Nitrile, Aldehyde, and Halonitroalkane Formation during Chlorination

and toxicity of disinfection byproducts in chlorinated swimming pools: An overview .... Tom Bond , Michael R. Templeton , Nurul Hana Mokhtar Kamal...
0 downloads 0 Views 362KB Size
Environ. Sci. Technol. 2007, 41, 1288-1296

Nitrile, Aldehyde, and Halonitroalkane Formation during Chlorination/Chloramination of Primary Amines SUNG HEE JOO AND WILLIAM A. MITCH* Department of Chemical Engineering, Yale University, Mason Lab 313b, 9 Hillhouse Avenue, New Haven, Connecticut 06520

The decreasing availability of pristine water supplies is prompting drinking water utilities to exploit waters impacted by wastewater effluents and agricultural runoff. As these waters feature elevated organic nitrogen concentrations, the pathways responsible for transformation of organic nitrogen into toxic nitrogenous disinfection byproducts during chlorine and chloramine disinfection are of current concern. Partially degraded biomolecules likely constitute a significant fraction of organic nitrogen in these waters. As primary amines occur in important biomolecules, we investigated formation pathways for nitrile, aldehyde, and halonitroalkane byproducts during chlorination and chloramination of model primary amines. Chlorine and chloramines transformed primary amines to nitriles and aldehydes in significant yields over time scales relevant to drinking water distribution systems. Yields of halonitroalkanes were less significant yet may be important because of the high toxicity associated with these compounds. Our results indicate that chloramination should reduce nitrile concentrations compared to chlorination but may increase the formation of aldehydes and halonitroalkanes at high oxidant doses.

Introduction Population growth has encouraged utilities to exploit waters impaired by agricultural runoff or wastewater effluents (1). Whether via direct dissolved organic nitrogen (DON) inputs from wastewater effluents or via algal blooms instigated by fertilizer runoff, such waters often feature higher DON concentrations (2). The occurrence of DON in water supplies results in organic chloramine formation during chlorine or chloramine disinfection. There are two primary concerns regarding organic chloramine formation in drinking water. First, because they are less effective disinfectants than inorganic chloramines, organic chloramine formation raises concerns whether monitoring total chlorine residual overestimates the extent of disinfection (3, 4). Second, the formation of toxic nitrogenous disinfection byproducts (N-DBPs) from reactions between chlorine and DON represents an emerging concern. In addition to nitrosamines, N-DBPs featuring nitro (-NO2; e.g., halonitromethanes) and nitrile (-CN; e.g., cyanogen chloride and haloacetonitriles) functional groups are of particular concern because of their toxicity. The mammalian cell genotoxicity of halonitromethanes exceeded even the halofuranone, MX * Corresponding author phone: (203)432-4386; fax: (203)432-4387; e-mail: [email protected]. 1288

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007

(5). Haloacetonitriles constituted 10% of the 50 DBPs predicted to be the most carcinogenic (6). Evaluating the potential to form N-DBPs is timely because utilities are considering switching from chlorination to chloramination to meet more stringent effluent limitations on trihalomethanes and haloacetic acids. Whether the use of an inorganic nitrogen-based disinfectant enhances toxic N-DBP formation is relevant to whether chloramination reduces the toxicity associated with disinfected drinking water. Our current understanding of organic chloramine decay and N-DBP formation is limited to specific low molecular weight analogues of N-DBP families under particular conditions. For example, halonitroalkane research has focused on trichloronitromethane (chloropicrin) formation during chlorination in the presence of nitrite. Trichloronitromethane may form by base-catalyzed addition of chlorine to low molecular weight nitrated organics (e.g., nitromethane) (79) formed by the reaction of ClNO2 with phenols (9) and humic acids (10). However, the importance of this pathway is unclear because free chlorine and nitrite are unlikely to coexist in significant quantities during disinfection. Two pathways have been proposed for nitrile formation. The “aldehyde pathway” involves the nearly quantitative reaction between monochloramine (NH2Cl) and an aldehyde (11). For example, cyanogen chloride forms by a nucleophilic attack of NH2Cl on formaldehyde to form chloroaminomethanol (HOCH2NHCl) that eliminates water and hydrochloric acid to yield cyanide. While this pathway may explain increased nitrile formation during chloramination of waters containing aldehydes (e.g., downstream of ozonation), this pathway does not address the effects of elevated DON concentrations. The focus of the “decarboxylation pathway” is the N-terminal amino group in amino acids or peptides (1221). When the N-terminal amino group is dichlorinated, elimination of hydrochloric acid followed by a concerted decarboxylation (i.e., decarboxylation coupled with chloride loss) forms a nitrile (Scheme 1 for cyanogen chloride formation from glycine). Alternatively, concerted decarboxylation of a mono-N-chlorinated amino group releases an aldehyde and ammonia. The importance of this pathway is unclear because free amino acids constitute only ∼10% of total amino acids (22, 23). Moreover, while nitriles have been synthesized via dehydrohalogenation of chlorinated primary amines (24), the importance of dehydrohalogenation under environmentally relevant conditions and time scales for primary amines lacking the respective β-carboxyl or carbonyl groups present in amino acid or polypeptide backbones should be clarified. Such primary alkylamines occur in important biomolecules, including the side chain of the amino acid lysine and the structural fatty acid phosphatidylethanolamine. A general understanding of the transformation of DON moieties into N-DBPs in the presence of chlorine and chloramines may allow the prediction of N-DBP products likely to form from particular DON functionalities in higher yields than the low molecular weight N-DBPs of current interest. If members of N-DBP families are found to exhibit similar levels of toxicity, such information would greatly facilitate risk assessment for this DBP family. In addition, as with nitrosamines (25-27), a greater understanding of formation pathways may enable the design of disinfection systems that minimize N-DBP formation. To initiate this process, we report the results of experiments describing the conversion of primary amines to nitriles, aldehydes, and halonitroalkanes during chlorination and 10.1021/es0612697 CCC: $37.00

 2007 American Chemical Society Published on Web 01/06/2007

SCHEME 1

chloramination. Because the biomass-derived organic matter in algal- or wastewater-impacted waters has been subjected to less processing than the organic matter in pristine waters, we expect important nitrogenous biochemicals to serve as significant DON constituents. Organic nitrogen moieties in biochemicals are dominated by amine (N(R)3; e.g., lysine, adenine) or amide (N(R)2-(CdO)R; e.g., peptide bonds, cytosine) functionalities. Amines are much more reactive with chlorine than amides (28), likely because the electronwithdrawing nature of the amide carbonyl deactivates the amide nitrogen toward electrophilic substitution. Monomethylamine (MMA) and n-propylamine (PA) were used as models for short- and long-chain primary amines in biochemicals because of the availability of some standards for their nitrile (cyanogen chloride and propionitrile), aldehyde (formaldehyde and propionaldehyde), and halonitroalkane (chloro-, dichloro-, and trichloronitromethane and 1-chloroand 1,1-dichloro-1-nitropropane) products.

Materials and Methods Materials. Cambridge Isotope Laboratories d3-MMA (98%) and 15N-ammonium chloride (98%), Sigma-Aldrich sodium cyanide (99.98%) and propionaldehyde (97%), Protocol Analytical trichloronitromethane (1 g/L), Orchid Helix Biotech chloronitromethane (90-95%) and dichloronitromethane (95%), TCI America propionitrile (99%) and 1-nitropropane (98%), Fisher Scientific formaldehyde (37%), VWR sodium hypochlorite (2-3%), and Alfa Aesar nitromethane (98+%) and n-propylamine (99+%) were used as received. CarboxenPDMS (75 µm) solid-phase microextraction (SPME) fibers were obtained from Supelco. All experiments were conducted in deionized water (18 MΩ) produced with a Millipore Elix 10/Gradient A10 water purification system. All glassware was baked for 3 h at 400 °C in a muffle furnace to remove organic contaminants prior to use. Analyses. Preparation of stock oxidant solutions was described previously (25) and is summarized in the Supporting Information. Most experiments were conducted under headspace-free conditions in 25-mL glass screw-cap vials with PTFE-lined septa. A typical experiment involved the syringe injection of MMA or PA through the septa into 25-mL buffered deionized water containing appropriate concentrations of chlorine or chloramines. Samples were quenched with potassium iodide and ascorbic acid at 2 and 3 times the initial normality of oxidant added, respectively; iodide behaves as a catalyst for organic chloramine dechlorination. For MMA reactions, glacial acetic acid was injected to lower the pH to 3.5-4.0, where cyanogen chloride and trichloronitromethane are stable (6). MMA samples were extracted for analysis within 1 h, while PA samples were extracted ∼2 h after quenching to permit adequate time for dechlorination of chlorinated propylamine. Dechlorination

was verified by observing the loss in gas chromatography mass spectrometry (GC-MS) peaks corresponding to the organic chloramines (see Figure SI-1 in the Supporting Information). For cyanogen chloride and halonitromethane analyses, most samples were transferred to 40-mL glass vials and were extracted by shaking for 10 min into 4-mL MtBE containing 1,2-dibromopropane as an internal standard. Samples were analyzed by gas chromatography (Alltech AT-1701 30 m × 0.25 mm × 1 µm column) with electron capture detection (GC-ECD) against authentic standards of halonitromethanes or standards of cyanogen chloride made freshly in 25-mL deionized water by mixing sodium hypochlorite with sodium cyanide in a 1:20 molar ratio. Method detection limits were 0.4 µM for cyanogen chloride and 1 nM for halonitromethanes. The column temperature program was 40 °C held for 15 min and ramping to 120 °C at 20 °C/min held for 4 min. Injection port and detector temperatures were 220 °C and 280 °C, respectively. For MMA samples treated with 15N-labeled chloramines, cyanogen chloride and trichloronitromethane were analyzed by GC-MS in the electron impact mode using the same column, oven, and injection port temperatures used for GCECD detection. After transferring a sample to a 40-mL glass vial, a Teflon-lined magnetic stir bar was added and the vial was capped with either Parafilm or a Teflon-lined septum. Results obtained using Parafilm were not significantly different from those obtained using a Teflon-lined septum. While stirring, the headspace was extracted with a CarboxenPDMS SPME fiber for 20 min at room temperature. The SPME fiber was desorbed in the injection port for 3 min. A similar SPME procedure was used for all nitromethane, propionitrile, nitropropane, and halonitropropane analyses. Propionitrile, nitropropane, chloronitropropane, and dichloronitropropane were analyzed in the electron impact mode, while nitromethane was analyzed by methanol chemical ionization. For propionitrile, nitropropane, chloronitropropane, and dichloronitropropane, the injection port was cooled to 110 °C. Standards for chloronitropropane and dichloronitropropane were not commercially available. These compounds were identified by comparing retention times and mass spectra against standards obtained by mixing nitropropane with an excess of sodium hypochlorite at pH 11. For standards exhibiting no chloronitropropane formation, dichloronitropropane concentrations in standards were calculated assuming 100% yield on the basis of the loss in nitropropane concentrations. Mass spectral interpretations for chloronitropropane and dichloronitropropane, the dichloronitropropane standard curve, and quantification ions for all analytes are provided in the Supporting Information. Additional aliquots were quenched with thiosulfate and were analyzed for formaldehyde and propionaldehyde folVOL. 41, NO. 4, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1289

lowing derivatization by o-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride by GC/MS in the chemical ionization mode using methanol (29). For MMA and PA analyses, experiments were conducted in 1-L reagent bottles. Aliquots were periodically removed, the residual chlorine was quenched, and MMA and PA were derivatized with 4-methoxybenzenesulfonyl chloride for analysis by a GC/MS procedure described previously (30) and summarized in the Supporting Information.

Results Byproduct Stability in the Presence of Oxidants. Unlike chloramines, hypochlorite catalyzes cyanogen chloride hydrolysis with half-lives under an hour under typical chlorine disinfection conditions; hydrolysis may proceed by elimination of cyanogen chloride’s chloride via a nucleophilic attack of hypochlorite on cyanogen chloride’s carbon (31). Because chloride elimination does not apply to nitriles like propionitrile, we assessed the stability of propionitrile in the presence of chlorine and chloramines. Neither free chlorine nor chloramines significantly enhanced propionitrile degradation (Figure 1A). The decreased degradation of propionitrile in the presence of dichloramine at pH 5 may result from partial reformation of propionitrile from the reaction of dichloramine with propionaldehyde, the likely hydrolysis product of propionitrile, by the “aldehyde pathway”. Regardless, after 2 days, approximately 50% of propionitrile had degraded at pH 9, but degradation was less significant at pH 5. We confirmed previous research (32) indicating that trichloronitromethane degrades with a half-life of ∼3 days over pH 5-9 in the absence of oxidants (Figure 1B). However, while we observed an ∼3 day half-life for trichloronitromethane in the presence of NH2Cl or free chlorine at pH 9, trichloronitromethane was stable in the presence of free chlorine or dichloramine at pH 5. Because of the lack of standards, the stability of dichloronitropropane was not assessed. Byproduct Formation from MMA. Various free chlorine concentrations were applied to 50 µM MMA in deionized water buffered at pH 7. Because Cl[+1] transfer from free chlorine to MMA is relatively fast (Table SI-3 in the Supporting Information), Cl:N ratios (i.e., the free chlorine to amine molar ratio) of 1 and 2 rapidly formed mono- and dichlorinated MMA, respectively (see Figure SI-1 in Supporting Information for their mass spectra). Although these species predominated under most experimental conditions, standards were not available. To quantify their degradation, these compounds were measured indirectly by measuring MMA after dechlorination of the solutions with 1.2 M ascorbic acid. While this procedure did not distinguish mono- and dichlorinated MMA, it enabled quantification of total residual MMA species (chlorinated and unchlorinated). Figure 2A provides firstorder observed degradation rate constants (i.e., kobs) for total MMA species as a function of the Cl:N molar ratio; kobs values were determined from plots of ln(C/Co) versus time, where C/Co represents total MMA species concentrations normalized by the initial MMA concentration (see Figure SI-4 in the Supporting Information for example plots). As the Cl:N molar ratio increased from 1 to 2 (where mono- and dichlorinated MMA predominates, respectively), kobs increased by a factor of 4 (t1/2 for MMA ∼12 days) but remained constant at higher Cl:N ratios. While monochlorinated MMA degraded more slowly than dichlorinated MMA, the lack of increase in kobs at Cl:N > 2 indicates that the rate-limiting step for the decay of dichlorinated MMA must not involve free chlorine. No phosphate buffer catalysis was observed over 10-50 mM phosphate buffer at pH 7. In addition, for solutions treated with free chlorine at Cl:N ) 20 at pH 5-9, kobs did not vary significantly with pH (Figure 2B). 1290

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007

FIGURE 1. Propionitrile and trichloronitromethane destruction in the presence and absence of free chlorine and chloramines in deionized water buffered with 20 mM phosphate buffer. (A) 200 µM propionitrile subjected to 2 mM free chlorine, 2 mM NH2Cl, or 0.9 mM NHCl2 for 2 days. (B) Trichloronitromethane. Circles ) 0.1 µM trichloronitromethane with no oxidant, squares ) 1 mM free chlorine applied to 1 µM trichloronitromethane at pH 5 or 400 µM free chlorine applied to 0.1 µM trichloronitromethane at pH 9, triangles ) 200 µM NHCl2 applied to 1 µM trichloronitromethane at pH 5 or 400 µM NH2Cl applied to 0.1 µM trichloronitromethane at pH 9. Unfilled ) pH 5 (NHCl2 for chloramines), filled ) pH 9 (NH2Cl for chloramines). Error bars represent the standard deviation of experimental replicates (n ) 2-3). The decay of total residual MMA was assessed after application of 500 µM dichloramine and 1000 µM NH2Cl to 50 µM MMA at pH 5 and pH 9, respectively, where the inorganic chloramine speciation was stable over the twoweek course of the experiment (Figure 2B). For NH2Cl, the kobs for MMA decay (8.8 ((2.4) × 10-7 s-1) was similar to that observed during chlorination (6.9 ((0.7) × 10-7 s-1). However, for dichloramine, kobs was lower by a factor of 3 (2.3 ((0.5) × 10-7 s-1). Total cyanogen chloride (i.e., 14N- and 15N-labeled) and halonitromethane yields were assessed by GC-ECD 1, 3, 5, and 7 days after applying free chlorine or chloramines to 200 µM MMA. In general, product yields calculated on the basis of 200 µM MMA varied by no more than a factor of 3 over time, reaching a maximum at 3 days (Figure SI-5 in the Supporting Information). For chlorination at pH 7, cyanogen chloride yields after 3 days increased with Cl:N molar ratio up to a maximum yield of ∼1.5% at Cl:N ) 2 (Figure 2C), where dichlorinated MMA predominates in the absence of

FIGURE 2. Results for MMA reactions with chlorine/chloramines. Clear ) free chlorine, black ) chloramines (NHCl2 at pH 5, NH2Cl at pH 7 and pH 9). First-order observed decay rates (kobs) of total MMA species for 50 µM MMA in deionized water buffered at pH 7 with 10-50 mM phosphate buffer as a function of (A) Cl:N molar ratio at pH 7 and (B) pH at Cl:N ) 20. Product yields from 200 µM MMA for application of free chlorine and chloramines after 3 days. At pH 7 as a function of Cl:N molar ratio: (C) cyanogen chloride and (D) trichloronitromethane. At Cl:N ) 1.75 as a function of pH: (E) cyanogen chloride and (F) trichloronitromethane. Error bars represent the standard error of replicate measurements (n ) 2-12). free chlorine. At higher Cl:N ratios, no cyanogen chloride was detected, presumably because of rapid hydrolysis of cyanogen chloride catalyzed by the residual free chlorine. Although cyanogen chloride yields generally were lower when

NH2Cl was applied at Cl:N e 2, yields continued to increase with Cl:N molar ratios up to Cl:N ) 10, where the yield was 3.4% (Figure 2C). Cyanogen chloride yields after 3 days at Cl:N ) 1.75 were higher by at least a factor of 4 at pH 7 VOL. 41, NO. 4, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1291

compared with pH 5 or 9 during chlorination or chloramination (Figure 2E). Trichloronitromethane yields were 2 orders of magnitude lower than those of cyanogen chloride during application of free chlorine to MMA at pH 7, and yields increased only slightly with Cl:N molar ratio (Figure 2D). When NH2Cl was applied at pH 7, trichloronitromethane yields always exceeded those observed during chlorination and increased dramatically with Cl:N molar ratio (Figure 2D). Nevertheless, trichloronitromethane yields were always at least 30 times lower than those of cyanogen chloride at pH 7. Between pH 5 and 9, trichloronitromethane yields increased with pH during chlorination, and to a lesser extent, during chloramination (Figure 2F). Except at pH 7, trichloronitromethane yields were higher during chlorination than during chloramination. Chloronitromethane and dichloronitromethane concentrations were negligible in all cases except two. During chlorination at Cl:N ) 1.75 at pH 5, the chloronitromethane yield was 0.05% after 7 days. During application of NH2Cl at Cl:N ) 4 at pH 7, the chloronitromethane yield was 0.04% after 1 day. When 2400 µM free chlorine or NH2Cl was applied to 600 µM MMA at pH 5, no significant nitromethane concentrations were observed after 1 or 5 days. Similar to chloronitromethane and dichloronitromethane, nitromethane concentrations likely were unimportant compared with trichloronitromethane; however, the 1 µM detection limit for nitromethane was high. During chlorination of 200 µM MMA, formaldehyde concentrations were low, exceeding the 0.5 µM concentration observed in blanks only under conditions of elevated pH or free chlorine doses. After 7 days of treatment with free chlorine at Cl:N ) 10 at pH 7 and at Cl:N ) 1.75 at pH 9, formaldehyde yields were 1.1 and 1.3%, respectively. Because chloramines react with aldehydes to form nitriles by the aldehyde pathway (11), 15N-labeled cyanogen chloride was measured by GCMS during application of 15N-labeled chloramines to assess the importance of aldehyde formation during chloramination. 15N-labeled cyanogen chloride generally was not detected. However, when 2400 µM 15NH2Cl was applied to 600 µM MMA at pH 5, 15N-labeled cyanogen chloride concentrations approximately doubled between 1 and 7 days to reach 1 µM, constituting 23% of the total cyanogen chloride; no significant concentrations of 15N-labeled cyanogen chloride were detected in 15N-chloraminated, buffered deionized water blanks. Byproduct Formation from PA. The decay of total PA species was determined by measuring PA after quenching chlorinated solutions. When 200 µM and 1000 µM free chlorine were applied to 50 µM PA at pH 7 (i.e., Cl:N ) 4 and 20, respectively), the first-order observed degradation rate constants (Figure 3A) were not significantly different between the two treatments (i.e., kobs ) 2.3 × 10-6 ((0.5 standard deviation) s-1. These kobs were ∼3.5 times faster than those for the decay of total MMA species. Compared with pH 7, kobs was slightly higher when 1000 µM free chlorine was applied at pH 9. However, when 1000 µM NH2Cl was applied at pH 9, kobs was nearly 5 times lower than during chlorination. Yields of 14N-propionitrile, propionaldehyde, and halonitropropanes were assessed by GC-MS 1, 4, and 7 days after chlorination or 15N-chloramination of 600 µM PA. Yields calculated on the basis of 600 µM PA generally varied by no more than a factor of 4 over time (Figure SI-6 in Supporting Information). While most byproduct concentrations reached a maximum at 4 days, 14N-propionitrile and dichloronitropropane accumulated over 7 days during chloramination. At pH 7, 14N-propionitrile yields after 4 days were 25-38% during chlorination but only 3-8% when 15NH2Cl was applied (Figure 3B). In both cases, yields increased as the Cl:N molar ratio increased from 1 to 2. However, during chlorination, yields declined slightly at higher Cl:N molar ratios, while they stayed 1292

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007

constant for 15N-chloramination. For both chlorination and 15N-chloramination at Cl:N ) 2, yields were highest at pH 7 (Figure 3E). For chlorination at pH 7, propionaldehyde yields after 4 days ranged between 9 and 29%, with the highest yield at a Cl:N molar ratio of 2 (Figure 3C). During application of 15NH Cl at pH 7, propionaldehyde yields after 4 days increased 2 with Cl:N molar ratios up to 4 but then leveled off at 23% (Figure 3C). As for 14N-propionitrile, propionaldehyde yields 4 days after chlorination or 15N-chloramination at Cl:N ) 2 were highest at pH 7 (Figure 3F). Yields from chlorination exceeded those from 15N-chloramination over pH 5-9. We measured 15N-propionitrile to examine whether the lower yields of propionaldehyde observed during 15N-chloramination were associated with 15N-propionitrile formation via the aldehyde pathway. No significant concentrations of 15Npropionitrile were observed under any conditions after 4 days. However, after 7 days, 15N-propionitrile yields were 2.3% and 4.0% at Cl:N molar ratios of 4 and 8 at pH 7 and 3% at Cl:N ) 2 at pH 9. These values corresponded to 16%, 31%, and 53% of the total propionitrile, respectively, and were the only conditions under which significant concentrations of 15N-propionitrile were observed. Both the absolute yields of dichloronitropropane and their trends with Cl:N molar ratios and pH were similar to those of trichloronitromethane. During chlorination at pH 7, dichloronitropropane yields increased modestly with Cl:N molar ratio after 4 days, to reach a maximum of 0.02% (Figure 3D). Dichloronitropropane yields 4 days after application of 15NH Cl at pH 7 were always higher than yields observed 2 during chlorination and increased more significantly with Cl:N molar ratio; the yield reached 0.13% at Cl:N ) 8. As with trichloronitromethane (Figure 2F), the dichloronitropropane yield 4 days after application of free chlorine or 15Nchloramines at Cl:N ) 2 was highest for chlorination at pH 9 (Figure 3G). Nitropropane and chloronitropropane were never significant.

Discussion Byproduct Formation Pathway. We propose Scheme 2A and 2B for byproduct formation during chlorination or chloramination of MMA and PA, respectively. Table SI-3 (Supporting Information) presents rate constants obtained from the literature for some of the MMA reactions. Following rapid transfer of two Cl[+1] from the oxidants to the amines, we propose that a key branching point occurs at which the dichlorinated amine either eliminates hydrochloric acid to form a chlorinated imine (k2) or is oxidized to form a nitroalkane (k12). When oxidation of dichlorinated amines occurs, halonitroalkanes rapidly form via sequential additions of Cl[+1] to the nitronate anion formed by deprotonation of the nitroalkane (7). Cl[+1] additions proceed more rapidly with increasing chlorine substitution as the electron-withdrawing chlorines enhance the acidity of halonitroalkanes. Indeed, only the maximally chlorinated analogues of the halonitroalkane byproducts (i.e., trichloronitromethane and dichloronitropropane) were observed in significant concentrations in our experiments. However, the low yields of halonitroalkanes indicate that the oxidation pathway is minor. On the basis of analogy to the decarboxylation pathway, a series of two hydrochloric acid eliminations from dichlorinated amines forms nitriles (i.e., hydrogen cyanide or propionitrile). In the case of MMA, cyanogen chloride forms from the chlorination of cyanide. Cyanogen chloride hydrolyzes rapidly to cyanate in the presence of free chlorine (31). For both amines, hydrolysis of the chlorinated imine formed via elimination of a single hydrochloric acid from the dichlorinated amine (k7) releases an aldehyde and NH2Cl. During 15N-chloramination, the aldehyde pathway leads

FIGURE 3. Results for PA reactions with chlorine/chloramines. (A) First-order observed decay rates (kobs) of total PA species for 50 µM PA in deionized water buffered at pH 7 with 20 mM phosphate buffer. Clear ) 200 µM free chlorine, black ) 1000 µM free chlorine, gray ) 1000 µM NH2Cl. Byproduct yields 4 days after application of free chlorine or NH2Cl to 600 µM PA as a function of Cl:N molar ratio: (B) 14N-propionitrile, (C) propionaldehyde, and (D) dichloronitropropane. Byproduct yields 4 days after application of free chlorine or chloramines (NHCl2 at pH 5, NH2Cl at pH 7 and pH 9) to 600 µM PA as a function of pH: (E) 14N-propionitrile, (F) propionaldehyde, and (G) dichloronitropropane. Clear ) free chlorine, black ) NH2Cl. Yields calculated on the basis of 600 µM PA. Error bars represent the standard deviation of experimental replicates (n ) 2-6). VOL. 41, NO. 4, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1293

SCHEME 2

to formation of 15N-labeled nitriles from aldehydes. The combined yields of propionitrile and propionaldehyde indicate that hydrochloric acid elimination from dichlorinated propylamine (k2) is much more important than oxidation (k12). Subsequent hydrolysis (k7) or hydrochloric acid elimination (k3) was approximately equally important during chlorination. However, hydrolysis was nearly twice as important during chloramination. The potential instability of the chlorinated imines over longer time periods is notable because previous research involving chlorination of amino acids indicated that chlorinated imines were stable over a 0.5-h time frame (16). Moreover, the relatively high propionaldehyde yields in the presence of excess 15NH2Cl (Figure 3C) and low 15N-propionitrile yields indicate that nitrile formation via a nucleophilic attack of NH2Cl on propionaldehyde (i.e., the aldehyde pathway) is less significant. Presumably, the electron-donating propyl group decreases the susceptibility of propionaldehyde to this pathway. 1294

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 4, 2007

In the case of MMA, the lack of significant formaldehyde formation indicates that the hydrolysis pathway is not important. However, the relatively low cyanogen chloride yields may underestimate the importance of the elimination pathway. Preliminary modeling (see Supporting Information) indicates that nearly all of the dichlorinated MMA may degrade via the elimination pathway but that low yields of cyanogen chloride are observed because of the instability of cyanogen chloride. Free chlorine catalyzes the hydrolysis of cyanogen chloride (Figure 2C). Even though chloramines do not catalyze cyanogen chloride hydrolysis (31), catalysis of hydrolysis by our phosphate buffer would occur with an ∼1 day half-life (Table SI-3). Overall, the half-lives of our dichlorinated amines (3.512 days) were much higher than those associated with the N-terminus of free amino acids (i.e.,