lable at ScienceDirect

Water Research 87 (2015) 403e411

Contents lists avai

Water Research

journal homepage: www.elsevier .com/locate/watres

Effect of oxidation on amine-based pharmaceutical degradation andN-Nitrosodimethylamine formation

Xiaofeng Wang b, Hongwei Yang a, *, Beihai Zhou b, Xiaomao Wang a, Yuefeng Xie a, c

a School of Environment, Tsinghua University, Beijing 100084, Chinab Department of Environmental Engineering, School of Civil and Environmental Engineering, University of Science and Technology Beijing, Beijing 100083,Chinac Civil and Environmental Engineering Programs, Pennsylvania State University, Middletown, PA 17057, USA

a r t i c l e i n f o

Article history:Received 17 May 2015Received in revised form24 July 2015Accepted 27 July 2015Available online 29 July 2015

Keywords:NDMAPPCPsOzoneChlorineChlorine dioxidePotassium permanganate

* Corresponding author.E-mail address: yanghw@tsinghua.edu.cn (H. Yang

http://dx.doi.org/10.1016/j.watres.2015.07.0450043-1354/© 2015 Published by Elsevier Ltd.

a b s t r a c t

Four pharmaceuticals (ranitidine, nizatidine, doxylamine, and carbinoxamine) were selected as modelcompounds to assess the efficiency of four oxidants (ozone (O3), chlorine (Cl2), chlorine dioxide (ClO2)and potassium permanganate (KMnO4)) on the removal of amine-based pharmaceutical and personalcare products (PPCPs), as well as the reduction of their N-Nitrosodimethylamine formation potentials(NDMAFPs). The changes in PPCPs and their NDMAFPs during oxidation were quantified using variousoxidants and dosages. The relationship between oxidation product structures and NDMAFP changes wasalso analyzed. The results showed that oxidation with O3, Cl2 and ClO2 were effective in removing theselected PPCPs. However, only ozonation was effective in reducing their NDMAFPs. Ozonation at 6 mg/Lremoved approximately 90% PPCPs and 90% NDMAFPs for all PPCPs. In addition, the results indicated thatozonation products made little contribution to NDMAFPs. In contrast, some PPCP products had higherNDMAFPs than PPCPs after oxidation with Cl2, ClO2 and KMnO4. There were two possible reactionpathways that led to decrease in NDMAFPs after oxidation. One was oxygen transfer to nitrogen at thetertiary amine site and the other was N-dealkylation from the tertiary amine site.

© 2015 Published by Elsevier Ltd.

1. Introduction

N-Nitrosodimethylamine (NDMA), one of the disinfectionbyproducts (DBPs) is of emerging concern due to its extremely hightoxicity. The United States Environmental Protection Agency(USEPA) (1993) has issued a cleanup level of 0.7 ng/L for NDMAbased on an increased lifetime cancer risk of 10�6 in drinking water.The USEPA (2009) also placed six nitrosamines on the UnregulatedContaminant Monitoring Rule List 3. The Ontario Ministry of theEnvironment (MOE) (2003) in Canada established an interimmaximum acceptable concentration of 9 ng/L for NDMA in drinkingwater. The Canadian Office of Environmental Health HazardAssessment (OEHHA) (2006) also implemented a public health goalof 3 ng/L for NDMA. Although the typical NDMA concentration isabout 10 ng/L (Mitch et al., 2003), NDMA was detected in drinkingwater at concentrations as high as 100 ng/L (Boyd et al., 2011). Thehigher NDMA concentration usually occurred in the water that was

).

disinfected by chloramines (Zhao et al., 2008; Krasner et al., 2013).Chloramine is commonly used as a continuous disinfectant in pipenetwork system. It also served as an alternative disinfectant (Guayet al., 2005) to reduce the formation of trihalomethanes, haloaceticacids and other regulated DBPs.

Initial studies on NDMA precursors mainly focused on dime-thylamine (DMA) (Choi and Valentine, 2002; Mitch and Sedlak,2002; Schreiber and Mitch, 2005, 2006). Nevertheless, the contri-bution of DMA is not significant because DMA concentrations inwastewaters were generally lower than 200 ng/L and the NDMAmolecular conversion of DMAwas low (0.082e3%) (Lee et al., 2007).Some researchers subsequently focused their attention on otherprecursors such as quaternary amines and tertiary amines (Kemperet al., 2010; Mitch and Schreiber, 2008). Model quaternary aminesformed NDMA at yields of about 0.2%. This was likely via a pathwayinvolving the degradation of quaternary amines to secondaryamines. In contrast, some tertiary amines, which have aromaticgroup at the b-position to nitrogen, had higher NDMA formationpotential (NDMAFP) during chloramination (Roux et al., 2011).Many amine-based pharmaceuticals and personal care products(PPCPs) have the special tertiary amine structure. For example,

X. Wang et al. / Water Research 87 (2015) 403e411404

ranitidine, which is often used to prevent gastritis, can form NDMAat yields up to 90%. Andrews's group (2011) found that NDMA couldbe formed through reaction between all the 20 amine-based PPCPsand monochloramine, and that most of these PPCPs had higherNDMAFPs than DMA. Previous surveys also showed widespreadoccurrence of ng/Lemg/L of PPCPs in an aquatic environment (Liuand Wong, 2013; Loos and Carvalho, 2013; Ying et al., 2004;Snyder et al., 2003). Ranitidine was detected ranging from 70 to540 ng/L in primary effluents of wastewater treatment plants inSpain and at 10 ng/L in several surface waters (Radjenovic et al.,2009; Kolpin et al., 2002). Around 300 ng/L doxylamine was re-ported in a reclamation plant effluent (Farr�e et al., 2011). In addi-tion, many PPCPs were hardly removed by the conventionalcoagulation/sedimentation/filtration process in drinking watertreatment plant (Farr�e et al., 2011; Westerhoff et al., 2005).Therefore, PPCPs containing DMA groups are an important group ofNDMA precursors.

Over the last few decades, oxidation has been widely used fordrinking water treatment. The most commonly used oxidantsinclude ozone (O3), chlorine (Cl2), chlorine dioxide (ClO2), ferrateFe(VI) and potassium permanganate (KMnO4) (Lee and von Gunten,2010). Oxidation is very effective in controlling taste and odor aswell as iron/manganese problems (Lin et al., 2013; Ellis et al., 2000).It is also very effective in aiding coagulation and in removing DBPprecursors (Ma and Liu, 2002; Shah et al., 2012). Recently, re-searches were conducted to study PPCP degradation duringoxidation processes (Westerhoff et al., 2005; Lee and von Gunten,2010). The results showed that O3 was the optimal oxidant forPPCP degradation. Oxidation was also used to minimize NDMAformation by oxidizing the amine-based precursors upstream ofchloramination. Among all the oxidants, O3 and Cl2 were found tobe most effective in reducing NDMA formation. More than 50%NDMA precursors were deactivated by 140 mg/L � min for Cl2 and2 mg/L � min for O3 (dosage for Giardia control) in river water andwastewater (Shah et al., 2012). However, an increase in NDMAFPswas observed in some samples after Cl2 and ClO2 oxidation (Shahet al., 2012; Chen and Valentine, 2008). This could indicate thatsome oxidation products had high contributions to NDMAFPsbecause of uncompleted mineralization at common oxidantdosages.

Some oxidation products of tertiary amines by O3 (Lange et al.,2006; Khan et al., 2010; Mu~noz et al., 2001), Cl2 (Debordea andVon Gunten, 2008) and Fe(VI) (Zimmermann et al., 2012) havebeen reported. Degradation pathways of amine-based PPCPs by O3and Fe(VI) were discussed in particular. A primary ozonationpathway was oxygen transfer to the lone electron pair at nitrogen,yielding N-oxide (Mu~noz et al., 2001). A primary ferrate oxidationpathway was N-dealkylation at the tertiary amine site(Zimmermann et al., 2012). However, little was found on amine-based PPCP degradation by other traditional oxidants such as Cl2,ClO2 and KMnO4. In addition, a relationship between the degra-dation products and NDMAFP changes needs to be betterunderstood.

The first aim of this study is to evaluate the potential of fouroxidants (O3, Cl2, ClO2 and KMnO4) to remove PPCPs containing theDMA group, and the second is to better understand oxidationproducts and their NDMAFPs. Four common PPCPs were selected asmodel compounds because of their high NDMAFPs (89.9e94.2% forranitidine, 4.5e4.8% for nizatidine, 8.0e9.7% for doxylamine, and1.0e1.4% for carbinoxamine) (Shen and Andrews, 2011). As shownin Fig. 1, the structures of ranitidine and nizatidine were similar,except for the difference in the five-member rings. The structures ofdoxylamine and carbinoxamine were similar except for the twoaromatic rings. Selection of target compounds was also based ontheir prevalence in Chinese pharmaceutical markets and their

frequent presence in the water environment.

2. Material and methods

2.1. Chemicals

PPCPs (HPLC grade), NDMA (HPLC grade), formic acid (LC-MSgrade), sodium hypochlorite (NaClO) solution (reagent grade) andsodium chlorite (NaClO2) powder (80%) were purchased from Sig-maeAldrich in US. Methanol (HPLC grade) and acetonitrile (HPLCgrade) were obtained from Fisher Scientific in US. KMnO4,ammonium chloride, disodium hydrogen phosphate, sodiumdihydrogen phosphate, sodium chloride, sodium thiosulfate (all inAR grade) were purchased from Xilong Chemical Industry in China.Indigo carmine (AR grade), phosphoricacid (85%, AR grade) andsulfuric acid (98%, AR grade) were purchased from Sinopharm inChina. Ultrapure water (18 MU) was produced with a Milliporesystem.

2.2. Natural water systems

To simulate real treatment conditions, experiments were per-formed using source water from a drinking water treatment plantin Jinan, Shandong Province, China. The basic quality parameters ofthe natural water are summarized in the Supporting InformationTable S1. The natural water were filtered through a 0.45 mmmembrane filter (Millipore, HAWP04700, USA) within 48 h aftersampling and stored at 4 �C until use.

2.3. Oxidants preparation

A saturated O3 solution (40 mg/L) was prepared by bubblinggaseous O3 into ultrapure water in a 2 L airtight glass container inan ice bath. The gaseous O3 was produced from extra dry grade(minimum purity of 99.99%) oxygen using an O3 generator (New-land, NLO-A-10, China). Aqueous O3 concentration was determinedby the modified indigo colorimetric method with a UV-VIS spec-trophotometer (Meipuda, UV-1800, China). ClO2 stock solution(300 mg/L as ClO2) was prepared by adding sulfuric acid to NaClO2solution (Jin et al., 2006). The final yellow ClO2 solution was storedin an amber glass bottle and refrigerated. The concentration of ClO2was determined by the DPD colorimetry (Hach, PCII 58700-51,USA). Cl2 stock solution (1500mg/L as Cl2) was prepared by dilutinga commercial NaClO solution and the concentration determined bythe DPD colorimetry (Hach, PCII 58700-00, USA). KMnO4 stocksolution (2000 mg/L) was prepared by dissolving 1 g KMnO4 in500 mL of ultrapure water. Monochloramine (NH2Cl) solution wasprepared freshly (1600 mg/L as Cl2) by slow addition of 5% NaClOsolution to ammonium chloride solution (Cl/N molar ratio of 1/1.2)with continuous stirring. After being stirred for 30min, the solutionwas aged in the dark for 1 h and used within 1 d (Padhye et al.,2013). NH2Cl concentrations were determined prior to eachexperiment using the DPD colorimetry (Hach, PCII 58700-26, USA).

2.4. Experimental procedure

In order to better understand oxidation reaction pathways, eachPPCP was studied individually. The initial PPCP water solutioncontaining one PPCP, phosphate buffer and sodium chloride wasprepared in a 500 mL volumetric flask.

All oxidation experiments were carried out at 20 �C, pH 7.0 usinga 10 mM phosphate buffer and a certain ionic strength using a5 mM sodium chloride was maintained. The initial concentration ofthe PPCP in the reaction solutionwas 25 mM. Each oxidation dosagewas conducted at three concentration levels (e.g., [O3] ¼ 1, 3, 6 mg/

Fig. 1. Structures of the selected PPCPs.

X. Wang et al. / Water Research 87 (2015) 403e411 405

L). For each concentration level, the four oxidant dosages wereexpressed by equal electron accepting ability. The oxidant dosagesare showed in Table 1.

For the samples that were treated with O3, 200 mL initial PPCPsolutionwas poured into a reactor (250mL Erlenmeyer flask) whichhas a true volume of 290 mL. Then, X mL ultrapure water wasadded. Finally, Y mL (X þ Y ¼ 90 mL) saturated O3 solution wasinjected into the reactor (Kuang et al., 2013). For samples treatedwith the other three oxidants, oxidant stock solution was addeddirectly into 40 mL amber vials containing sample solution, usingoxidant dosages shown in Table 1. After 2 h oxidation reaction, acalculated sodium thiosulfate was added to quench the reaction.Then, a part of the sample was prepared for residual PPCP andtransformation product detection. The other part of the sample wasprepared for NDMAFP tests. The MnO2 suspension, generated as aresult of KMnO4, was removed by 0.45 mm syringe filter.

NDMAFP tests were conducted in 20 mL amber vials with Tefloncaps. 140 mg/L of monochloramine was added to ensure that therewas enough residual monochloramine to predict the ultimate for-mation potential (Lee et al., 2007). The chloramination was carriedout at 20 �C. Reactions were halted after 10 d by the addition ofexcess sodium thiosulfate powder (at approximately 1 g/L). Thenthe samples were prepared for NDMA detection.

Oxidation of natural water was performed with 62.5 mM O3

(3 mg/L), 93.5 mM Cl2 (6.7 mg/L), 75 mM ClO2 (5 mg/L) and 75 mMKMnO4 (11.8 mg/L) in 1 L amber vials with Teflon caps. After filterwith 0.45 mmmembrane, the pH of the natural waters was adjustedto 7.0 using H2SO4 prior to treatment. 25 nM PPCP was added to thenatural water to achieve enough NDMA precursors. Oxidation withdifferent oxidants and NDMA incubation tests were similar to thosein ultrapure water system. All experiments were conducted induplicates.

2.5. PPCPs and NDMA analysis

NDMA, PPCPs and oxidation products were analyzed usingliquid chromatography tandem mass spectrometry (LC-MS/MS)

Table 1The dosages of four oxidants.

Oxidants Dosage mM (mg/L)

O3 (as O3) 20.8 (1.0) 62.5 (3.0) 125.0 (6.0)NaClO (as Cl2) 31.2 (2.2) 93.7 (6.7) 187.5 (13.3)ClO2 (as ClO2) 25.0 (1.7) 75.0 (5.0) 150.0 (10.0)KMnO4 (as KMnO4) 25.0 (3.9) 75.0 (11.8) 150.0 (23.7)

(Agilent 1290LC coupled with 6460 MS/MS). The HPLC columnsused were Agilent Poroshell 120, SB-C18 (2.1 mm � 100 mm,2.7 mm) for NDMA and Agilent Zorbax SB-C18 (2.1 mm � 100 mm,1.8 mm) for PPCPs and oxidation products. The mobile phase wascomposed of acetonitrile (solvent A) and 0.1% formic acid in ul-trapure water (v/v, solvent B). The flow rate was 0.3 mL/min. Thesample injection volume was 10 mL. The multiple reaction moni-toring (MRM) MS detection mode was used to quantify NDMA andPPCPs. The MS detection modes for product identification wereMS2 scan and MS2 product ion scan. The detailed instrument pa-rameters for the compounds detected using LC-MS/MSwere shownin Table S2. PPCPs and NDMA in natural water were enriched bysolid phase extraction. The extraction programswere shown in TextS2 and Text S3.

3. Results and discussion

3.1. Degradation of PPCPs

Doxylamine and carbinoxamine seemed to be more sensitive toO3 than ranitidine and nizatidine (Fig. 2). When the O3 dosage wasas low as 1.0 mg/L, 50% of doxylamine and carbinoxamine wereremovedwhile only 25% of ranitidine and nizatidinewere removed.When the O3 dosage was increased to 3.0 mg/L, doxylamine andcarbinoxamine were completely removed, while only about 50% ofranitidine and nizatidine were removed. All PPCPs were removedby almost 90% when the O3 dosage was increased to 6.0 mg/L.

In contrast, oxidation of PPCPs with Cl2 showed an oppositedegradation pattern. Ranitidine and nizatidine were more sensitiveto Cl2 than carbinoxamine and doxylamine. When the Cl2 dosagewas as low as 2.2 mg/L, 50% of ranitidine and nizatidine wereremoved while only 30% of doxylamine and carbinoxamine wereremoved. When the Cl2 dosage was increased to 6.7 mg/L, 90% ofranitidine and nizatidine were removed while only 70% of doxyl-amine and carbinoxamine were removed. Almost 10% of carbi-noxamine and doxylamine remained at 13.3 mg/L Cl2.

Degradation curves of four PPCPs paralleled with each otheralong with increasing dosage of ClO2, which indicated that theylikely degraded through the same pathway. When the ClO2 dosagewas as low as 1.7 mg/L, they were removed about 30%. When theClO2 dosage was increased to 5 mg/L, they were completelyremoved, proving ClO2 to be an ideal oxidant for PPCP degradation.

Similar to oxidation with Cl2, ranitidine and nizatidine weremore sensitive to KMnO4 than carbinoxamine and doxylamine. At12 mg/L KMnO4, 90% of ranitidine and nizatidine were removed.

X. Wang et al. / Water Research 87 (2015) 403e411406

However, only 10% of carbinoxamine and doxylamine wereremoved. A possible reason was low reaction activity betweenKMnO4 and the two PPCPs. Similar experimental results have beenreported with low reactivity of KMnO4 (Rodríguez et al., 2007) foruracil (kapp ¼ 1.00 M�1 s�1) and cylindrospermopsin(kapp ¼ 0.3 M�1 s�1).

3.2. Effects of oxidation on NDMAFPs

ThemaximumNDMAFPs of the PPCPs without oxidation rangedfrom 0.1 to 17.5 mM, and the molar conversion rate was 0.5%e80%(carbinoxamine < nizatidine < doxylamine < ranitidine), which areconsistent with the value found in previous studies (Shen andAndrews, 2011). These oxidants could affect NDMAFPs of PPCPs tovarying degrees (Fig. 3). Only ozonation was effective in removingNDMAFPs of the PPCPs.

The NDMAFP of doxylamine was completely (>95%) removedwith 3 mg/L O3. When the O3 dosage was doubled from 3 mg/L to6 mg/L, approximately 90% NDMAFPs were removed for all PPCPs.Even though the product, DMA, was considered as a residual NDMAprecursor after complete ozonation (Lee et al., 2007), less than 5%DMA formed after ozonation of several tertiary amines andNDMAFP of DMA was 0.082% ~ 3%. The product DMA made littlecontribution to NDMAFP.

Cl2 had a weak ability in removing NDMAFPs. NDMAFP ofnizatidine increased with increasing Cl2 dosage. When the Cl2dosage was as low as 2.2 mg/L, NDMAFPs of doxylamine, carbi-noxamine and ranitidine were removed by 22%, 5% and 3% whileNDMAFP of nizatidine increased 21%. When the Cl2 dosage wasincreased to 6.7 mg/L, NDMAFPs of doxylamine, carbinoxamine andranitidine were removed by 68%, 24% and 18% while NDMAFP ofnizatidine increased 47%. Almost 98% of NDMAFP of nizatidineremainedwith 13.3mg/L Cl2. It was demonstrated that the products

Fig. 2. Degradation of PPCPs with increasing oxidant dosage (C ranitidine, B nizatidine

from nizatidine had a higher NDMAFP. Be different from ranitidine,a sulfur was contained in the five-membered heterocycle of niza-tidine. Sulfur was an active site that is easy to be attacked by Cl2.The products that chlorine transferred to the five-membered het-erocycle may made a larger contribution to NDMAFP than nizati-dine itself.

For ClO2 oxidation, the higher the NDMAFP of the PPCP, thehigher the NDMAFP removal was observed. NDMAFPs of ranitidine,nizatidine and doxylamine decreased while an increase wasobserved in samples containing carbinoxamine with increasingClO2 dosage. Shah et al. (2012) reported both reduced andenhanced NDMAFPs in natural water samples after oxidation withClO2. Product of DMA (Lee et al., 2007) was treated as the mainresidual NDMA precursor after complete ClO2 oxidation. NDMAFPincreased in the cabinoxamine solution was because NDMAFP ofcarbinoxamine was lower than DMA and more than 90% carbi-noxamine transformed into DMA after ClO2 oxidation.

The reaction between KMnO4 and precursors was too slow toreduce NDMAFP significantly at limited reaction time. Even thoughit had some advantages as it formed particles that can be easilyremoved by flocculation and did not formed disinfection byprod-ucts during KMnO4 oxidation (Andrzejewski and Nawrocki, 2009).Using KMnO4 for oxidation and disinfection would be difficultbecause of its long contact time requirement and low effectivenessin NDMAFP removal and pathogen inactivation.

In order to understand the relationships between the degrada-tion products and NDAMFPs, the measured NDMAs were comparedwith the calculated NDMAs from residual PPCPs under variousoxidant dosages. The latter was calculated through the followingFormula 1.

½NDMA�i ¼ ½PPCP�i � ½NDMA�0�½PPCP�0;

, ; doxylamine, D carbinoxamine, [PPCP]0 ¼ 25 mM, oxidation contact time ¼ 2 h).

X. Wang et al. / Water Research 87 (2015) 403e411 407

[NDMA]0 and [PPCP]0 were the measured results withoutoxidation,[PPCP]i was the measured result after i mg/L oxidant oxidation,[NDMA]i was the calculated result with residual PPCP after i mg/L oxidant oxidation.

The results demonstrated that ozonation products from allPPCPs made little contribution to NDMAFPs (Fig. 4). Products fromnizatidine and ranitidine made a large contribution to NDMAFPswhile products from doxylamine and carbinoxamine made a smallcontribution to NDMAFPs after Cl2 oxidation. In comparison toozonation products, products after ClO2 oxidation made highercontribution to NDMAFPs.

3.3. Relationship between oxidation product structure andNDMAFP

To further explore the mechanisms of PPCP degradation withoxidation and the relationships between oxidation products andNDMAFPs, the products were characterized by LC-MS/MS. The MS2scanmode andMS2 product ion scanmodewere used to obtain themass spectrums. The products were named as m/z Mþ 1. Similarproducts were detected from doxylamine and carbinoxamine, aswell as ranitidine and nizatidine (Fig. S1eS16). Only doxylamineand ranitidine were discussed in this chapter.

Oxygen transfer product m/z 287 [271 þ O] from doxylaminewas observed after ozonation (Fig. 5). Previous studies suggestedthat oxygen was transferred to the lone pair electrons on the ni-trogen thereby forming an ozonide ammonium zwitterion aftertertiary amines ozonation. The ozonide ammonium zwitterion laterlost the dioxygen yielding the N-oxide (Lange et al., 2006; Mu~nozet al., 2001). Oxygen transfer destroyed DMA structure in tertiaryamines leading to NDMAFP decrease after ozonation. Meanwhile,

Fig. 3. Changes of NDMAFPs with increasing oxidant dosage (C ranitidine, B nizatidine,;incubation time ¼ 10 d).

m/z 106 [90 þ O] was found in samples containing doxylamine. InFarr�e's study, (2012) doxylamine also degraded to m/z 106 [90 þ O]by UV treatment. Our results indicated that the production of m/z106 [90þO] came from the oxidation of hydroxyl radical existing inboth O3 and UV treatment.

N-dealkylation product m/z 257 [271-CH2] and NeC bondcleavage product m/z 200 [271-NC2H6eC2H3] were detected in thedoxylamine solution after Cl2 oxidation. Under weak oxidant con-ditions, it was difficult to open the two aromatic rings. Therefore,for doxylamine, N-dealkylation took place in the DMA functionalgroup where two branched chain methyl existed. Zimmermann'sgroup (2012) found that an N-centered radical cation was an in-termediate in the oxidation of amines by ferrate, and the N-centered radical cation intermediate could lead to N-dealkylation.N-dealkylation destroyed DMA structure that was necessary forNDMA formation. This might be the main reason for NDMAFPdecrease after Cl2 oxidation. In addition, the odd molecular mass ofthis product m/z 200 [271-NC2H6eC2H3] indicated an odd numberof nitrogen atoms (Ning, 2000), reflecting the loss of the DMAgroup. Product m/z 257 [271-CH2] was also detected in doxylaminesolution after ClO2 oxidation and after KMnO4 oxidation. The yieldsof product m/z 200 [271-NC2H6eC2H3] were higher than m/z 257[271-CH2] after ClO2 oxidation, whereas product m/z 200 [271-NC2H6eC2H3] was not detected due to low reactivity of KMnO4with doxylamine.

Ranitidine with long chain structure might be easier to bedestroyed than doxylamine with steady ring structures afterozonation. Oxygen transfer product m/z 331 [315 þ O] were notdetected in the ranitidine solution. However, oxygen transferproduct m/z 172 [156 þ O] from a fragment m/z 156 was detected.Roux et al. found that the rupture of CeS bond in ranitidinegenerated the product m/z 156 (Roux et al., 2012). MS2 product ionscan mode was used to characterize the structures of m/z 156 and

doxylamine, D carbinoxamine, [PPCP]0 ¼ 25 mM, oxidation contact time ¼ 2 h, NDMA

X. Wang et al. / Water Research 87 (2015) 403e411408

m/z 172 for confirming whether m/z 172 came from the fragmentm/z 156. Themass spectrums showed thatm/z 156 andm/z 172 haduniform product fragments m/z 111, m/z 83 and m/z 55. This resultreflected the loss of the DMAm/z 45 and DMA þ Om/z 61 [45 þ O]from m/z 156 and m/z 172. The result also demonstrated that ox-ygen was added to the lone pair electrons on nitrogen in DMAgroups. We could concluded that oxygen transfer reaction at thetertiary amine site caused NDMAFP decrease after ozonation.

Oxygen and chlorine transfer products m/z 365 [þOCleH], m/z399 [þOCl2eH2] were detected from ranitidine after Cl2 oxidation.The amount of chlorine in the products was confirmed according tothe special ratio of isotope abundance (m/z 35:37 ¼ 3:1) (Pretschet al., 2009). Unlike O3, Cl2 oxidation tended to add chlorine andoxygen to sulfur, primary and secondary amines, not tertiary amine(Debordea and Von Gunten, 2008). Generally, rate constants for thereaction Cl2 with sulfur are typically 1e2 orders of magnitudehigher than amines. The Cl2 attack on sulfur ether would form achlorosulfonium cation intermediate. After hydrolysis, sulfoxidecompounds were formed. For primary and secondary amines, Cl2reactivity constants are in the range 107e108 M�1 s�1; for tertiaryamines, lower rate constants of about 103e104 M�1 s�1. It indicatedthat DMA structure in ranitidine had not been destroyed after Cl2oxidation. Some of these products still had high NDMAFPs. It isknown that chlorine transfer occur in both Cl2 and NH2Cl oxidation.If sulfur, primary and secondary site were occupied firstly by

Fig. 4. Comparison between measured NDMAFPs (white bar) and calculated NDMAFPs (s

chlorine from Cl2, it would be easy that NH2Cl attack tertiary aminesite in ranitidine. Maybe this is a reason for chlorine transferproducts still had a high NDMAFP after Cl2 oxidation.

Oxygen transfer products m/z 331 [315 þ O] and m/z 286[315þOH-NC2H6] were detected after ClO2 oxidation in ranitidinesolution. However, the yields of m/z 331 [315 þ O] were muchlower, but the yields of m/z 286 [315þOH-NC2H6] were muchhigher with increasing ClO2 dosage. The change of m/z odd-evenparity suggested the cleavage of DMA functional groups. Thisresult indicated ranitidine and doxylamine had a common NeCbond cleavage reaction pathway under ClO2 oxidation. With ClO2pretreatment, an unbroken DMA group would drop off from thePPCP, while it was still acting as a NDMA precursor. The change ofNDMAFP before and after ClO2 pretreatment was determined bywhich had a higher NDMAFP, PPCP or DMA.

Oxygen transfer product m/z 347 [315 þ O2] was detected in theranitidine solution after KMnO4 oxidation. Reduced sulfur was themost probable oxygen transfer site. Sulfur ether was very easy to beoxidized to sulfone or sulfoxide by KMnO4 (Cremlyn, 1996). Therewas a symmetry-adapted configuration between empty 3d orbitalin sulfur and 2p orbital in oxygen. A feedback coordination bondwas formedwhen the sulfur accepted lone pair electrons of oxygen.DMA structures and neighboring aromatic rings in ranitidine hadnot been destroyed after KMnO4 oxidation. These products still hadhigh NDMAFPs.

triped bar) from residual PPCPs ([PPCPs]0 ¼ 25 mM, NDMA incubation time ¼ 10 d).

X. Wang et al. / Water Research 87 (2015) 403e411 409

3.4. The changes of PPCPs and NDMAFPs in natural water

To investigate the effect of oxidation with O3 Cl2, ClO2 andKMnO4 on the PPCP degradation and NDMA formation in naturalwater, experiments were performed using source water from adrinking water treatment plant with spiking of 25 nM model PPCP.The results of PPCP degradation and NDMA formation in naturalwater with oxidation treatment were shown in Fig. 6.

As expected from the results with model PPCP in ultrapurewater, nizatidine in natural water could be removed by four oxi-dants, and doxylamine could be removed by three oxidants exceptKMnO4. Comparing to the other three oxidants, O3 was the mosteffective in reducing NDMAFPs. And increasing NDMAFP was alsoobserved in natural water with nizatidine after oxidation of Cl2. Theresults were similar to those in ultrapure waters indicating that theselected PPCP had similar reactivity during oxidation in naturalwater and in ultrapure water. A result worth noting was that thechange of NDMAFP in natural water was lower than in ultrapurewater. One possible explanation is that the natural waters con-tained a significant amount of compounds, which may producetertiary amine NDMA precursors by reactionwith oxidation. In fact,it is very difficult to trace the definite origin of this behavior due tothe complex matrix of the natural waters.

Fig. 5. Oxidation products related to NDM

4. Conclusions

The results suggest that utilities employing chloramination caneffectively minimize NDMA formation by incorporating a pre-treatment with oxidation. O3, Cl2 and ClO2 were effective inremoving the selected amine-based PPCPs in aqueous solution.However, only O3 was effective in reducing NDMAFPs from theselected PPCPs. Enhanced NDMAFPs were observed in some sam-ples after Cl2 and ClO2 oxidation. It was notable that almost allNDMAFPs of the PPCPs decreased to 1% after ClO2 oxidation. DMA isthe main NDMA precursor produced from oxidation of ClO2 andDMA is degradable. ClO2 combined biologically activated carbonwould be an ideal technology for both PPCPs degradation andNDMA formation.

NDMAFP could be controlled significantly if only the DMAstructures of PPCP parents were destroyed. It was proposed thatoxygen adds to the lone pair electrons in nitrogen through anozonation pathway, thereby forming an ozonide ammoniumzwitterion after tertiary amines ozonation. Then it loses dioxygenyielding the N-oxide. Methyl was cleaved from tertiary amine afterCl2 and KMnO4 oxidation. However, oxygen or chlorine transferredto the reduced sulfur, primary and secondary amines after Cl2 andKMnO4 oxidation. A whole DMA group dropped off from the PPCP

AFPs of doxylamine and ranitidine.

Fig. 6. The changes of PPCPs and NDMAFPs in natural water ([PPCP]0 ¼ 25 nM, oxidation contact time ¼ 2 h, NDMA incubation time ¼ 10 d).

X. Wang et al. / Water Research 87 (2015) 403e411410

and the DMA was the final NDMA precursor after ClO2

pretreatment.

Acknowledgments

This work was supported by National Natural Science Founda-tion of China (No. 51278268 and No. 51290284).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2015.07.045.

References

Andrzejewski, P., Nawrocki, J., 2009. N-nitrosodimethylamine (NDMA) as a productof potassium permanganate reaction with aqueous solutions of dimethylamine(DMA). Water Res. 43 (5), 1219e1228.

Boyd, J.M., Hrudey, S.E., Li, X., Richardson, S.D., 2011. Solid-phase extraction andhigh-performance liquid chromatography mass spectrometry analysis of ni-trosamines in treated drinking water and wastewater. TrAC Trends Anal. Chem.30 (9), 1410e1421.

Choi, J., Valentine, R.L., 2002. Formation of N-nitrosodimethylamine (NDMA) fromreaction of monochloramine: a new disinfection by-product. Water Res. 36 (4),817e824.

Chen, Z., Valentine, R.L., 2008. The influence of the pre-oxidation of natural organicmatter on the formation of N-nitrosodimethylamine (NDMA). Environ. Sci.Technol. 42 (14), 5062e5067.

Cremlyn, R.J., 1996. An Introduction of Organosulfur Chemistry. John Wiley &SonsLTD, London, U.K.

Debordea, M., Von Gunten, U., 2008. Reactions of chlorine with inorganic andorganic compounds during water treatment e kinetics and mechanisms: acritical review. Water Res. 42 (1e2), 13e51.

Ellis, D., Bouchard, C., Lantagne, G., 2000. Removal of iron and manganese fromgroundwater by oxidation and microfiltration. Desalination 130 (3), 255e264.

Farr�e, M.J., Reungoat, J., Argaud, F.X., Rattier, M., Keller, J., Gernjak, W., 2011. Fate ofN-nitrosodimethylamine, trihalomethane and haloacetic acid precursors intertiary treatment including biofiltration. Water Res. 45 (17), 5695e5704.

Farr�e, M.J., Radjenovic, J., Gernjak, W., 2012. Assessment of degradation byproductsand NDMA formation potential during UV and UV/H2O2 treatment of doxyl-amine in the presence of monochloramine. Environ. Sci. Technol. 46 (23),12904e12912.

Guay, C., Rodriguez, M., S�erodes, J., 2005. Using ozonation and chloramination to

reduce the formation of trihalomethanes and haloacetic acids in drinking water.Desalination 176 (1e3), 229e240.

Jin, Y., E, X., Chen, Y., Zhang, L., Chen, C., Chen, S., Xing, D., Wang, Z., Wei, J., Yang, Y.,Zhang, H., Ai, Y., Zhuang, L., Jiang, S., Lu, Y., Zhou, M., 2006. Standard Exami-nation Methods for Drinking Water Disinfectants Parameters. GB-5750.11e2006. Standardization Administration of the People's Republic ofChina.

Krasner, S.W., Mitch, W.A., McCurry, D.L., Hanigan, D., Westerhoff, P., 2013. For-mation, precursors, control, and occurrence of nitrosamines in drinking water:a review. Water Res. 47 (13), 4433e4450.

Kemper, J.M., Walse, S.S., Mitch, W.A., 2010. Quaternary amines as nitrosamineprecursors: a role for consumer products? Environ. Sci. Technol. 44 (4),1224e1231.

Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B.,Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewatercontaminants in U. S. streams, 1999e2000: a national reconnaissance. Environ.Sci. Technol. 36 (6), 1202e1211.

Khan, M.H., Bae, H., Jung, J., 2010. Tetracycline degradation by ozonation in theaqueous phase: proposed degradation intermediates and pathway. J. Hazard.Mater. 181 (1e3), 659e665.

Kuang, J., Huang, J., Wang, B., Cao, Q., Deng, S., Yu, G., 2013. Ozonation oftrimethoprim in aqueous solution: identification of reaction products and theirtoxicity. Water Res. 47 (8), 2863e2872.

Lee, C., Schmidt, C., Yoon, J., Von Gunten, U., 2007. Oxidation of N-nitro-sodimethylamine (NDMA) precursors with ozone and chlorine dioxide: kineticsand effect on NDMA formation potential. Environ. Sci. Technol. 41 (6),2056e2063.

Liu, J., Wong, M., 2013. Pharmaceuticals and personal care products (PPCPs): a re-view on environmental contamination in China. Environ. Int. 59, 208e224.

Loos, R., Carvalho, R., 2013. EU-wide monitoring survey on emerging polar organiccontaminants in wastewater treatment plant effluents. Water Res. 47 (17),6475e6487.

Lee, Y., von Gunten, U., 2010. Oxidative transformation of micropollutants duringmunicipal wastewater treatment: comparison of kinetic aspects of selective(chlorine, chlorine dioxide, ferrate VI, and ozone) and non-selective oxidants(hydroxyl radical). Water Res. 44 (2), 555e566.

Lin, J., Huang, C., Pan, J.R., Wang, Y., 2013. Fouling mitigation of a dead-endmicrofiltration by mixing-enhanced preoxidation for Fe and Mn removal fromgroundwater. Coll. Surf. A Physicochem. Eng. Asp. 419, 87e93.

Lange, F., Cornelissen, S., Kubac, D., Sein, M.M., Von Sonntag, J., Hannich, C.B.,Golloch, A., Heipieper, H.J., Moder, M., von Sonntag, C., 2006. Degradation ofmacrolide antibiotics by ozone: a mechanistic case study with clarithromycin.Chemosphere 65 (1), 17e23.

Mitch, W.A., Schreiber, I.M., 2008. Degradation of tertiary alkylamines duringchlorination/chloramination: Implications for formation of aldehydes, nitriles,halonitroalkanes, and nitrosamines. Environ. Sci. Technol. 42 (13), 4811e4817.

Mitch, W.A., Sedlak, D.L., 2002. Formation of N-nitrosodimethylamine (NDMA) fromdimethylamine during chlorination. Environ. Sci. Technol. 36 (4), 588e595.

X. Wang et al. / Water Research 87 (2015) 403e411 411

Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., Alvarez-Cohen, L., Sedlak, D.L.,2003. N-nitrosodimethylamine (NDMA) as a drinking water contaminant: areview. Environ. Eng. Sci. 20 (5), 389e404.

Ma, J., Liu, W., 2002. Effectiveness and mechanism of potassium ferrate(VI) pre-oxidation for algae removal by coagulation. Water Res. 36 (4), 871e878.

Mu~noz, F., Mvula, E., Braslavsky, S.E., Von Sonntag, C., 2001. Singlet dioxygen for-mation in ozone reactions in aqueous solution. J. Chem. Soc. Perkin Trans. 2 (7),1109e1116.

Ning, Y., 2000. Structure Determination of Organic Compounds and Spectroscopy inOrganic Chemistry. Science Press, Beijing, P.R.C.

Ontario Ministry of the Environment, 2003. Ontario Regulation 268/03 made underthe Safe Drinking Water Act, 2002.

Office of Environmental Health Hazard Assessment (OEHHA), 2006. Public HealthGoal for N-nitrosodimethylamine and Cadmium in Drinking Water.

Padhye, L.P., Kim, J.H., Huang, C., 2013. Oxidation of dithiocarbamates to yield N-nitrosamines by water disinfection oxidants. Water Res. 47 (2), 725e736.

Pretsch, E., Bühlmann, P., Badertscher, M., 2009. Tables of Spectral Data StructureDetermination of Organic Compounds. Springer, Berlin Heidelberg, GER.

Roux, J.L., Gallard, H., Crou�e, J.-P., 2011. Chloramination of nitrogenous contaminants(pharmaceuticals and pesticides): NDMA and halogenated DBPs formation.Water Res. 45 (10), 3164e3174.

Radjenovic, J., Petrovic, M., Barcelo, D., 2009. Fate and distribution of pharmaceu-ticals in wastewater and sewage sludge of the conventional activated sludge(CAS) and advanced membrane bioreactor (MBR) treatment. Water Res. 43 (3),831e841.

Rodríguez, E., Onstad, G.D., Kull, T.P.J., Metcalf, J.S., Acero, J.L., Von Gunten, U., 2007.Oxidative elimination of cyanotoxins: comparison of ozone, chlorine, chlorinedioxide and permanganate. Water Res. 41 (15), 3381e3393.

Roux, J.L., Gallard, H., Croue, J.P., Papot, S., Deborde, M., 2012. NDMA formation bychloramination of ranitidine: kinetics and mechanism. Environ. Sci. Technol. 46(20), 11095e11103.

Shen, R., Andrews, S.A., 2011. Demonstration of 20 pharmaceuticals and personal

care products (PPCPs) as nitrosamine precursors during chloramine disinfec-tion. Water Res. 45 (2), 944e952.

Schreiber, I.M., Mitch, W.A., 2005. Influence of the order of reagent addition onNDMA formation during chloramination. Environ. Sci. Technol. 39 (10),3811e3818.

Schreiber, I.M., Mitch, W.A., 2006. Nitrosamine formation pathway revisited: theimportance of chloramine speciation and dissolved oxygen. Environ. Sci.Technol. 40 (19), 6007e6014.

Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personalcare products, and endocrine disruptors in water: implications for the waterindustry. Environ. Eng. Sci. 20 (5), 449e469.

Shah, A.D., Krasner, S.W., Lee, C.F.T., Von Gunten, U., Mitch, W.A., 2012. Trade-offs indisinfection byproduct formation associated with precursor preoxidation forcontrol of N-nitrosodimethylamine formation. Environ. Sci. Technol. 46 (9),4809e4818.

U.S. Environmental Protection Agency, 1993. Intergrated Risk Information System(IRIS) N-nitrosodimethylamine.

U.S. Environmental Protection Agency, 2009. Contaminant Candidate List 3.Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor,

pharmaceutical, and personal care product chemicals during simulated drink-ing water treatment processes, 39 (17), 6649e6663.

Ying, G., Kookana, R., Waite, T., 2004. Endocrine Disrupting Chemicals (EDCs) andPharmaceuticals and Personal Care Products (PPCPs) in Reclaimed Water inAustralia. Australian Water Conservation and Reuse Research Program, Sydney,Aus.

Zhao, Y., Boyd, J.M., Woodbeck, M., Andrews, R.C., Qin, F., Hrudey, S.E., Li, X., 2008.Formation of N-Nitrosamines from eleven disinfection treatments of sevendifferent surface waters. Environ. Sci. Technol. 42 (13), 4857e4862.

Zimmermann, S.G., Schmukat, A., Schulz, M., Benner, J., Von Gunten, U., Ternes, T.A.,2012. Kinetic and mechanistic investigations of the oxidation of tramadol byferrate and ozone. Environ. Sci. Technol. 46 (2), 876e884.