Environ. Sci. Technol. 2007, 41, 6059-6065
Formation of N-Nitrosodimethylamine (NDMA) from Humic Substances in Natural Water ZHUO CHEN AND RICHARD L. VALENTINE* Civil & Environmental Engineering, 4105 Seamans Center for the Engineering Arts and Sciences, The University of Iowa, Iowa City, Iowa 52242-1527
N-nitrosodimethylamine (NDMA) formation in chloraminated Iowa River water (IRW) is primarily attributed to reactions with natural organic matter (NOM) generally classified as humic substances. Experiments were conducted to determine the contribution of various NOM humic fractions to the NDMA formation potential (NDMA FP) in this drinking water source. NOM was concentrated by reverse osmosis (RO) and humic fractions were obtained by a series of resin elution procedures. Mass balances showed that nearly 90% of the NDMA formation potential could be recovered in the NOM concentrate and in water reconstituted using additions of the various humic fractions. Generally, the hydrophilic fractions tended to form more NDMA than hydrophobic fractions, and basic fractions tend to form more NDMA than acid fractions when normalized to a carbon basis. Overall, the hydrophobic acid fraction was the dominant source of NDMA when both formation efficiency and water composition were considered. The amount of NDMA formed in a sample was found to correlate linearly with an oxidation-induced decrease in specific UV absorbance (SUVA) value at 272 nm. This is consistent with a mechanism in which precursors are formed as the direct consequence of the oxidation of NOM. The NDMA FP estimated using the slope of this relationship and the initial SUVA value compared closely to the value obtained by measuring the NDMA formed in solutions dosed with excess concentrations of monochloramine that presumably exhaust all potential precursor sources. However, the NDMA FP could not be correlated to the SUVA value of the individual humic fractions indicating that the relationship of the NDMA FP to SUVA value is probably a waterspecific parameter dependent on the exact composition of humic fractions. It is hypothesized that either specific NDMA precursors are distributed among the various humic fractions or that the humic material itself represents a “generic” nonspecific precursor source that requires some degree of oxidation to eventually produce NDMA. The nonmonotonic behavior of NOM fluorescence spectra during chloramination and lack of correlation between NOM fluorescence characteristics and NDMA formation limited the usage of fluorescence spectra into probing NDMA formation. * Corresponding author phone: (319) 335-5653; e-mail:
[email protected]. 10.1021/es0705386 CCC: $37.00 Published on Web 08/07/2007
2007 American Chemical Society
Introduction Recent studies have shown that the use of monochloramine as a secondary disinfectant may lead to the formation of N-nitrosodimethylamine (NDMA), sometimes exceeding 30 ng/L in distribution systems (1, 2). NDMA is a concern because it is recognized as a potent carcinogen (3). Risk assessments from California’s Office of Environmental Health Hazard Assessment and U.S. EPA identify lifetime de minimis (i.e., 10-6) risk levels of cancer from NDMA exposures as 0.002 ppb (2 ng/L) and 0.0007 ppb, respectively. The state of California also established an “action level” for NDMA at 10 ng/L (3). Little is known about the nature of the naturally occurring organic precursors involved in the formation of NDMA, although they must contain organic nitrogen. Laboratory studies have shown that dimethylamine (DMA), a ubiquitous substance in water, can react with monochloramine to produce NDMA (2, 4). While DMA may be an important precursor in natural waters if in sufficient concentration, recent studies have shown that DMA can usually account for only a rather small portion of the total NDMA formation potential leading to the hypothesis that proteins or other polymeric forms of natural organic matter (NOM) may be important (5). It is logical to hypothesize that humic substances (the bulk of which makes up conventional natural organic matter “NOM”) may serve as a source of precursors, since the organic nitrogen content of humic substances is generally up to several percent (6), more than sufficient to account for the ng/L levels of NDMA that are usually associated with the practice of chloramination. Numerous studies have also shown that humic substances are major precursors for DBP formation during drinking water treatment (7, 8). The reaction mechanisms leading to NDMA formation may therefore have some similarities to reactions leading to the formation of halogenated DBPs. Indeed, Chen and Valentine (9) recently modeled the formation of NDMA from a reaction of unfractionated NOM with monochloramine largely adopting a model developed by Duirk et al. (10) to predict the formation of haloacetic acids (HAAs). Key to this model was the finding that HAAs formation was linearly related to the amount of NOM oxidized by monochloramine. An improved understanding of the nature of NDMA precursors in natural waters is critical to the development of effective strategies to minimize NDMA formation as a consequence of chloramination such as selective removal of organic matter corresponding to those components which are most reactive (11). Unfortunately, the nature of NOM is particularly poorly understood mostly due to its heterogeneity and complexity of its structure. Indeed, the very definition of NOM and what is considered a humic substance is operationally defined. A major goal of this study was to investigate the formation of NDMA from NOM by quantifying the importance of various humic fractions. This included comparing the recovery of the NDMA formation potential in the whole water sample to that obtained in water reconstituted from the various humic fractions. The UV absorbance and fluorescence spectra of NOM were also examined to investigate correlations between NOM spectral characteristics and NDMA formation, which led to development of a novel approach to measuring the NDMA formation potential that is not directly dependent on reaction conditions but only on measurement of NDMA formation and the oxidative-induced change in specific UV absorbance (SUVA). VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6059
Materials and Methods Batch NDMA formation studies were conducted using NOM derived from the Iowa River. NOM from this source was also used in a previous study focusing on the kinetics of NDMA formation (9). Water samples studied included the following: (1) original whole water samples from the Iowa River (IRW), (2) water reconstituted from NOM obtained as a reverse osmosis (RO) concentrate from the Iowa River using a RealSoft PROS/2S RO unit (Stone Mountain, GA), and (3) selected humic fractions derived by resin separation applied to the RO concentrate (12). It should be pointed out that the exact nature of the concentrate and humic fractions is operationally defined and could include a variety of nonhumic constituents such as proteinaceous material associated with agricultural or domestic waste discharges. However, the NDMA formation characteristics of the Iowa River water NOM were previously shown to be very similar to those of NOM obtained from a pristine source. All experiments were conducted in batch reactors (1-L capacity clear Pyrex bottles with PTFE screw caps) utilizing additions of preformed monochloramine stock solutions. This was done to avoid artifacts caused by reactions between NOM and free chlorine which may exist if monochloramine was formed in situ. Two types of NDMA formation experiments were conducted. One was to measure the NDMA formation potential (NDMA FP) using a high dosage of monochloramine, a long reaction time, and reaction conditions presumed to exhaust all NDMA precursors. This approach was based upon a similar logic behind the THM formation potential test (13-15) and was first proposed by Mitch et al. (16) for determination of the NDMA formation potential in water. Briefly, the buffered water samples (10 mM NaHCO3 at pH 7.0 ( 0.2) containing NOM were allowed to contact with 1 mM monochloramine. The bottles were kept at 20 °C in the dark, and the reaction was quenched by addition of excess ascorbic acid after 7 days. NDMA formation was also studied using reaction conditions that more realistically simulate an actual distribution system. In this case, monochloramine dosage of 0.05 mM was utilized in buffered water samples that contained 4 mM sodium bicarbonate and were adjusted to pH 7.0 by addition of concentrated HCl or NaOH. These experiments focused on correlation between NDMA formation and NOM spectral and fluorescence characteristics. Ionic strength was adjusted at 8 mM by sodium perchlorate. NDMA concentrations, SUVA values, and monochloramine residuals were monitored for up to 5 days. Monochloramine residuals were determined using Standard Method 4500-Cl F DPD-FAS titrimetric method (17). DOC was determined using a Shimadzu TOC 5000 (Shimadzu Scientific, Columbia, MD). UV absorbance and spectral characteristics of the natural organic mater were obtained with a Shimadzu UV1601 dual-beam spectrophotometer. Fluorescence spectra were obtained using a Perkin-Elmer LS55 luminescence spectrometer. Excitation-emission matrix (EEM) spectra were collected with subsequent scanning emission spectra from 290 to 600 nm at 10 nm increments by varying the excitation wavelength from 200 to 400 nm at 10 nm increments. NDMA analysis was accomplished using solid-phase extraction, isotope dilution gas chromatography/mass spectrometer (GC/MS) methodology similar to that originally proposed by Taguchi et al. (18) and used by others (19, 20). The detailed procedure has been described elsewhere (9). Dimethylamine (DMA) was analyzed using a method involving derivatization using p-toluenesulphonylchloride and alkali to form N, N-dimethyl-p-toluenesulphonamide (21), which has a molecular weight that is over four times that of DMA and can be effectively extracted from water 6060
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
FIGURE 1. Standard addition test on NDMA formation potential using Iowa River water (pH ) 7.0, DOC ) 3.4 mg/L, T ) 20 °C, [NH2Cl]0 ) 1 mM, [DMA] ) 0-1.0 µg/L, contact time ) 7d). using methylene chloride and analyzed by GC/MS after preconcentration. Generally 500 mL of sample is required for the DMA analysis. After addition of ascorbic acid to quench any sodium hypochlorite residual, internal standard d6-DMA was added at a concentration of the 10 µg/L. This was followed by addition of 2 mL of 50% (w/w) K2CO3 and 1 mL of p-toluenesulphonylchloride (60 mg/mL). The alkyl condition of the solution favors the formation of the derivatized product: N,N-dimethyl-p-toluenesulphonamide (DMPTSA). After the sample was shaken for 2.5 h at 180 oscillation/min, glacial acetic acid was added to obtain pH of about 3-4. The addition of acid enables more efficient extraction of the DMPTSA with methylene chloride. An aliquot of 25 mL of methylene chloride was added into the sample. After 40 min of shaking, the organic phase was withdrawn with a graduated pipet and passed through a plug of glass wool into a 250 mL round-bottom flask. The glass wool was rinsed using 10 mL of methylene chloride and the organic phase was evaporated using Rotovapping method to about 4 mL. A 2 µL aliquot of the sample was injected into the GC/MS/MS in CI mode. Varian CP-3800 gas chromatography coupled with Saturn 2200 MS/MS was used for DMA analysis. A DB1701 capillary column was used for DMA determination. The inlet temperature of the GC system was kept at 200 °C and the MS temperature was set at 300 °C. The oven temperature ramp started at 90 °C and was held for 1 min. It started increasing at 10 °C/min to 260 °C and was held for 7 min. The carrier gas was Helium and the flow rate was about 1 mL/min. The selected ions m/z 200 (M + 1) for DMA and m/z 206 (M + 1) for d6-DMA were used for the quantification of DMA analysis. The MDL of this method is about 0.1 µg/L.
Results and Discussion Contribution of DMA and Uncharacterized NOM to NDMA Formation Potential. An initial activity was to determine the extent to which dimethylamine (DMA) (one naturally occurring NDMA precursor identified) might contribute to NDMA formation in Iowa River water. This was accomplished by measuring the NDMA formation potential (FP) in a series of whole Iowa River water samples containing increasing amounts of DMA (a standard addition type of approach). Figure 1 shows a linear relationship between the increasing amounts of NDMA formed with the increasing amount of added DMA. Approximately 120 ng/L of NDMA was formed without added DMA. DMA analysis showed that the water contained a DMA concentration of 0.5 µg/L. Based on the slope of 39 ng NDMA/µg DMA, a formation of only approximately 20 ng/L of NDMA would be expected with the naturally occurring amount of DMA. Consistent with this
FIGURE 2. (a) Fluorescence Excitation-Emission Matrix (EEM) of original Iowa River water (DOC ) 3.4 mg/L, pH ) 7.0); (b) Fluorescence Excitation-Emission Matrix (EEM) of reconstituted water sample from RO collected water (DOC ) 3.4 mg/L, pH ) 7.0). expectation, 22.5 ng/L of NDMA was formed in organic free water with an addition of 0.5 µg/L of DMA. Clearly the naturally occurring DMA content cannot account for the majority of the NDMA formed, a finding similar to that of Gerecke and Sedlak (22). Therefore DMA must be considered a comparatively minor NDMA precursor in this water, at least under these reaction conditions. Some other component(s) of NOM such as humic material must account for the majority of NDMA formed. Characteristics of Iowa River NOM. Natural organic matter from Iowa River was concentrated using reverse osmosis (RO). Determining how well this concentrate represents the NDMA precursors in the whole water sample was an important objective. Kitis et al. (23) observed that RO isolation of NOM had almost no impact on the integrity and reactivity of NOM. They were able to recover approximately 98% of the DOC and obtained similar recoveries in the formation potential of the DBPs such as THMs and HAAs. The DOC and SUVA recoveries during the RO concentration procedure were summarized previously in a modeling study using the same NOM source (9). Briefly, DOC recovery of 86% was realized with a resulting concentration by a factor of approximately 15. SUVA is the UV absorbance divided by the amount of DOC concentration. Both SUVA254 and SUVA272 are considered to be effective indexes of the reactivity and aromaticity of NOM (24-26). Specifically, they are widely
considered as accurate predictors of reactivity with disinfectants in reactions leading to the formation of halogenated DBPs (27-29). The original SUVA254 and SUVA272 were 2.30 L/mg-m and 2.14 L/mg-m, respectively. After concentration, the SUVA recoveries were approximately 96% and 90%, respectively. Figure 2a shows the fluorescence excitation-emission matrix (EEM) graph for original Iowa River containing 3.4 mg/L DOC while Figure 2b is the EEM graph of a reconstituted water sample containing the RO concentrate at the same DOC concentration of 3.4 mg/L. Both water samples were adjusted to neutral pH 7.0. As shown in Figure 2, y-axis is the excitation wavelength (ranging from 200 to 400 nm), while x-axis is the emission wavelength (ranging from 290 to 600 nm). The 45° ridge line in the upper left corner is the first order Raleigh scattering of the incident light from the excitation grating, and the 22.5° ridge line on the lower right corner is associated with second order Raleigh light scattering, where excited light was emitted at an emission wavelength twice that of the excitation wavelength (30). It is the region between these two ridge lines that represents the fluorescence characteristics of the NOM. The contour lines are the distribution of fluorescence intensity at each excitationemission wavelength pair. The colors on EEM spectra provide a complementally and visible indication of the fluorescence intensity. The brighter the color, the higher the intensity. VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6061
FIGURE 3. DOC recovery of NOM fractionation (IRW NOM, fractionation using procedure proposed by Leenheer (14)). The EEMs of original Iowa River water and reconstituted NOM samples showed strong similarity in both the shapes and the intensities of the contour lines. Both of them captured two broad-shaped major peaks in the EEM spectra. One peak is located around excitation-emission pair (240 nm, 425 nm) (Peak A) and the other at excitation-emission pair (330 nm, 424 nm) (Peak B). Generally, different peaks in an EEM spectrum suggest the existence of different fluorophores. But the interference and overlapping of different fluorophores make the EEM spectra very difficult to interpret. Some researchers associated peak B with bigger molecules originating from humic substances, while associating peak A with smaller molecules (31, 32). The high recovery in DOC, SUVA, and the strong similarity in fluorescence spectra shown in Figure 2 all indicate that the RO concentrate is representative of the original Iowa River water insofar as these characteristics are concerned. RO-concentrated Iowa River NOM (IRW NOM) was further fractionated into six operationally defined fractions using the procedure proposed by Leenheer (12). Figure 3 is a pie chart of DOC recovery of each fraction showing that hydrophobic acids (HPOA), which usually contain both humic acids and fulvic acids, are the major contributor of total DOC (72%). Hydrophobic bases (HPOB) and hydrophobic neutrals (HPON) contributed 1.98% and 1.1%, respectively. Hydrophilic fractions generally make up a smaller fraction of DOC than that due to hydrophobic fractions. Hydrophilic acids (HPIA) contributed 7.9%, while hydrophilic bases (HPIB) and hydrophilic neutrals (HPIN) contributed 4.6% and 0.8%, respectively, of the total DOC. A relatively small, 11% loss in total DOC was realized, probably due to the washing and elution of columns and evaporation during the rotovapping procedure. The EEM spectra of all six NOM fractions at equivalent DOC level (3.4 mg/L) including hydrophilic acids, hydrophilic bases, hydrophilic neutrals, hydrophobic acids, hydrophobic bases, and hydrophobic neutrals are shown in the Supporting Information in Figures S.1-S.6. All six EEM spectra captured basically the two peaks identified in the EEM spectrum of un-pretreated IRW NOM: peak A located around excitationemission pair of 240 nm and 425 nm, which is associated with higher excitation energy, and peak B located around excitation-emission pair of 330 nm and 424 nm, which is associated with lower excitation energy. Peak A was always more intense than peak B. Some researchers have associated peak A with tryptohan-like or protein-like compounds and 6062
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
peak B with humic substances (30, 31), but the exact identification of each fluorophore is still far from clear. Even though both peak A and peak B were found in all the EEM spectra of the six fractions, a significant shift in the location of the two peaks was observed in the EEM spectra of each fraction. The fluorescence intensity peak locations in EEM of each fraction are listed in Table S.1 in the Supporting Information. Most EEM spectra of the NOM fractions showed fluorescence intensity peaks across the excitation wavelength of 240 and 330 nm. The emission spectra of each fraction at these two excitation wavelengths were collected and are shown in Figures S.7-S.10 in the Supporting Information. All the spectra were normalized by their respective maximum emission intensities. And the peak positions of the emission spectra are listed inTable S.2. Significant blue-shifting in the locations of both peak A and peak B was observed in hydrophilic fractions compared to hydrophobic fractions, which probably indicate smaller molecular sizes in hydrophilic humic fractions. Specific UV absorbance spectra of all six NOM fractions at equivalent DOC level (3.4 mg/L) are shown in the Supporting Information in Figures S.11 and S.12. The spectra of the fractions did not show any distinctive features, and the shapes of all the UV spectra appeared quite similar, especially in the hydrophobic and hydrophilic categories. All the spectra showed a broad absorbance band at wavelength smaller than 250 nm, which is probably due to inorganic species interference. Hydrophilic fraction spectra showed some inflection points between 240 and 300 nm. Although the shapes of spectra shared a certain extent of similarity, the SUVA values at 254 and 272 nm of different fractions varied significantly (Table 1). Hydrophobic acids showed the biggest SUVA254 and SUVA272 values, while hydrophilic neutrals had the smallest values in both categories. Basic portions tended to have smaller SUVA values compared to acid portions of NOM, and generally hydrophobic fractions showed larger SUVA values than hydrophilic fractions. NDMA Formation Potential. NDMA formation potential tests (16) were conducted using both whole Iowa River water, and reconstituted river water prepared with RO concentrate and humic fractions which replicated the whole-water DOC concentration of 3.4 mg C/L. A previous study using the same NOM concentrate showed the NDMA formation potential in the original unprocessed whole water sample as a function of time (9). NDMA formation steadily increased with time
TABLE 1. List of NDMA Formation Potential of Fractionated NOM Fractions with Their SUVA Values
a
fraction
SUVA254 (L/mg-m)
SUVA272 (L/mg-m)
DOC contribution (%)
NDMA FPa (ng/(mgDOC))
total NDMA FP contribution (%)
HPOA HPOB HPON HPIA HPIB HPIN
2.14 1.79 0.44 1.76 1.32 0.20
1.53 1.42 0.28 1.19 0.83 0.17
72.00 1.98 1.10 7.90 4.60 0.80
27.47 31.43 22.44 43.50 77.50 25.76
71 2.2 0.9 12.3 12.8 0.7
FP: Formation potential.
TABLE 2. 3.4 mg/L DOC Equivalent NDMA Formation Potential sample
NDMA formation potential (ng/L)
original river water reconstituted for RO water reconstituted from humic fractions determined by SUVA method
112 100 95 90
attaining a value of 112 ng/L at the end of the 7-day test period. In comparison, approximately 100 ng/L of NDMA was formed in the laboratory-prepared water containing RO concentrate at the same DOC concentration as the unprocessed water. In addition, the concentration-time profiles were similar. This nearly 90% recovery of NDMA formation potential indicates that the RO concentration method was very effective in preserving reactivity. In other words, most NDMA precursors were captured and concentrated during this RO concentration procedure. The NDMA formation potential of each humic fraction was also determined. Table 1 tabulates the NDMA formation potential normalized to carbon content (ng of NDMA formed per mg of DOC). For example, 77.5 ng of NDMA were formed from every 1 mg of carbon in the hydrophilic bases (HPIB) fraction. Obviously HPIB showed the highest NDMA formation potential compared to other NOM fractions. Two trends are evident from the data in Table 1. First, hydrophilic fractions tend to form more NDMA than hydrophobic fractions. Second, basic fractions tend to have a larger NDMA formation potential than acid fractions. This may be due to the higher nitrogen content in the polar hydrophilic and basic fractions (6). Table 2 summarizes the expected contribution of all fractions to the NDMA formation potential measured in the whole water. In this table, the NDMA formation potential was calculated based upon what is expected for the DOC concentration found in original river water (3.4 mg/L). The predicted NDMA formation potential from the sum of the six fractions is expected to be approximately 95 ng/L, based on the DOC percentage contribution (Figure 3) and NDMA formation potential of each fraction (Table 1). This value can be compared to the value of 112 ng/L measured for the whole original Iowa River water. The relatively small difference indicates excellent recovery of NDMA formation potential in both the RO-concentrated water and in the individual humic fractions. Last, the observation of the formation of NDMA in all humic fractions suggests that individual precursors exist with properties that do not allow them to be separated by the same approach taken to fractionate the NOM. If discrete precursors are indeed responsible for NDMA formation, an alternative approach to their separation is needed. Relationship of NDMA Formation to Changes in SUVA. SUVA has been extensively used as indicator of the reactivity of NOM with disinfectants (33) both as a source of demand and to form halogenated DBPs. This reactivity has been
attributed to the aromatic moieties in NOM that are believed to largely influence its UV absorbance spectrum (34, 35). The SUVA values at 254 and 272 nm are considered especially good indicators of the aromaticity and reactivity of NOM (as measured by chlorine or monochloramine demand) and the potential to form some halogenated DBPs (28, 36). For example Duirk et al. (37) demonstrated a linear relationship between a decrease in SUVA254 and the amount of NOM oxidized by monochloramine. Furthermore the formation of dichloroacetic acid was also found to be linearly related to the amount of NOM oxidized (38). Korshin et al. have also shown that using differential spectroscopy, the formation of several halogenated DBPS produced by a reaction of free chlorine and NOM is linearly related to the decrease in SUVA that accompanies the reaction (36). Experiments were conducted to investigate the relationship between the amount of NDMA formed and the oxidative loss of SUVA in chloraminated water. This was done using reconstituted IRW NOM at a DOC level of 3.4 mg/L and a dosage of 0.05 mM preformed NH2Cl. At intervals of every 1 h, the monochloramine residual was quenched by addition of excess ascorbic acid and both NDMA concentration and SUVA272 value were determined. The formation of NDMA was found to correlate linearly with the reduction in SUVA272 occurring during the course of the reaction (Figure 4). This suggests the possibility of tracking NDMA formation by measuring changes in SUVA272 for a particular supply although it is quite likely that the exact relationship will be source water dependent. The relationship between NDMA formation and changes in SUVA caused by NOM oxidation calls into question the notion that NDMA is formed from one or more simple discrete and independently reacting precursor substances. The spectral results support an alternative hypothesis that the
FIGURE 4. Relationship between change in specific UV absorbance of NOM and NDMA formation in the reaction of NOM with monochloramine (pH 7.0, T ) 20 °C, IRW NOM DOC ) 3.5 mg/L, [NH2Cl]0 ) 0.05 mM; slope and intercept are shown with their 95% confidence intervals). VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6063
FIGURE 5. NDMA formation potential of each NOM fraction vs their SUVA values. majority of NDMA “precursors” in this source are produced as a consequence of the general oxidation of NOM. Humic material possibly represents a “generic” nonspecific precursor source that requires pre-oxidation to eventually produce NDMA. The amount of NDMA formed by the reaction of monochloramine with NOM was indeed recently reported to be rate limited by the oxidation of NOM and linearly related to the amount of NOM oxidized in several source waters including that used in this study (9). The relationship between the NDMA formation potentials and the individual SUVA values of each NOM fraction is shown in Figure 5. The lack of a consistent trend indicates that the SUVA value of NOM is not a universal index for NDMA formation potential. It may represent the aromaticity of NOM, but apparently is not linearly related to potential precursor content which may be more related to the exact nature of specific functional groups or non-humic compounds also present. Taken together these finding indicate that while the NDMA formation potential of the bulk water may be directly related to the SUVA value of the water, this relationship will differ depending on the composition of the water with regard to the various humic fractions. The absolute value of SUVA might be a good predictor of NDMA formation potential only if all water sources had the same distribution of each humic fraction, and presumably the same distribution of potential NDMA precursors. The correlation with SUVA, similar to that observed for the formation of some halogenated DBPs, does support the hypothesis that humic fractions obtained by a classical resin fractionation approach (12) are indeed an important NDMA precursor. A contribution by non-humic substances cannot, however, be ruled out. The linearity of the relationship between NDMA formation and the change in SUVA suggests that the NDMA formation potential might be proportional to the bulk water SUVA value as the product of the initial SUVA value and the slope of the regression line:
NDMA FP ) Slope × SUVA0
(1)
Indeed, the predicted formation potential of approximately 90.54 ( 3.72 ng/L determined from the slope of 42.31 ( 1.74 (in the unit of ng‚mg‚m/L2) in Figure 4 and the initial SUVA value of 2.14 L/mg-m is very close to the value of 100 ng/L determined in the “conventional” NDMA formation potentialtestwhichrequiresverylargedosagesofmonochloramine and long reaction times necessary to presumably exhaust all NDMA precursors. Certainly additional work on other supplies is warranted to determine if this observation and calculation based upon nearly opposite reaction conditions is indeed general. This provocative finding does provide some 6064
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 17, 2007
new insight into the possible nature of NDMA (and perhaps other DBP) precursors. Relationship of Fluorescence Characteristics to NDMA Formation. It has been shown that chlorination of NOM could cause a significant intensity change in fluorescence spectra and the position of emission bands (39). Figure S.13 in the Supporting Information shows the effect of chloramination on the fluorescence emission spectra of NOM at excitation wavelength of 240 nm. After 1 day of contact with 0.05 mM NH2Cl, the peak position of the emission spectra shifted slightly to the lower wavelength. The intensity and the width of the emission peak increased significantly. This is probably due to the attack of the monochloramine on the NOM reaction sites and the subsequent decomposition of larger molecules to smaller ones (25). After 4 days, the intensity and the width of the emission peak diminished, which is consistent with the loss of aromaticity in NOM structure. In contrast, Figure S.14 shows the effect of chloramination on fluorescence emission spectra of NOM at excitation wavelength of 330 nm. Both the intensity and the width of the emission peak decreased continuously with time in the presence of 0.05 mM NH2Cl. As mentioned earlier, some researchers associated the emission peak B at excitation wavelength of 330 nm with larger molecule units in NOM structure. The continuous decrease in terms of the intensity and width of the emission peak is probably caused by the breaking down of the molecules and continuous loss of aromaticity. No quantitative relationship was found between the decrease of fluorescence intensity and NDMA formation. Figure S.15 shows this lack of relationship for studies conducted in IRW NOM over a period of 4 days. Also the non-monotonic behavior in fluorescence characteristics of NOM makes measurement of fluorescence a poor quantitative probe of NDMA formation in water with time.
Supporting Information Available Table S.1 lists peak positions in fluorescence EEM graphs of aquatic IRW NOM fractions, table S.2 lists the fluorescence peak positions of IRW NOM fractions in emission scans at excitation wavelength of 240 and 330 nm. Figures S.1-S.6 show the fluorescence EEM graphs of IRW NOM humic fractions. Additional UV spectral and fluorescence characteristics of IRW NOM humic fractions are shown in Figures S.7-S.15. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Valentine, R. L.; Choi, J.; Chen, Z.; Barrett, S. E.; Hwang, C. J.; Guo, Y.; Wehner, M.; Fitzsimmons, S.; Andrews, S. A.; Werker, A. G.; Brubacher, C. M.; Kohut, K. D. Factors affecting the formation of NDMA in water and occurrence. AWWA Research Foundation: Denver, CO, 2005. (2) Mitch, W. A.; Sharp, J. O.; Trussell, R. R.; Valentine, R. L.; AlvarezCohen, L.; Sedlak, D. L. N-Nitrosodimethylamine (NDMA) as a drinking water contaminant: A review. Environ. Eng. Sci. 2003, 20 (5), 389-404. (3) California Department of Health Services. California Drinking Water: Activities Related to NDMA and other Nitrosamines; CDNH: Sacramento, CA, 2005. (4) Choi, J.; Valentine, R. L. Studies on the formation of Nnitrosodimethylamine (NDMA) in drinking water: a new chloramination disinfection by-product. In Proceedings - Annual Conference American Water Works Association; AWWA: Denver, CO, 2001; pp 47-55. (5) Mitch, W. A.; Sedlak, D. L. Characterization and fate of N-nitrosodimethylamine precursors in municipal wastewater treatment plants. Environ. Sci. Technol. 2004, 38 (5), 14451454. (6) Croue, J. P.; Korshin, G. V.; Benjamin, M. Characterization of natural organic matter in drinking water; AWWA Research Foundation: Denver, CO, 1999.
(7) Wei, Y.; Wu, J.; Henriques, C.; Reavis, W.; Page, K.; Thomas R. How organic nitrogen compounds in surface water relate to DBP formation in the drinking water treatment process. In Proceedings - Water Quality Technology Conference; 2002; pp 107-120. (8) Wu, W. W.; Chadik, P. A.; Delfino, J. J. The relationship between disinfection by-product formation and structural characteristics of humic substances in chloramination. Environ. Toxicol. Chem. 2003, 22 (12), 2845-2852. (9) Chen, Z.; Valentine, R. L. Modeling the formation of Nnitrosodimethylamine (NDMA) from the reaction of natural organic matter (NOM) with monochloramine. Environ. Sci. Technol. 2006, 40 (23), 7290-7297. (10) Duirk, S. E.; Gombert, B.; Croue, J.-P.; Valentine, R. L. Modeling monochloramine loss in the presence of natural organic matter and relationship to haloacetic acid formation. Prepr. Ext. Abstr. ACS Natl. Meet. Am. Chem. Soc. Div. Environ. Chem. 2001, 41 (2), 799-803. (11) Gonzalez-Velasco, J. R.; Iriarte, U.; Lopez-Fonseca, R.; AlvarezUriarte, J. I. Innovative quality control strategies in controlling drinking water production. Prog. Water Resour. 2003, 8 (Water Resources Management II), 119-124. (12) Leenheer, J. A. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environ. Sci. Technol. 1981, 15 (5), 578-87. (13) Bruchet, A.; Anselme, C.; Marsigny, O.; Mallevialle, J. THM formation potential and organic content: a new analytical approach. Aqua (Oxford) 1987, 2, 102-9. (14) Chu, H. P.; Wong, J. H. C.; Li, X. Y. Trihalomethane formation potentials of organic pollutants in wastewater discharge. Water Sci. Technol. 2002, 46 (11-12, Water Quality and Environmental Management in Asia), 401-406. (15) Mega, J. Water quality assessment and trihalomethane formation potential. Vodni Hospodarstvi 2005, 55 (2), 33-35. (16) Mitch, W. A.; Gerecke, A. C.; Sedlak, D. L. A N-nitrosodimethylamine (NDMA) precursor analysis for chlorination of water and wastewater. Water Res. 2003, 37 (15), 3733-3741. (17) APHA, AWWA, and WEF. Standard Methods for the Examination of Water and Wastewater, 18th ed.; Washington, DC, 1992. (18) Taguchi, V. Y.; Jenkins, S. W. D.; Wang, D. T.; Palmentier, J.-P.F. P.; Reiner, E. J. Determination of N-nitrosodimethylamine by isotope dilution, high-resolution mass spectrometry. Can. J. Appl. Spectrosc.1994, 39 (3), 87-93. (19) Luo, X.; Clevenger, T. E.; Gang, D. NDMA analytical methods comparisons and its occurrence in Missouri. In Proceedings Water Quality Technology Conference; American Water Works Association: Denver, CO, 2003; pp 131-142. (20) Choi, J.; Valentine, R. L. Formation of N-nitrosodimethylamine (NDMA) from reaction of monochloramine: a new disinfection by-product. Water Res. 2002, 36 (4), 817-824. (21) Graham, J. E. Factors affecting NDMA formation during drinking water treatment. In Civil Engineering; University of Waterloo: Waterloo, Ontario, Canada, 1996. (22) Gerecke, A. C.; Sedlak, D. L. Precursors of N-Nitrosodimethylamine in Natural Waters. Environ. Sci. Technol. 2003, 37 (7), 1331-1336. (23) Kitis, M.; Kilduff, J. E.; Karanfil, T. Isolation of dissolved organic matter (DOM) from surface waters using reverse osmosis and its impact on the reactivity of DOM to formation and speciation of disinfection by-products. Water Res. 2001, 35 (9), 2225-2234. (24) Traina, S. J.; Novak, J.; Smeck, N. E. An ultraviolet absorbance method of estimating the percent aromatic carbon content of humic acids. J. Environ. Quality 1990, 19 (1), 151-3. (25) Korshin, G. V.; Kumke, M. U.; Li, S.-W.; Frimmel, F. H. Influence of Chlorination on Chromophores and Fluorophores in Humic
Substances. Environ. Sci. Technol. 1999, 33 (8), 1207-1212. (26) Weishaar, J. L.; Aiken, G. R.; Bergamaschi, B. A.; Fram, M. S.; Fujii, R.; Mopper, K. Evaluation of Specific Ultraviolet Absorbance as an Indicator of the Chemical Composition and Reactivity of Dissolved Organic Carbon. Environ. Sci. Technol. 2003, 37 (20), 4702-4708. (27) Clark, R. M.; Thurnau, R. C.; Sivaganesan, M.; Ringhand, P. Predicting the formation of chlorinated and brominated byproducts. J. Environ. Eng. (Reston, Virginia) 2001, 127 (6), 493501. (28) Kitis, M.; Karanfil, T.; Kilduff, J. E.; Wigton, A. The reactivity of natural organic matter to disinfection by-products formation and its relation to specific ultraviolet absorbance. Water Sci. Technol. 2001, 43 (2, 1st World Water Congress, Part 2: Industrial Wastewater and Environmental Contaminants, 2000), 9-16. (29) Korshin, G. V.; Wu, W. W.; Benjamin, M. M.; Hemingway, O. Correlations between differential absorbance and the formation of individual DBPs. Water Res. 2002, 36 (13), 3273-3282. (30) Westerhoff, P.; Chen, W.; Esparza, M. Fluorescence analysis of a standard fulvic acid and tertiary treated wastewater. J. Environ. Qual. 2001, 30 (6), 2037-2046. (31) Alberts, J. J.; Takacs, M.; Egeberg, P. K. Total luminescence spectral characteristics of natural organic matter (NOM) size fractions as defined by ultrafiltration and high performance size exclusion chromatography (HPSEC). Org. Geochem. 2002, 33 (7), 817-828. (32) Swietlik, J.; Sikorska, E. Application of fluorescence spectroscopy in the studies of natural organic matter fractions reactivity with chlorine dioxide and ozone. Water Res. 2004, 38 (17), 37913799. (33) Westerhoff, P.; Chao, P.; Mash, H. Reactivity of natural organic matter with aqueous chlorine and bromine. Water Res. 2004, 38 (6), 1502-1513. (34) Chen, J.; LeBoeuf, E. J.; Choi, S.; Gu, B. Spectroscopic characterization of structural and functional properties of natural organic matter fractions. In Abstracts of Papers, 220th ACS National Meeting, August 20-24, 2000; American Chemical Society: Washington, DC, 2000; ENVR-322. (35) Vogt, R. D.;Akkanen, J.; Andersen, D. O.; Bruggemann, R.; Chatterjee, B.; Gjessing, E.; Kukkonen, J. V. K.; Larsen, H. E.; Luster, J.; Paul, A.; Pflugmacher, S.; Starr, M.; Steinberg, C. E. W.; Schmitt-Kopplin, P.; Zsolnay, A. Key site variables governing the functional characteristics of dissolved natural organic matter (DNOM) in Nordic forested catchments. Aquat. Sci. 2004, 66 (2), 195-210. (36) Korshin, G. V.; Li, C. W.; Benjamin, M. M. The decrease of UV absorbance as an indicator of TOX formation. Water Res. 1997, 31 (4), 946-949. (37) Duirk, S. E.; Gombert, B.; Croue, J. P.; Valentine, R. L. Modeling monochloramine loss in the presence of natural organic matter. Water Res. 2005, 39 (14), 3418-3431. (38) Duirk, S. E.; Valentine, R. L. Modeling dichloroacetic acid formation from the reaction of monochloramine with natural organic matter. Water Res. 2006, 40 (14), 2667-2674. (39) Fabbricino, M.; Korshin, G. V. Probing the mechanisms of NOM chlorination using fluorescence: formation of disinfection byproducts in Alento River water. Water Sci. Technol.: Water Supply 2004, 4 (4), 227-233.
Received for review March 2, 2007. Revised manuscript received June 11, 2007. Accepted June 22, 2007. ES0705386
VOL. 41, NO. 17, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6065
