Environ. Sci. Technol. 2008, 42, 7778–7785
Evaluation of Functional Groups Responsible for Chloroform Formation during Water Chlorination Using Compound Specific Isotope Analysis W I L L I A M A . A R N O L D , * ,†,§ J A K O V B O L O T I N , ‡ U R S V O N G U N T E N , §,‡ A N D T H O M A S B . H O F S T E T T E R * ,‡ Department of Civil Engineering, University of Minnesota, 500 Pillsbury Dr. Southeast, Minneapolis, Minnesota 55455, Eawag, Swiss Federal Institute of Aquatic Science and ¨ berlandstrasse, 8600, Du Technology, 133 U ¨ bendorf, Switzerland, and Institute of Biogeochemistry and Pollutant Dynamics, ETH Zu ¨ rich, CH-8092, Zu ¨ rich, Switzerland
Received February 8, 2008. Revised manuscript received May 12, 2008. Accepted May 15, 2008.
Compound-specific isotope analysis was used to monitor the δ13C signature of chloroform produced upon the chlorination of model compounds representing natural organic matter functional groups (resorcinol, acetylacetone, acetophenone, phenol, and 2,4,6-trichlorophenol) and a natural water sample. For each model compound, a different apparent kinetic isotope effect was found for chloroform formation. Normal isotope effects were found for resorcinol, acetylacetone, and acetophenone, and ranged from 1.009 ( 0.002 to 1.024 ( 0.004. For the two phenols, an inverse effect was found (0.980 ( 0.004). Lake Zu¨rich water also had a inverse effect (0.997 ( 99%), acetylacetone (Sigma-Aldrich, >99%), 1,1,1trichloropropanone (1,1,1-TCP, Sigma-Aldrich; library product), sodium hypochlorite (NaOCl; Riedel-de Hae¨n, 6-14%, determined to be 68 mM by direct UV detection at 230nm, ) 100 M-1 cm-1), sodium sulfite (Merck, >98%), tetradecane (Alfa Aesar, >99%), methyl t-butyl ether (MTBE; Mallinckrodt, 99%), chloroform (Fluka; >99.5%), sodium dihydrogenphosphate monohydrate (Merck, >99%), disodium hydrogenphosphate dodecahydrate (Fluka, >99%). Ultrapurified water (18.2 MΩ; Barnstead) was used to prepare 5 mM pH 8 or 100 mM pH 9 phosphate buffer from
¨ rich water (LZW) was NaH2PO4 and Na2HPO4. Lake Zu collected at the Lengg drinking water treatment plant from a depth of 30 m. The dissolved organic carbon (DOC) concentration of the sample was 1.3 mg C/L, the alkalinity was 2.5 mM, and the pH was 8.1. Analytical Methods. Chloroform was analyzed using gas chromatography (Fisons GC8000) with electron capture detection (ECD). After incubating at 80 °C for 30 min, headspace samples (250 µL) were injected in split mode (1: 10) by a Fisons HS800 autosampler onto a 30 m × 0.23 mm × 1.8 µm film thickness RTX-624 column (Restek). The injector temperature was 170 °C and the detector temperature was 320 °C. The oven temperature program was 40 °C for 5 min, ramp 5 °C/min to 140 °C, 20 °C/min to 240 °C, and hold 10 min. For the 1,1,1-TCP hydrolysis experiments, both the parent compound disappearance and chloroform formation were measured in MTBE extracts using the method and instrumentation described by ref 17. CSIA of aqueous samples was performed on a GC-C-IRMS system (Trace GC/GC Combustion III interface/DeltaPlus XL or Delta V Plus isotope ratio mass spectrometer, Thermo Electron Corporation) coupled to a purge and trap concentrator (Velocity XPT, Tekmar Dohrmann) equipped with a liquid autosampler (AquaTek 70, Tekmar Dohrmann) according the procedure described in ref 18. Briefly, aqueous samples (40 mL) were purged for 11 min with an N2 flow of 40 mL/min and trapped on a VOCARB 3000 trap (Supelco) at room temperature. By heating the trap to 250 °C for 1 min, analytes were thermally desorbed and transferred to the GC at 200 mL min-1. The GC oven temperature program was 40 °C for 5 min, ramp 10 °C/min to 100 °C (hold 2 min), ramp 40 °C/min to 200 °C (hold 5 min) using He as carrier gas (200 kPa). Oxidation and reduction reactors of the combustion interface were maintained at 940 and 640 °C, respectively. CSIA of 1,1,1-trichloropropanone and chloroform from tetradecane extracts were performed with a split-splitless injector (inlet temperature 200 °C, split flow 50 mL min-1, split time 0.5 min). The GC oven temperature program was 40 °C for 5 min, ramp 10 °C/min to 140 °C (hold 2 min), ramp 40 °C/min to 200 °C (hold 10 min) using He as carrier gas (125 kPa). Solvent diversion via backflushing prevented tetradecane from entering the combustion interface between 10 and 25 min. All other parameters were kept as described above. All δ13C-values were derived from triplicate measurements with good precision (standard deviations usually < ( 0.5‰) and are reported relative to Vienna PeeDee Belemnite (VPDB). To avoid uncertainty of δ13C due to instrument nonlinearity (19), aqueous samples were diluted to the level of the least concentrated solution prior to analysis and measured at VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7779
constant peak amplitudes between 0.4 and 2V. Nevertheless, instrument response was linear within ( 1.4‰ over a range of 0.06-5 V (n ) 17). δ13C signatures of pure compounds used as starting material were analyzed by elementalanalyzer-IRMS. Chloroform Formation Experiments. All experiments were conducted at room temperature (22 °C). For the slow reacting systems (phenol, 2,4,6-trichlorophenol (an intermediate in chloroform production from phenol), acetophenone, Lake Zu ¨rich water), batch experiments were conducted following previously published procedures (6, 7). The required volume (1-2 L) of 5 mM pH 8 phosphate buffer was prepared and spiked with the desired concentration (4-10 µM) of the target compound. To initiate the reaction, NaOCl was added to the stirred solution to achieve a concentration of 150 µM (in excess to obtain pseudo-first order kinetics). For Lake Zu ¨ rich water, only the spiking of NaOCl was necessary. The solution was then rapidly transferred to glass vials using a dispenser. The vials were filled so that no headspace remained and sealed with Teflon lined caps. Vial sizes ranged from 10 to 40 mL nominal volume depending on the amount of sample needed for chloroform quantification (0.1-5 mL) and δ13C measurements (2-40 mL) as determined from preliminary experiments. These volumes were dictated by instrument detection limits/linear ranges and considerations that at earlier time points, less chloroform has been produced, so larger sample volumes are needed. At selected time points, samples were quenched with an aqueous sodium sulfite solution to halt the reaction. For chloroform concentration measurements, the sample was collected using a glass syringe and dispensed into a 10 mL (nominal volume) headspace autosampler vial. The vial contained 100 µL of 100 g/L sodium sulfite solution and, if required, a predetermined volume of phosphate buffer to dilute the sample such that the concentration was in the linear response range of the ECD. The total aqueous volume in the vial was always 5 mL. Triplicate samples for GC-IRMS analysis were either directly quenched by injection of 500-800 µL of 100 g/L sodium sulfite solution (for 40 mL vials containing expected levels of chloroform between 1-10 nmol) or by rapidly pouring the sample into a 40 mL purge-and-trap vial containing the quencher and phosphate buffer solution to dilute the sample such that ∼10 nmol C (chloroform) was in each vial. For resorcinol and acetylacetone (the fast reacting compounds), a continuous quench flow system described by ref 20 was used for the chlorination experiments in pH 8, 5 mM phosphate buffer. A solution containing 20 µM resorcinol or acetylacetone and one containing 300 µM sodium hypochlorite were fed into a mixing T at 5 mL/min each. The reaction solution after mixing thus contained 10 µM of the target compound and 150 µM sodium hypochlorite. After passing through a loop of known volume, the solution was quenched with 5.67 g/L sodium sulfite flowing at 5 mL/min. Knowing the loop volumes (eight in total) and flow rates allows calculation of the reaction times (0.4, 0.9, 1.6, 1.9, 3.2, 6.0, 11.9, and 23.5 s). Samples for chloroform concentration measurements were collected in 2 mL vials and sealed with PTFE lined septa. Based on preliminary experiments, subsamples were later diluted to a total volume of 5 mL in the 10 mL headspace autosampler vials for analysis via GC-ECD. Triplicate samples for IRMS analyses were collected directly into 40 mL purge-and-trap vials leaving no headspace. Some sets of vials contained a predetermined amount of phosphate buffer such that samples collected at all time points contained ∼10 nmol chloroform. Hydrolysis of 1,1,1-trichloropropanone (which leads to production of chloroform; trichloropropanone is also a potential intermediate, Scheme 1) was performed in 100 mM pH 9 phosphate buffer in a 50 mL glass syringe containing glass beads and a Teflon coated stirbar to facilitate mixing. 7780
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
To obtain the concentrations necessary for CSIA, the initial concentration of the 1,1,1-trichloropropanone was 50 mM. Samples (2 µL) were collected and injected into 1 mL MTBE for GC-ECD analysis. Samples (1 mL) for GC-IRMS analysis were extracted with 1 mL tetradecane. Data Analysis. As shown in Scheme 1, the formation of chloroform upon chlorination of organic precursors is a multistep process. In this study, only the formation of chloroform was measured. Thus, a simplified kinetic scheme was used to analyze the kinetic and isotopic data. First, using the concentration versus time data, a pseudo-first order rate constant for chloroform production and the final concentration of chloroform were fit using Scientist for Windows (v.2.01; Micromath Scientific Software). The chloroform concentration and the δ13C as a function of time data were then provided as input to Scientist for Windows, for the following equations: d[R - 12C] ) -k1[R - 12C] dt
(1)
k1 d[R - 13C] )[R - 13C] dt AKIECHCl3
(2)
d[12CHCl3] ) k1[R - 12C] dt
(3)
k1 d[13CHCl3] ) [R - 13C] dt AKIECHCl3
(4)
[CHCl3] ) [12CHCl3] + [13CHCl3] δ13C ) 1000
[
[13CHCl3]/[12CHCl3] 13
RVPDB
-1
]
(5)
(6)
where R-12C is the precursor reactant compound (for which isotopic fractionation occurs) leading to 12CHCl3, R-13C is the precursor compound leading to 13CHCl3, k1 is a pseudo-first order rate constant for chloroform formation, 13RVPDB is the isotope ratio of standard material (Vienna PeeDee Belemnite), and AKIECHCl3 is the apparent kinetic isotope effect for chloroform formation. The initial conditions for the set of differential equations are as follows. The initial concentration of R-12C is selected to match the final concentration of the chloroform predicted from the concentration versus time fit. For R-13C, the initial concentration is R-12C × (δ13C∞/1000 + 1) × (13RVPDB), where δ13C∞ is the δ13C of the chloroform upon completion of the reaction (i.e., t ) ∞). For 12CHCl3 and 13CHCl the initial concentrations are zero. The chloroform 3 concentration and the δ13C (which is an measurement of the relative amounts of 12C and 13C in the chloroform) as a function of time are then fit via a least-squares method to numerically integrated solutions of eqs 1–4, eqs 5 and 6, and the initial R-13C concentration formula using AKIECHCl3 and δ13C∞ as fitting parameters. k1 is fixed at the value determined for the fit of the chloroform concentration versus time data. Errors reported for these values are 95% confidence intervals. For the 1,1,1-trichloropropanone hydrolysis, both the parent disappearance and product formation could be analyzed. The chloroform data was analyzed as described above. The 1,1,1-trichloropropanone data was analyzed using eq. 7 (21): lnRC ) εC ⁄ 1000 × ln(C) + ln(RC,0 ⁄ C ε0c
⁄1000
)
(7)
where εc is the bulk isotopic enrichment factor, C is the concentration of the reactant, and RC, the isotope ratio of the reactant is RC ) 13RVPDB × (δ13C ⁄ 1000 + 1)
(8)
FIGURE 1. Chloroform formation upon the chlorination of (a) resorcinol and (b) phenol in 5 mM pH 8 phosphate buffer dosed with 150 µM sodium hypochlorite. Initial concentrations are the targeted values and the reported pseudo-first order rate constants and final chloroform concentrations are derived from the model fit (solid line).
TABLE 1. Rate Constants, Yields, δ13C at t=∞ (δ13C∞) and Apparent Kinetic Isotope Effects (AKIECHCl3) of Chloroform Produced from Model Precursors and a Natural Water Sample k 1 (s-1) phenol 2,4,6-trichlorophenol acetophenone resorcinol acetylacetone 1,1,1-trichloropropanone Lake Zu¨rich water
1.67 ((0.33) × 10-4 2.00 ((0.83) × 10-4 6.39 ((2.50) × 10-6 0.26 ( 0.11 0.46 ( 0.04 3.17 ((0.33) × 10-5d 1.72 ((0.85) × 10-4
CHCl3 yield per CHCl3 yield per molea mole Ca 0.051 0.075 1.1 0.84 0.95 1.0
0.0085 0.013 0.14 0.14 0.19 0.33 0.0061
δ13 C of substrate (‰)
δ13C ∞CHCl3 (‰)
AKIECHCl3 (–)
-28.3 ( 0.2b -28.8 ( 0.1 -31.8 ( 0.1 -28.1 ( 0.1 -28.1 ( 0.1 -20.5 ( 0.6e nmf
-44.0 ( 3.2c -52.7 ( 2.6 -22.6 ( 3.1 -30.6 ( 1.3 -31.1 ( 2.7 -61.3 ( 1.6 -37.0 ( 0.2
0.980 ( 0.004 0.983 ( 0.004 1.024 ( 0.004 1.009 ( 0.002 1.016 ( 0.005 1.014 ( 0.002 0.9971 ( 0.0004
a Yields are expressed as per mole of substrate and per mole of carbon in the substrate. The latter allows Lake Zu¨rich to be included based on the DOC. b Errors are (1 standard deviation. c Errors for the fitted δ13C and AKIE values are 95% confidence intervals as determined by Scientist for Windows. d For 1,1,1-trichloropropanone, the pseudo-first order rate constant for hydrolysis, rather than chlorination is given. This value was measured at pH 9 vs 8 for the other experiments. e Measured by GC/C/IRMS. f Not measured. For the NOM in Lake Zu¨rich water, such an analysis is not possible.
The AKIETCP of 1,1,1-TCP hydrolysis via an addition/ elimination mechanisms was then found using AKIETCP )
1 1 + λ · εC ⁄ 1000
(9)
where λ equals 3 as a correction for isotopic dilution of the reaction at the carbonyl-C by two additional, nonreactive C atoms in terms of the addition step to the carbonyl (see details in ref 15).
Results Chloroform Formation from Model Precursors. Chloroform production as a function of time for phenol and resorcinol upon chlorination is shown in Figure 1 together with fitted chloroform final concentrations. Fitted values for the rate constants are given in Table 1. Similar plots for chlorination of the other selected model compounds and the hydrolysis of 1,1,1-trichloropropanone are given in the Supporting Information (SI, Figure S1). The reactions of resorcinol and acetylacetone were complete in seconds, whereas those for phenol, 2,4,6-trichlorophenol, and acetophenone occurred over hours to days. The second-order rate constant for 1,1,1trichloropropanone hydrolysis was 3.8 ( 0.4 M-1 s-1. For phenol and 2,4,6-trichlorophenol, chloroform yields (determined by predicted chloroform concentrations at t ) ∞) were 0.051 and 0.075 mol CHCl3 per mole of substrate, respectively. For the chlorination of resorcinol, acetylacetone, and acetophenone, and the hydrolysis of 1,1,1-trichloro-
propanone, yields were essentially 1 mol (0.85-1.1; see Table 1) CHCl3 per mol of substrate. Chloroform Isotope Signature from Model Precursors. The δ13C signatures of the chloroform produced upon chlorination of the five model precursors and the hydrolysis of the 1,1,1-trichloropropanone are shown in Figure 2. Given in Table 1 are the calculated AKIECHCl3 values and the calculated δ13C∞ values for chloroform. The δ13C of the parent compounds determined by elemental analysis is also shown. Resorcinol, acetylacetone, and acetophenone display normal kinetic isotope effects (i.e., AKIECHCl3 > 1, indicating that 12C reacts faster than 13C, see eqs 3 and 4). The chloroform arising from the hydrolysis of 1,1,1-trichloropropanone, also displays a normal AKIECHCl3. The process giving rise to chloroform from phenol and 2,4,6-trichlorophenol has an inverse kinetic isotope effect (i.e., AKIECHCl3 < 1, indicating that 13C reacts faster than 12C, see eqs 3 and 4). The values for phenol and 2,4,6-trichlorophenol are statistically identical. During the hydrolysis of 1,1,1-trichloropropanone, the δ13C values of the parent compound were also measured. The resulting enrichment factor was -13.9 ((3.6)‰, giving an AKIETCP of 1.044 ((0.012) (eq 9). The relevant plot is shown in Figure S2. Note that this value reflects the KIE for the carbonyl carbon, whereas the 1.014 ( 0.002 measured for chloroform is for the leaving group carbon. Chloroform from Lake Zu ¨rich Water. Chloroform production and the δ13C signature for the chloroform produced upon chlorination of Lake Zu ¨ rich water are shown in Figure VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7781
FIGURE 2. The δ13C of chloroform produced upon chlorination of (a) resorcinol, (b) acetylacetone, (c) acetophenone, (d) 1,1,1-trichloropropanone, (e) phenol, and (f) 2,4,6-trichlorophenol. All experiments were conducted at pH 8 in 5 mM phosphate buffer with 150 µM sodium hypochlorite, except the hydrolysis of 1,1,1-trichloropropanone (100 mM pH 9 phosphate buffer). Lines are model fits, and error bars are one standard deviation. Note the different time scales for panels a-b and c-f. 3. For Lake Zu ¨ rich water, 0.66 µM (79 µg/L) of chloroform was produced. The other relevant fitted parameters are given in Table 1. An inverse apparent kinetic isotope effect is observed for the chloroform produced from Lake Zu ¨ rich water.
Discussion Chloroform Formation. The rates of chloroform production were comparable to those found in previous studies (3, 6, 8). Similarly, the high yields for acetophenone, acetylacetone, and resorcinol were expected based on previous studies (3–6). The yields for phenol, 2,4,6-trichlorophenol, and Lake Zu ¨rich water were 12-50% lower than those reported previously (0.10 mol CHCl3 per mol phenol or 2,4,6-trichlorophenol; 97 7782
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
µg/L chloroform for Lake Zu ¨ rich water with 1.4 mg/L DOC 6, 7). The present study monitored the formation over shorter time scales (to obtain the key time frame for δ13C measurements) and predicted the expected concentrations for long reaction times, whereas the previous studies measured over longer time scales. On a per mole C basis, the chloroform production from the Lake Zu ¨ rich water and phenols are similar (Table 1). The hydrolysis of 1,1,1-trichloropropanone (k ) 3.8 M-1 -1 s ) occurred more slowly than expected based on previous studies of this compound (k ) 13-23 M-1 s-1; refs 17, 22). Preliminary experiments at concentrations similar to those used previously gave a similar rate constants, suggesting the
FIGURE 3. Chloroform formation (a) and chloroform δ13C (b) upon the chlorination of Lake Zu¨rich water. The water was dosed with 150 µM sodium hypochlorite. Lines are model fits used to determine KIE and δ13C∞ values. The error bars are one standard deviation. high concentration of 1,1,1-trichloropropanone in the experiment (50 mM) may be somehow responsible for the difference. Kinetic Isotope Effects for Precursor Identification. Individual precursor compounds give specific AKIECHCl3 values suggesting that, even in the absence of detailed knowledge of the reaction mechanisms, precursors can be identified by trends in chloroform 13C enrichment or depletion. This indicates that AKIECHCl3 values may reveal the functional groups responsible for chloroform produced upon chlorination of NOM. The 13C trends observed for chloroform might also be indicative for other THMs or DBPs. The two ketones tested (acetophenone and acetylacetone) react at very different rates, but both of the AKIECHCl3 values are near 1.02. Thus, this may be a typical isotope effect for ketones as precursors of chloroform. Phenol and 2,4,6-trichlorophenol also have identical AKIECHCl3 values of about 0.98. Another potential use is to identify which type of functional group is responsible for the production of chloroform from NOM over a specific time scale. Both resorcinol and acetylacetone rapidly produce chloroform, but their AKIECHCl3 values are significantly different at the 95% confidence interval. Similarly, the AKIECHCl3 values and isotopic enrichment trends
are opposite for acetophenone and the phenols, providing a potential diagnostic tool. 1,1,1-Trichloropropanone is a potential intermediate in chloroform formation (Scheme 1), so studying this compound may lead to mechanistic insights and information about competing rate-limiting steps. The AKIECHCl3 for the chloroform produced from the hydrolysis of 1,1,1-trichloropropanone is statistically different from all of the other species tested except for acetylacetone, and these reactions are over markedly different time scales. This observation is discussed in further detail below. The fitted AKIECHCl3 resulting from 1,1,1-trichloropropanone hydrolysis (1.014 ( 0.002) for the breaking of a C-C bond is somewhat higher than those for 15N/14N (1.004) and 18O/16O (1.006-1.009) in amide (C-N cleavage; leaving group sNR2) or ester (C-O cleavage; leaving group sOR) leaving groups during base catalyzed hydrolysis (23). Given the similar masses of C, N, and O, these would be expected to be similar, but we cannot comment on any trend between the three systems. The hydrolysis AKIE determined for the carbonyl C of 1,1,1-trichloropropanone is larger than that for the leaving group, consistent with values for amide (1.032) and ester (1.034-1.043) hydrolysis reported previously (23). Lake Zu ¨rich Water. Chlorination of Lake Zu ¨ rich water gives rise to a small inverse AKIECHCl3 (0.9971 ( 0.0004, Table 1). A qualitative comparison of this isotope effect with those of NOM model compounds suggests that phenolic moieties may be responsible for chloroform production in this system. This observation is in agreement with previous interpretations of kinetic data regarding chloroform formation from Lake Zu ¨ rich water and other natural waters (6, 7, 11). The significant difference in the AKIECHCl3 values (0.980 ( 0.004 vs 0.9971 ( 0.0004), however, may be an indication for chloroform formation from simultaneous reactions of phenols and other slowly reacting THM precursors with normal AKIEs (e.g., ketones). Although the AKIECHCl3 is statistically different from 1, it is very close to 1. Thus, further work is necessary with additional natural water samples to verify the inverse AKIE and the potential role of phenolic moieties in chloroform formation. Even though it was concluded from previous investigations based on kinetic information that phenolic moieties might be responsible for the slow chloroform formation over longer time scales, the isotopic data indicate that this is an oversimplification of the underlying processes. In future studies other factors such as parallel reactions or substituents on the aromatic moieties need to be investigated. Mechanistic Insights. A comparison of δ13C values of chloroform calculated for t ) ∞ with the initial δ13C of NOM or precursor material gives additional insight into the reactions contributing to the decomposition of the precursor molecules versus pathways responsible for chloroform formation. For a reaction with a complete carbon mass balance, the final δ13C of the product should equal that of the reactant. For resorcinol and acetylacetone, the δ13C values for the substrate and δ13C∞ of chloroform are similar. While there is not a complete carbon mass balance (only one of the carbons in the parent molecule ends up in the chloroform), this observation is consistent with the high yields of chloroform formation per mol of precursor (chloroform yields approaching unity, Table 1), if one assumes that the isotope ratio at the C-atom becoming chloroform is, on average, the same as that for the whole molecule. Deviations of δ13C∞ of chloroform toward more enriched and more depleted 13C contents compared to the initial precursor δ13C-values, in contrast, are indicative of an isotope sensitive branching of competing reaction pathways (24). For example, a chlorinated intermediate reacting to chloroform can also be transformed in parallel to another compound. The rate and isotope fractionation of such a VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7783
parallel reaction will determine both the concentration of the intermediate available for chloroform formation as well as the intermediate’s isotopic composition from which the δ13C of chloroform will finally evolve. In the case of phenols, the chloroform produced is much more depleted in 13C (i.e., δ13C∞ is more negative than the δ13C of the starting material, Table 1) than the original material. This suggests that the isotope fractionating reaction leading to chloroform was accompanied by a faster transformation of the chlorinated intermediate to another product which preferentially removed 13C-containing intermediates, thus leaving an isotopically light intermediate behind for chloroform formation. Note that such isotope sensitive branching is determining δ13C∞ of chloroform regardless of whether chloroform is produced via a normal or inverse isotope effect. From the present data, however, it is neither possible to speculate on the isotope effect of the parallel reactions that do not yield chloroform nor on their rate relative to the one of THM production. The hydrolysis rate constant and 13C fractionation of chloroform derived from 1,1,1-trichloropropanone hydrolysis provide important reference values for the addition/ elimination mechanism proposed as the predominant pathway to chloroform. In this mechanism, the -CCl3 is the leaving group after base-catalyzed attack at the neighboring carbonyl C, and its rate can be determined by both nucleophilic attack and departure of the -CCl3 leaving group (25). If chloroform arises from addition/elimination reactions of 1,1,1-trichloropropanone-like functional groups for all substrates, one would expect similar rates of chloroform formation for all substrates. The range of chloroform formation rates from different model NOM functional groups (Table 1) indicates that the chloroform rate is not solely dictated by addition/elimination mediated breakdown of similar functional groups. In the experiments with 1,1,1-trichloropropanone, OH- was the predominant nucleophile. In all the other experiments, chloroform formation is initiated by OCl-, which is a much better nucleophile than hydroxide (26, 27). This may partially explain the more rapid rates in chlorinated experiments but cannot wholly account for the variability among the compounds tested. Leaving the (chloro)phenol experiments aside, the very different rates of chloroform formation (kl ranges from 10-6 to 10-1 s-1) and normal AKIECHCl3 values between 1.009 and 1.024 seem to argue against a mechanism in which all NOM precursors react via a common intermediate. This supports the hypothesis that rates of THM formation originate from the reactivity of intermediates, rather than the starting material. The variability of AKIECHCl3 can be taken as evidence that -CCl3 elimination is not equally rate limiting (i.e., there is masking from other non- or less-fractionating steps in the sequence) in all experiments and that the molecular structures of chloroform precursors might also differ in structure from 1,1,1-trichloropropanone. Using a AKIECHCl3 value of 1.014 ( 0.002 from the hydrolysis of 1,1,1-trichloropropanone as a benchmark, the elimination of the leaving group from chlorinated resorcinol breakdown products (AKIECHCl3 ) 1.009 ( 0.002) would not be as rate limiting as from chlorinated intermediates from acetylacetone (AKIECHCl3)1.016 ( 0.005). This behavior is reflected in a smaller AKIECHCl3 measured with resorcinol as starting material. Nevertheless, the observed isotope effects would still agree with the addition/ elimination pathway determining the isotope signatures of THMs for resorcinol and acetylacetone. The acetophenone AKIECHCl3, however, is rather high to be associated with a leaving group isotope effect. We hypothesize that in this case a slow reaction preceding the addition/elimination step, that is the tauterimization/chlorination of the methyl C group, is responsible for the observed isotope fractionation in chloroform. 7784
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 21, 2008
Isotope effects for chloroform produced from (chloro)phenol, in contrast, are markedly distinct from the other model functional groups. The moderately large inverse AKIECHCl3 indicates a fundamental difference in the reaction kinetics leading to chloroform. This inverse isotope effect can be taken as qualitative evidence for a reaction in which the bond(s) to the C atom winding up in chloroform experience looser binding in the transition state than in the ground state (24). A typical example is a weakening of bonds to C atoms in transitions from C-sp2 (CdC) to C-sp3 (CsC) as occurs during the cleavage of the aromatic ring. Such a pathway would be consistent with the identical AKIECHCl3 values reported for phenol and 2,4,6-trichlorophenol and consistent with earlier observations that both compounds exhibit the same rate-limiting reaction step in reaction to chloroform (6). Implications for Chloroform Formation from Natural Precursor Material. The AKIECHCl3 values were determined in this study using a simplified model. This was done because the chlorination of NOM moieties is a complex, multistep process for which, with one exception, only a reaction product, chloroform, was monitored. The fitting procedure does appear to give reasonable fits to the data, which demonstrates that such an analysis can provide measures of isotope effects for chloroform formation from intermediates arising during NOM chlorination. Clearly identifying such transient chlorinated compounds and monitoring their δ13C signatures throughout the reaction sequence are necessary to gain further insight into the mechanisms of chloroform formation and rate limiting and isotope fractionating steps. Once this information is available, CSIA of chloroform is likely to provide a new avenue to assess possible precursor structures and their reactive functional groups beyond those obtained from kinetic, spectroscopic, or organic matter fractionation studies alone or in combination. The inverse ¨ rich water AKIECHCl3 found for (chloro)phenols and Lake Zu provides support for previous interpretations of kinetic studies of chlorination. The ability of CSIA to qualitatively identify phenols as the moiety leading to chloroform upon NOM chlorination also lends credibility to our approach. With additional effort, CSIA may provide mechanistic understanding of DBP formation from NOM and enhanced predictive capabilities beyond those allowed by current techniques.
Acknowledgments Thanks to Lisa Salhi for technical assistance and to Werner Angst and Rene´ Schwarzenbach for their comments on the manuscript. Financial support of WAA by Eawag is acknowledged. We also thank the reviewers for their helpful comments.
Supporting Information Available Chloroform formation upon chlorination the acetophenone, acetylacetone, and 2,4,6-trichlorophenol, and the hydrolysis of 1,1,1-trichloropropanone; plot of ln RC vs ln C for 1,1,1trichloropropanone. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Rook, J. J. Formation of haloforms during chlorination of natural waters. Water Treat. Exam. 1974, 23 (2), 234–243. (2) Krasner, S. W.; McGuire, M. J.; Jacangelo, J. G.; Patania, N. L.; Reagan, K. M.; Aieta, E. M. The occurrence of disinfection byproducts in U.S. drinking water. J. Am. Water Works Assoc. 1989, 81 (8), 41–53. (3) Boyce, D. D.; Hornig, J. F. Reaction pathways of trihalomethane formation from the halogenation of dihydroxyaromatic model compounds for humic acid. Environ. Sci. Technol. 1983, 17, 202–211.
(4) De Laat, J.; Merlet, N.; Dore, M. Chlorination of organic compounds: chlorine demand and reactivity in relationship to the trihalomethane formation. Water Res. 1982, 16, 1437–1450. (5) Dore, M.; Goichon, J. Etude d’une methode d’evaluation globale des precurseurs del la reaction haloforme. Water Res. 1980, 14, 657–663. (6) Gallard, H.; von Gunten, U. Chlorination of phenols: Kinetics and formation of chloroform. Environ. Sci. Technol. 2002, 36, 884–890. (7) Gallard, H.; von Gunten, U. Chlorination of natural organic matter: kinetics of chlorination and of THM formation. Water Res. 2002, 36, 65–74. (8) Deborde, M.; von Gunten, U. Reactions of chlorine with inorganic and organic compounds during water treatmentKinetics and mechanisms: A critical review. Water Res. 2008, 42, 13–51. (9) Rios, R. V. R. A.; da Rocha, L. L.; Vierira, T. G.; Lago, R. M.; Augusti, R. On-line monitoring by membrane introduction mass spectrometry of chlorination of organic in water. Mechanistic and kinetic aspects of chloroform formation. J. Mass Spectrom. 2000, 35, 618–624. (10) Hua, G.; Reckow, D. A. Characterization of disinfection byproduct precursors based on hydrophobicity and molecular size. Environ. Sci. Technol. 2007, 41, 3309–3315. (11) Korshin, G. V.; Benjamin, M. M.; Chang, H.-S.; Gallard, H. Examination of NOM chlorination reactions by conventional and stop-flow differential absorbance spectroscopy. Environ. Sci. Technol. 2007, 41, 2776–2781. (12) Elsner, M.; Cwiertny, D. M.; Roberts, A. L.; Sherwood Lollar, B. 1,1,2,2-Tetrachloroethane reactions with OH-, Cr(II), granular iron, and a copper-iron bimetal: insights from product formation and associated carbon isotope fractionation. Environ. Sci. Technol. 2007, 41, 4111–4117. (13) Hofstetter, T. B.; Neumann, A.; Arnold, W. A.; Hartenbach, A. E.; Bolotin, J.; Cramer, C. J.; Schwarzenbach, R. P. Substituent effects on nitrogen isotope fractionation during abiotic reduction of nitroaromatic compounds. Environ. Sci. Technol. 2008, 42, 1997– 2003. (14) VanStone, N.; Elsner, M.; Lacrampe-Couloume, G.; Mabury, S.; Sherwood Lollar, B. Potential for identifying abiotic chloroalkane degradation mechanisms using carbon isotopic fractionation. Environ. Sci. Technol. 2008, 42, 126–132. (15) Elsner, M.; Zwank, L.; Hunkeler, D.; Schwarzenbach, R. P. A new concept linking observable stable isotope fractionation to transformation pathways of organic pollutants. Environ. Sci.
Technol. 2005, 39, 6896–6916. (16) Hofstetter, T. B.; Reddy, C. M.; Heraty, L. J.; Berg, M.; Sturchio, N. C. Carbon and chlorine isotope effects during abiotic reductive dechlorination of polychlorinated ethanes. Environ. Sci. Technol. 2007, 41, 4662–4668. (17) Chun, C. L.; Hozalski, R. M.; Arnold, W. A. Degradation of drinking water disinfection byproducts in the presence of synthetic goethite and magnetite. Environ. Sci. Technol. 2005, 39, 8525–8532. (18) Zwank, L.; Berg, M.; Schmidt, T. C.; Haderlein, S. B. Compoundspecific carbon isotope analysis of volatile organic compounds in the low-microgram per liter range. Anal. Chem. 2003, 75, 5575–5583. (19) Sherwood Lollar, B.; Hirschorn, S. K.; Chartrand, M. M. G.; Lacrampe-Couloume, G. An approach for assessing total instrumental uncertainty in compound-specific carbon isotope analysis: Implications for environmental remediation studies. Anal. Chem. 2007, 79, 3469–3475. (20) Buffle, M.-O.; Schumacher, J.; Salhi, E.; Jekel, M.; von Gunten, U. Measurement of the initial phase of ozone decomposition in water and wastewater by means of a continuous quenchflow system: Application to disinfection and pharmaceutical oxidation. Water Res. 2006, 40, 1884–1894. (21) Scott, K. M.; Lu, X.; Cavanaugh, C. M.; Liu, J. S. Optimal methods for estimating kinetic isotope efects from different forms of the Rayleigh distillation equations. Geochim. Cosmochim. Acta 2004, 68, 433–442. (22) Croue, J. P.; Reckow, D. A. Destruction of chlorination byproducts with sulfite. Environ. Sci. Technol. 1989, 23 (11), 1412–1419. (23) Marlier, J. F. Multiple isotope effects on the acyl group transfer reactions of amides and esters. Acc. Chem. Res. 2001, 34, 283– 290. (24) Melander, L.; Saunders, W. H. Reaction Rates of Isotopic Molecules; John Wiley & Sons: New York, 1980. (25) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; 2nd ed.; John Wiley & Sons: New York, 2003. (26) Edwards, J. O.; Pearson, R. G. The factors determining nucleophilic reactivities. J. Am. Chem. Soc. 1962, 84, 16–24. (27) Jencks, W. P.; Carriulo, J. Reactivity of nucleophilic reagents toward esters. J. Am. Chem. Soc. 1960, 82, 1778–1786.
ES800399A
VOL. 42, NO. 21, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7785
