Article pubs.acs.org/est
Relative Importance of N‑Nitrosodimethylamine Compared to Total N‑Nitrosamines in Drinking Waters Ning Dai and William A. Mitch* Department of Chemical and Environmental Engineering, Yale University, Mason Lab 313b, 9 Hillhouse Ave., New Haven, Connecticut 06520, United States S Supporting Information *
ABSTRACT: A U.S.-wide occurrence survey conducted as part of the Unregulated Contaminant Monitoring Rule 2 found that N-nitrosodimethylamine (NDMA) was present in 34% of chloraminated drinking water samples but was the most prevalent of the six N-nitrosamines evaluated using U.S. Environmental Protection Agency (EPA) Method 521. If the U.S. EPA considers limiting exposures to N-nitrosamines as a group, a critical question is whether NDMA is the most prevalent N-nitrosamine or whether significant concentrations occur for N-nitrosamines other than those captured by EPA Method 521. A total N-nitrosamine assay was developed and applied to 36 drinking water plant effluents or distribution system samples from 11 utilities, including 9 utilities that practiced chloramination for secondary disinfection. Concurrent application of EPA Method 521 indicated that NDMA was the most prevalent of the Method 521 N-nitrosamines yet accounted for ∼5% of the total Nnitrosamine pool on a median basis. Among eight plant influent waters, NDMA was detected once, while total N-nitrosamines were detected in five samples, suggesting the importance of source water protection. Similar to NDMA, total N-nitrosamine concentrations in source waters increased after chloramination. Chloramines were applied to organic precursors serving as models for pristine natural organic matter, algal exudate, wastewater effluent, and polyDADMAC quaternary amine-based coagulation polymers. While high yields of NDMA were restricted to the wastewater effluent and polyDADMAC, high yields of total N-nitrosamines were observed from the algal exudate, the wastewater effluent, and polyDADMAC. The results suggest that N-nitrosamines as a class may be more prevalent than suggested by occurrence surveys conducted using EPA Method 521.
■
INTRODUCTION
optimization to minimize the overall risk associated with exposure to the resulting disinfection byproduct mixtures.7 N-Nitrosamines have raised particular concerns. The formation of some N-nitrosamines has been associated with chloramination.8−10 Therefore, their concentrations in tap waters may increase with the increasing use of chloramination for secondary disinfection. Additionally, the declining availability of pristine water supplies has forced utilities to exploit lowergrade waters impaired by wastewater effluents or algal blooms. Compared to pristine waters, such waters feature higher dissolved organic nitrogen concentrations11 that may serve as precursors for nitrogenous disinfection byproducts (N-DBPs), including N-nitrosamines. Toxicology studies have indicated that, among potential carcinogens, N-nitrosamines are particularly potent, with the U.S. EPA’s Integrated Risk Information System (IRIS) database indicating that nanogram per liter drinking water concentrations of N-nitrosamines were associated with 10−6 lifetime cancer risks.12 As a result, the California Department of Health Services has established a 10 ng/L
Disinfection of drinking water has significantly reduced risks associated with pathogen exposures over the past century. However, the discovery during the 1970s that chlorine, the predominant disinfectant in the United States at that time, reacts with natural organic matter (NOM) in water supplies to form halogenated byproducts (e.g., trihalomethanes and haloacetic acids)1 raised concerns about the chronic exposures to these potential carcinogens. Epidemiology studies have since suggested a significant association between consumption of chlorinated drinking water and bladder cancer risk2 and potential associations with adverse reproductive outcomes.3 The Disinfection Byproduct Rules established by the U.S. Environmental Protection Agency (U.S. EPA) attempt to reduce these risks by limiting the concentrations of trihalomethanes and haloacetic acids in distribution systems.4 Because these byproducts are associated with chlorination rather than chloramination,5 utilities increasingly are experimenting with new disinfectant combinations combining strong preoxidants (e.g., chlorine, ozone, chlorine dioxide, or UV) with chloramines for maintenance of a disinfectant residual in the distribution system.6 Unfortunately, each of these disinfectants has associated disinfection byproducts, such that their combination may require © 2013 American Chemical Society
Received: Revised: Accepted: Published: 3648
December 20, 2012 March 4, 2013 March 18, 2013 March 18, 2013 dx.doi.org/10.1021/es305225b | Environ. Sci. Technol. 2013, 47, 3648−3656
Environmental Science & Technology
Article
applies to other constituents of the total N-nitrosamine pool. A second objective is to apply a chloramine formation potential assay to a range of representative organic precursor pools to compare their propensity to form total N-nitrosamines.
notification level for N-nitrosodimethylamine, N-nitrosodiethylamine, and N-nitrosodipropylamine.13 To begin to evaluate their nationwide occurrence, the U.S. EPA developed Method 521, covering seven of the eight N-nitrosamines on their IRIS database (i.e., N-nitrosodimethylamine, N-nitrosomethylethylamine, N-nitrosodiethylamine, N-nitrosodipropylamine, Nnitrosodibutylamine, N-nitrosopyrrolidine, and N-nitrosopiperidine).14 Having listed five N-nitrosamines in the Candidate Contaminant List 3,15 the U.S. EPA is considering promulgation of regulations to limit N-nitrosamine concentrations in tap waters. A nationwide survey of six of the N-nitrosamines measured by Method 521 (i.e., excluding N-nitrosopiperidine), conducted as part of the Unregulated Contaminant Monitoring Rule 2 (UCMR2)16 found that N-nitrosodimethylamine (NDMA) was highly associated with chloramination, occurring in 34% of chloraminated distribution system samples at a median concentration of 4.1 ng/L where detected.10 NDMA was also the most prevalent N-nitrosamine, with the other five N-nitrosamines rarely detected. The U.S. EPA’s IRIS database indicates that N-nitrosamines ranging from hydrophilic N-nitrosodiethanolamine to hydrophobic N-nitrosodibutylamine exhibit comparable cancer potencies.12 If the U.S. EPA seeks to limit exposures to Nnitrosamines as a group, the size of the total N-nitrosamine pool is an important question. The rare detection during UCMR2 of Method 521 N-nitrosamines other than NDMA would seem to suggest that NDMA dominants the total N-nitrosamine pool. If true, regulating NDMA concentrations in drinking waters would have a similar effect as regulating all N-nitrosamines to limit lifetime cancer risks associated with exposure to these compounds in drinking waters. However, it is unclear whether the N-nitrosamines targeted for detection in Method 521 are important components of the total N-nitrosamine pool. The first objective of this study is to evaluate whether the seven specific N-nitrosamines measured by Method 521 constitute a significant fraction of the total pool of N-nitrosamines. A total Nnitrosamine (TONO) method developed for quantifying the total concentration of N-nitrosamines in recreational waters17 was adapted to the analysis of drinking waters. The method was applied to a range of drinking water plant influents, effluents, and distribution system samples. Simultaneous analysis of the seven specific N-nitrosamines by Method 521 permitted an evaluation of the fraction of the total pool of N-nitrosamines constituted by Method 521 N-nitrosamines. Though the determination of the absolute concentrations of total N-nitrosamines was hindered by the range in recoveries of model N-nitrosamines under the current method, the results demonstrate that NDMA, the dominant Method 521 N-nitrosamine, was only a minor component of the total N-nitrosamine pool. NDMA formation generally has been associated with the use of quaternary amine-based anion exchange resins18,19 or cationic coagulant polymers (e.g., polydiallyldimethylammonium chloride (polyDADMAC))20,21 or with the use of wastewaterimpacted source waters.9,22,23 Specific precursors in wastewater-impacted water supplies have not been characterized but may include tertiary amine-based pharmaceuticals24 or quaternary amine-based constituents of shampoos and other personal care products.25 Algal-impacted water supplies were much less prone to NDMA formation.26−28 While these results may suggest that chloraminating utilities exploiting pristine or algalimpacted waters would be less concerned with NDMA formation, provided they do not employ cationic polymers or anion-exchange resins, it is unclear whether a similar association
■
MATERIALS AND METHODS Chemicals. Acros sulfamic acid (99%), sodium nitrite (98.5%), potassium iodide (99%), iodine resublimed and sulfanilamide (98%), and ammonium chloride; Baker sodium phosphate monobasic monohydrate, sodium phosphate dibasic, sulfuric acid (95.8%), sodium hydroxide (98.5%), glacial acetic acid (99.9%), methylene chloride (99.8%), magnesium sulfate (99%), and 2-propanol (reagent grade); Alpha Aesar ascorbic acid (99%); Fisher sodium hypochlorite solution; Pfaltz & Bauer mercuric chloride; Fluka hydrochloric acid (1 N); Chem Service N-nitrosodimethylamine (99.5%); Sigma−Aldrich N-nitrosodiethanolamine (99.8%); Macron ethyl acetate (99.9%); and Accustandard EPA Method 521 nitrosamine mix were used as received. UCT Enviro-Clean cartridges were used for EPA Method 521 analysis for specific nitrosamines. Sample Collection and Analysis. Grab samples were collected from a range of drinking water and municipal wastewater treatment plants and shipped in fluorinated high density polyethylene containers overnight to the laboratory at 4 °C. If collected downstream of disinfectant application, sample collection bottles contained 33 mg/L ascorbic acid to quench residual disinfectants. Nitrite was measured by the N-(1naphthyl)ethylenediamine dihydrochloride colorimetric method,29 and specific N-nitrosamines were measured by EPA Method 521.14 The concentrations of EPA Method 521 Nnitrosamines were below the detection limit (2 ng/L) in deionized water blanks. The total N-nitrosamine (TONO) analysis was adapted from a previously reported method for recreational waters.17 Briefly, evaluation of a range of structurally similar model compounds indicated that only N-nitrosamines, nitrite, and S-nitrosothiols exhibit a response from this method. In previous research, we had demonstrated that even low micrograms of nitrogen per liter concentrations of nitrite can produce N-nitrosamine artifacts via reaction with amine precursors during sample extraction.17 Accordingly, nitrite was reduced to nitrogen gas by application of 2 g/L sulfamic acid to 1 L water samples, adjustment of sample pH to 2, and storage overnight;30 for most samples, the sulfamic acid dose was sufficient to reduce sample pH to 2 directly. Samples were readjusted to pH 7 with sodium hydroxide and 5 mM phosphate buffer and extracted by continuous liquid−liquid extraction with 400 mL of ethyl acetate for 24 h; previous research indicated that this extraction was capable of extracting N-nitrosamines ranging in polarity from N-nitrosodiethanolamine to N-nitrosodibutylamine.17,30 To avoid interference from contamination, extraction glassware was thoroughly rinsed with deionized water and methanol prior to each use, and the concentration flasks were baked at 400 °C overnight in a muffle furnace. Additionally, if the extraction glassware had not been used the previous night or if the previous sample featured high concentrations of total N-nitrosamines, extractions of deionized water (1 L) were conducted overnight to clean the extraction glassware. A blank control of deionized water was subsequently extracted, prior to the extraction of authentic water samples. The ethyl acetate extracts of the blank control and authentic water samples were concentrated to 1 mL by rotary evaporation and blowdown under nitrogen gas. To improve solubility in subsequent steps, the 1 mL concentrate was supplemented 3649
dx.doi.org/10.1021/es305225b | Environ. Sci. Technol. 2013, 47, 3648−3656
Environmental Science & Technology
Article
reporting limit of 0.5 nM (37 ng/L as NDMA) was used hereafter. Total N-nitrosamine concentrations in deionized water blanks were below the 0.5 nM reporting limit. The detection limit calculation was referenced to NDMA, because it was the nitrosamine spiked into the tap water aliquots and because it was spiked into the reaction chamber to generate standard curves on the chemiluminescence detector. Additionally, it is the N-nitrosamine of highest current concern, and so, it serves as a benchmark against which to compare concentrations of total N-nitrosamines. As indicated above, previous work with an array of structurally similar model compounds indicated that the method was specific to N-nitrosamines, after pretreatments to eliminate nitrite and Snitrosothiols.17 However, pending complete characterization of all the specific N-nitrosamine constituents of the total Nnitrosamine signal, there remains some uncertainty regarding whether unanticipated water constituents other than N-nitrosamines contribute to the total N-nitrosamine signal. Nnitrosamines tend to be subject to UV photolysis.30 As a partial validation that the total N-nitrosamine assay truly captured Nnitrosamines, a municipal wastewater sample collected upstream of any disinfectant application from the secondary clarifier at a Connecticut utility employing nitrification and denitrification, was exposed to 1500 mJ/cm2 of incident UV fluence from a semicollimated beam apparatus containing low pressure mercury bulbs, described previously.32 As a result of UV exposure, the total N-nitrosamine signal dropped by 97% from 34 (±0.3) nM to 1.0 (±0.1) nM. Municipal wastewater was used here to represent a challenge condition, because it likely contains a wide array of potentially interfering constituents compared to drinking waters. The conversion efficiency of various N-nitrosamines to nitric oxide in the reaction chamber was comparable to that of NDMA, with the exception of N-nitrosodiethanolamine (75% of that of NDMA).17 The largest uncertainty of the current total Nnitrosamine assay originates from the variation in extraction efficiencies for various model N-nitrosamines. Extraction with ethyl acetate was pursued in the current method to permit partial capture of highly polar N-nitrosamines (e.g., N-nitrosodiethanolamine), which pass through common solid-phase extraction cartridges (e.g., Method 521 cartridges).17,30 Further method development focusing on sample extraction and concentration is needed to reduce the uncertainty of the total N-nitrosamine assay. Despite the uncertainty in the quantification of total Nnitrosamines resulting from the uncertainty in extraction efficiency, the results permitted significant conclusions regarding the fraction of the total N-nitrosamine pool constituted by Method 521 N-nitrosamines. Chloramine Formation Potential Assay. To evaluate the propensity of different organic precursor sources to form total Nnitrosamines, a chloramine formation potential assay was applied to an array of organic precursor model solutions. A stock solution of Elliott Soil humic acid, obtained from the International Humic Substances Society, was formed by dissolving 1 g/L humic acid in 1 g/L sodium hydroxide, stirring the solution overnight, and filtering through a 0.7 μm borosilicate glass fiber filter. An algal exudate stock solution was obtained by culturing Neochloris oleabundans (UTEX no. 1185) in modified Bold 3N medium at room temperature. The culture was harvested after 21 days, in the initial stationary phase. The algae solution was centrifuged at 2000 rpm for 15 min, and the supernatant was filtered through a 0.7 μm borosilicate glass fiber filter. A sample of a 10% (w/w)
with 1 mL of 2-propanol, and the mixture was further concentrated to 0.4 mL (2500-fold overall concentration factor) under nitrogen gas blowdown. The extract was treated with 40 μL of 20 g/L HgCl2 for 30 min in the dark to destroy potential interference from S-nitrosothiols and then with 40 μL of 50 g/L sulfanilamide in 1 N HCl for 15 min to destroy any nitrite formed via destruction of Snitrosothiols. The two nitrite destruction steps enabled initial sample concentration prior to treatment with HgCl2, thereby minimizing the use of HgCl2. The treated extract (100 μL) was injected into a heated (80 °C) reaction chamber containing 55 mL of glacial acetic acid and 4 mL of tri-iodide reducing solution (540 g/L potassium iodide and 114 g/L iodine), where fragmentation of the N−N bond in N-nitrosamines releases nitric oxide (NO). Continuous bubbling of the reaction chamber with nitrogen gas purges NO from the chamber through a base trap (1 N sodium hydroxide) and into a chemiluminescence detector (EcoPhysics CLD 88sp). Reaction of NO with ozone in the chemiluminescence chamber forms excited-state nitrogen dioxide (NO2). Photons emitted during relaxation of NO2 are recorded. The standard curve for total N-nitrosamines was constructed against NDMA standards for the chemiluminescence analysis. In previous research using an array of model nitrogen-containing compounds, only nitrite, S-nitrosothiols, and N-nitrosamines provided a response with this method.17,30 Among N-nitrosamines, N-nitrosomorpholine, N-nitrosodiphenylamine, N-nitrosodibutylamine, 1-nitro-4-nitrosopiperazine, 1,4-dinitrosopiperazine, and the N-nitrosamines in an EPA Method 521 standard mix gave signal responses within 10% of that of NDMA; the signal responses reflect the efficiency associated with cleavage of the N−N bond to liberate NO by the acidic tri-iodide solution. However, the signal response for Nnitrosodiethanolamine and 1-nitrosopiperazine were lower, at 75% and 66%, respectively, of that of NDMA. Anticipating lower total N-nitrosamine concentrations than observed previously in recreational waters, the sample concentration factor was increased to 2500-fold compared to the previous method for recreational waters (1000-fold).17 Recoveries of model N-nitrosamines under this higher concentration factor were re-evaluated by comparing the signal response for 1 L deionized water samples buffered at pH 7, spiked with model N-nitrosamines, and extracted and analyzed as described above to the signal response from the same mass of model N-nitrosamines injected directly to the reaction chamber. The change resulted in reduced recoveries of model Nnitrosamines. For example, recoveries were reduced from 37% to 24% (±2.7% standard deviation; n = 7) for NDMA, from 39% to 19% (±0.3% standard deviation; n = 2) for N-nitrosodiethanolamine, and from 75% to 52% (±7.9% standard deviation; n = 2) for a 1.2 nM total N-nitrosamine concentration from a Method 521 standard mixture.17 However, the increased concentration factor more than offset the lower recoveries to improve the method detection limit (see below). The method detection limit of the total N-nitrosamine assay was determined according to standard protocol.31 Seven aliquots of tap water samples (1 L) were spiked with 1.2 nM NDMA, treated with sulfamic acid, extracted, and analyzed as described above over a period of 5 days. The total N-nitrosamine concentration in the 0.4 mL extracts was 0.73 (±0.08) μM. With a 2500-fold concentration factor, NDMA recovery was determined to be 24%. The standard deviation of the seven replicates was used to derive a method detection limit of 0.43 nM,31 using the 24% extraction efficiency relevant to NDMA. A 3650
dx.doi.org/10.1021/es305225b | Environ. Sci. Technol. 2013, 47, 3648−3656
Environmental Science & Technology
Article
Table 1. NDMA and Total N-Nitrosamines in Distribution System Samples NDMAa
utility
EPA region
A
5
August
B
5
October
C D
3 5
December December
month
May
E
9
January
April
F
4
March
G
5
March
II
9
April
I
10
April
I
9
April
K
4
August
disinfection scheme
category distribution distribution distribution distribution distribution distribution plant effluent plant effluent distribution distribution influent distribution distribution distribution distribution influent plant effluent distribution distribution influent distribution distribution distribution distribution influent distribution distribution distribution distribution influent distribution distribution influent distribution influent distribution distribution distribution distribution distribution distribution influent distribution distribution
Cl2 O3/Cl2
Cl2/NH2Cl Cl2/NH2Cl
Cl2/NH2Cl
ClO2/NH2Cl
O3/NH2Cl
Cl3/NH2Cl NH2Cl/O3/Cl2/ NH2Cl Cl2/NH2Cl
Cl2/NH2Cl
countd minimum median 75th percentile maximum
total N-nitrosaminesab (nM)
molar fraction of NDMA in total N-nitrosamines (%) NAc NA