Article pubs.acs.org/est
Trade-Offs in Disinfection Byproduct Formation Associated with Precursor Preoxidation for Control of N-Nitrosodimethylamine Formation Amisha D. Shah,† Stuart W. Krasner,‡ Chih Fen Tiffany Lee,‡ Urs von Gunten,§,⊥ and William A. Mitch*,† †
Department of Chemical and Environmental Engineering, Yale University, Mason Lab 313b, 9 Hillhouse Avenue, New Haven, Connecticut 06520, United States ‡ Metropolitan Water District of Southern California, Water Quality, 700 Moreno Avenue, La Verne, California 91750, United States § Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611, CH-8600 Duebendorf, Switzerland ⊥ School of Architecture, Civil and Environmental Engineering (ENAC), Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland S Supporting Information *
ABSTRACT: Chloramines in drinking water may form N-nitrosodimethylamine (NDMA). Various primary disinfectants can deactivate NDMA precursors prior to chloramination. However, they promote the formation of other byproducts. This study compared the reduction in NDMA formation due to chlorine, ozone, chlorine dioxide, and UV over oxidant exposures relevant to Giardia control coupled with postchloramination under conditions relevant to drinking water practice. Ten waters impacted by treated wastewater, poly(diallyldimethylammonium chloride) (polyDADMAC) polymer, or anion exchange resin were examined. Ozone reduced NDMA formation by 50% at exposures as low as 0.4 mg×min/L. A similar reduction in NDMA formation by chlorination required ∼60 mg×min/L exposure. However, for some waters, chlorination actually increased NDMA formation at lower exposures. Chlorine dioxide typically had limited efficacy regarding NDMA precursor destruction; moreover, it increased NDMA formation in some cases. UV decreased NDMA formation by ∼30% at fluences >500 mJ/cm2, levels relevant to advanced oxidation. For the selected pretreatment oxidant exposures, concentrations of regulated trihalomethanes, haloacetic acids, bromate, and chlorite typically remained below current regulatory levels. Chloropicrin and trichloroacetaldehyde formation were increased by preozonation or medium pressure UV followed by postchloramination. Among preoxidants, ozone achieved the greatest reduction in NDMA formation at the lowest oxidant exposure associated with each disinfectant. Accordingly, preozonation may inhibit NDMA formation with minimal risk of promotion of other byproducts. Bromide >500 μg/L generally increased NDMA formation during chloramination. Higher temperatures increased NDMA precursor destruction by preoxidants but also increased NDMA formation during postchloramination. The net effect of these opposing trends on NDMA formation was water-specific.
■
INTRODUCTION Utilities in the United States have been exploring different disinfection/oxidation scenarios to minimize the formation of trihalomethanes (THMs) and haloacetic acids (HAAs), while achieving pathogen and micropollutant removal. Of particular interest is the increasing reliance on chloramination for secondary disinfection and the use of alternative primary disinfectants.1 Unfortunately, formation of N-nitrosodimethylamine (NDMA) has been associated with chloramination.2 A drinking water concentration of 0.7 ng/L NDMA is associated with a 10−6 lifetime excess cancer risk.3 Together with several other N-nitrosamines, NDMA has a 10 ng/L notification level in California,4 a 40 ng/L maximum acceptable concentration in Canada,5 was listed in the USEPA’s Contaminant Candidate List 3,6 and is being considered for federal regulation in the © 2012 American Chemical Society
U.S. The World Health Organization lists a 100 ng/L guideline value for NDMA.7 Substantial progress has been made delineating the formation pathways for nitrosamines and other nitrogenous disinfection byproducts during disinfection.8 Research has suggested three NDMA formation pathways that are potentially important during drinking water disinfection. The most important mechanism involves the slow formation during chloramination,9−11 arising from the reaction of inorganic dichloramine with amine precursors via a chlorinated unsymmetrical Received: Revised: Accepted: Published: 4809
January 3, 2012 March 21, 2012 March 30, 2012 March 30, 2012 dx.doi.org/10.1021/es204717j | Environ. Sci. Technol. 2012, 46, 4809−4818
Environmental Science & Technology
Article
Table 1. Water Quality Information for Tested Waters water
location
Wastewater-Impacted A nitrified wastewater effluent B ozone contactor influent C filtered raw water D softened water recarbonation effluent E filtered raw water F filtered raw water PolyDADMAC-Impacted G ozone contactor influent H pilot plant filter effluent I filter effluent Anion Exchange-Impacted J anion exchange effluent
pH
TOC mg/L
Br‑ μg/L
NH4+-N mg/L
NO2−-N μg/L
NO3−-N mg/L
UV254 cm‑1
SUVA L (mg*m)‑1
NDMAFP ng/L
NDMAa ng/L
6.7 6.0 8.3 8.0−8.5c
3.3 1.9 5.0 5.4
310 40 170 20
NAb ND 0.06 0.05
290 ND 13 56
0.7 2.0 0.6 2.7
0.094 0.026 0.084 0.089
2.84 1.36 1.68 1.65
NA 64 28 59
9 18 25 14
7.3 7.1
4.9 3.9
50 61
0.12 0.08
31 43
1.1 1.4
0.138 0.086
2.82 2.21
35 NA
13 10
6.7 8.0 7.2
3.3 2.5 3.0
50 80 246
NA NA 0.06
ND ND 9
NA 0.4 0.6
0.066 0.038 0.048
2.02 1.52 1.60
74 57 44
40 20 6.4
7.0
ND
40
ND
ND
0.6
ND
NA
6
3.5
a
NDMA formed for 0 mg×min/L prechlorine exposure at room temperature. bNA = not analyzed. ND = not detected. cpH adjusted to this range prior to preoxidation.
dimethylhydrazine intermediate (yield ∼2%).12 Although less prevalent than monochloramine under typical chloramination conditions, dichloramine always coexists according to the equilibrium: 2 NH2Cl + H+ ↔ NHCl2 + NH4+. Nitrosamines can also form when chlorination in the presence of nitrite forms the nitrosating agent, dinitrogen tetraoxide (N2O4).13,14 Lastly, NDMA can form during ozonation of N,N-dimethylsulfamide, a metabolite of the fungicide tolylfluanid (yield ∼50%),15 certain nitrogen-containing pharmaceuticals,16 dyes,17 and dimethylamine itself.18 While most research has employed dimethylamine as a model precursor,9,10,12−14 further work indicated that tertiary amines could form NDMA at yields comparable to that from dimethylamine,19,20 while yields from quaternary amines were much lower.21−23 Formation potential (FP) analyses revealed that treated wastewaters exhibit NDMA precursor concentrations far higher than that for pristine or algal-impacted waters.19,24−27 With dimethylamine concentrations in drinking waters and wastewaters generally 2 ng/L in waters E and F (from an effluent-impacted river) at concentrations of ∼8 ng/L prior to treatment. N-Nitrosomorpholine is a commonly detected nitrosamine in wastewater treatment plant effluents, but is not a DBP per se.49,50 With the exception of reduction associated with LP or MP UV photolysis in water F, Nnitrosomorpholine concentrations did not change with treatment. Although N-nitrosomorpholine formation was significant 4812
dx.doi.org/10.1021/es204717j | Environ. Sci. Technol. 2012, 46, 4809−4818
Environmental Science & Technology
Article
results suggested a role for the chlorine−nitrite reaction pathway. For the other three waters, nitrite concentrations were significantly lower (i.e., 31−56 μg N/L; Table 1). Spiking of water E with 35−140 μg N/L nitrite did not significantly increase NDMA formation for the 6.4 mg×min/L prechlorination exposure. These results suggest that the chlorine−nitrite pathway was not important for this water, perhaps because of lower concentrations of nitrite and NDMA precursors expected for this wastewater-impacted raw water, compared to water A, a nitrified wastewater effluent. For these three waters, it is likely that the increase in NDMA formation is attributable to conversion of NDMA precursors to more potent forms at low exposure, an effect also observed by Chen and Valentine.31 NDMA formation declined at higher prechlorine exposure likely because of deactivation even of these more potent NDMA precursors. Preoxidation with chlorine dioxide increased NDMA formation with increasing exposure for waters D, I, and J (Figure SI-4, Supporting Information); these waters were impacted by wastewater, polyDADMAC, or anion exchange resin-related precursors, respectively. Unlike pretreatment with chlorine, NDMA formation did not decline at higher exposure. It is likely that pretreatment with chlorine dioxide converted precursors to more potent forms, but the nature of these reactions is unclear and may be utility-specific. Comparison of Preoxidant Efficacy. To compare the ability of preoxidants to control NDMA formation, the NDMA concentrations following postchloramination for various preoxidant exposures were normalized to that formed at 0 exposure (i.e., C/C0) and then were divided into exposure bins compared to the highest exposure relevant to Giardia control. For example, for chlorine, where the maximum exposure was taken to be 140 mg×min/L, the 0.2−0.4 exposure/exposuremax bin refers to exposures between 28 and 56 mg×min/L. The results for different waters (nine waters for chlorination, eight waters for ozonation, eight waters for chlorine dioxide, six waters for MP UV) in these bins were averaged (Figure 3). Ozone achieved the greatest reduction in NDMA at the lowest exposure/exposuremax, although the performance of chlorine was comparable at higher exposure. This figure incorporated data for waters where pretreatment with chlorine or chlorine dioxide increased NDMA formation. However, excluding data for waters A, D, E, and F, where prechlorine treatment at low exposure increased NDMA formation, the C/C0 value for the 0−0.2 exposure/exposuremax bin would drop from 1.2 (±0.6 standard deviation) to 0.8 (±0.2 standard deviation), still above the 0.5 (±0.2 standard deviation) observed for preozone treatment for this exposure/exposuremax bin. Pretreatment with chlorine dioxide achieved negligible reduction, but, as indicated by the wide error bars, these results incorporate data from waters D and I (although not J), where pretreatment increased NDMA formation. Excluding these data, the C/C0 value for the 0.6−0.9 exposure/ exposuremax bin would drop from 1.2 (±0.8 standard deviation) to 0.8 (±0.6 standard deviation), higher than seen for chlorine or ozone. For the limited fluences that were evaluated for MP UV, NDMA precursor deactivation was negligible at low fluence, but some precursor destruction was observed at high fluence. Although UV was less effective than ozone and chlorine, it is important to account for it, as installation of UV is expected to increase for disinfection, and higher fluences
concentration promoted the formation of brominated THMs, increasing the overall concentration of the four regulated THMs. The native pH of the water was relatively high (8.3). High pH promotes the formation of THMs during chlorination, and bromate during ozonation. Even for water C, several factors reduce the risk of exceeding DBP regulatory limits. Water C was a laboratory-filtered raw water. Although some utilities apply oxidants to raw waters (as this plant did), a portion of the organic DBP precursors would likely be removed during treatment (e.g., coagulation/filtration). Moreover, the water was treated at room temperature. Formation of other DBPs can be reduced at lower temperatures, and regulatory limits for DBPs such as THMs are based on running annual average values. Preoxidation-Induced Increases in NDMA Formation. Preoxidation with chlorine at low exposure followed by chloramination increased NDMA formation for wastewaterimpacted waters A, D, E, and F (E and F represent the same wastewater-impacted river), representing 33% of the waters sampled in this study. Figure 2 provides NDMA formation as a
Figure 2. NDMA formation versus prechlorine CT and postchloramination for waters A and E.
function of prechlorination exposure for waters A and E. In both cases, NDMA formation increased significantly at the lowest exposure, representing a 3 min contact time with chlorine before ammonia addition but declined at higher exposures. All of these waters contained nitrite (Table 1). It was hypothesized that NDMA formation could have arisen from the reaction of chlorine with nitrite to form N2O4.13,14 For water A, a nitrified municipal wastewater effluent with 290 μg N/L nitrite and pH 6.7, NDMA formation increased from 9 to 28 ng/L as exposure increased from 0 to 4.9 mg×min/L. As noted in Materials and Methods, the 0 exposure sample actually involved a 30 s free chlorine contact time at pH ∼8 prior to ammonia addition (i.e., chloramines were formed by addition of chlorine first), whereas the 4.9 mg×min/L exposure sample was chlorinated at pH 6.7 for 3 min before adjustment to pH ∼8 for postchloramination. Water A was stored until the nitrite concentration decayed to a nondetectable level. Repetition of the 0 exposure test at either pH 6.7 or 8 resulted in 5 ng/L of NDMA. Spiking of 280 μg N/L nitrite and evaluation of the 0 exposure condition at pH 6.7 formed 25 ng/L NDMA but only 9 ng/L at pH 8.2. As the pH increased from 6.7 to 8.2, the free chlorine speciation changes to the less reactive OCl−. These 4813
dx.doi.org/10.1021/es204717j | Environ. Sci. Technol. 2012, 46, 4809−4818
Environmental Science & Technology
Article
(i.e., preoxidation at the native sample pH) suggest that preozonation exhibits the greatest efficacy. The high efficiency of ozone pretreatment should enable ozone to control NDMA formation while minimizing the promotion of other DBPs. For each preoxidant, Table 2 Table 2. Formation of Other DBPs for Exposure/ Exposuremax = 0.60−0.95
Pre-Chlorine THMsa HAAsb trichloroacetaldehyde Pre-Ozone bromatec chloropicrin trichloroacetaldehyde Pre-Chlorine Dioxide chlorited Pre-MP UV chloropicrin trichloroacetaldehyde
Figure 3. Reduction in postchloramination NDMA concentrations for preoxidation exposure compared to no pretreatment. Pretreatment exposures expressed as a percentage of the range relevant to Giardia control: 140 mg×min/L for chlorine, 2 mg×min/L for ozone, 25 mg×min/L for chlorine dioxide, and 200 mJ/cm2 germicidal fluence for MP UV treatment. X-axis values reflect the upper end of bins combining data for different preoxidant exposures for each water. Error bars reflect one standard deviation for results from different waters. Reprinted with permission from Krasner et al. Development of a Protocol to Predict the Formation of Nitrosamines; Water Research Foundation: Denver, CO, in press.
minimum μg/L
median μg/L
maximum μg/L
n
6 12
