Journal AWWA In Press Date of submission: 09/11/2016 Date of acceptance: 12/21/2016 Editor’s disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our readers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to Journal AWWA pertain.

The Removal of N-Nitrosodimethylamine Formation Potentials in Drinking Water Treatment Plants

Habibullah Uzun, Daekyun Kim, and Tanju Karanfil*

Department of Environmental Engineering and Earth Sciences, Clemson University, Anderson, SC 29625, USA

*Corresponding author: Phone: 1-864-656-1005, Fax: 1-864-656-0672, email: [email protected]

Submitted to Journal of American Water Works Association

ABSTRACT A long term (18+ months) systematic investigation was conducted to examine i) the removal of N-nitrosodimethylamine (NDMA) formation potentials (FPs) by different treatment processes under various operational conditions at nine water treatment plants, and ii) the occurrence of NDMA in their distributions systems. Average NDMA FP removal by alum clarification process ranged from 12% to 30% during different seasons and weather conditions. PAC addition improved the removal of NDMA FP, especially at high doses (>4 mg/L). The use of oxidants (i.e., Cl2 and or ClO2), especially simultaneous application, enhanced the removal of NDMA FPs and lowered the NDMA concentration in the distribution systems to below 10 ng/L. However, simultaneous application of ClO2 and Cl2 led to the formation of elevated level of ClO3. The average NDMA FP reduction by RO and MF filtration was 81% and 7%, respectively. The overall NDMA FP removal efficiencies between raw and finished water ranged 40-59%.

Keywords: Disinfection by-products, NDMA, alum clarification, PAC, oxidation, membranes

INTRODUCTION Nitrogenous disinfection by-products (N-DBPs) are far more cytotoxic and genotoxic than carbonaceous DBPs (Plewa & Wagner 2009). Among such N-DBPs, nitrosamines have been shown to be probable human carcinogens associated with 10-6 lifetime cancer risk at ng/L levels (USEPA 2002). An increasing number of water treatment plants (WTPs) in the United States (US)

2

have been employing or considering chloramination over chlorination to comply with the stringent regulations for halogenated DBPs such as trihalomethanes (THMs) and haloacetic acids (HAAs) (Li 2011). However, nitrosamines have been detected at notable levels in chloraminated waters (Choi & Valentine 2002b, Mitch et al. 2003, Charrois & Hrudey 2007, Nawrocki & Andrzejewski 2011, Russell et al. 2012). N-nitrosodimethylamine (NDMA) has been the most commonly detected nitrosamine species in the US drinking water distribution systems (Bond et al. 2011, Wang et al. 2011, Russell et al. 2012, Woods & Dickenson 2015). There are currently no federal regulations for nitrosamines in drinking water systems. However, five of them were included in the United States Environmental Protection Agency’s (USEPA) Contaminant Candidate List 3 (CCL3) (USEPA 2009), and USEPA is expected to make a regulatory determination for the nitrosamines during the six year review of disinfectants and disinfection byproducts rules (DBPRs).

BACKGROUND AND OBJECTIVES One of the strategies to reduce NDMA formation in the distribution systems is to remove or deactivate its precursors during water treatment before chloramination. Since the known adverse health effects of NDMA occur at ng/L levels, US water utilities have had a strong interest in understanding i) the robustness of their processes/operations on the removal/deactivation of NDMA precursors, while complying with The Stage 2 DBPRs, and ii) the effect of seasonal and weather changes on the removal of NDMA precursor. A few previous studies have shown that NDMA FP changes were negligible by alum (Sacher et al. 2008) and ferric chloride clarifications (Knight et al. 2012) at full scale WTPs. In contrast, some other studies have reported increasing NDMA FPs after conventional clarification

3

(i.e., coagulation, flocculation and sedimentation) compared to raw water (Krasner et al. 2008, Krasner et al. 2012, Sacher et al. 2008, Mitch et al. 2009). Such increases have been attributed to polymers used during conventional clarification processes. It has been reported that some polymers (e.g., aminomethylated polyacrylamide, poly[epichlorohydrin dimethylamine] [polyamine] and poly[diallyldimethylammonium chloride] [polyDADMAC]) used commonly for water treatment could increase NDMA FP levels in water (Kohut & Andrews 2003, Wilczak et al. 2003, Mitch and Sedlak 2004, Najm et al. 2004, Park et al. 2007, Park et al. 2009). A previous study conducted by Krasner et al. (2016) reported that there was a seasonal variability when the 75th percentile values were considered in terms of NDMA FP in both source and treated waters. However, the median NDMA FP levels of those collected in different seasons were in general comparable at most of the full-scale WTPs (Krasner et al. 2016). The seasonal/temporal removal of NDMA FP during water treatment has not been further investigated in this study. Powdered activated carbon (PAC) application resulted in increases in the NDMA FP removals from surface waters and wastewater effluents (Sacher et al. 2008; Hanigan et al. 2012; Beita-Sandi et al. 2016). 17-34% of watershed-derived NDMA precursors were removed by 3 mg/L of a bituminous-based PAC when blended with treated secondary wastewater effluent, and up to 20 mg/L of PAC application was effective to deactivate NDMA precursors (Hanigan et al. 2015). Beita-Sandi et al. (2016) showed that PACs with hybrid characteristics (micro and mesoporous), higher surface areas, and basic surface chemistry are more likely to be effective for NDMA precursor control by PAC adsorption. They also found that PAC adsorption was more effective in reducing NDMA FP in fresh wastewater effluents than in surface waters. Microfiltration (MF) and ultrafiltration (UF) were not effective on the removal of NDMA precursors (Pehlivanoglu-Mantas & Sedlak 2008, Krauss et al. 2010), while nanofiltration (NF)

4

and reverse osmosis (RO) achieved 57-98% (Miyashita et al. 2009, Ersan et al. 2016) and >98% (Schmidt et al. 2008, Krauss et al. 2010) of NDMA FP removals, respectively. It has also been reported that the bulk portion of NDMA precursors consists of small molecular weight compounds that can pass through 3000 Da membranes (Mitch & Sedlak, 2004, Pehlivanoglu-Mantas & Sedlak, 2008). In addition, Chen and Valentine (2008) reported that appreciable amount of NDMA precursors were deactivated by chlorine (Cl2) addition for 10-20 minutes of contact time. Shah et al. (2012) observed that chlorination with the Ct (oxidant concentration × contact time) value of 37 mg*min/L reduced the NDMA formation up to 80% prior to chloramination; further increases in Ct did not result in additional reduction. They also found that chlorine dioxide (ClO2) had little impact on the NDMA formation. On the other hand, ClO2 was reported to have both positive and negative impacts on formation on NDMA during subsequent chloramination (Lee et al. 2007, Sacher et al. 2008, Selbes et al. 2014). Although these previous works have provided useful information, the fate of NDMA precursors during water treatment processes under dynamic operational conditions has not been extensively reported in the literature due to insufficient data and lack of long term and systematic monitoring studies at full-scale WTPs. Therefore, the main objectives of this study were to: i) examine the fate of NDMA precursors during different treatment processes (e.g., alum clarification, PAC adsorption, pre-oxidation and post-oxidation) for an extended monitoring period (~2 years) at full-scale WTPs, and ii) evaluate NDMA occurrence levels in distribution systems. A long term sampling strategy allowed investigating the impacts of different seasons and temporal weather related conditions (e.g., high/low rainfall periods) on the NDMA FP removal efficiency. THM FP and THM occurrence were also monitored in selected systems due to their regulatory importance and compared with NDMA results.

5

MATERIALS AND METHODS WTPs. Nine conventional WTPs located in the Southeastern US using twelve different surface source waters participated in this study. Source waters (e.g., SW A, SW B, and etc.) and the WTPs were named (e.g., WTP 1, WTP 2 and etc.) by letters and numbers, respectively. Detailed information for each WTP`s process configuration is summarized in Table 1. Alum was used for coagulation at all WTPs, and application doses depended on DOC and/or turbidity levels in influents. WTPs 4 and 5 applied 2-3 mg/L of PAC continuously, while WTPs 2 and 3 used 110 mg/L of PAC occasionally to control taste and odor problems. WTPs 1, 4, 5, and 8 applied different pre-oxidation strategies including Cl2 and/or ClO2, and NH2Cl prior to conventional clarification, and WTPs 1, 2, 3, 4, and 5 employed three different types of polyacrylamide polymers (i.e., Optimer Nalco Pulv 8110, Nalco 8170, and Sedifloc 400C) in their flocculation basins and/or rapid mixing units. A RO system at WTP 1 was operated occasionally to supply water to a nearby power plant. WTP 7 used MF unit in addition to conventional clarification, and dissolved air flotation (DAF) system was used for solid separation instead of sedimentation at WTP 9. Seven out of nine WTPs used chloramines, while WTPs 6 and 7 utilized Cl2 to maintain residual in the distribution systems. Sample collection and analysis. At each WTP, water samples were collected from the influent (i.e., raw water), clarifier or DAF effluents (only for WTP 9), after post-oxidation with Cl2 and/or ClO2 (i.e., entry point [EP]), and the longest point in the distribution system. The NDMA FP removal efficiencies for conventional clarification and post-oxidation were calculated using following equations; Conventional treatment removal

(%) = [(Craw - Ceff)/Craw] x 100

6

Eq 1)

Post-oxidation removal

(%) = [(Ceff - Cfin)/Ceff] x 100

(Eq 2)

where Craw= NDMA FP (ng/L) in raw water, Ceff= NDMA FP (ng/L) in treated water (i.e., clarifier effluent, before filters and any oxidant addition), Cfin= NDMA FP (ng/L) after post-oxidation (i.e., finished water collected at EP). To determine NDMA and THM occurrences and to conduct FP tests, water samples were collected in 1000 mL (for NDMA), and 125 mL (for THMs) pre-washed borosilicate amber glass bottles, filled head space free, and transferred immediately to the laboratory. For the occurrence of NDMA, sodium thiosulfate (~30 mg) and sodium sulfite (~2 mg) were added immediately as quenching agents to avoid further formation of NDMA and THMs, respectively, and stored in the refrigerator at 4 °C in the dark until analysis. FP tests were conducted to determine the maximum NDMA and THM precursor concentrations by adding excess amount of monochloramine (NH2Cl) and Cl2, respectively. During FP tests, pH was maintained at 7.8 with 20 mM phosphate buffer. Pre-formed NH2Cl stock solution at Cl2/N mass ratio of 4:1 was generated by dropping sodium hypochlorite slowly to (NH4)2SO4 solution at pH ~9 (Hong et al. 2007), and pre-determined volume of NH2Cl stock solution was spiked to samples to achieve 100 mg/L of initial NH2Cl. For the THM FP tests, pre-determined amount of Cl2 stock solution (5-6% available free Cl2) was spiked to achieve 50 mg/L initial Cl2 concentration. After 5 days of contact time at room temperature (21-22 oC), the residual oxidants were measured (NH2Cl [>25 mg/L] during NDMA FP tests and Cl2 [>20 mg/L] during THM FP tests) and quenched with sodium thiosulfate (Na2S2O3) and sodium sulfite (Na2SO3), respectively. NDMA was analyzed using a Varian GC 3800-MS/MS 4000 equipped with RTX-5MS (Restek 30 m × 0.25 mm × 0.25 μm) column under the chemical ionization mode and an Agilent 6890 GC-ECD equipped with Phenomenon ZB-1 (30 m × 0.25 mm × 1 μm) was used for THM analysis. Sample extraction and analysis methods have been described in detail elsewhere (Uzun et al. 2015, Uzun

7

2016, Uzun et al. 2016). Concentrations of Cl2 and chloramines as Cl2 were determined following Standard Method 4500-Cl F (APHA/AWWA/WEF 2005). Jar tests. A PHIPPS & BIRDTM PB-700TM Jar tester equipped with six paddles rotating in a six beakers was used for jar tests. First, pH was adjusted to 6.0 with HCl and NaOH. A predetermined amount of alum was added to the tested water and mixed rapidly (100 rpm=1 min, and 60 rpm=3 min). Then water was mixed at 20 rpm for 30 min. for flocculation. During the test, pH changes were kept ±0.3 unit, and were recorded at the end of the mixing periods. Finally, the mixer was stopped and the flocks were allowed to settle out (>30 min), then the samples were collected from mid-point of the beakers and used for further tests.

RESULTS AND DISCUSSIONS Raw water quality. Measured NDMA FP, THM FP, and selected water quality parameters of WTP influent waters are listed in Table 2. The ranges of DOC, specific ultraviolet absorbance at 254 nm (SUVA254), dissolved organic nitrogen (DON), Br-, and boron values were 0.7-16.2 mg/L, 0.6-5.7 L/mg-m,