OCCURRENCE AND FORMATION OF N-DBPS IN RHODE ISLAND DRINKING WATERS

N-nitrosamines are toxic compounds that have persistently been associated with water treatment processes since the 1970’s. There are currently no federal regulations for N-nitrosamines in drinking water, however few states have established their own guidelines. Many studies have identified major mechanisms of N-nitrosamine formation during water treatment, however a gap in knowledge still exists regarding the formation of select N-nitrosamines from treatment of clean water sources. Performance of such critical research is often an expensive process, leaving many facilities and institutions resorting to other approaches for analysis. In the case of this research, efforts were made to develop a lower-cost, and widely applicable method for N-nitrosamine analysis utilizing standard liquid-liquid extraction techniques, coupled with common GCMS analytics. This study also focused on identifying the formation potential of select N-nitrosamines during treatment of seasonally and spatially varying source water, using a bench-top water treatment system. Results from the method development section show that the method was capable of detecting 9 target Nnitrosamines at a concentration of 2 μg/L, suggesting that this method could be applied to N-nitrosamine formation pathway studies. To perform N-nitrosamine analysis in the water treatment study, a lower limit of detection was required, therefore analysis was outsourced to a laboratory at Kagoshima University, Japan. Results from the water treatment section show that the system design did not reduce the likelihood of forming select N-nitrosamines during pre-treatment, and the formation of those N-nitrosamines was significantly dependent on factors such as disinfection contact time, precursors, and source water type. All results from this research will supplement the science of previous N-nitrosamine studies, and promote future N-nitrosamine research as it relates to water treatment.

ix LIST OF TABLE  LIST OF TABLE  LIST OF TABLE  LIST OF TABLES S S S Table 3. ANOVA table comparing   however, the exact processes and environments in which nitrosamines form are still not well understood. Organic nitrogen precursors react within the WTP and distribution system, forming the toxic by-products during chloramination, or while in distribution. To best control the formation potential of nitrosamines, precursors must be removed from source water prior to chloramine disinfection.
These nitrosamine forming precursors are abundant in source waters worldwide, presenting a need for further study of the mechanisms that reduce the formation potential of nitrosamines in chloramination WTPs.  [2]. The degree of influence on formation of nitrosamines varies between all WTP processes.
Chloramine disinfection is the most important pathway for nitrosamine formation [4]. Findings from [3,5]  Oxidation of Cl-UDMH by dissolved oxygen forms NDMA and other nitrosamines. This particular formation pathway has a slow reaction process, often days, indicating that nitrosamines continue formation and accumulation within a chloraminated distribution system [6,7].
3 Nitrosamine precursors in source water 3 Nitrosamine precursors in source water 3 Nitrosamine precursors in source water 3 Nitrosamine precursors in source waters s s s Source waters utilized for consumption are extremely influential to nitrosamine formation potential. Quality of source water is largely dependent on factors such as watershed and source water type. Seasonal variation is also known to have significant impact on disinfection byproducts [8]. Through proper assessment of source water before disinfection, WTPs can provide potable water while preventing the formation of nitrosamines.
Evaluation of nitrosamine formation potential begins with proper source water assessment. Watershed variation generates differences in precursor type and relating concentrations. Amines are expected to be the major nitrosamine forming precursor during chloramination [2]. Although the reaction time is much slower, amides are the other major category of organic nitrogen precursors [9].
Signatures of amine and amide precursors exist multiple watershed types, including urban and agricultural. Source water containing high concentrations of precursors is likely impaired by treated wastewater, industrial effluents, or herbicides diuron and dimethyldithiocarbamate [9][10][11][12][13].
Surface runoff enriched with heavy metals, nutrients and sediments, rubber fragments, and other contaminants is an essential source of non-point source pollution to receiving water bodies such as drinking water reservoirs [14,15]. Forested watersheds naturally offer more protection to source water, rather than urban or agricultural watersheds. Forested buffers located around a reservoir system limit the direct influence of contaminated runoff on quality of source water. Buffer areas change the quantity of water available for runoff through interception, evapotranspiration, infiltration, percolation, and absorption, resulting in different physical, chemical, and biological processes in the receiving water bodies [14]. Removal of nitrosamines following drinking water treatment is a difficult task, as many nitrosamines are hydrophilic (log Kow = -0.57 for NDMA), and will poorly sorb to activated carbon, and other sorbents [7,16]. NDMA has a relatively high vapor pressure at 2.7 mm Hg at 20˚C [17]. The estimated Henry's Law constant for NDMA is low at 2.6 x 10-7 atm-m3/mol at 20˚C, due to the high water solubility of NDMA [16,18]. Due to the chemical and physical properties of NDMA, volatilization from air stripping during water treatment is unlikely to result in significant removal from solution [7].
Removal of nitrosamine precursors before disinfection is a vital process to control nitrosamine formation during drinking water treatment. Furthermore, nitrosamines will typically not be present in drinking waters treated by activated carbon prior to chloramination [2]. Sorption of precursors exposed to powdered activated carbon (PAC) at a dose of 5 mg/L for 7 days, showed 50% reduction of NDMA formation potential [19]. During the same study, water exposed to a PAC dose of 20 mg/L for 7 days produced an NDMA formation potential reduction of 90%. Water was in contact with PAC for 7 days to assure establishment of adsorption equilibrium, even though conventional treatment contact times typically last hours [19].
A study conducted by [20] demonstrated that by using granular activated carbon (GAC) to treat a mixture of 90% surface water and 10% wastewater at a 10-minute simulated empty bed contact time, NDMA formation potential breakthrough was less than 20% after 10,000 bed volumes. Also, GAC demonstrated 60-80% reduction of NDMA formation potential in surface waters during pilot-and full-scale studies [20].
5 Global occurrence of nitros 5 Global occurrence of nitros 5 Global occurrence of nitros 5 Global occurrence of nitrosamines amines amines amines The presence of nitrosamines is worldwide and relatively similar among all detection locations. Given the expectations from known formation pathways, North American studies found that NDMA formation is closely associated with chloramination than with chlorination [15,[21][22][23]. Water treatment plants with long disinfection chloramine contact times (12-18 hours) tended to have greater NDMA concentrations in the plant effluent than those with short (0.5-2 hours) contact times, due to the long time-scales of nitrosamine formation [24]. One large study collected drinking water samples under the second Unregulated Contaminants Monitoring Rule (UCMR2). NDMA was detected in 34% of chloramination plant effluents [23]. Other nitrosamines N-nitrosodiethylamine (NDEA), N-nitrosopyrrolidine (NPYR), N-nitrosodi-n-butylamine (NDBA), and Nnitrosomethylethylamine (NMEA) were also detected, but each at less than 1% occurrence [23].
[25] performed a nitrosamine occurrence study in England and Wales.
Out of 41 surveyed plants, only 3 had detectable concentrations of NDMA; however, the levels were always below 6 ng/L. Another UK study conducted by [26] found NDMA concentrations just above the method detection limit (0.9 ng/L) in a few isolated samples from one distribution system. WTP practices in the UK typically operate with a set 30 minute pre-chlorine contact time, and low chloramine disinfection dose (0.5 mg/L), explaining why such low NDMA concentrations are found in chloraminated drinking waters of the UK [26].
High nitrosamine occurrence was seen in Australia due to the high prevalence of chloramination WTPs. One study detected NDMA in 75% of chloraminated waters, where 37% of the detections had NDMA concentrations >10 ng/L [27]. Besides the high rate of chloramination WTPs, wastewater recycling, and high source water ammonia concentrations are accountable for such high levels of NDMA in drinking water in Australia [27,28].  to achieve target N-nitrosamine concentrations during analysis. 10 N laboratory grade sodium hydroxide (NaOH) and sodium sulfate (Na2SO4) anhydrous were also purchased from Fisher Scientific.

Sample preparation
1 L amber bottles were rinsed with DCM, then detergent washed, followed by rinsing with deionized water. Bottles were drained, then baked in a muffle furnace at 270˚C for three hours. After bottles were baked and cooled, the bottles and PTFE caps were rinsed with DCM and allowed to dry before use.
Milli-Q ultrapure deionized water was used as the sample matrix in this detection limit study to avoid signal interferences during analysis. Without prior rinsing with sample matrix, each 1 L sample bottle was filled to no head space then capped. Since the samples did not require to be stored for later processing, all filled sample bottles were immediately extracted.

Sample extraction
Individual Extract was transferred to a PTFE-lined screw cap vial using a clean borosilicate glass syringe then stored at 4˚C until analyzed.

Analytical techniques
N-nitrosamines from sample extracts were chromatographically separated and analyzed using a Shimadzu GC-MS -QP2010SE equipped with an AOC -20S auto sampler. The analytical column used for this experiment was a Restek®  (Table 1).

Chromatogram of the N-nitrosamines
The GC-MS chromatogram for the nine N-nitrosamines (2 µg/L) is shown in Fig. 1. All nine N-nitrosamines were completely separated, respectively. Other prominent peaks are shown in Fig. 1 and were identified as hydrocarbons.
Further evaluation is required to determine possible sources of the additional peaks.
Results show that the lower molecular weight compounds tend to have shorter retention times, with NDMA exhibiting the shortest and NDPhA the longest (Table 1). It is apparent that some of the N-nitrosamines, e.g., NDMA, NMEA, and NMOR, produced weak signals, therefore it can be declared that limit

Method evaluation
As previously mentioned, several of the target analytes produced weak signals at 2 µg/L, signifying LOD. Another important finding was the occurrence of coelution with other prominent peaks, and poor separation, e.g., NDEA, and NPYR. In Fig. 2, we saw poor separation of NDEA and a hydrocarbon, where in  Although some implications were present with peak separation, the proposed method was still capable of identifying other N-nitrosamines without any interferences. In fact, Fig. 1 demonstrates the potential for identification and quantification of all target analytes listed in Table 1. To avoid future interferences from unwanted chromatographic artifacts, it is recommended that all nonvolumetric experimental glassware be baked at a higher temperature (400 ˚C) for 1 hr. Should interferences still persist, an evaluation must be made on the quality of the experimental sample matrix.

Method application
Application of this method to quantify unknown levels of N-nitrosamines is possible. Further calibrations, and extractions are required to establish method detection limits (MDL) for each target analyte. After a preliminary assessment, 1 -experimental glassware is suggested. If unwanted compounds still persist, the sample matrix quality must be evaluated.
Application of this method in other research is feasible, particularly in formation potential studies where it is presumed high levels of N-nitrosamines will form from known reactants. In regards capturing ng/L levels of Nnitrosamines, this method will not perform effectively. Therefore, any studies targeting the occurrence of N-nitrosamines in natural water, treated water, or wastewater, should resort to lower detection limit analytical methods.
International  N-nitrosamines are a group of contaminants of emerging concern that may be present in drinking water as by-products from water treatment plant (WTP) operations [1-6]. Significant influences on the formation of N-nitrosamines in drinking water include source water impairment before treatment, e.g., industrial, wastewater, and septic system effluents [6,7]. Presence of N-nitrosamines in drinking water is of particular concern because of their carcinogenic, mutagenic, and teratogenic properties [8,9]. N-nitrosodimethylamine (NDMA, C2H6N2O), and several other N-nitrosamines are classified as probable carcinogens based on domestic and international assessments [10,11] Traditionally, formation of disinfection by-products (DBPs) such as trihalomethanes (THMs) and haloacetic acids (HAAs) can be controlled by precursor removal before disinfection [27][28][29]. Conventional WTP processes used for removing precursors include coagulation and flocculation, sedimentation, and filtration [30]. Other studies have been conducted to address the impact of using additional filtration techniques for controlling DBP precursors [31,32]. Little information is known about the impact of water treatment processes on potential NDMA precursor removal, and the relationship it has with NDMA formation following monochloramine disinfection.
To address this issue, we investigated the formation potential of NDMA in a modelled chloramination WTP fed by seasonally and spatially varying source water. To enhance the reduction of potential NDMA precursors, conventional water treatment methods were combined with additional filtration techniques.
This work will also address the impact of regional WTP operations and their potential to form NDMA under specific treatment scenarios. Results from this study will serve as the foundation for further NDMA research as it relates to drinking water treatment of regionally sourced waters. The field sites identified in Figure 1 were chosen for this study to demonstrate impact of spatial variability on source water quality. Both locations are headwater streams for major reservoirs in Newport and Scituate, Rhode Island. Source water entering Newport WTPs, e.g., Bailey Brook, has received higher loads of nutrients from a variety of sources due to its location in an urban area [33]. Cork Brook, although forested [34], has experienced seasonally high loadings of precursors, particularly during intense precipitation events. It is expected that the different land use activities associated with each field site will 37 produce varying levels of precursors [19,24,26], later affecting the formation of NDMA upon water treatment.

2.3 3 3 3 Sample Sample Sample Sample P P P Processing rocessing rocessing rocessing
Each field sample was processed two times using a benchtop water treatment system (Figure 2)  Pump Drive peristaltic pump was used to pump effluent from the dual-media column filter into a borosilicate glass column filled with 20-40 mesh granular activated carbon (GAC). Addition of the GAC column filter to the treatment system was to enhance precursor removal before disinfection. Both columns were selected to have an empty bed contact time (EBCT) of 10 min [31]. During each run of processing, two samples were collected in 500 mL amber glass jars with no headspace from raw water influent and post-filtration effluents, and one sample was collected in a 300 mL amber glass jar from the same influent and effluents. All processed samples were kept in refrigerator storage (4 ˚C) until disinfection. Figure 2. Figure 2. Figure 2. Figure 2. Flow-scheme of benchtop water treatment system.

2.4 4 4 4 Monoc Monoc Monoc Monochloramine hloramine hloramine hloramine D D D Disinfection isinfection isinfection isinfection
All 500 mL processed samples were removed from refrigeration and brought to room temperature (25 ˚C), and then pH was adjusted to between 9 and 10 using 1 N laboratory grade sodium hydroxide (NaOH). Pre-formed monochloramine (NH2Cl) stock solution was prepared from diluted solutions of sodium hypochlorite (NaOCl) and ammonium chloride (NH4Cl). The Cl2/N ratio was 1:1.2, and pH was adjusted to ≥8 with 1 N laboratory grade NaOH to prevent the decay of NH2Cl into dichloramine (NHCl2) and trichloramine (NCl3) from excess free chlorine, and low pH values [28,36] where the magnitude of any observation Yijkt can be affected by several possible influences. µ is the overall mean, αi is the influence of the i th category of the column variable, βj is the influence of the j th category of the column variable, and γk is the influence of the k th category of the column variable. Interaction effects from the combination of column variables are denoted by terms (αβ)ij, (αγ)ik, and (βγ)jk. The term (αβγ)ijk is called a three-way interaction term, and εijkt is the residual error term.
Significance level α = 0.05 was set for all calculations to define the probability of concluding that a difference between groups exists when there is no actual difference. Limitations of ANOVA excluded a fourth main effect from the analysis. Adjusting for this confine, separate ANOVA tables were generated to represent NDMA formed from treatment of the two source waters. This study was designed to model the formation of NDMA from treatment of local source waters using chloramine water treatment techniques.
The aim behind the study design was to demonstrate that if local WTPs used chloramine as a primary disinfectant, then the product water sent into distribution would contain low levels of NDMA. Seasonal and spatial variations were considered as possible effects, and precursor removal before disinfection was also considered. Based on previous NDMA formation potential studies, it was expected that after longer monochloramine CT higher concentrations of NDMA would form [18,21,22,40]. In this experiment, we found that only Cork Brook produced higher concentrations of NDMA during the first 24 hr of CT when compared to the 72 hr CT (Figure 3). Scavenging of NDMA precursors by other DBPs is a possibility given that residual chloramine decreased consistently over 72 hr; however, it is speculated that if scavenging were to occur, there would not be a spike of NDMA in the Cork Brook 24 hr CT samples [41].
There was a substantial difference in average NDMA formed between Cork Brook (12.2 ng/L) and Bailey Brook (4.2 ng/L) in the 24 hr CT samples ( Figure 3a): however, the difference between the two averages noticeably decreased in the 72 hr CT samples (Figure 3b). This effect could be explained by NDMA precursors being site-specific and influenced by several factors [42][43][44][45].
To determine the processes leading to NDMA reduction over longer CT, further studies are required. It is also noteworthy that the 72 hr CT samples still produced higher average levels of NDMA in Cork Brook (4.1 ng/L) than Bailey Brook (3.4 ng/L), suggesting that the precursors associated with NDMA formation are more frequently associated with forested areas rather than areas of urban influence [46][47][48][49][50].
In previous NDMA formation studies, known precursors were used as reactants with varying doses of monochloramine, resulting in increasing NDMA concentrations with respect to time [18,21,22,40].

Seasonal Seasonal Seasonal Seasonal P P P Precursor recursor recursor recursor P P P Presence resence resence resence
Seasonal influence on concentrations of TOC and TN was substantial ( Figure 4). However, TOC and TN concentrations had no apparent effect on the formation of NDMA ( Figure 5). This finding supports claims that no significant relationships exist between NDMA and dissolved organic carbon (DOC), natural organic matter (NOM), or TN, and provides further evidence that NDMA has a very complex formation pathway [40,[42][43][44]. The NDMA precursor fingerprinting study by [45] explained that certain aliphatic, as well as peptide and lipid-like compounds are responsible for the majority of NDMA formation in natural waters, and the origin of those constituents is likely from wastewater effluents.

The likelihood that both Cork Brook and Bailey Brook source water is being
impacted by wastewater effluent is low, therefore creating a need for future investigation into precursor identification at the selected field sites.

Precursor removal
In order to assess bench-top treatment system efficacy, precursor concentrations were quantified in samples collected from three main points in the system, e.g., raw water influent (R), post-dual-media filtration effluent (F), and post-dual-media filtration + GAC filtration effluent (GAC). At alpha level of 0.05, Figure 6 shows significant differences in precursors concentrations as source water passes through the treatment system. Also noted in Figure 6, there is a negative correlation between the precursors concentration and the place in the treatment system where the sample was collected, demonstrating that the bench-top system was effective at removing traditional DBP precursors [51,52].
The addition of a GAC column filter following dual-media filtration proved to be highly effective at reducing TOC and TN to concentrations below 0.25 mg/L. identify what particular group of precursors were driving the reactions that generated NDMA (Table 1). The most significant finding was that NDMA formation has a strong positive correlation with aliphatic, as well as peptide and lipid-like compounds [45]. Since the presence of the aforementioned group of precursors is frequently associated with wastewater impacts [45], other routes of exposure were considered.
The findings from [53,54]    NDMA formation has weak correlation to DOC content (R 2 = 0.41), however strength of correlation is source dependent. NDMA precursors are a suite of compounds associated with humic substances and other high molecular weight polymers [40] NDMA formation in water and wastewater Untreated natural water (U. S.) No significant relationship between NDMA formation and natural organic carbon or nitrogen [42] Survey of NDMA occurrence in drinking water distribution systems Lakes, rivers, creeks, and groundwater (Canada) No apparent trends between NDMA concentrations and DOC, NH3-N, NO3 -, total Kjedldahl nitrogen (TKN), and organic N [43] NDMA formation in natural water Rivers and lakes in (U.S. & Canada) No significant relationships between NDMA formation and total organic carbon (TOC) [44] NDMA formation in natural water, and precursor fingerprinting Natural reservoirs (Spain) After fingerprinting dissolved organic matter (DOM), a positive correlation was found between NDMA formation and aliphatic as well as peptide and lipid-like compounds (r 2 = 0.88) [45] 3.3.3. Acceptable NDMA concentrations NDMA formed as a by-product from the bench-top water treatment system varied significantly with changing CT (Figure 3). There were also significant differences in NDMA produced from each source water site during both CTs ( Table 2,  Drinking water leaving a WTP is typically pumped to a storage facility where it resides for days before reaching the consumer [55]. For the case of NDMA formation potential in Rhode Island based source water (Figure 1), these findings are essential. In this experiment, both source waters exposed to 72 hr monochloramine CT resulted in NDMA concentrations below 10 ng/L, complying with regulatory limits already established by California and Massachusetts [12,13].
With the information provided by this study, and future NDMA formation potential tests of the selected source waters, Rhode Island based WTPs may consider switching to chloramine disinfection to comply with established DBP regulations effectively, and N-DBP regulations to come in the near future. Table  Table Table  Table 2  * estimate of population variance based on the variability among a given set of measures Table  Table Table  Table 3 Fig. Fig.  Fig. A4.
A4. A4. A4. Magnified view of NPIP (2 µg/L). The peak at 20.36 min was identified as a hydrocarbon.  Table  Table Table  Table A Table  Table Table  Table A