Environ. Sci. Technol. 2007, 41, 6732-6739
Volatile Disinfection Byproduct Formation Resulting from Chlorination of Organic-Nitrogen Precursors in Swimming Pools JING LI AND ERNEST R. BLATCHLEY III* School of Civil Engineering, Purdue University, West Lafayette, Indiana 47907-2051
Clinical studies have documented the promotion of respiratory ailments (e.g., asthma) among swimmers, especially in indoor swimming pools. Most studies of this behavior have identified trichloramine (NCl3) as the causative agent for these respiratory ailments; however, the analytical methods employed in these studies were not suited for identification or quantification of other volatile disinfection byproducts (DPBs) that could also contribute to this process. To address this issue, volatile DBP formation resulting from the chlorination of four model compounds (creatinine, urea, L-histidine, and L-arginine) was investigated over a range of chlorine/precursor (Cl/P) molar ratios. Trichloramine was observed to result from chlorination of all four model organic-nitrogen compounds. In addition to trichloramine, dichloromethylamine (CH3NCl2) was detected in the chlorination of creatinine, while cyanogen chloride (CNCl) and dichoroacetonitrile (CNCHCl2) were identified in the chlorination of L-histidine. Roughly 0.1 mg/L (as Cl2) NCl3, 0.01 mg/L CNCHCl2, and 0.01 mg/L CH3NCl2 were also observed in actual swimming pool water samples. DPD/ FAS titration and MIMS (membrane introduction mass spectrometry) were both employed to measure residual chlorine and DBPs. The combined application of these methods allowed for identification of sources of interference in the conventional method (DPD/FAS), as well as structural information about the volatile DBPs that formed. The analysis by MIMS clearly indicates that volatile DBP formation in swimming pools is not limited to inorganic chloramines and haloforms. Additional experimentation allowed for the identification of possible reaction pathways to describe the formation of these DBPs from the precursor compounds used in this study.
Introduction Chlorination is the most common method for disinfection of recreational water, such as swimming pools. Although +1 valent chlorine has been demonstrated to be effective for the control of many microbial pathogens, it is also known to react with aqueous constituents to yield disinfection byproducts (DBPs) (1). Many common aqueous constituents can react with chlorine to yield DBPs; however, the focus of this investigation is on the chlorination of organic-nitrogen compounds. Important sources of organic-nitrogen compounds in recreational waters include sweat and urine, and relevant * Corresponding author phone: (765)494-0316; fax: (765)494-0395; e-mail:
[email protected]. 6732
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007
organic-nitrogen precursors may include urea, creatinine, and amino acids. These compounds will consume free chlorine (2, 3) and also act as precursors in the formation of DBPs, including some that are sufficiently volatile to be transferred to the gas phase (4). For example, trihalomethanes (THMs) have been reported to be formed by chlorination of materials of human origin (hair, lotion, saliva, skin, and urine) (5) and by chlorination of human urine analogues containing urea, creatinine, and citric acid (6). Past studies have suggested that the smell and irritant properties of swimming pool air are largely attributable to inorganic chloramines. These compounds, particularly trichloramine (NCl3), are volatile (7, 8). Trichloramine has been reported to function as an irritant to the eyes and upper respiratory tract, and it may contribute to acute lung injury in accidental, occupational, or recreational exposures to chlorine-based disinfectants (9-11). However, relatively little information is available to describe the specific chemistry responsible for NCl3 formation in recreational waters. Additionally, most available literature related to DBPs in recreational water has focused on inorganic chloramines. Comparatively little information is available to describe the formation of other volatile DBPs that may be found in swimming pool settings. The objective of this study was to identify volatile DBPs that will result from the chlorination of organic-nitrogen compounds that are likely to be present in recreational waters. Four model organic-nitrogen compounds, urea, creatinine, L-histidine, and L-arginine, were selected as representative organic pollutants of recreational water (refer to Figure 1). Urea is the major nitrogenous end product of protein metabolism and is the chief nitrogenous component of urine and sweat in mammals. Creatinine is a constituent of perspiration and urine formed by the metabolism of creatine; it is found in muscle tissue and blood and is normally excreted in urine and sweat as a metabolic waste product. L-Histidine and L-arginine are amino acids that are commonly found in human sweat. Urea, creatinine, and L-histidine are also the primary constituents of “body fluid analogue” (BFA), which has been used as a surrogate mixture of organic-nitrogen compounds in previous studies involving chlorination of recreational waters (12, 13). However, the focus of the previously published research involving the chlorination of BFA was on the measurement of THMs and chloramines. Moreover, these investigations were based on analytical measurements that are known to be susceptible to common forms of analytical interference (e.g., diethyl-p-phenylenediamine/ferrous ammonium sulfate (DPD/FAS) titration). Membrane introduction mass spectrometry (MIMS) was employed to monitor the DBPs of chlorination because MIMS has been shown to be a suitable method for the analysis of volatile compounds in aqueous samples. Also, MIMS-based analytical procedures have the potential to be used to characterize many aspects of chlorine-based water treatment applications. An important advantage of the method is that it can yield quantitative and structural information about volatile DBPs present in an aqueous sample. MIMS measurements were also conducted in parallel with measurements by DPD/FAS titration, so as to allow for comparisons between the two methods. In conjunction with previously published information, the results of these measurements allowed for the development of hypothesized reaction mechanisms to describe volatile DBP formation from chlorination of several model organic-nitrogen compounds. Last, pool water samples were collected and analyzed using these same methods so that comparisons between DBP formation 10.1021/es070871+ CCC: $37.00
2007 American Chemical Society Published on Web 09/05/2007
FIGURE 1. Illustration of the structures of model organic-nitrogen compounds and possible volatile DBPs that result from chlorination of recreational waters. in actual recreational water could be made with the results of the more controlled experiments involving model organicnitrogen compounds.
Experimental Procedures Materials and Methods. All chemicals used in this study, unless otherwise noted, were reagent-grade, purchased from Sigma-Aldrich, and used without further purification. Dilution to target aqueous-phase concentrations was accomplished with distilled, deionized water. Free chlorine stock solutions and standard solutions of inorganic chloramines were prepared in the same manner as described previously (14). Standard solutions of cyanogen chloride (CNCl) were prepared daily from a CNCl stock solution (2000 mg/L Protocol Analytical Supplies, Inc.). Standard solutions of chloroform (CHCl3) and dichloroacetonitrile (CNCHCl2) were prepared gravimetrically from pure compounds. Standard dichloromethylamine (CH3NCl2) solutions were prepared by chlorination of methylamine (CH3NH2) at a Cl/N molar ratio of 2.0; the concentration of CH3NCl2 was defined by DPD/FAS titration as apparent dichloramine. The MIMS system was based on a modification of an HP 5892 benchtop GC/MS comprising an HP 5972A mass selective detector (MSD) equipped with electron (70 ev) ionization (EI). A mass spectrum scan mode (49 e m/z e 200) coupled with EI was used to identify possible DBPs, while a selected ion monitoring (SIM) mode was used for quantification of volatile DBPs. Other details of the configuration and setup for the MIMS system and operational conditions can be obtained from ref 14. The concentration of volatile DBPs was determined by comparison of ion abundance measurements with those developed from a series of standard solutions. Ions at m/z 61, 74, 85, 98, and 119 amu were monitored for quantification of CNCl, CNCHCl2, CHCl3, CH3NCl2, and NCl3, respectively. All compound identifications by MIMS were confirmed by analysis of standard solutions. All experiments were conducted using a bicarbonate buffer system (120 mg/L CaCO3) at pH 7.5; this water chemistry was selected to provide an aquatic matrix that was representative of swimming pool chemistry. Chlorination experiments were carried out in well-mixed, glass-stoppered, 250 mL flasks in the dark. For experiments in which the objective was the identification of volatile DBPs, organic-nitrogen compounds were added to achieve a concentration of 1.8 × 10-4 M. Solutions were chlorinated at a chlorine/precursor molar ratio (Cl/P) of 5.0 for 24 h. This Cl/P molar ratio was selected because it is well beyond the breakpoint based on reduced N (in the form of NH3) and is representative of the conditions of chlorination that are often used in swimming pools. The 24 h reaction period was selected because previous experiments had demonstrated that most (detectable) changes
in water chemistry were complete in a 24 h period. The resulting solutions were then subjected to MIMS analysis. For experiments in which the objective was the quantification of volatile DBPs, aqueous precursor solutions were freshly prepared by the addition of 1.0 mL of a model compound stock solution to 200 mL of an aqueous bicarbonate buffer solution to achieve a target precursor concentration of 1.8 × 10-5 M. Aliquots of a standardized sodium hypochlorite (NaOCl) stock solution were then added to the flasks. The 0.01 M HCl and 0.01 M NaOH were used to adjust the initial pH of the solution to 7.5. The reaction vessels had little headspace, and they were sealed to avoid volatilization and kept in the dark at room temperature. The concentrations of free and (apparent) combined chlorine were measured by DPD/FAS titration; MIMS was used to measure residual chlorine and volatile DBPs. Recreational water samples were collected from indoor and outdoor municipal swimming facilities. Pool water samples were immediately transported to the laboratory in sealed plastic bottles with little headspace for analysis by DPD/FAS titration and MIMS. The bottles used to collect and transport the pool water samples were opaque to ultraviolet radiation.
Results and Discussion Identification of Volatile DBPs. Initial experiments were conducted using relatively high concentrations of free chlorine and model organic-nitrogen compounds to aid in volatile DBP identification. A summary of the volatile DBPs identified in this study is provided in Figure 1. Figure 2 illustrates typical mass spectra obtained by MIMS analysis of chlorinated solutions of the four organic-nitrogen precursors. Trichloramine was identified in all four cases by the existence of a peak cluster at m/z 84 (N35Cl2•+), 86 (N35Cl37Cl•+), and 88 (N37Cl2•+) at an abundance ratio of 9:6:1, as well as a molecular ion peak at m/z 119 (N35Cl3•+). This spectral pattern was in agreement with spectra that were developed from NCl3 standard solutions, as well as previously published spectra developed using MIMS (14). As shown in Figure 2, urea appears to be the most active NCl3 precursor among these four model organic-nitrogen compounds. Chlorination of creatinine and L-histidine also yielded other volatile DBPs. Figure 2a presents the MIMS mass spectrum resulting from chlorination of aqueous creatinine. In addition to the peak clusters described previously that were associated with NCl3, three additional clusters were observed. The peak clusters located at m/z 98 (CH2N35Cl2•+)-100 (CH2N35Cl37Cl•+)-102 (CH2N37Cl2•+) and 99 (CH3N35Cl2•+)-101 (CH3N35Cl37Cl•+)-103 (CH3N37Cl2•+), each with abundance ratios of 9:6:1, and m/z 62 (CHN35Cl•+), 63 (CH2N35Cl•+), 64 (CHN37Cl•+ and CH3N35Cl•+), 65 (CH2N37Cl•+), and 66 (CH3N37Cl•+) suggested the formation of N,N-dichloromethylamine (dichloromethylamine). MIMS analysis of a standard solution of dichloromethylamine confirmed this finding. Although no reports of dichloromethylamine formation from the chlorination of creatinine were found in the literature, methylamine has been reported as a disinfection byproduct and has also been detected in swimming pool water samples (15). Methylamine will react with free chlorine to generate monochloromethylamine immediately by chlorine substitution; however, the results of these experiments described herein indicate that the formation of dichloromethylamine by chlorination of methylamine requires several hours to complete, as well as a Cl/P ratio of greater than 1. Unlike many other organic chloramines, which often yield DPD/FAS signals that correspond to inorganic dichloramine (NHCl2), monochloromethylamine and dichloromethylamine behave much like their inorganic chloramine analogues (NH2Cl and NHCl2, respectively) in the DPD/FAS titrimetric VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6733
FIGURE 2. Mass spectra of chlorinated samples of aqueous solutions of organic-nitrogen compounds: (a) creatinine, (b) L-histidine, (c) urea, and (d) L-arginine after 24 h. In each case, organic-nitrogen compounds were present at an initial concentration of 1.8 × 10-4 M; initial Cl/P ) 8.0 for L-arginine and Cl/P ) 5.0 for all other cases; pH ) 7.5. Note that the scales on the vertical (abundance) axes in these mass spectra have been adjusted to coincide with the largest abundance value in each spectrum. method. Specifically, monochloromethylamine yields an apparent NH2Cl signal in the DPD/FAS method, while dichloromethylamine yields an apparent NHCl2 signal in the same method. The health effects of dichloromethylamine are not known; however, exposure to this chemical during this research indicated a characteristic malodor, with a smell similar to that of trichloramine. In the case of L-histidine (Figure 2b), a peak cluster at m/z 74 (CNCH35Cl•+) and 76 (CNCH37Cl•+), with an abundance ratio of 3:1, along with a second cluster at m/z 82 (C35Cl2•+)84(C35Cl37Cl•+)-86(C37Cl2•+) indicated the formation of dichloroacetonitrile. A peak cluster at m/z 61 (CN35Cl•+) and 63 (CN37Cl•+), with an abundance ratio of 3:1, suggested the possible formation of cyanogen chloride. These peaks showed good agreement with mass spectra developed from CNCHCl2 and CNCl standard solutions, respectively. However, the peak corresponding to trichloramine was not obvious in this spectrum because trichloramine was present at a relatively low concentration under this condition as compared to the other volatile compounds. CNCHCl2 and CNCl have both been reported as DBPs that result from the reaction of free chlorine with natural organic-nitrogen compounds (16). CNCHCl2 has been identified as an irritant of the respiratory system and skin and a possible mutagen in humans (17). CNCl is also highly toxic, even at very low concentrations. The World Health Organization (WHO) has recommended that a maximum concentration of 70 µg/L (as cyanide) be used as a guideline for total cyanogen compounds in drinking water. As shown in Figure 2b, small amounts of chloroform were formed as a result of chlorination of L-histidine, as evidenced by the peak cluster at 83 (CH35Cl2•+)-85(CH35Cl37Cl•+)87(CH37Cl2•+). CNCHCl2 and NCl3 will also yield peaks at these same m/z values; however, the abundance signature at 83:85:87 was different than expected on the basis of any of 6734
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007
these compounds individually. This suggests that some combination of CHCl3, CNCHCl2, and NCl3 is formed as a result of the chlorination of L-histidine. The MIMS configuration used in this research yields mass spectral signals only for compounds that are able to pervaporate through a tubular silicone membrane. Therefore, this method allows for selective identification and quantification of volatile and semi-volatile constituents. As such, it is well-suited for analysis of volatile DBPs that may develop in a recreational water setting. Given that all the DBPs identified in these experiments are volatile, it is reasonable to hypothesize that they could also affect air quality in recreational water settings. Quantification of Volatile DBPs and Residual Chlorine. DPD/FAS titration and MIMS were employed in parallel to measure the residual chlorine and volatile DBPs that resulted from the chlorination of aqueous solutions containing organic-nitrogen precursors at a concentration of 1.8 × 10-5 M. These experiments were conducted over a range of Cl/P molar ratios (1.6 e Cl/P e 9.6) that is believed to be representative of those employed in recreational water settings. DPD/FAS titration consistently yielded false-positive measurements of inorganic combined chlorine residuals, which were attributed to the formation of organic chloramines. The volatile DBPs that resulted from reaction of these four organic-nitrogen precursors and free chlorine responded in the DPD/FAS titration predominantly as apparent dichloramine, with smaller amounts of apparent monochloramine; these observations were consistent with previous findings (14, 18). Free chlorine measurements by DPD/FAS titration and MIMS were in good agreement. The four volatile N-containing DBPs that were identified in the experiments involving relatively high precursor concentrations (1.8 × 10-4 M) were also detected by MIMS at low precursor concentrations (1.8 × 10-5 M).
FIGURE 3. Residual chlorine concentration as a function of Cl/P molar ratio after 1 h (bottom) and 96 h (top) chlorination of an aqueous solution containing creatinine (1.8 × 10-5 M) at pH 7.5. Residual chlorine (here defined as the sum of the concentrations of free chlorine and [organic or inorganic] chloramines) and volatile DBPs generated in an aqueous solution of creatinine (1.8 × 10-5 M) treated with free chlorine at pH 7.5 were quantitatively analyzed by both DPD/FAS titration and MIMS. Figure 3 illustrates the results of these measurements for 1 and 96 h of chlorination of creatinine. As described previously, dichloromethylamine yielded a signal that corresponded with NHCl2 in the DPD/FAS titration. According to the DPD/FAS titration results, the concentration of apparent dichloramine did not change substantially from 1 to 96 h post-chlorination. However, the MIMS results showed a significant increase of dichloromethylamine from 1 to 96 h chlorination. The concentration of dichloromethylamine measured by MIMS accounted for 80-90% of apparent dichloramine by the DPD/FAS method after 96 h of chlorination. These data also suggest that there are some nonvolatile (or perhaps low volatility) intermediates that are formed in the production of CH3NCl2. These data support previous research efforts that have revealed that the DPD/FAS method does not discriminate between volatile and nonvolatile aqueous constituents, nor does it discriminate among many organic chloramines or inorganic chloramines. In contrast, the MIMS configuration used in this work allows selective measurement of constituents that will diffuse through a silicon polymer membrane (i.e., volatile and semi-volatile compounds). However, this method does not allow measurement of low volatility compounds. The MIMS results for total residual chlorine were consistently lower than the corresponding measurements from DPD/ FAS titration. The difference between these total residual chlorine measurements was assumed to be attributable to low volatility DBPs. The time-course concentrations of free chlorine, trichloramine, and dichloromethylamine are shown in Figure 4 for chlorination of an aqueous solution containing creatinine. MIMS measurements indicated a dichloromethylamine concentration of less than 10 µg/L (as Cl2) after 1 h chlorination; however, the dichloromethylamine concentration increased steadily during the 24 h chlorination period, ultimately yielding a concentration of approximately 1.5 mg/L Cl2. The maximum yield of dichloromethylamine was approximately 66% of the initial creatinine, on a molar basis. The experiment was repeated at different Cl/P molar ratios, and dichloromethylamine formation was observed to be directly related to the initial concentration free chlorine (see Figure S1 in Supporting Information). Free chlorine decayed more rapidly at the beginning of the experiment than at the
FIGURE 4. Residual chlorine concentration as a function of time at Cl/P ) 8.0 chlorination of an aqueous solution containing creatinine (1.8 × 10-5 M) at pH 7.5.
FIGURE 5. Residual chlorine concentration as a function of Cl/P molar ratio after 1 h (bottom) and 96 h (top) chlorination of an aqueous solution containing L-histidine (1.8 × 10-5 M) at pH 7.5. end. Approximately 0.15 mg/L (Cl2) trichloramine was formed after 4 h chlorination of creatinine; the concentration of trichloramine was fairly stable after that point. In general, roughly 0.1 mg/L trichloramine (Cl2) was detected by MIMS after 96 h creatinine chlorination at Cl/P > 3.2. At these low concentrations, DPD/FAS titration generally does not yield accurate measurements of NCl3 because it is near the detection limit and because of interference by organic chloramines. As will be described next, the concentration of NCl3 in recreational water is often in the vicinity of 0.1 mg/L Cl2. Therefore, accurate quantification and identification of NCl3 in recreational water by DPD/FAS is likely to be difficult. In contrast, the detection limit of NCl3 by MIMS is less than 0.06 mg/L Cl2, and the NCl3 signal provided by a mass spectrometer is unlikely to be interfered with by common constituents in recreational waters. Among the organic precursors investigated in this study, L-histidine was found to be the most reactive toward free chlorine, as defined by the rate of free chlorine consumption (compare Figure 5 with Figure 3 and Supporting Information Figure S2). Two volatile DBPs, CNCHCl2 and CNCl, were detected by MIMS after L-histidine chlorination. At the end of the experiment (t ) 96 h), the MIMS signal for CNCHCl2 accounted for roughly 50-80% of the apparent dichloramine signal from DPD/FAS, depending on the Cl/P ratio. These results imply that CNCHCl2 was the major volatile DBP to VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6735
FIGURE 6. Residual chlorine concentration as a function of time for Cl/P ) 8.0 in chlorination of an aqueous solution containing L-histidine (1.8 × 10-5 M) at pH 7.5. Inset illustrates the dynamics of volatile DBP formation in greater detail. result from L-histidine chlorination. Approximately 0.2 mg/L (Cl2) CNCHCl2 and 0.02 mg/L (Cl2) CNCl were found to be generated at a molar ratio Cl/P of 8.0. As shown in Figure 6, the free chlorine concentration decreased steadily over the course of 96 h chlorination, while the CNCHCl2 concentration generally increased. NCl3 and CNCl concentrations both increased early in the reaction, reached a plateau, and then decreased (see inset in Figure 6). Roughly 0.22 mg/L (Cl2) NCl3 was formed after 4 h chlorination of L-histidine, followed by decay. Urea was found to yield relatively high concentrations of NCl3, even at molar ratios as low as Cl/P ) 1.6 (see Figure S2A in Supporting Information). For example, roughly 0.1 mg/L (Cl2) NCl3 was detected by MIMS after 1 h chlorination of urea at Cl/P ) 1.6. This observation was consistent with results from experiments involving high precursor concentrations. No other forms of residual chlorine were evident in chlorinated urea samples. Trichloramine was the only DBP that was identified by MIMS to result from chlorination of L-arginine (see Figure S2B in Supporting Information). No NCl3 was detected after 1 h chlorination of L-arginine at a Cl/P molar ratio from 1.6 to 9.6, and NCl3 was detectable only under conditions of Cl/P g 6.4. This behavior suggested that L-arginine is less active as a precursor to volatile DBP production than the other three model precursors. In general, chlorination of all four organic-nitrogen precursors was shown to yield volatile DBPs, even at low precursor concentrations that are believed to be representative of recreational waters. Inorganic chloramine, which was always observed as trichloramine (always in the presence of free chlorine), was found as a common byproduct in all cases. This finding, together with previously published results (19-27), suggests that many organic-nitrogen compounds can act as precursors to NCl3 formation. The formation of trichloramine was strongly dependent on the Cl/P molar ratio, the structure of the precursor, and the reaction time. It is also important to recognize that inorganic monochloramine (NH2Cl) and dichloramine (NHCl2) were always present below the detection limits for MIMS in these laboratory experiments. The MIMS detection limits for NH2Cl and NHCl2 were approximately 0.1 and 0.02 mg/L (Cl2), respectively (14). The low concentrations of these inorganic chloramines were probably attributable to the relatively high Cl/P molar ratios used in these experiments. Mechanisms of Volatile DBP Formation. Reactions between free chlorine and organic-nitrogen precursors have 6736
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007
been generalized to proceed via electrophilic substitution (19), combined with hydrolysis. In the initial stage, organicnitrogen precursors are generally chlorinated to N-monoor N,N-di-chlorinated forms by electrophilic attack of OCl- or HOCl. When the molar ratio of free chlorine to organic precursor exceeds the stoichiometric requirement for dichloro substitution, the organic N-mono- or N,Ndichloramine will often decompose to yield carbon dioxide, inorganic chloramines, nitriles, aldehydes, and chloroaldimines (20-22). Some generalizations can be identified in chlorination mechanisms. For example, electron-withdrawing groups can increase the acidity of a proton of C or N, thereby promoting chlorination of C or N at those positions by substitution (20). Good leaving groups (e.g., carboxylate) can also enhance electrophilic substitution. Furthermore, a leaving group near a chlorinated N atom in an organic-nitrogen precursor (e.g., carboxylate) will enhance the formation of nitriles or chloroaldimines. As an illustration of this behavior, it has been reported that nitriles and chloroaldimines are formed as a result of chlorination of R-amino acids (23-25). The mechanism of trichloramine formation can be assumed to involve simple substitution of Cl+. Most organic amines, including primary, secondary, and tertiary amines, can act as precursors to trichloramine formation. Nitriles can also function as trichloramine precursors. For example, trichloramine can be produced when cyanogen chloride reacts with free chlorine in water (26). As part of this study, we also found that detectable trichloramine was formed as a volatile byproduct when dichloroacetonitrile reacted with free chlorine at a molar ratio Cl/P ) 50 (see Figure S3 in Supporting Information). This observation suggests that NCl3 formation could be a minor reaction pathway of chlorination of dichloroacetonitrile. A proposed mechanism for the formation of dichloromethylamine from the chlorination of creatinine is illustrated in Scheme 1. The mechanism begins with chlorine substitution, followed by hydrolysis to yield urea and N-chlorosarcosine (CH3NClCH2COOH) as intermediates. Urea then reacts through a sequence of chlorine substitution and hydrolysis steps to yield trichloramine. N-Chloro-sarcosine is hypothesized to first undergo decarboxylation and dehydrochlorination to yield CH3NdCH2 as an intermediate, which itself undergoes hydrolysis to yield methylamine, which in turn undergoes N-dichloro substitution to yield dichloromethylamine. To test the validity of this hypothesized mechanism, an aqueous solution of sarcosine was exposed to free chlorine at a molar ratio of Cl/P ) 5.0, and dichloromethylamine was found as the major DBP (see Figure S4 in Supporting Information). Urea was observed as a byproduct of chlorination of creatinine by Tachikawa et al. (15). Dichlorourea was found to be formed in the chlorination of urea, and when a solution of dichlorourea was allowed to stand, it decomposed, yielding trichloramine as one of the products (27). A proposed mechanism for the formation of CNCl and CNCHCl2 from the chlorination of L-histidine (I) is illustrated in Scheme 2. As with other R-amino acids (22, 28), an initial chlorine substitution step is followed by dechlorination and decarboxylation to yield a nitrile (II). It is hypothesized that the two electron-withdrawing groups, sCtN and a heterocycle, will promote electrophilic attack of OCl- on the β-carbon of the intermediate (II). As a result, one of the protons of the β-carbon of nitrile (II) is substituted by chlorine to generate intermediate (III), and then with a further electrophilic attack of OCl- on this position, the C-C bond is broken to form dichloroacetonitrile and the heterocycle (IV). To test the validity of this hypothesized mechanism, an experiment was conducted involving aspartic acid (HOOC-
SCHEME 1. Proposed Mechanism for Formation of Trichloramine and Dichloromethylamine from Chlorination of Creatinine
SCHEME 2. L-Histidine
Proposed Mechanism for the Formation of Dichloroacetonitrile and Cyanogen Chloride from Chlorination of
TABLE 1. Volatile DBP Measurement in Samples of Recreational Water
c
samplea
NCl3b (mg/L Cl2)
CHCl3b (mg/L)
CNCHCl2b (mg/L)
free chlorinec (mg/L Cl2)
combined chlorinec (mg/L Cl2)
free chlorine to combined chlorine (molar ratio)
A B C D E F
0.08 0.07 0.09 0.16 0.1 0.07
0.07 0.13 0.14 0.08 0.13 0.08
0.01 0.03 0.01 0.02 0.01 0.01
1.5 1.95 0.68 6.52 5.92 1.72
1.34 0.25 1.36 1.76 1.28 0.76
1.12 7.80 0.50 3.70 4.63 2.26
a A, C, E, and F: indoor lap swimming pool; B: outdoor general use swimming pool; and D: outdoor recreation park. Analysis by DPD/FAS titration.
CH2CHNH2COOH) as the precursor, for which the carboxylate group on the β-carbon would also be expected to act as an electron-withdrawing group and a good leaving group. Chlorination of aspartic acid also resulted in the formation of dichloroacetonitrile. On the other hand, no dichloroacetonitrile was detected by chlorination of alanine
b
Analysis by MIMS.
(CH3CHNH2COOH) (see Figure S5 in Supporting Information). With free chlorine, dichloroglycine (V) could be formed by the cleavage of the heterocycle (IV) through the rupture of two C-N bonds. Several studies have documented CNCl formation to result from the chlorination of glyVOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6737
cine (16, 23) by the mechanism illustrated at the end of Scheme 2. Analysis of Recreational Water Samples. MIMS was also employed to analyze volatile DBPs in samples collected from public recreational water facilities. Water samples were collected from indoor and outdoor swimming pools in screwcapped bottles and transported back to the Environmental Engineering Laboratories at Purdue University to allow initiation of MIMS analyses within 1 h of collection. Table 1 provides a summary of volatile DBP measurements for samples collected from six public pools. These data are accompanied by measurements of residual chlorine concentration by DPD/FAS titration. Three volatile DBPs (chloroform [CHCl3], trichloramine, and dichloroacetonitrile) were detected and quantified from all six recreational water samples by MIMS. Among them, trichloramine and dichloroacetonitrile were also detected in the chlorination of model organic-nitrogen compounds. Dichloromethylamine was detected only in sample A and was present at a concentration of approximately 10 µg/L. Interestingly, sample A was collected from a natatorium that is used almost exclusively for lap swimming and high-level competitive swimming. Although no measurements of precursor concentrations were performed for the swimming pool samples, the use pattern for swimming pool A is consistent with a scenario where relatively high creatinine concentrations might be expected. Creatinine was the only precursor in this research that yielded dichloromethylamine as a result of chlorination. More generally, the volatile DBPs that were identified in the experiments with model organicnitrogen compounds were also detected in municipal pool water samples. However, no CNCl was detected in swimming pool samples, which may be attributed to the short half-life (about 1 h in the presence of 0.5 mg/L free chlorine, at 25 °C, pH 7) of CNCl in the presence of free chlorine (26). The data presented in Table 1 also suggest that the concentration ranges of these volatile products in actual recreational water facilities were relatively narrow, regardless of the concentrations of free chlorine and combined chlorine. This suggests that these volatile DBPs may be ubiquitous in chlorinated swimming pools, even in well-maintained facilities. These compounds are difficult to eliminate by simple chlorination, even shock chlorination. Therefore, additional treatment operations may be needed in recreational water settings to improve water and air quality, relative to these volatile DBPs. Collectively, the experiments involving chlorination of organic precursors and the analysis of swimming pool samples by MIMS indicate the presence of volatile DBPs, including NCl3 and several organic chloramines. Interestingly, NH2Cl and NHCl2 were not detected in samples from laboratory experiments or swimming pools. While the results of these experiments do not allow for the prediction of the rates of volatile DBP formation in swimming pools, the analysis by MIMS provides a clear indication that DBP formation in swimming pools is not limited to inorganic chloramines and haloforms. Therefore, future research should be aimed at further defining the reactions that are responsible for volatile DBP formation in swimming pools, as well as processes that may be used for removal or control of these compounds.
Acknowledgments The authors are grateful to the DuPont Experimental Station (Wilmington, DE) and the National Swimming Pool Foundation for financial support of this research.
Supporting Information Available Detailed descriptions of volatile DBP formation as well as additional information to support the hypothesized reaction 6738
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007
mechanisms. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Zwiener, C.; Richardson, D. S.; De Marini, M. D.; Grummt, T.; Glauner, T.; Frimmel, H. F. Drowning in disinfection byproducts? Assessing swimming pool water. Environ. Sci. Technol. 2007, 41, 363-372. (2) Yoon, J.; Jensen, J. N. Chlorine transfer from inorganic monochloramine in chlorinated wastewater. Water Environ. Res. 1995, 67, 842-847. (3) Scully, F. E., Jr.; Hartman, A. C.; Rule, A.; Leblanc, N. Disinfection interference in wastewaters by natural organic nitrogen compounds. Environ. Sci. Technol. 1996, 30, 1465-1471. (4) Beech, J. A.; Diaz, R.; Ordaz, C.; Palomeque, B. Nitrates, chlorates, and trihalomethanes in swimming pool water. Am. J. Publ. Health 1980, 70, 79-82. (5) Kim, H.; Shim, J.; Lee, S. Formation of disinfection byproducts in chlorinated swimming pool water. Chemosphere 2002, 46, 123-130. (6) Judd, S. J.; Jeffrey, J. A. Trihalomethane formation during swimming pool water disinfection using hypobromous and hypochlorous acids. Water Res. 1995, 29, 1203-1206. (7) Holzwarth, G.; Balmer, R. G.; Soni, L. The fate of chlorine and chloramines in cooling towers: Henry’s law constants for flashoff. Water Res. 1984, 18, 1421-1427. (8) Blatchley, E. R., III; Johnson, R. W.; Alleman, J. E.; McCoy, W. F. Effective Henry’s constants for free chlorine and free bromine. Water Res. 1992, 26, 99-106. (9) Bernard, A.; Carbonnelle, S.; Nickmilder, M.; de Burbure, C. Non-invasive biomarkers of pulmonary damage and inflammation: Application to children exposed to ozone and trichloramine. Toxicol. Appl. Pharmacol. 2005, 206, 185190. (10) Lagerkvist, J. B.; Bernard, A.; Blomberg, A.; Bergstrom, E.; Forsberg, B.; Holmstrom, K.; Karp, K.; Lundstrom, N.; Segerstedt, B.; Svensson, M.; Nordberg, G. Pulmonary epithelial integrity in children: Relationship to ambient ozone exposure and swimming pool. Environ. Health Perspect. 2004, 112, 17681771. (11) Carbonnelle, S.; Francaux, M.; Doyle, I.; Dumont, X.; de Burbure, C.; Morel, G.; Michel, O.; Bernard, A. Changes in serum pneumoproteins caused by short-term exposures to nitrogen trichloride in indoor chlorinated swimming pools. Biomarkers 2002, 7, 464-478. (12) Judd, S. J.; Black, S. Disinfection byproduct formation in swimming pool waters: A simple mass balance. Water Res. 2000, 34, 1611-1619. (13) Judd, S. J.; Bullock, G. The fate of chlorine and organic materials in swimming pools. Chemosphere 2003, 51, 869-879. (14) Shang, C.; Blatchley, E. R., III. Differentiation and quantification of free chlorine and inorganic chloramines in aqueous solution by MIMS. Environ. Sci. Technol. 1999, 33, 2218-2223. (15) Tachikawa, M.; Aburada, T.; Tezuka, M.; Sawamura, R. Occurrence and production of chloramines in the chlorination of creatinine in aqueous solution. Water Res. 2005, 39, 371379. (16) Shang, C.; Gong, W. L.; Blatchley, E. R., III. Breakpoint chemistry and volatile byproduct formation resulting from chlorination of model organic-nitrogen compounds. Environ. Sci. Technol. 2000, 34, 1721-1728. (17) Osgood, C.; Sterling, D. Dichloroacetonitrile, a byproduct of water chlorination, induces aneuploidy in drosophila. Mutat. Res. 1991, 261, 85-91. (18) White, G. C. Handbook of Chlorination and Alternative Disinfectants, 3rd ed.; Van Nostrand Reinhold: New York, 1992. (19) Armesto, X. L.; Canle, M. L.; Santaballa, J. A. R-Amino acid chlorination in aqueous media. Tetrahedron 1993, 49, 275284. (20) Young, M. S.; Uden, P. C. Byproducts of the aqueous chlorination of purines and pyrimidines. Environ. Sci. Technol. 1994, 28, 1755-1758. (21) Nweke, A.; Scully, F. E., Jr. Stable N-chloroaldimines and other products of the chlorination of isoleucine in model solutions and in a wastewater. Environ. Sci. Technol. 1989, 23, 989994. (22) Conyers, B.; Scully, F. E., Jr. Chloramines V: Products and implications of the chlorination of lysine in municipal wastewaters. Environ. Sci. Technol. 1997, 31, 1680-1685.
(23) Na, C.; Olson, T. M. Mechanism and kinetics of cyanogen chloride formation from the chlorination of glycine. Environ. Sci. Technol. 2006, 40, 1469-1477. (24) McCormlck, E. F.; Conyers, B.; Scully, F. E., Jr. N-Chloroaldimines. 2. Chlorination of valine in model solutions and in wastewater. Environ. Sci. Technol. 1993, 27, 255-261. (25) Conyers, B.; Scully, F. E., Jr. N-Chloroaldimines. 3. Chlorination of phenylalanine in model solutions and in wastewater. Environ. Sci. Technol. 1993, 27, 261-266. (26) Na, C.; Olson, T. M. Stability of cyanogen chloride in the presence of free chlorine and monochloramine. Environ. Sci. Technol. 2004, 38, 6037-6043.
(27) Dowell, C. T. The action of chlorine upon hydrazine, hydroxylamine, and urea. J. Am. Chem. Soc. 1919, 41, 124-125. (28) Mehrsheikh, A.; Bleeke, M.; Brosillon, S.; Laplanche, A.; Roche, P. Investigation of the mechanism of chlorination of glyphosate and glycine in water. Water Res. 2006, 40, 30033014.
Received for review April 13, 2007. Revised manuscript received July 20, 2007. Accepted July 30, 2007. ES070871+
VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6739
