Jinwei
Zhang
a,
David
Hanigan
b,
Paul
Westerhoff
b and
Pierre
Herckes
*a
aSchool of Molecular Sciences, Arizona State University, Tempe, AZ 85287-1604, USA. E-mail: pierre.herckes@asu.edu; Fax: +1 480 965 2747; Tel: +1 480 965 4497
bSchool of Sustainable Engineering and the Built Environment, Arizona State University, Tempe, AZ 85287-3005, USA
First published on 5th January 2016
N-Nitrosodimethylamine (NDMA) is a chloramination disinfection by-product (DBP) with an uncertain regulatory future. While extensive literature exists on NDMA formation potentials (FP) for natural waters and for model compounds considered as NDMA precursors, less data exists on the kinetics of NDMA formation in surface and wastewaters. NDMA formation kinetics experiments were conducted in seven source waters at two monochloramine doses. NDMA formation was modeled by a simple, second-order model, using the measured NDMAmax and monitored monochloramine concentrations at selected reaction times. The model fits NDMA formation well (R2 > 0.88) in all source waters. While the extent of NDMA formation was highly variable, the rate constant (kapp) values from different waters fell in a narrow range (0.01–0.09 M−1 s−1). This suggests that a common precursor or rate limiting step for NDMA formation likely exists despite the differences in matrices. Although further studies are needed to validate the model over a wider range of water conditions such as pH and N:Cl2 ratios, the model could help water utilities to predict NDMA formation in distribution systems.
Water impactN-Nitrosodimethylamine (NDMA) has drawn extensive attention because of its rodent carcinogenicity and occurrence in drinking water. Kinetics information for NDMA formation is lacking for wastewaters and drinking waters. In this work, chloramination of waters from various sources (e.g. surface waters and wastewaters) was performed and a second order model of NDMA formation is proposed to describe NDMA formation kinetics. The proposed model presents a practical way for water utilities to predict NDMA formation in their waters. The results offer new insights on possible NDMA formation mechanisms in real waters. |
A recent review indicates that most studies have found that NDMA formation is more associated with chloramination than with chlorination.1 Systems using chloramines as the primary, rather than secondary, disinfectant have high NDMA Formation Potential (FP) (i.e., >50 ng L−1) in plant effluent,3 indicating the potential for precursor deactivation by strong pre-oxidants such as chlorine. Since nitrosamine formation is a kinetically slow process, plants using chloramine with long hydraulic contact times in plant plus distribution system (e.g., 12–18 h) tend to have more NDMA in the effluent than those using chloramine for short (e.g., 0.5–2 h) contact times.7 NDMA concentrations tend to increase throughout chloraminated distribution systems.7–11
NA formation in drinking water requires an organic nitrogen-based precursor plus an oxidant (e.g., inorganic chloramine, ozone).12,13 Oxidation chemistry, including inorganic reactions with bromide and ammonia,14,15 is important; however, little information is available regarding which organic precursors control the rate and extent of NA formation in drinking water. Mechanistic studies indicate that yields of NDMA from chloramination of most secondary and tertiary model amines are ∼0–2% but can be >80% for certain tertiary amines with β-aryl functional groups.16,17 Wastewater-impaired source waters contain NDMA precursors, suggesting the importance of anthropogenic constituents. Specific precursors have not been characterized outside of a select few18 in wastewater-impaired source waters but could include either tertiary amine-based microconstituents that form NDMA at high yield or quaternary amine-based macroconstituents of consumer products that form NDMA at low yield.
Different NDMA formation pathways during chloramination are briefly illustrated in the ESI† section (Text S1). Despite rich literature on pathways and yields of NDMA formation using model compounds,19–23 less information exists related to the kinetics of NDMA formation in surface and wastewaters. Many studies rely upon NDMA FP measurements which are akin to THM-FP measurements and, while useful, lack information suitable for managing DBP formation in complex hydraulic systems. Simulated distributed system (SDS) test methods for NA's have been developed, but often include a short free chlorine period (before NH3 addition) to mimic common drinking water treatment plant (DWTP) disinfection processes. We believe a focus on NDMA formation kinetics in raw water samples will expand our understanding (i.e. profiling) of NDMA precursors.
The aim of this paper is to investigate NDMA formation kinetics in waters with lower (surface waters) and higher (treated wastewater effluents) levels of NDMA precursors. In experiments conducted with seven different waters, the decay of monochloramine and formation of NDMA were monitored. Experimental data were fit using a second-order reaction model. We observed similar magnitudes of the fitted second order apparent rate constant for NDMA formation across a range of water sources, suggesting the model represented a possible common rate limiting step that exists in most raw waters.
The extracted samples were analyzed using an Agilent 6890N/5973 inert GC/MS operated in positive chemical ionization mode with ammonia as the reagent gas.25 In brief, the chromatographic column used was an Agilent DB-1701P (30 m × 0.250 mm × 0.25 μm) (Santa Clara, CA) and followed a pulsed splitless injection (initial pulse 15 psi for 45 s and then 10 psi) set at 250 °C with a reduced diameter SPME inlet liner (Sigma Aldrich, St. Louis, MO). The helium carrier gas was initially pulsed at 1.9 mL min−1 for 45 s and then reduced to 1.3 mL min−1 for the rest of the run. 4 μL of sample was injected into GC through the inlet, with oven temperature of 40 °C held for 3 min, increased by 4 °C min−1 to 80 °C and increased to 120 °C at 20 °C min−1. The column interface temperature was set at 200 °C. The mass selective detector was set to analyze for mass-to-charge 92 (NDMA + NH4+) and 98 (NDMA-d6 + NH4+). The GC/MS was calibrated using a series of NDMA standards ranging from 1 μg L−1 to 1 mg L−1 and NDMA-d6 (100 μg L−1) as internal standard.
Fig. 1 (A) Monochloramine (NH2Cl) decay kinetics in WW1 for two initial monochloramine doses. (B) NDMA formation observed (symbols) and fitted by eqn (2) & (3) (lines). Error bars represent one standard deviation (n = 3) for select time points (pH = 8.2, temperature = 23 ± 1 °C). |
Table 1 summarizes NDMAmax values for each experiment. In the wastewater effluent samples, NDMAmax ranged from 4 to 12 nM. There was no correlation found between NDMAmax and DOC or UV254, similar to statistical analyses presented elsewhere.26,27 In all cases, higher monochloramine doses led to 30% to >50% higher NDMAmax values. This was unexpected because even at very long reaction times there was adequate oxidant residual present to react with NDMA precursors. Although monochloramine is the dominant chloramine species in our test (pH = 8, N:Cl2 molar ratio 1.2:1), dichloramine, the disproportionation product from monochloramine, was still present according to the equilibrium:
2NH2Cl + H+ ↔ NHCl2 + NH4+ | (1) |
It has been reported that dichloramine is responsible for greater NDMA formation from NDMA precursors such as DMA.16,21 Additionally, NDMA precursors were found to react preferably with either monochloramine or dichloramine.17,28 Thus, in our experiment, even trace levels of dichloramine formed could affect the maximum NDMA formation. At higher doses of monochloramine there would be more dichloramine enhancing the NDMA formation and in contrast a low monochloramine concentration solution would contain less dichloramine, resulting in less NDMA formation.
Sample ID | Source | DOC (mgC L−1) | UV254 (cm−1) | Upon monochloramine addition | ||||
---|---|---|---|---|---|---|---|---|
Monochloramine dose (mgCl2 L−1 [μM]) | pH | NDMAmax (nM) | k app (M−1 s−1) | R 2 | ||||
a Data of NDMA formation and monochloramine were from Chen and Valentine, 2006. | ||||||||
WW1 | Wastewater | 4.6 | 0.12 | 18 [254 μM] | 8.2 | 6 | 0.04 | 0.99 |
6 [85 μM] | 4 | 0.04 | 0.95 | |||||
WW2 | Wastewater | 6.17 | 0.15 | 20 [282 μM] | 8.0 | 7 | 0.04 | 0.97 |
7 [99 μM] | 2 | 0.04 | 0.88 | |||||
WW3 | Wastewater | 5.83 | 0.14 | 20 [282 μM] | 8.0 | 12 | 0.07 | 0.95 |
7 [99 μM] | 9 | 0.05 | 0.98 | |||||
WW4 | Wastewater | 5.32 | 0.17 | 20 [282 μM] | 8.0 | 12 | 0.08 | 0.96 |
7 [99 μM] | 8 | 0.08 | 0.92 | |||||
WW5 | Wastewater | — | — | 20 [282 μM] | 8.0 | 8 | 0.08 | 0.97 |
6 [85 μM] | 5.5 | 0.09 | 0.96 | |||||
SW1 | Lake | 3.88 | 0.08 | 36 [507 μM] | 8.0 | 0.7 | 0.01 | 0.91 |
12 [169 μM] | 0.4 | 0.02 | 0.91 | |||||
GW1 | Groundwater | 1.78 | 0.05 | 20 [282 μM] | 8.0 | 0.2 | 0.04 | 0.91 |
7 [99 μM] | 0.2 | 0.06 | 0.89 | |||||
SW2a | River | 3.4 | — | 3.5 [50 μM] | 8.0 | 0.3 | 0.09 | 0.96 |
Fig. 2 NDMA formation observed (symbols) and fitted by eqn (2) & (3) (line) in SW at two initial monochloramine doses. Error bars represent one standard deviation (n = 3) for select time points (pH = 8.0, 23 ± 1 °C). |
The groundwater had NDMAmax (15–20 ng L−1) on the same order of magnitude with, but lower than the surface water (30–50 ng L−1). It is possibly because it contained less DOC (1.78 mgC L−1) in groundwater. No two kinetic parts (fast vs. slow) were observed and NDMA formation reached maximum in less than 100 hours (Fig. 3) at two monochloramine doses.
Fig. 3 NDMA formation observed (symbols) and fitted by eqn (2) & (3) (line) in GW at two initial monochloramine doses (pH = 8.0, 23 ± 1 °C). |
In summary, in all water samples the overall NDMA formation at high monochloramine dose was higher than at low monochloramine dose. Wastewaters showed higher NDMA formation (200–950 ng L−1) than surface waters (30–50 ng L−1) and groundwater (15–20 ng L−1). The reaction times for NDMA formation reaching its maximum were shorter when higher monochloramine doses were applied in wastewater and groundwater. No significant difference of such reaction times was found between the two doses of chloramine in surface water. The differences in NDMA formation potential are due to the various types and concentrations of precursors in the different source waters. Precursors such as DMA, which is more reactive with dichloramine, may explain the higher and faster NDMA formation at higher doses of monochloramine.21 Recent research found that pharmaceutical compounds such as methadone, which was found in wastewater effluents have high yields of NDMA and could contribute to large fractions of total NDMA formed.30 It is possible that such high yield precursors are also more reactive to chloramines making the NDMA formation in wastewater relatively faster. So higher amine concentrations and higher levels of these known high yield NDMA precursors may be responsible for a higher overall NDMA formation in wastewater than in other water sources. Compared to the differences in NDMA yields in source waters, the differences in kinetics or rates were rather small. Molecular identification of NDMA precursors and NDMA formation kinetics of these precursors are needed to improve our understanding of NDMA formation in real waters.
(2) |
[NDMA(t)] = [P]0 − [P]t | (3) |
Although dichloramine was thought to react with precursors forming NDMA, the model did not use dichloramine as a reactant variable. One reason is that the measurement of dichloramine is time consuming and more complicated compared to that of monochloramine.31 Additionally, under our experimental conditions (pH = 8.0) the NHCl2/NH2Cl ratio would be constant because the system is in equilibrium during the reaction time.32,33 Thus dichloramine could be represented as ratio of monochloramine and its reaction could also be fit with our model empirically, leading to an apparent rate constant kapp.
Using this modeling approach, NDMA formation in treated wastewaters and in surface waters was well fit with correlation coefficients (R2) greater than 0.9 in most tests. Optimized data fits were achieved using Kintecus.34 Model fits of experimental data are shown in Fig. 1–3. In wastewater effluents tests, the model overestimated the NDMA in the beginning of the test (<50 h) at low doses of monochloramine, showing a ‘lag period’ of NDMA formation. Such a ‘lag-period’ was observed in previous studies of NDMA formation from pharmaceutical compounds (e.g. ranitidine) as NDMA precursors at similar monochloramine dose (6 mg Cl2 L−1) in surface waters.16 In work conducted by Shen and Andrews, the lag and a subsequent initiation of NDMA formation from selected pharmaceuticals were successfully modeled with a dose–response curve.16 The lag and the rate constants were found correlated to TOC and SUVA values for certain pharmaceuticals. It was suggested that there was NOM-pharmaceutical binding that inhibited the initial NDMA formation.16 A similar lag-period was interpreted as indicating the possible interactions of pharmaceutical compounds with wastewater organics. A similar dose–response curve model was applied to our kinetics data. The NDMA formation was well represented with a similar correlation coefficient than in our kinetic model (R2 > 0.9) (Table S1†). The wastewater effluent samples with the ‘lag-period’ were better fit with the dose–response curve model than our proposed model. However, the rate constant k (h−1) derived from the dose–response model showed broad variations, ranging from 0.007 h−1 to 0.175 h−1 for surface water and wastewater respectively. No correlation was found between water quality (e.g. TOC and SUVA) and model parameters (e.g. lag and rate constant k). The dose–response model also did not take into account the role of monochloramine dose, which in our kinetic experiments, impacted NDMA FP and the reaction kinetics. Finally the dose–response model is purely descriptive and does not provide insight into the underlying formation mechanisms, especially in complex water matrices.
Table 1 summarizes fitted kapp and R2 values for all experiments. The magnitude of kapp values fall within a relatively narrow range of less than one order of magnitude (0.01–0.09 M−1 s−1). The goodness of fit of the model, represented by eqn (2) and (3), was evaluated by comparing observed NDMA formation data and model prediction in Fig. 4. The small 95% confidence intervals suggested significant correlation between model-predicted and observed NDMA formation. Data in Chen and Valentine's 2006 work on NDMA formation from NOM was also fit with our model and the kapp (0.09 M−1 s−1) was similar to that of our wastewater and surface waters. A narrow range in kapp values was surprising given the very different precursors expected in varying source waters and the order of magnitude differences in NDMAmax between sources.
Fig. 4 Linear correlation between model predictions and observations of NDMA concentrations in all waters. Data from all reaction time periods are included. |
Most mechanistic work on NDMA formation during chloramination has been conducted with secondary amines such as DMA.21 It has been suggested that during chlorination some tertiary amines could decay to secondary amines forming NDMA upon subsequent chloramination.22 It was still unclear if such transformation reaction between amines or NDMA formation from amines was a rate limiting step. The small variance of rate constants suggests that reactions like degradation of higher order amines to secondary amine precursors, if there any were present, are rapid compared to NDMA formation. This is in agreement with the short half-life (<14 h) of trimethylamine (TMA) decomposing to DMA in presence of monochloramine and indicates that amine precursor groups, including secondary amines and tertiary amines with either low or high yields, probably have a similar rate limiting step forming NDMA in our tests.
We propose a general reaction pathway of NDMA formation from chloramination of NDMA precursors in treated wastewaters and surface waters. Precursors that include anthropogenic chemicals or natural biomolecules rapidly react with chloramines to produce intermediates. The yield of intermediates depends upon types of precursors (e.g. secondary or tertiary amines, β-aryl amines) and monochloramine dose. The intermediates then function as the precursors (P) and the subsequent conversion from intermediates (P) to NDMA undergoes a second order reaction mechanism in the presence of oxidant such as monochloramine.
(4) |
Eqn (4) offers another way of quantifying the apparent second order rate constant kapp. In addition, it shows ([P]/[P]0), the relative conversion of NDMA precursor, is related to the NH2Cl exposure. Fig. 5 shows the plots of [P]/[P]0 against NH2Cl exposure for water samples at low and high NH2Cl doses. In different water samples, the reaction required different NH2Cl exposure for the same conversion of precursors. The rate constants varied by water sources, possibly due to the varying precursor groups and their different reactivity with NH2Cl or NHCl2. From the previous discussion in this work NH2Cl dose determines the NDMAmax even in the same water sample, but does not affect the rate constants in wastewater and groundwater. The relative change of precursors (or increase of NDMA formation) had the same kinetics for the same water at different NH2Cl doses (Fig. 5a–c, S5–S7†). Surface water represents an exception (Fig. 5d). The surface water had a lower kapp than all other waters tested. The relative change of precursors in surface water differed with low and high dose NH2Cl, especially when NH2Cl exposure is less than 5 × 106 mgCl2 min L−1, possibly due to the ‘fast and slow’ reaction mechanism in surface water.
Fig. 5 Plots of P/P0versus monochloramine exposure for water samples (a) WW1, (b) WW4, (c) GW1, (d) SW1. L = lower, H = higher, represent samples with lower or higher NH2Cl concentrations. |
In finished drinking water NH2Cl concentrations (<0.06 mM) are typically lower than those used in our test (0.09–0.51 mM). From our results, NDMA formation reaction kinetics are dependent upon NH2Cl exposure, not NH2Cl concentrations. The rate constants could be applied to waters with lower NH2Cl concentration and longer contact time ranges. NH2Cl exposure and the rate constants are the key parameters for the prediction of the transformation efficiency of NDMA precursors into NDMA. With a measurement of NDMAmax in site-specific sample and chloramine conditions, our proposed model provides a practical way to predict NDMA formation in drinking water influenced by wastewater effluents and surface waters in water plants and drinking water distribution systems.
A simple second order NDMA formation model of reactions between amine precursors and monochloramine was developed. NDMA formations were well predicted by the model with correlation coefficients higher than 0.9 in most cases. The modeled rate constants for different water samples were found surprisingly within a narrow range (0.01–0.09 M−1 s−1), indicating a possible rate limiting step of NDMA formation for different amine precursor groups. With only two simple measurements (NDMA formation potential and monochloramine exposure), our model provides a practical way to predict NDMA concentrations in distribution system.
Our proposed model was validated at pH 8 and monochloramine doses between 0.09 mM to 0.51 mM with wastewater effluents and surface water samples. It would be of value to extend the work further in a larger variety of water matrices and reaction conditions such as pH, N:Cl2 ratio and dissolved oxygen level, to simulate a larger variety of water treatment plants or distribution systems. Additionally, characterization or profiling of the precursors of NDMA and chloramination kinetic studies of more model precursors are needed to better understand different pools of precursors and how they could interact or contribute to NDMA formation in different water systems.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ew00234f |
This journal is © The Royal Society of Chemistry 2016 |