact that oxidation intermediates can also be possible NDMA precursors and they cannot be explained by the DMA measurement. Although we did not measure DMA in this study, its possible formation would not be able to fully explain the observed enhancement in NDMA formation potential, as one molecule of DOX can only generate one molecule of DMA. The reported molar NDMA formation potential yield of DMA is 0.5−0.76%.24,29 This value was confirmed in our experiments as 0.5 ± 0.1%, in comparison to 1.2 ± 0.2% for DOX. Conversion of DOX to DMA should therefore lower the NDMA formation potential of the solution. Structural Elucidation of Degradation Byproducts of DOX. SI Table S1 summarizes the proposed molecular structures, tR's and molecular weights (MWs) of DOX and the detected oxidation byproducts, while Figure S3 illustrates their extracted ion chromatograms (XICs). The recorded mass spectra and their detailed descriptions are given in Figures S4− S21. The fragmentation of the molecular ion of DOX, m/z 271, exhibited a loss of 45 Da (i.e., fragment ion m/z 226), and yielded N,N-dimethylethyleneamine ion m/z 72 and N,Ndimethylethanolamine ion m/z 90, originating from the scission of the ether bond (Figure S4). The observed mass spectral pattern of DOX was used as an indicator of the presence of an N,N-dimethylamino (−NH(CH3)2) group and an intact side chain of DOX, as explained further in the text. Along with the degradation of DOX (tR = 8.5 min), several new peaks appeared very early in the total ion chromatograms (TICs), and were identified as products P89 (tR = 2.5 min), P105 (tR = 2.4 min), P131 (tR = 3.0 min), and P135 (tR = 2.7, 3.2, 3.5, and 4.0 min). Their MS2 and MS3 spectra are depicted in Figure 2. For product P89 only the MS2 spectrum could be obtained (Figure 2 A). Neutral loss of water from the molecular ion m/z 90 resulted in a base peak ion m/z 72, also observed in the spectrum of DOX (Figure S4), and the trimethylamine radical fragment ion m/z 57. Therefore, P89 was determined to be a product of the oxidative cleavage of the ether bond in the side chain of DOX. In the MS2 spectrum of P105, molecular ion m/z 106 and base peak ion m/z 88 (Figure 2 B) were shifted upward for 16 Da relative to the ions m/z 90 and m/z 72 in the spectrum of P89. Thus, formation of P105 was traced back to further hydroxylation of P89. The occurrence of loss of a second water molecule (i.e., product ion m/z 70) in both MS2 and MS3 spectrum m/z 106→m/z 88 (insert in Figure 2B) implied that both −OH groups were located at the aliphatic chain. Product P135 was detected in four chromatographic peaks likely corresponding to two pairs of stereoisomers of P135, in accordance with the proposed structure containing two chiral carbon atoms. The MS2 and MS3 spectra of P135 (Figure 2C) exhibited product ions of even nominal mass. Considering that CID in the ESI source is unlikely to generate radical fragment ions,30 these fragment ions were assumed to contain a nitrogen atom of the dimethylamino group.

ammonium as byproducts as previously studied by Li and Blatchley.27 Ammonia, nitrite, and nitrate were also measured and the results are plotted in Figure S2. Nitrate was not detected during UV photolysis but it could be measured at a maximum concentration of 0.6 and 0.7 mg/L N-NO3− during UV/NH2Cl and UV/NH2Cl/H2O2, respectively. Ammonia was quickly released into the solution in the presence of NH2Cl at a concentration of 3.0 and 3.9 mg/L N-NH4+ during UV/NH2Cl and UV/NH2Cl/H2O2, respectively, but only at 0.3 mg/L NNH4+ during UV photolysis (Figure S2). Nitrite was not detected above the limit of detection (i.e., 0.02 mg/L) in any experiment. Hydrogen peroxide was only degraded up to 24% after 2250 mJ/cm2 (Figure S1), indicating that the concentration of this reagent could be optimized further, which may change the process’ kinetics. However, the aim of this work was to identify the degradation byproducts of DOX that could explain a change in its NDMA formation potential. Therefore, no further efforts were undertaken to optimize the reagent concentrations. Total organic carbon (TOC) was used to evaluate the mineralization of DOX by the three studied processes (Figure S2). Less than 10% of TOC was mineralized during UV and UV/NH2Cl treatment. Moreover, TOC decay reached a plateau at 250 mJ/cm2 indicating the stability of the generated byproducts under the applied oxidation conditions. The percentage of mineralization doubled when hydrogen peroxide was added to the system as TOC was reduced by 20% at the end of the experiment. Similarly, in the UV/NH2Cl/ H2O2 process most of the degradation byproducts formed were degraded again during the course of the experiment, whereas many of these byproducts were quite stable in the UV and UV/ NH2Cl processes. Detailed information on the degradation profiles of the byproducts can be found in Figures S22−S24. NDMA Formation Potential. Samples were taken at increasing UV doses and subjected to NDMA formation potential tests in all three oxidation processes studied. Figure 1 shows the NDMA formation potential of the initial DOX solution and after different UV doses in the experiments conducted. The initial NDMA molar conversion of DOX was 1.2 ± 0.2%. This value is lower than the one previously reported by Shen and Andrews15 (i.e., 7.5 ± 2.0%), but still in the same order of magnitude considering that the two studies did not follow exactly the same protocol for the NDMA formation potential test. NDMA formation potential was approximately halved during UV photolysis in the presence and absence of NH2Cl (i.e., 47.2% and 54.1% NDMA formation potential removal after 2250 mJ/cm2 UV and UV/ NH2Cl, respectively). As described earlier, more than 90% of DOX was degraded after this UV dose. Consequently, DOX was not the only source for NDMA formation, and other degradation byproducts acted as NDMA precursors, but the entire sum of UV and UV/NH2Cl generated byproducts always had a lower NDMA formation potential than the blank with the parent compound for the UV doses applied here. In fact, as explained further in the text, very stable byproducts were generated after 250 mJ/cm2 of DOX treatment (Figures S22− S24), which is in agreement with the stable NDMA formation potential measured after this UV dose in the presence and absence of NH2Cl. Contrarily, NDMA formation potential increased during UV/NH2Cl/H2O2 oxidation of DOX, which indicated the generation of byproducts that are more amenable to reaction with chloramines to generate NDMA. In a previous paper investigating the effect of UV and UV/H2O2 on the NDMA FP of tramadol an increase was observed in all cases 12908



Article

Figure 3. Scheme of the proposed oxidation pathway of doxylamine (DOX). *Related to the NDMA formation potential increase observed during UV/NH2Cl/H2O2. **Detected in UV/NH2Cl and UV/NH2Cl/H2O2 process.

expulsion of CO, typical of phenol moieties,32 and phenol product ion m/z 93. In accordance with the oxidation and opening of a pyridine ring, spectra of products P235, P237, P275, P293, and P291 (Figures S13−S17, respectively) lacked the product ions characteristic for a pyridine moiety (i.e., signal at m/z 78, loss of HCN). However, fragment ions m/z 72 and m/z 90 implied a preserved backbone of DOX, while benzene (m/z 77) and/or toluene (m/z 91) product ions evidenced the presence of an aromatic ring. The mass spectral pattern of products P327 (Figure S18) and P325 (Figure S19), with intensity ratios of molecular ions m/z 328:330 and m/z 326:328 of 3:1, evidenced the presence of one chlorine atom. Further scrutiny of their mass traces revealed a shift of +34 Da of several aromatic fragment ions, which was the basis for identifying P327 and P325 as monochlorinated derivatives of P293 and P291, respectively. Similarly, product P304 was determined to be monochlorinated DOX (Figure S20). Finally, relative abundances of the molecular ions m/z 304, m/z 306, and m/z 308, and mass traces in the spectra of P303 (Figure S21) shifted for +68 Da relative to the spectra of P235 (e.g., dichlorinated benzene ion m/z 145) demonstrated the substitution of the aromatic ring in P235 with two chlorine atoms. Relation between DOX Degradation and NDMA Formation Potential Variation. In total, 21 oxidation byproducts were identified in UV, UV/NH2Cl, and UV/ NH2Cl/H2O2 oxidation, which were interpreted to support a tentative degradation pathway of DOX illustrated in Figure 3 and to understand the variations in NDMA formation potential observed during different treatments (Figure 1). The oxidation byproducts are divided into three categories to facilitate the discussion: (i) formation of byproducts without the dimethylamine group, (ii) formation of byproducts containing the dimethylamine group, and (iii) formation of products P135, P131, P105, and P89 that were the only products found after 2250 mJ/cm2 during the UV/NH2Cl/H2O2 process. Figures

Contrarily to the spectrum of DOX, fragmentation of the molecular ion m/z 136 did not result in the cleavage of the entire −NH(CH3)2 group. Plausible structures of fragment ions m/z 118, m/z 108, and m/z 106 were derived from the expulsion of water, cleavage of two −CH2 groups, and scission of the ether bond, respectively, while signal at m/z 94 possibly corresponded to the C2H8NO3+ ion. Sequential fragmentation m/z 136→m/z 106 (insert in Figure 2C) resulted in two successive losses of a −CH2 group and fragment ions m/z 92 and m/z 78. In the MS2 spectrum of product P131 (Figure 2 D), the presence of diagnostic ions m/z 72 and m/z 90 suggested an intact side chain of DOX. On the other hand, molecular ion m/z 132 underwent a neutral loss of 28 Da, which coincides with the cleavage of −C2H4 or CO group. Considering possible oxidative transformations of DOX, it was assumed that P131 contained a terminal CO group. Isolation of the base peak ion m/z 87 in the MS3 experiments (insert in Figure 2 D) gave rise to another prominent loss of 28 Da and fragment ion m/z 59. Absence of signals at m/z 72 and m/z 90, and loss of 45 Da in the spectra of products P95, P169, P155, P181, P199, and P197 (Figures S5−S10, respectively) implied that these byproducts were formed by the oxidative cleavage of the side chain in DOX. On the other hand, the spectra of the abovementioned products contained fragment ions m/z 78 and m/z 96, corresponding to a pyridine and hydroxylated pyridine ion, respectively. Furthermore, product ion m/z 110 identified as a hydroxylated methyl pyridine (picoline) ion C6H8NO+ was noted for products P169 and P199. This ion, as well as lack of a neutral loss of 18 Da characteristic of aliphatic −OH groups, assisted in elucidating the structure of a hydroxylated DOX, product P286, with an −OH substituent located at the pyridine ring (Figure S11). Another important fragmentation indicative of the pyridine ring was a characteristic loss of HCN (27 Da), also observed for DOX.31 Hydroxylation of a benzene moiety in product P209 (Figure S12) was diagnosed based on the 12909



Article

Figure 4. Peak areas of degradation byproducts formed during (●) UV, (○) UV/NH2Cl, and (▼) UV/NH2Cl/H2O2, normalized to the initial value of the peak area of DOX (t = 0) presented vs UV dose. P89 solid line, P135 dashed line, P105 dotted line. Inset corresponds to P131. Axis units of the inset are the same as for the main graph.

determine the ammonia released from the DOX oxidation in the presence of NH2Cl. A similar mechanism has been described in photoinduced Ndemethylation, through generation of singlet- and/or tripletexcited state of a molecule and addition of a superoxide radical ion (O2●−).36,37 In this reaction, H2O2 and an iminium ion are produced, which is further hydrolyzed to N-demethylated amine and formaldehyde. Furthermore, in all cases hydroxylation on the pyridine ring occurred, likely by the attack of •OH. Considering the acid dissociation constant (pKa) of DOX (pKa = 5.8 and 9.3, American Hospital Formulary Service (2000) AHFS Drug Information 2000, Bethesda, MD, American Society of Health-System Pharmacists), both dimethylamino and nitrogen atom at pyridine ring will be protonated in the working pH range of the experiments (Figure S1). It should be noted here that the reaction of pyridine with •OH and other free radicals is expected to be faster in the acidic environment.38 Addition of •OH at pyridine ring was reported during photolysis of the herbicide quinclorac.39 Free radicals substitution and addition reactions occur more readily in pyridine than in benzene rings,38 which can explain the intact aromatic benzene moiety in the aforementioned products. Similar degradation profiles were also obtained for oxidation byproducts P286, P291, P293, P275, P235, P237, and P209 during both UV and UV/NH2Cl processes (Figure S23). The byproducts plotted in Figure S23 still contain the initial benzyl group (that may have been hydroxylated) and the dimethylamino moiety, and were formed via oxidation of the pyridine group. Benzi et al.40 observed the opening of the pyridine ring in photolysis of nicosulfuron. Also, as mentioned earlier, photolytic and photocatalytic degradation of quinclorac involved initial hydroxylation of the pyridine moiety of the molecule, followed by the ring-opening.39 Byproducts containing the dimethylamino group, namely P286, P291, P293, P275, P235, P237, and P209, were also found during the first stages of UV/NH2Cl/H2O2 oxidation, where the NDMA formation potential measured was also lower than that of the parent compound (Figure 1). P275, P291, and P293 are intermediates

S22−S24 illustrate the qualitative profiles of the identified DOX degradation byproducts. While trying to be comprehensive, this study cannot exclude other byproducts that were potentially generated and not identified by this analytical method. For example, very small and polar compounds such as DMA may not be retained by the chromatographic column. Nevertheless, the identification of P89 and P105, fairly small and polar compounds, in the same method as the parent compound shows that the range of measured compounds is quite ample. Considering that potential NDMA precursors will include a dimethylamino group in their structure, they should be amenable to protonation in the ESI interface and thus analysis in the (+)ESI with the employed analytical procedure. Similar degradation profiles were obtained in UV and UV/ NH2Cl oxidation for the byproducts P181, P197, P199, and P169, P155, and P95 characterized by the loss of the dimethylamino group (Figure S22) (see Figure 3 for chemical structures). These structures appeared very fast during both treatments and remained stable until the end of the experiment. However, they were completely oxidized after 1750 mJ/cm2 UV dose, when hydrogen peroxide was added to the system. The cleavage of the dimethylamino group by •OH has been previously described by an initial attack of the •OH on the αCH bond, abstraction of hydrogen, and further addition of •OH, leading to a cleavage of the C−N bond and release of formaldehyde. The presence of •OH in absence of H2O2 could be explained by energy transfers from singlet or triplet excited states of the parent compound and/or its byproducts as oxygen was present in solution during the experiment.33 After the initial attack of the •OH on the α-CH bond, the second methyl group undergoes the same mechanism, and finally ammonia is released.34,35 Around 0.3 mg/L of N-NH4+ could be measured at the end of the UV process which corresponds to 29% of nitrogen recovery assuming this degradation pathway. In the presence of NH2Cl, the concentration of N-NH4+ was 3 mg/L at the end of the treatment. Ammonia is a common byproduct of NH2Cl photolysis,27 and hence it was not possible to 12910



Article

related to the increase of NDMA formation potential observed. However, this study cannot exclude the presence of other nonidentified byproducts that may have been formed in the process but have not been identified. In addition, further studies with analytical reference standards of the newly identified byproducts would be required in order to determine their individual NDMA formation potentials. Finally, it needs to be noted that in order to elucidate the structures of DOX byproducts the experiments had to be performed at high initial concentration (i.e., 20 mg/L). It is likely that a lower initial concentration of DOX, as well as optimized reagent dosage and the presence of other organic and inorganic compounds in real waters will modify the degradation pathway and kinetics presented in this study. However, this work demonstrates the need to monitor NDMA formation potential when integrating oxidative treatments into a treatment train in order to consider the formation of oxidation byproducts and associated implications during the control of NDMA.

after the ring-opening of the pyridine moiety occurred and are abundantly found in UV and UV/NH2Cl but not in the UV/ NH2Cl/H2O2 process, likely to rapid further oxidation in the latter. P286 and P237 seem to react slower than the aforementioned products and are therefore found abundantly also at a UV dose of 250 and 750 mJ/cm2, but further transformation is nearly complete after 1750 mJ/cm2. As previously mentioned, pyridines react faster than benzene rings with free radicals.38 Indeed, hydroxylation of benzene ring occurs later than the formation of any of the previously described hydroxylated pyridines. Three halogenated byproducts were found in the presence of NH2Cl (P303, P304, P325), while P327 was only found during UV/NH2Cl treatment (Figure S24). Contrarily to free radical reactions, the benzene ring will be more reactive than the pyridine ring in electrophilic substitution reactions,38 and in all instances the chlorine atom was incorporated into the phenyl ring, as this is the most electron rich moiety. This was also observed during the oxidation of tramadol in the presence of NH2Cl.18 Preferential chlorination of the phenyl ring over nitrogen atoms of the dimethylamino group or pyridine ring is also a consequence of protonation of both groups in the working pH range, which makes them less reactive with active chlorine species. None of these byproducts could be found after 1750 mJ/cm2 of UV dose suggesting that •OH oxidized them further. Among the 21 byproducts identified in this study, only 4 could be detected at the end of the UV/NH2Cl/H2O2 treatment and could therefore be potentially related to an increase in NDMA formation potential. The byproducts identified after 2250 mJ/cm2 of UV/NH2Cl/H2O2 were P89, P105, P131, and P135. Figure 4 shows the qualitative profiles of P89, P105, P131, and P135, determined by the peak areas in full-scan ESI(+) experiments, and normalized to the initial peak area of DOX (t = 0 min), plotted versus UV dose. Product P89 (dimethylethanolamine) was present during UV and UV/ NH2Cl at a higher concentration than during UV/NH2Cl/ H2O2. In particular, formation of P89 was significantly higher during UV treatment. As seen in Figure 3, P89 can be directly generated from DOX in a single step by scission of the ether bond or by subsequent degradation of the phenyl and pyridine groups of the parent DOX. NDMA formation potential of dimethylethanolamine has been previously reported as around 0.5%16,41 and therefore cannot be related to the increase of NDMA formation potential observed during UV/NH2Cl/ H2O2. P131 (2-(allyoxy)-N,N-dimethylethanamine) is likely an intermediate during the formation of P89, P135, and P105, and as such could be identified at similar concentration during each of the three treatments. Because a substantial increase of NDMA formation potential was only observed in the presence of H2O2, it can therefore be assumed that P131 is not a main responsible compound for the observed increase. However, P105 (1-(dimethylamino)ethane-1,2-diol) and P135 (1-(dimethylamino)-2-methoxyethane-1,2-diol) were present at significantly higher concentrations during UV/NH2Cl/H2O2 than during UV in the presence or absence of NH2Cl as a result of •OH attack at the aliphatic side chain. The formation of diols by the cleavage of aromatic ether bond was previously observed during the electrochemical oxidation of metoprolol42 and tramadol.18 Hydroxyl groups are electron donating and increase the electron density at the nitrogen atom, thus facilitating the cleavage of the dimethylamino moiety to generate NDMA during disinfection with chloramines.15 Hence they can be

■

ASSOCIATED CONTENT

S Supporting Information *

Additional text and figures as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +61 7 3346 3233 (M. J. F.); +61 7 3346 3234 (J. R.). Fax: +61 7 3365 4726 (M. J. F.); +61 7 3365 4726 (J. R.). Email: [email protected] (M. J. F.); j.radjenovic@awmc. uq.edu.au (J. R.). Author Contributions †

The two corresponding authors contributed equally to this work. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Veolia Water Australia and Seqwater for funding received through the “Water Recycling Research Program” agreement within The University of Queensland and the Urban Water Security Research Alliance for funding the “NDMA formation potential project”. J.R. was supported by the Queensland Government Early Career Smart Future Fund grant. We acknowledge the help of Miss Hollie King for her assistance during the experiments.

■

REFERENCES

(1) U.S. EPA. Emerging Contaminant- N-Nitrosodimethylamine (NDMA) Fact Sheet; 2008. (2) U.S. EPA. Record of Decision for the Western Ground Water Operable Unit OU-3, Aerojet Sacramento Site; 2001. (3) OEHHA. Public Health Goal for N-Nitrosodimethylamine in Drinking Water; California, 2006. (4) Australian Drinking Water Guidelines; National Health and Medical Research Council, Commonwealth of Australia: Canberra, 2011. (5) WHO. Guidelines for Drinking Water Quality, 4th ed.; Geneva, 2011. (6) Schreiber, I. M.; Mitch, W. A. Nitrosamine formation pathway revisited: The importance of chloramine speciation and dissolved oxygen. Environ. Sci. Technol. 2006, 40 (19), 6007−6014. (7) Mitch, W. A.; Oelker, G. L.; Hawley, E. L.; Deeb, R. A.; Sedlak, D. L. Minimization of NDMA formation during chlorine disinfection of 12911



Article

(29) Sacher, F.; Schmidt, C.; Lee, C.; von Gunten, U. Strategies for Minimizing Nitrosamine Formation During Disinfection; American Water Works Association Research Foundation, American Water Works Association, and IWA Publishing: Denver, CO, 2008. (30) Xu, G.; Huang, T.; Zhang, J.; Huang, J. K.; Carlson, T.; Miao, S. Investigation of collision-induced dissociations involving odd-electron ion formation under positive electrospray ionization conditions using accurate mass. Rapid Commun. Mass Spec. 2010, 24 (3), 321−327. (31) Gross., J. H. Proteomics In Mass Spectrometry: A Textbook; Springer-Verlag: Heidelberg, Germany, 2005. (32) McLafferty, F. W. Interpretation of Mass Spectra; Sausalito, CA, 2005. (33) Blough, N. V.; Zeep, R. G. Reactive oxygen species in natural waters. In Active Oxygen Species in Natural Waters; Foote, C. S., Valentine, J. S., Greenberg, A. E., Liebman, J. F., Eds.; Blackie A & P: Glasgow, U.K., 1995. (34) Radjenović, J.; Sirtori, C.; Petrović, M.; Barceló, D.; Malato, S. Characterization of intermediate products of solar photocatalytic degradation of ranitidine at pilot-scale. Chemosphere 2010, 79 (4), 368−376. (35) Klare, M.; Scheen, J.; Vogelsang, K.; Jacobs, H.; Broekaert, J. A. C. Degradation of short-chain alkyl- and alkanolamines by TiO2- and Pt/TiO2-assisted photocatalysis. Chemosphere 2000, 41 (3), 353−362. (36) Belvedere, A.; Boscá, F.; Cuquerella, M. C.; De Guidi, G.; Miranda, M. A. Photoinduced N-demethylation of rufloxacin and its methyl ester under aerobic conditions. Photochem. Photobiol. 2002, 76 (3), 252−258. (37) Cuquerella, M. C.; Boscá, F.; Miranda, M. A.; Belvedere, A.; Catalfo, A.; De Guidi, G. Photochemical properties of ofloxacin involved in oxidative DNA damage: A comparison with rufloxacin. Chem. Res. Toxicol. 2003, 16 (4), 562−570. (38) Comins, D.; O’Connor, S.; Al-Awar, R. S. Pyridines and their Benzo Derivatives: Reactivity at the Ring. In Comprehensive Heterocyclic Chemistry III; Katritzky, A. R., Ramsden, C. A., Scriven, E. F. V., Taylor, R. J. K., Eds.; Elsevier: Oxford, 2008. (39) Pareja, L.; Pérez-Parada, A.; Agüera, A.; Cesio, V.; Heinzen, H.; Fernández-Alba, A. R. Photolytic and photocatalytic degradation of quinclorac in ultrapure and paddy field water: Identification of transformation products and pathways. Chemosphere 2012, 87 (8), 838−44. (40) Benzi, M.; Robotti, E.; Gianotti, V. HPLC-DAD-MSn to investigate the photodegradation pathway of nicosulfuron in aqueous solution. Anal. Bioanal. Chem. 2011, 399 (4), 1705−1714. (41) Mitch, W. A.; Sedlak, D. L. Characterization and Fate of NNitrosodimethylamine Precursors in Municipal Wastewater Treatment Plants. Environ. Sci. Technol. 2004, 38 (5), 1445−1454. (42) Radjenovic, J.; Escher, B. I.; Rabaey, K. Electrochemical degradation of the β-blocker metoprolol by Ti/Ru0.7Ir0.3O2 and Ti/ SnO2-Sb electrodes. Water Res. 2011, 45 (10), 3205−3214.

municipal wastewater by application of pre-formed chloramines. Environ. Eng. Sci. 2005, 22 (6), 882−890. (8) Schreiber, I. M.; Mitch, W. A. Influence of the order of reagent addition on NDMA formation during chloramination. Environ. Sci. Technol. 2005, 39 (10), 3811−3818. (9) Farré, M. J.; Döderer, K.; Hearn, L.; Poussade, Y.; Keller, J.; Gernjak, W. Understanding the operational parameters affecting NDMA formation at Advanced Water Treatment Plants. J. Hazard. Mater. 2011, 185 (2−3), 1575−1581. (10) Stefan, M. I.; Bolton, J. R. UV direct photolysis of Nnitrosodimethylamine (NDMA): Kinetic and product study. Helv. Chim. Acta 2002, 85 (5), 1416−1426. (11) Chitra, S.; Paramasivan, K.; Cheralathan, M.; Sinha, P. K. Degradation of 1,4-dioxane using advanced oxidation processes. Environ. Sci. Pollut. Res. 2011, 1−8. (12) Haag, W. R.; David Yao, C. C. Rate constants for reaction of hydroxyl radicals with several drinking water contaminants. Environ. Sci. Technol. 1992, 26 (5), 1005−1013. (13) Andrezejewski, P.; Nawrocki, J. N-nitrosodimethylamine formation during treatment with strong oxidants of dimethylamine containing water. Water Sci. Technol. 2007, 56 (12), 125−131. (14) Chen, H. W.; Chen, C. Y.; Wang, G. S. Performance evaluation of the UV/H2O2 process on selected nitrogenous organic compounds: Reductions of organic contents vs. corresponding C-N-DBPs formations. Chemosphere 2011, 85, 591−597. (15) Shen, R.; Andrews, S. A. Demonstration of 20 pharmaceuticals and personal care products (PPCPs) as nitrosamine precursors during chloramine disinfection. Water Res. 2011, 45 (2), 944−952. (16) Lee, C.; Schmidt, C.; Yoon, J.; Von Gunten, U. Oxidation of Nnitrosodimethylamine (NDMA) precursors with ozone and chlorine dioxide: Kinetics and effect on NDMA formation potential. Environ. Sci. Technol. 2007, 41 (6), 2056−2063. (17) Nawrocki, J.; Andrzejewski, P. Nitrosamines and water. J. Hazard. Mater. 2011, 189 (1−2), 1−18. (18) Radjenovic, J.; Farré, M. J.; Gernjak, W. Effect of UV and UV/ H2O2 in the presence of chloramines on NDMA formation potential of tramadol. Environ. Sci. Technol. 2012, 46 (15), 8356−8364. (19) Shen, R.; Andrews, S. A. NDMA formation kinetics from three pharmaceuticals in four water matrices. Water Res. 2011, 45 (17), 5687−5694. (20) Farré, M. J.; Reungoat, J.; Argaud, F. X.; Rattier, M.; Keller, J.; Gernjak, W. Fate of N-nitrosodimethylamine, trihalomethane and haloacetic acid precursors in tertiary treatment including biofiltration. Water Res. 2011, 45 (17), 5695−5704. (21) Valentine, R. L.; Brandt, K. I.; Jafvert, C. T. A spectrophotometric study of the formation of an unidentified monochloramine decomposition product. Water Res. 1986, 20 (8), 1067−1074. (22) Jin, S.; Mofidi, A. A.; Linden, K. G. Polychromatic UV fluence measurement using chemical actinometry, biodosimetry, and mathematical techniques. J. Environ. Eng. 2006, 132 (8), 831−841. (23) von Sonntag, C.; Schuchmann, H. P. UV disinfection of drinking water and by-product formation - some basic considerations. J. Water Supply Res. Technol. AQUA 1992, 41, 67−74. (24) Mitch, W. A.; Gerecke, A. C.; Sedlak, D. L. N-Nitrosodimethylamine (NDMA) precursor analysis for chlorination of water and wastewater. Water Res. 2003, 37 (15), 3733−3741. (25) Farré, M. J.; Keller, J.; Holling, N.; Poussade, Y.; Gernjak, W. Occurrence of N-nitrosodimethylamine precursors in wastewater treatment plant effluent and their fate during ultrafiltration-reverse osmosis membrane treatment. Water Sci. Technol. 2011, 63 (4), 605− 612. (26) Legrini, O.; Oliveros, E.; Braun, A. M. Potochemical processes for water treatment. Chem. Rev. 1993, 93 (2), 671−698. (27) Li, J.; Blatchley, E. R., III UV photodegradation of inorganic chloramines. Environ. Sci. Technol. 2009, 43 (1), 60−65. (28) Kim, J.; Clevenger, T. E. Prediction of N-nitrosodimethylamine (NDMA) formation as a disinfection by-product. J. Hazard. Mater. 2007, 145 (1−2), 270−276. 12912


Assessment of Degradation Byproducts and NDMA Formation

Recommend Documents