wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 7

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Occurrence and formation potential ofN-nitrosodimethylamine in ground water and river water inTokyo

Nguyen Van Huy a, Michio Murakami b,*, Hiroshi Sakai a, Kumiko Oguma a, Koji Kosaka c,Mari Asami c, Satoshi Takizawa a

aDepartment of Urban Engineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japanb “Wisdom of Water” (Suntory), Corporate Sponsored Research Program, Organization for Interdisciplinary Research Projects, The University

of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, JapancDepartment of Water Supply Engineering, National Institute of Public Health, 2-3-6 Minami, Wako, Saitama 351-0197, Japan

a r t i c l e i n f o

Article history:

Received 11 November 2010

Received in revised form

28 January 2011

Accepted 27 March 2011

Available online 5 April 2011

Keywords:

NDMA formation potential

NDMA precursors

Ground water

Chlorination

Chloramination

Disinfection byproducts

* Corresponding author. Tel.: þ81 3 5841 626E-mail address: michio@env.t.u-tokyo.ac

0043-1354/$ e see front matter ª 2011 Elsevdoi:10.1016/j.watres.2011.03.053

a b s t r a c t

N-nitrosodimethylamine (NDMA), a disinfection byproduct of water and wastewater

treatment processes, is a potent carcinogen. We investigated its occurrence and the

potential for its formation by chlorination ðNDMA� FPCl2 Þ and by chloramination ðNDMA�FPNH2ClÞ in ground water and river water in Tokyo. To characterize NDMA precursors, we

revealed their molecular weight distributions in ground water and river water. We

collected 23 ground water and 18 river water samples and analyzed NDMA by liquid

chromatography-tandem mass spectrometry. NDMA� FPCl2 was evaluated by chlorinating

water samples with free chlorine for 24 h at pH 7.0 while residual free chlorine was kept at

1.0e2.0 mgCl2/L. NDMA� FPNH2Cl was evaluated by dosing water samples with mono-

chloramine at 140 mgCl2/L for 10 days at pH 6.8. NDMA precursors and dissolved organic

carbon (DOC) were fractionated by filtration through 30-, 3-, and 0.5 kDa membranes.

NDMA concentrations were <0.5e5.2 ng/L (median: 0.9 ng/L) in ground water and

<0.5e3.4 ng/L (2.2 ng/L) in river water. NDMA concentrations in ground water were slightly

lower than or comparable to those in river water. Concentrations of NDMA� FPCl2 were not

much higher than concentrations of NDMA except in samples containing high concen-

trations of NH3 and NDMA precursors. The increased NDMA was possibly caused by

reactions between NDMA precursors and monochloramine unintentionally formed by the

reaction between free chlorine and NH3 in the samples. NDMA precursors ranged from 4 to

84 ng-NDMA eq./L in ground water and from 11 to 185 ng-NDMA eq./L in river water. Those

in ground water were significantly lower than those in river water, suggesting that NDMA

precursors were biodegraded, adsorbed, or volatilized during infiltration. The molecular

weight of NDMA precursors in river water was dominant in the <0.5 kDa fraction, followed

by 0.5e3 kDa. However, their distribution was inconsistent in ground water: one was

dominant in the <0.5 kDa fraction, and the other in 0.5e3 kDa. Molecular weight distri-

butions of NDMA precursors were very different from those of DOC. This is the first study

to reveal the widespread occurrence and characterization of NDMA precursors in ground

water.

ª 2011 Elsevier Ltd. All rights reserved.

3; fax: þ81 3 5841 8529..jp (M. Murakami).ier Ltd. All rights reserved.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 73370

1. Introduction

N-nitrosodimethylamine (NDMA) is a highly water-soluble

N-nitrosamine (WHO, 2008). NDMA had been used as an inter-

mediate in the production of rocket fuel, an inhibitor of nitrifi-

cation in soil, a plasticizer in the manufacture of rubber and

polymers, a solvent in the fiber and plastic industry, an antiox-

idant, a softener of copolymers, and an additive to lubricants

(WHO, 2002). Recently, NDMA was found to be a disinfection

byproduct of chlorination, chloramination (Najm and Trussell,

2001; Choi et al., 2002; Mitch and Sedlak, 2002; Chen and

Young, 2008; Zhou et al., 2009a), and ozonation (Andrzejewski

and Nawrocki, 2007; Andrzejewski et al., 2008; Oya et al., 2008).

Its occurrence in drinking water has been investigated

throughout Canada, the USA (Charrois et al., 2007), and Japan

(Asami et al., 2009). A survey of 20 municipal drinking water

systems in Alberta, Canada, showed concentrations of

<5e100ng/L (Charrois etal., 2007).Anational surveyof Japanese

drinking water treatment plants revealed concentrations in

finished water of<1e10 ng/L (Asami et al., 2009).

NDMA in waters has been causing concern because of its

risk to health. NDMA has been classified as a probable human

carcinogen (B2) by the Integrated Risk Information System

(IRIS) of the United States Environmental Protection Agency

(USEPA, 1987). WHO (2008) has set the guideline value for

NDMA in drinking water at 100 ng/L. Health Canada proposed

a maximum acceptable concentration of NDMA in drinking

water of 40 ng/L (Health Canada, 2010). In Japan, the Ministry

of Health, Labor andWelfare added NDMA to items for further

study in the setting of drinking water quality standards and

adopted the WHO’s guideline value as the target in April 2010.

NDMA likely reaches aquifers owing to its high polarity (log

octanol/water partition coefficient¼�0.57) (Singer et al., 1977)

and moderate biodegradation rate. Zhou et al. (2009b) esti-

mated that 90% of NDMA by mass recharged from surface

water to ground water was biodegraded over 7 years in Los

Angeles, USA. In the USA, the detection of NDMA in ground

water is commonly attributed to its use and release in asso-

ciation with asymmetrical dimethylhydrazine, a rocket fuel

component, at aerospace facilities, or to the infiltration of

effluent from municipal wastewater treatment plants

(Fleming et al., 1996; Gunnison et al., 2000; Zhou et al., 2009b).

The occurrence of NDMA in ground water in Japan has not

been investigated.

The use of groundwater by some sectors such as hospitals,

hotels, and small factories in Tokyo has increased recently.

However, some aquifers in Tokyo are heavily polluted by high

concentrations of NH3 and organic matter (Kuroda et al., 2007,

2008). Nakada et al. (2008) also revealed that ground water in

Tokyo is contaminated by pharmaceuticals and personal care

products, probably owing to leakage from decrepit sewer

pipes, the past practice of sewage disposal underground, and

infiltration by contaminated river water, and estimated that

the average composition of ground water is w1% sewage

across Tokyo. Since NDMA can be formed from precursors

such as dimethylamine and natural organic matter by chlor-

amination or chlorination in the presence of high concentra-

tions of NH3 (Choi and Valentine, 2002;Mitch and Sedlak, 2002;

Gerecke and Sedlak, 2003; Mitch et al., 2003; Chen and

Valentine, 2007), the formation of NDMA after disinfection of

these waters is a matter of concern. It is now required to

investigate NDMA and its potential for formation in ground

water to avoid detrimental impacts to health. AlthoughNDMA

precursors, normally estimated by monochloramine reaction

during 10 days (Mitch et al., 2003), have been measured in

surface water (Gerecke and Sedlak, 2003; Schreiber and Mitch,

2006) and wastewater (Mitch and Sedlak, 2004; Sedlak et al.,

2005; Pehlivanoglu-Mantas and Sedlak, 2006b), there are no

studies of the occurrence of NDMAprecursors in groundwater

over a wide area.

Our research had three aims. First, we investigated the

extent of the occurrence of NDMA in ground water in Tokyo,

comparing the results from river water. Second, we evaluated

the potential for NDMA formation by chlorination

ðNDMA� FPCl2 Þ and by chloramination ðNDMA� FPNH2ClÞ andinvestigated the factors influencing it. NDMA� FPCl2 was

analyzed to mimic a practical chlorination process in Japan.

NDMA� FPNH2Cl was analyzed to estimate total NDMA

precursors. Third, we revealed the molecular weight distri-

butions of NDMA precursors in ground water and river water.

Mitch and Sedlak (2004) investigated their molecular weight

distributions in municipal wastewater using a series of

ultrafiltrationmembraneswith cutoffs of 30, 10, and 3 kDa and

showed a dominant size of <3 kDa. So we used membranes

with cutoffs of 30, 3, and 0.5 kDa. To our knowledge, this is the

first study to reveal the widespread occurrence and charac-

terization of NDMA precursors in ground water.

2. Materials and methods

2.1. Ground water and river water sampling

We collected 23 samples from springs and from private and

public wells in Tokyo from October 2009 to May 2010. During

the daytime we also collected 18 samples from the surface of

5 rivers at 15 locations in Tokyodthe Iruma River (R.I), the Ara

River (R.A1e4), the Edo River (R.E1e3), the Tama River

(R.T1e4), and the Tsurumi River (R.TS1e3) (Fig. 1). Sampling

dates, aquifer type, and basic water quality parameters are

shown in Supplementary Tables S1 and S2.

The samples were filtered through pre-baked GF/F glass

fiber filters (pore size 0.7 mm, Whatman) and divided between

2 glass bottles for NDMA and NDMA-FP measurements.

Sodium thiosulfate, a quenching agent, was added into the

bottle for NDMA measurements. All samples were stored in

the dark at 4 �C before analysis.

2.2. Chemical analysis

2.2.1. ChemicalsNDMA was purchased from Supelco. NDMA-d6 (98%) was

purchased from Cambridge Isotope Laboratories. HPLC-grade

distilled water, formic acid, and acetonitrile were purchased

from Wako Pure Chemical. Methanol and dichloromethane

(DCM) of pesticide residue and PCB analysis grade were

purchased from Kishida. Special grade sodium bicarbonate,

sodium thiosulfate, and sulfuric acid, 1st grade monop-

otassium phosphate, and sodium hypochlorite and sodium

Fig. 1 e Sampling locations.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 7 3371

hydroxide were purchased from Kishida. Special grade

ammonium chloride was purchased from Wako. Distilled

water was passed through a tC18 cartridge (Waters) to remove

trace levels of NDMA precursors andwas then used to prepare

reagent solutions for NDMA-FP measurements. Monochlor-

amine (NH2Cl) solutions were prepared fresh daily following

Mitch and Sedlak (2002). The monochloramine concentration

was confirmed by the indophenol method using a colorimeter

(HACH).

GW

1807

GW

1808

GW

1904

GW

0402

GW

0907

GW

1204

GW

1207

GW

1401

GW

1402

GW

1406

GW

1801

GW

1806

GW

2102

GW

2103

GW

0303

GW

0307

GW

0308

GW

0404

GW

0502

GW

0905

GW

1206

GW

1213

GW

1810 R

.IR

.A1

R.A

2R

.A3

R.A

4R

.E1

R.E

2R

.E3

R.T

1R

.T2

R.T

3R

.T4

R.T

S1R

.TS2

R.T

S3

0

1

2

3

4

5

6

Confinedaquifer

Unconfinedaquifer River waterSpring

ND

MA

(ng/

L)

Fig. 2 e NDMA concentrations in ground water and river

water.

2.2.2. NDMA analysis by solid-phase extraction and LC-MS/MSNDMA in the filtrate was concentrated by a factor ofw2500 by

solid-phase extraction. To 500 mL of sample, 1 g of sodium

bicarbonate and 5 ng of NDMA-d6 were added. The samples

were passed through an EPA 521 method cartridge (Resprep)

preconditioned with 10 mL DCM, 10 mL methanol, and 20 mL

distilled water. The flow rate was 5mL/min. The cartridgewas

then dried under a gentle stream of nitrogen gas. NDMA was

GW

1807

GW

1808

GW

1904

GW

0402

GW

0907

GW

1204

GW

1207

GW

1401

GW

1402

GW

1406

GW

1801

GW

1806

GW

2102

GW

2103

GW

0303

GW

0307

GW

0308

GW

0404

GW

0502

GW

0905

GW

1206

GW

1213

GW

1810 R

.IR

.A1

R.A

2R

.A3

R.A

4R

.E1

R.E

2R

.E3

R.T

1R

.T2

R.T

3R

.T4

R.T

S1R

.TS2

R.T

S3

0

2

4

6

8

10

12NDMA

NDMA-FP

ND

MA

(ng/

L)

Confinedaquifer

Unconfinedaquifer River waterSpring

Cl2

Fig. 3 e NDMA concentrations between before and after

chlorination.

-2

-1

0

1

2

3

4

5

6

0 50 100 150 200 250

Ground water

River water

Incr

ease

d N

DM

A* (n

g/L)

NDMA precursors (ng-NDMA eq./L)

-2

-1

0

1

2

3

4

5

6

0 5 10 15

Ground water

River water

R.A2GW0502

GW0303In

crea

sed

ND

MA*

(ng/

L)

NH 3-N (mg/L)

a b

* Increase = NDMA-FPCl2 – NDMA.

Fig. 4 e Relationship between increased NDMA due to chlorination and (a) NH3 and (b) NDMA precursors.

*Increase [ NDMA� FPCl2 e NDMA.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 73372

elutedwith 10mLDCM. The eluate was concentrated to 200 mL

under nitrogen gas. NDMA was separated in an Acquity UPLC

system (Waters) with a BEH C18 column (Waters). The mobile

phase was composed of 0.1% formic acid aqueous solution

(eluent A) and 100% acetonitrile (eluent B). The flow rate was

0.2mL/min at all stages, and the sample injection volumewas

30 mL. NDMAwas detected with an Acquity TQD tandemmass

spectrometer (Waters) operated in electrospray/chemical

ionization positive-ion mode. Multiple reaction monitoring

transitions were m/z 74.9e43.1 (quantification) and m/z

74.9e57.9 (confirmation) for NDMA and m/z 81.0e46.0 for

NDMA-d6. NDMA concentrations were label-recovery

corrected.

The reproducibility and recovery rate were confirmed by

spiking sampleswith 5 ng-NDMA standard in 500mL. The rate

of recovery from distilled water (n ¼ 5) was 116% and the

GW

1807

GW

1808

GW

1904

GW

0402

GW

0907

GW

1204

GW

1207

GW

1401

GW

1402

GW

1406

GW

1801

GW

1806

GW

2102

GW

2103

GW

0303

GW

0307

GW

0308

GW

0404

GW

0502

GW

0905

GW

1206

GW

1213

GW

1810 R

.IR

.A1

R.A

2R

.A3

R.A

4R

.E1

R.E

2R

.E3

R.T

1R

.T2

R.T

3R

.T4

R.T

S1R

.TS2

R.T

S3

0

50

100

150

200

ND

MA

prec

urso

rs

(ng-

ND

MA

eq./L

)

Confinedaquifer

Unconfinedaquifer River waterSpring

Fig. 5 e NDMA precursor concentrations in ground water

and river water.

relative standard deviation (RSD) was 4%. The rate of recovery

from ground water (n ¼ 3) was 83%.

The rates of recovery of NDMA-d6 were 78% from ground

water and 72% from river water. The limit of quantification

(LOQ) was 0.5 ng/L. An operational blank was run with every

batch, and NDMA was normally less than LOQ.

2.2.3. Potential for NDMA-formation by chlorinationðNDMA� FPCl2 ÞNDMA� FPCl2 was analyzed by following the method for the

investigation of disinfection byproducts (Japan Water Works

Association, 2001). Briefly, 570 mL water was buffered with

30 mL 0.2 M monopotassium phosphate at pH 7.0 � 0.2,

chlorinated by free chlorine, and then incubated at 20 �C in the

dark for 24 h. The residual free chlorine was kept at

1e2mgCl2/L. Chlorine was analyzed by the DPDmethod using

a colorimeter. The reactions were halted by the addition of

sodium thiosulfate solution, and NDMA was measured. The

reproducibility was confirmed by using ground water (n ¼ 4;

RSD ¼ 10%). NDMA� FPCl2 in the operational blank ranged

from <0.5 to 0.9 ng/L.

2.2.4. Potential for NDMA-formation by chloraminationðNDMA� FPNH2ClÞNDMA� FPNH2Cl was analyzed according to Mitch et al. (2003).

Briefly, 510 mL water was mixed with 30 mL 0.2 M monop-

otassium phosphate, dosed with 60 mL 20 mM (1400 mgCl2/L)

monochloramine solution at pH 6.8� 0.2, and then incubated

at 20 �C in the dark for 10 days. The residual total chlorine

was analyzed by the DPD method using a colorimeter. The

reactions were halted by the addition of sodium thiosulfate,

and NDMA was measured. The reproducibility was

confirmed by using river water (n ¼ 4; RSD ¼ 4%).

NDMA� FPNH2Cl in the operational blank using distilled water

passed through a tC18 cartridge was 5.4� 0.7 ng/L (arithmetic

mean � standard error; n ¼ 7), whereas those in the opera-

tional blank using Milli-Q water and distilled water were

0

20

40

60

80

100

0 10 20 30 40 50 500

1000

1500

Ground water

ND

MA

prec

urso

rs

(ng-

ND

MA

eq./L

)

Crotamiton (ng/L)

0

50

100

150

200

0 5 10 15 20

Ground water

River waterN

DM

A pr

ecur

sors

(n

g-N

DM

A eq

./L)

TIN (mg/L)

a b

Fig. 6 e Relationship between NDMA precursors and (a) TIN and (b) crotamiton (Kuroda, 2010). (a) Ground water: r2 [ 0.02,

P > 0.1; river water: r2 [ 0.81, P < 0.01. (b) Ground water: r2 [ 0.003, P > 0.1.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 7 3373

13.9 � 2.6 ng/L (n ¼ 2) and 8.6 � 1.5 ng/L (n ¼ 2), respectively.

Use of the tC18 cartridge reduced NDMA� FPNH2Cl in the

operational blank, indicating that the cartridge removes

trace levels of NDMA precursors in distilled water.

NDMA� FPNH2Cl in the operational blank was not subtracted

from that in ground water and river water, because

a proportion of the NDMA precursors may have come from

the distilled water, even after it had been passed through the

cartridge.

We expected total NDMA precursors in the water samples

to convert to NDMA during the 10 days of chloramination

(Mitch et al., 2003; Mitch and Sedlak, 2004). Therefore, we

considered the increased concentration to represent total

NDMA precursors, and defined the precursors as

NDMA� FPNH2Cl minus initial NDMA. The concentration was

expressed as ng-NDMA equivalent/L (ng-NDMA eq./L).

Under these conditions, monochloramine decays mainly

by disproportionation and other autodecomposition reactions

(Valentine and Jafvert, 1988; Vikesland et al., 1998). The

0

50

100

150

200

0 2 4 6 8 10

Ground water

River water

ND

MA

prec

urso

rs

(ng-

ND

MA

eq./L

)

DOC (mg/L)

Fig. 7 e Relationship between NDMA precursors and DOC.

Ground water: r2 [ 0.16, P [ 0.06; river water: r2 [ 0.47,

P < 0.01.

residual total chlorine concentrations were similar among all

samples except two, and were approximately 9 mgCl2/L,

which was comparable to a previous result (Mitch and Sedlak,

2004). NDMA precursor concentrations were possibly under-

estimated in samples GW0307 and R.A4, which showed

�0.1 mg/L of residual total chlorine.

2.2.5. Other chemical analysesAfter the samples were filtered through CE membrane filters

(pore size 0.45 mm, Advantec), NH3 was analyzed by colorim-

etry using a salicylate method or indophenol blue absorpti-

ometry. After the samples were filtered through PTFE

membrane filters (0.45 mm, ADVANTEC), dissolved organic

carbon (DOC), NO2, NO3, and UV absorbance were analyzed.

DOC was analyzed with a total organic carbon analyzer (TOC-

V, Shimadzu). NO2 and NO3 were analyzed by ion chroma-

tography (761 Compact IC, Metrohm). The sum of NH3eN,

NO2eN, and NO3eN was regarded as total inorganic N (TIN).

UV absorbance at 254, 260, and 272 nm (UV254, UV260, UV272)

was analyzed by spectrophotometer (U-2010, Hitachi).

2.3. Fractionation of NDMA precursors

The NDMA precursors in two ground water samples (GW0905

and GW0907) and two river water samples (R.TS3 and R.T3)

were fractionated by molecular weight by the filtration-in-

series method (Lohwacharin et al., 2009). The samples were

sequentially fractionated through 30-kDa regenerated cellu-

lose (PLTK), 3 kDa regenerated cellulose (PLBC), and 0.5 kDa

cellulose acetate (YC05) membranes (Millipore) in a 2000 mL

Millipore Amicon stirred cell unit.

Before use, all membranes were immersed in distilled

water for 24 h to remove the wetting agent. Distilled water

passed through a tC18 cartridge was then flushed through the

membranes for 30 min to obtain pure water permeability in

a quasi-steady state. Dead-end filtration was operated at

a constant trans-membrane pressure. Constant trans-

membrane pressures of 100, 207, and 401 kPa for the 30-, 3-,

and 0.5 kDa membranes, respectively, were maintained

during fractionation at a constant stirring rate of 100 rpm.

Fig. 8 e Molecular weight distributions of DOC and NDMA precursors. (a) DOC concentration in each molecular fraction.

(b) Molecular weight distribution of DOC. (c) NDMA precursors in each molecular fraction. (d) Molecular weight distribution

of NDMA precursors.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 73374

After fractionation, water samples were subjected to

NDMA� FPNH2Cl, DOC, and UV absorbance measurements.

3. Results and discussion

3.1. Occurrence of NDMA in Tokyo

In ground water, NDMA was detected in 20 out of 23 samples,

at <0.5e5.2 ng/L, with a median of 0.9 ng/L (Fig. 2). NDMA

probably decreased during infiltration owing to biodegrada-

tion (Zhou et al., 2009b), gaseous diffusion, or volatilization

(Arienzo et al., 2006). In river water, NDMA was detected at 13

out of 15 locations (16/18 samples) at <0.5e3.4 ng/L (median:

2.2 ng/L). Because the river water samples were collected from

the surface during the daytime, direct photolysis of NDMA

might have occurred in these samples (Plumlee and Reinhard,

2007). These results are comparable to those of Asami et al.

(2009), who reported a maximum NDMA concentration in

surface water in Japan of 4.3 ng/L. NDMA concentrations in

ground water were slightly lower than or comparable to those

in river water. The highest concentrations were 5.2 ng/L in

ground water and 3.4 ng/L in river water, or <10% of the WHO

guideline for NDMA in drinking water.

No strong relationships were found between NDMA and

other water quality parameters (Table S3) in groundwater and

river water (Tables S4, S5).

3.2. NDMA� FPCl2

NDMA� FPCl2 concentrations in ground water ranged from

<0.5 to 10.8 ng/L, with a median of 1.8 ng/L (Fig. 3). Those in

river water ranged from 0.7 to 7.8 ng/L (median: 2.3 ng/L).

Concentrations were not much increased (<4.0 ng/L) in most

samples, but increased from 0.6 to 5.2 ng/L in GW0303, from

5.2 to 10.8 ng/L in GW0502, and from 3.3 to 7.8 ng/L in R.A2.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 7 3375

Samples containing high concentrations of NH3 and NDMA

precursors showed large increases in NDMA by chlorination

(>4.0 ng/L) (Fig. 4). In bench-scale chlorine disinfection

experiments, peak NDMA production occurred near the

theoretical monochloraminemaximum in the sub-breakpoint

region of the disinfection curve (Charrois and Hrudey, 2007).

The increased NDMA in our study was possibly caused by

reactions between NDMA precursors and monochloramine

that was unintentionally formed by the reaction between free

chlorine and NH3 in the samples.

3.3. Occurrence of NDMA precursors

Total NDMA precursors in ground water ranged from 4 to

84 ng-NDMA eq./L, with amedian of 10 ng-NDMA eq./L (Fig. 5).

Total precursors in river water ranged from 11 to 185 ng-

NDMA eq./L (median: 51 ng-NDMA eq./L). Concentrations in

ground water were significantly lower than those in river

water (t-test, P< 0.01). The difference indicates that the NDMA

precursors were biodegraded, adsorbed, or volatized during

infiltration. Although NDMA precursors are degraded only

slightly in wastewater-impacted river water (Schreiber and

Mitch, 2006; Pehlivanoglu-Mantas and Sedlak, 2006a), they

disappeared during the relatively long process of infiltration

into the ground water.

To identify the sources of NDMA precursors in the water,

we investigated the relationships between the precursors and

TIN (Fig. 6a). There was a significant strong relationship in

river water (r2 ¼ 0.81, P < 0.01). This relationship suggests that

NDMA precursors in river water are derived from wastewater

treatment plants, because most N in urban surface waters in

Japan comes fromwastewater treatment plants (Toyoda et al.,

2009; Ohte et al., 2010). In contrast, there was no significant

relationship in ground water (r2 ¼ 0.02, P > 0.1). Since N in

ground water in Tokyo comes from a wide variety of sources,

such as natural soils, fertilizers, and sewage (Kuroda et al.,

2007), we analyzed crotamiton, a conservative marker of

domestic sewage (Nakada et al., 2008). The crotamiton was

analyzed in ground water collected from the same well in

a different year (Kuroda, 2010). There was no significant

relationship (Fig. 6b), suggesting that the major source of

NDMA was not leakage of domestic sewage, or that there is

a large difference in infiltration behavior between NDMA

precursors and crotamiton.

There were weak but significant relationships between

NDMAprecursors and DOC in groundwater (r2¼ 0.16, P¼ 0.06)

and river water (r2 ¼ 0.47, P< 0.01) (Fig. 7), but no relationships

with UV absorbance or specific UV absorbance (SUVA) (Fig. S1).

The arithmetic mean � standard error of the NDMA

precursor-to-DOC ratio was 20� 5 ng-NDMA eq./mg in ground

water and 39 � 5 ng-NDMA eq./mg in river water. Again, this

result indicates that NDMA precursors in ground water were

less abundant than those in river water. After reacting water

samples with monochloramine at 70 mg/L Cl2 for 7 days at pH

7.2 � 0.2, Gerecke and Sedlak (2003) found 1.1 ng-NDMA eq./

mg in ground water (n ¼ 1) and 3.5 � 0.7 ng-NDMA eq./mg in

surface waters (n ¼ 7) in the USA. The NDMA precursor-to-

DOC ratios in our study were approximately one order of

magnitude higher than those of Gerecke and Sedlak (2003),

although the measurement of NDMA precursors differed

between the two studies. Our results indicate that ground

water and river water in Tokyo are heavily contaminated by

NDMA precursors, possibly owing to urban activities.

3.4. Molecular size distributions of NDMA precursors

The distributions of DOC were inconsistent between the two

ground water samples tested (Fig. 8a and b, Table S6). DOC in

GW0907 was distributed substantially in all four fractions:

highest in the <0.5 kDa fraction (45%), followed by 3e30 kDa

(27%), 0.5e3 kDa (21%), and >30 kDa (7%). But DOC in GW0905

was dominant in the >30 kDa (58%) and <0.5 kDa (38%) frac-

tions, and very small in the 3e30 kDa (3%) and 0.5e3 kDa (1%)

fractions. On the other hand, DOCwas similarly distributed in

the two river water samples tested (Fig. 8a and b, Table S6):

dominant in the 3e30 kDa fraction (35%e40%), followed by

0.5e3 kDa (28%e32%), <0.5 kDa (18%e29%), and >30 kDa

(7%e10%).

The distributions of NDMA precursors (Fig. 8c and d, Table

S7) were very different from those of DOC (above) and UV

absorbance (Fig. S2, Tables S8eS10). There were also no clear

relationships between NDMA precursors and SUVA (Fig. S3).

The molecular weight distributions of NDMA precursors were

inconsistent between the two ground water samples tested

(Fig. 8c and d): NDMA precursors in GW0905 were dominant in

the <0.5-kDa fraction (73%), whereas those in GW0907 were

dominant in the 0.5e3 kDa fraction (53%). NDMAprecursors in

both river water samples were dominant in <0.5 kDa fraction

(49%e57%), followed by 0.5e3 kDa (24%e43%). In general,

NDMA precursors were dominantly (>70%) distributed in the

<3 kDa fraction in both groundwater and river water samples.

This result is consistent with the same finding in municipal

wastewater (Mitch and Sedlak, 2004).

4. Conclusions

(1) NDMA concentrations in groundwater ranged from<0.5 to

5.2 ng/L and were slightly lower than or comparable to

those in river water.

(2) NDMA was not greatly increased (<4.0 ng/L) by chlorina-

tion, except in two ground water and one river water

samples. NDMA was increased greatly in samples con-

taining high concentrations of NH3 and NDMA precursors.

(3) Concentrations of NDMA precursors ranged from 4 to

84 ng-NDMA eq./L in ground water and from 11 to 185 ng-

NDMAeq./L in riverwater. Therewereweak but significant

relationships between NDMA precursors and DOC in both

sources, and the NDMA precursor-to-DOC ratios were

20 � 5 ng-NDMA eq./mg in ground water and 39 � 5 ng-

NDMA eq./mg in river water. NDMA precursors in ground

water were less abundant than those in river water, indi-

cating that the NDMA precursors were biodegraded,

adsorbed, or volatized during infiltration.

(4) The molecular weight of NDMA precursors in river water

was dominant in the <0.5 kDa fraction, followed by

0.5e3 kDa. However, their distribution was inconsistent in

ground water: one was dominant in the <0.5 kDa fraction,

and the other in 0.5e3 kDa. Molecular weight distributions

of NDMA precursors were different from those of DOC.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 73376

Acknowledgments

This research was partially supported by JSPS Grants-in-Aid

for Scientific Research (22760406, 21860018, and 19360237)

and a CREST project grant for ‘Development of Well-balanced

Urban Water Use System Adapted for Climate Change’ from

the Japan Science and Technology Agency.

Appendix. Supplementary data

Supplementary data related to this article can be found online

at doi:10.1016/j.watres.2011.03.053.

r e f e r e n c e s

Andrzejewski, P., Kasprzyk-Hordern, B., Nawrocki, J., 2008.N-nitrosodimethylamine (NDMA) formation during ozonationof dimethylamine-containing waters. Water Res. 42 (4e5),863e870.

Andrzejewski, P., Nawrocki, J., 2007. N-nitrosodimethylamineformation during treatment with strong oxidants ofdimethylamine containing water. Water Sci. Technol. 56 (12),125e131.

Arienzo, M., Gan, J., Ernst, F., Qin, S., Bondarenko, S., Sedlak, D.L.,2006. Loss pathways of N -nitrosodimethylamine (NDMA) inturfgrass soils. J. Environ. Qual. 35 (1), 285e292.

Asami, M., Oya, M., Kosaka, K., 2009. A nationwide survey ofNDMA in raw and drinking water in Japan. Sci. Total Environ.407 (11), 3540e3545.

Charrois, J.W.A., Boyd, J.M., Froese, K.L., Hrudey, S.E., 2007.Occurrence of N-nitrosamines in Alberta public drinking-water distribution systems. J. Environ. Eng. Sci. 6 (1), 103e114.

Charrois, J.W.A., Hrudey, S.E., 2007. Breakpoint chlorination andfree-chlorine contact time: Implications for drinking water N-nitrosodimethylamine concentrations. Water Res. 41 (3),674e682.

Chen, Z., Valentine, R.L., 2007. Formation of N-nitrosodimethylamine (NDMA) from humic substances innatural water. Environ. Sci. Technol. 41 (17), 6059e6065.

Chen, W.H., Young, T.M., 2008. NDMA formation duringchlorination and chloramination of aqueous diuron solutions.Environ. Sci. Technol. 42 (4), 1072e1077.

Choi, J., Duirk, S.E., Valentine, R.L., 2002. Mechanistic studies ofN-nitrosodimethylamine (NDMA) formation in chlorinateddrinking water. J. Environ. Monit. 4 (2), 249e252.

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

Fleming, E.C., Pennington, J.C., Wachob, B.G., Howe, R.A., Hill, D.O., 1996. Removal of N-nitrosodimethylamine from watersusing physical-chemical techniques. J. Hazard. Mater. 51(1e3), 151e164.

Gerecke, A.C., Sedlak, D.L., 2003. Precursors of N-nitrosodimethylamine in natural waters. Environ. Sci.Technol. 37 (7), 1331e1336.

Gunnison, D., Zappi, M.E., Teeter, C., Pennington, J.C., Bajpai, R.,2000. Attenuation mechanisms of N-nitrosodimethylamine atan operating intercept and treat groundwater remediationsystem. J. Hazard. Mater. 73 (2), 179e197.

Health Canada, 2010. N-nitrosodimethylamine (NDMA) inDrinking Water Document for public comment.

Japan Water Works Association, 2001. Standard Methods for theExamination of Water (in Japanese).

Kuroda, K., 2010. Ph.D Thesis. The University of Tokyo (inJapanese).

Kuroda, K., Fukushi, T., Takizawa, S., Aichi, M., Hayashi, T.,Tokunaga, T., 2007. Source estimation of nitrogencontamination in groundwaters in Tokyo metropolitan area.Environ. Eng. Res. 44, 31e38 (in Japanese).

Kuroda, K., Fukushi, T., Takizawa, S., Murakami, M., Takada, H.,Nakada, N., Aichi, M., Hayashi, T., Tokunaga, T., 2008.Groundwater Qual.: securing Groundwater Qual. UrbanIndustr. Environ.. Sources of and Factors InfluencingGroundwater Contamination in the Tokyo Metropolitan Area,vol. 324. IAHS Publ.. pp16e23.

Lohwacharin, J., Oguma, K., Takizawa, S., 2009. Ultrafiltration ofnatural organic matter and black carbon: factors influencingaggregation and membrane fouling. Water Res. 43 (12),3076e3085.

Mitch, W.A., Gerecke, A.C., Sedlak, D.L., 2003. A N-nitrosodimethylamine (NDMA) precursor analysis forchlorination of water and wastewater. Water Res. 37 (15),3733e3741.

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

Mitch, W.A., Sedlak, D.L., 2004. Characterization and fate ofN-nitrosodimethylamine precursors in municipal wastewatertreatment plants. Environ. Sci. Technol. 38 (5), 1445e1454.

Najm, I., Trussell, R.R., 2001. NDMA formation in water andwastewater. J. Am. Water Works Assoc. 93 (2), 92e99.

Nakada, N., Kiri, K., Shinohara, H., Harada, A., Kuroda, K.,Takizawa, S., Takada, H., 2008. Evaluation of pharmaceuticalsand personal care products as water-soluble molecularmarkers of sewage. Environ. Sci. Technol. 42 (17), 6347e6353.

Ohte, N., Tayasu, I., Kohzu, A., Yoshimizu, C., Osaka, K.,Makabe, A., Koba, K., Yoshida, N., Nagata, T., 2010. Spatialdistribution of nitrate sources of rivers in the Lake Biwawatershed, Japan: controlling factors revealed by nitrogen andoxygen isotope values. Water Resour. Res. 46 (7), W07505.

Oya, M., Kosaka, K., Asami, M., Kunikane, S., 2008. Formation ofN-nitrosodimethylamine (NDMA) by ozonation of dyes andrelated compounds. Chemosphere 73 (11), 1724e1730.

Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006a. The fate ofwastewater-derived NDMA precursors in the aquaticenvironment. Water Res. 40 (6), 1287e1293.

Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006b. Wastewater-deriveddissolved organic nitrogen: analytical methods,characterization, and effects - a review. Environ. Sci. Technol.36 (3), 261e285.

Plumlee, M.H., Reinhard, M., 2007. Photochemical attenuation ofN-nitrosodimethylamine (NDMA) and other nitrosamines insurface water. Environ. Sci. Technol. 41 (17), 6170e6176.

Schreiber, I.M., Mitch, W.A., 2006. Occurrence and fate ofnitrosamines and nitrosamine precursors in wastewater-impacted surface waters using boron as a conservative tracer.Environ. Sci. Technol. 40 (10), 3203e3210.

Sedlak, D.L., Deeb, R.A., Hawley, E.L., Mitch, W.A., Durbin, T.D.,Mowbray, S., Carr, S., 2005. Sources and fate ofnitrosodimethylamine and its precursors in municipalwastewater treatment plants. Water Environ. Res. 77 (1),32e39.

Singer, G.M., Taylor, H.W., Lijinsky, W., 1977. Liposolubility as anaspect of nitrosamine carcinogenicity: quantitativecorrelations and qualitative observations. Chem.Biol. Interact.19 (2), 133e142.

Toyoda, S., Iwai, H., Koba, K., Yoshida, N., 2009. Isotopomericanalysis of N2O dissolved in a river in the Tokyo metropolitanarea, rapid Commun. Mass Spectrom. 23 (6), 809e821.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9e3 3 7 7 3377

USEPA, 1987. N-Nitrosodimethylamine (CASRN 62-75-9). http://www.epa.gov/ncea/iris/subst/0045.htm (accessed 18.07.10.).

Valentine, R.L., Jafvert, C.T., 1988. General acid catalysis ofmonochloramine disproportionation. Environ. Sci. Technol. 22(6), 691e696.

Vikesland, P.J., Ozekin, K., Valentine, R.L., 1998. Effect of naturalorganic matter on monochloramine decomposition: pathwayelucidation through the use of mass and redox balances.Environ. Sci. Technol. 32 (10), 1409e1416.

WHO, 2002. N-Nitrosodimethylamine (Concise InternationalChemical Assessment Document, 38) Geneva: 52.

WHO, 2008. N-Nitrosodimethylamine in Drinking-waterBackground Document for Development of WHO Guidelinesfor Drinking-water Quality.

Zhou,W.J., Boyd, J.M., Qin, F., Hrudey, S.E., Li X-, F., 2009a. Formationof N-nitrosodiphenylamine and two newN-containingdisinfectionbyproducts fromchloramination ofwater containingdiphenylamine. Environ. Sci. Technol. 43 (21), 8443e8448.

Zhou, Q.L., McCraven, S., Garcia, J., Gasca, M., Johnson, T.A.,Motzer, W.E., 2009b. Field evidence of biodegradation of N-nitrosodimethylamine (NDMA) in groundwater with incidentaland active recycled water recharge. Water Res. 43 (3), 793e805.