Environ. Sci. Technol. 2007, 41, 2387-2394
Formation of Chloroform and Other Chlorinated Byproducts by Chlorination of Triclosan-Containing Antibacterial Products E. MATTHEW FISS, KRISTA L. RULE, AND PETER J. VIKESLAND* The Charles E. Via, Jr. Department of Civil and Environmental Engineering, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
Triclosan is a widely used antibacterial agent found in many personal hygiene products. Although it has previously been established that pure triclosan and free chlorine readily react, interactions between triclosan-containing consumer products and free chlorine have not previously been analyzed in great depth. Sixteen double-blinded solutions including both triclosan-containing (1.14-3.12 mg triclosan/g product) and triclosan-free products were contacted with free chlorine at pH 7. Products detected included (chlorophenoxy) phenols, 2,4-dichlorophenol, 2,4,6trichlorophenol, and chloroform. The daughter product yields were found to be highly variable and were dependent on the antimicrobial product investigated, the free chlorine to triclosan ratio, and the temperature at which the study was conducted. Lowering the temperature from 40 to 30 °C resulted in a decreased average chloroform yield from 0.50 to 0.37 mol chloroform/mol triclosan consumed after 1 min of reaction time for an initial free chlorine concentration of 4.0 mg/L as Cl2. At 40 °C the average molar chloroform yields decreased to 0.29 and 98% purity). Triclosan stocks were prepared by dissolving 100 mg of triclosan in 50 mL of reagent grade methanol. Free chlorine stock solutions were prepared using purified grade NaOCl (Fisher Scientific). Free chlorine concentrations were measured by the DPD/FAS titrimetric method (17). A Fisher Scientific model 60 pH meter coupled with a Thermo-Orion Ross PerpHect Combination Electrode was utilized for pH measurements. Seven common triclosan-containing hygienic products and five triclosan-free products (e.g., lotions, soaps, and bodywash) were purchased at a local store. See Supporting Information Table S1 for a list of the tested products. Laboratory Experiments. To eliminate potential bias, all products to be tested (both triclosan-containing and triclosan-free, some in duplicate) were placed in 40 mL amber screw top vials and labeled using an alphabetic code. To further reduce bias and to ensure anonymity, each vial was relabeled by a second individual using Roman numerals. Both code-keys were kept confidential by the individuals VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2387
FIGURE 1. Reaction scheme showing reaction mechanisms and chemical structures for triclosan and its decay products. As detailed in ref 14, all species were identified either by mass spectral analysis ((chlorophenoxy)phenols and chlorophenols) or comparison of retention times of the analyte to known standards (chloroform). responsible for the vial labeling. Using GC-MS analysis, ten vials were shown to contain triclosan-laden products, while six vials contained triclosan-free products that served as negative controls. At the conclusion of the experimental phase of the study, the contents of each vial were revealed. The ingredients present in each numbered soap vial are listed in Table S2. Experimental solutions were made by mixing a given unknown product at a concentration of 0.25 g/L with reagent grade water containing 2 mM NaHCO3 buffer. Although a typical soap dispenser dispenses ∼7× this amount with a single pump (1.84 ( 0.002 g/pump based on 3 replicates), this soap concentration falls within the concentration range of 0.25-0.55 g/L calculated based on typical soap (∼40 g/person-day (18)) and water (72-160 L/person-day (19)) usage rates. NaOH and H2SO4 were used to adjust the solution to pH 7. Triplicate 40 mL amber screw top vials were filled headspace free with soap solutions and were used to perform experiments. Prior to the start of an experiment, temperature was controlled by submerging PTFE-septa sealed reaction vials in a water bath until their temperature stabilized. Reactions were initiated by spiking an aliquot of free chlorine into a reaction vial using a Cheney Adaptor equipped syringe. Solutions were then removed from the water bath to enable continuous mixing for a given reaction period. Reaction periods of less than 5 min were used, and this period was insufficient for any measurable temperature change to occur. Triclosan, Chlorophenol, and (Chlorophenoxy)phenol Analysis. In experiments where triclosan decay and phenolicdaughter-product formation were monitored, reactions were quenched using a 3× molar excess of sodium sulfite. Triclosan and phenolic daughter-products were quantified using solid-phase extraction followed by derivatization with 2388
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
pentafluorobenzyl bromide (PFBBr) and GC-MS analysis, as described previously (14). As illustrated in Supporting Information Figure S1, a control experiment established the validity of the PFBBr technique for the quantification of triclosan concentrations in soap. Trihalomethane (THM) Analysis. Samples were analyzed for both chloroform and the brominated THMs after quenching residual free chlorine with sodium sulfite. A 5 mL aliquot was transferred from each reaction vessel and crimp sealed in a 20 mL headspace vial. THMs were quantified via GCECD (Thermo-Finnegan TraceGC Ultra; Agilent GS-GASPRO column) using a HS2000 headspace autosampler. Individual THMs were identified based on their elution times and quantified using calibration curves. Field Experiments. Free chlorine concentrations were measured using a Hach field chlorine kit, and pH was determined using a Fisher Scientific portable pH probe. The accuracy of the field measurements was verified by comparison to the methods used in the laboratory (Supporting Information Figures S2 and S3). Reactions were initiated by the addition of a given volume of soap into 1 L of tap water. The solution was mixed, and the reaction was quenched by pouring the soap solution into a 40 mL amber vial containing excess sulfite after 1 and 4 min of reaction time. Samples were refrigerated and shipped to Virginia Tech for triclosan, THM, and phenol analysis.
Results and Discussion Triclosan concentrations in the ten samples that contained triclosan ranged from 1.14 to 3.48 mg triclosan/g product (Table 1). Although only four of the tested triclosancontaining products reported the triclosan content, this range roughly matches the concentration range reported by the
TABLE 1. Triclosan Content of Each Unknown Soap and the Corresponding Amount of Chloroform Produced after 1 min of Reaction Time (in µg/L and nM)a reaction conditions temperature ) 40 °C, [HOCl]i ) 4.0 mg/L triclosan content soln (mg triclosan/ no. g soap)
(µM)
chloroform concentration µg/L
temperature ) 30 °C, [HOCl]i ) 4.0 mg/L
µM
yield (mol CHCl3/mol triclosan)
0.963 ((0.098) 0.957 ((0.287) 1.270 ((0.284) 0.873 ((0.036) 0.207 ((0.100) 1.010 ((0.226) 1.016 ((0.226) 0.337 ((0.164) 0.637 ((0.143) 0.853 ((0.199)
0.587 0.692 0.783 0.324 0.137 0.419 0.420 0.343 0.605 0.633
chloroform concentration µg/L
µM
temperature ) 40 °C, [HOCl]i ) 2.0 mg/L yield (mol CHCl3/mol triclosan)
chloroform concentration
temperature ) 40 °C, [HOCl]i ) 1.0 mg/L
µg/L
µM
yield(mol CHCl3/mol triclosan)
71.0 ((11.2) 36.9 ((5.5) 63.1 ((3.7) 22.3 ((6.0) 0.0 ((0.0) 92.5 ((3.2) 124.3 ((17.1) 20.7 ((6.2) 59.8 ((3.7) 81.8 ((17.3)
0.595 ((0.094) 0.309 ((0.046) 0.529 ((0.031) 0.187 ((0.050) 0.0 ((0.0) 0.775 ((0.026) 1.041 ((0.143) 0.173 ((0.052) 0.501 ((0.032) 0.685 ((0.145)
0.363 0.224 0.326 0.069 0.000 0.320 0.431 0.176 0.475 0.509
0.0 ((0.0) 3.7 ((3.2) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
0.0 ((0.0) 0.031 ((0.027) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
chloroform concentration µg/L
µM
yield(mol CHCl3/mol triclosan)
2.31 ((0.779) 0.0 ((0.0) 2.11 ((0.116) 24.5 ((1.51) 0.0 ((0.0) 11.6 ((0.208) 22.2 ((0.10) 0.0 ((0.0) 7.75 ((0.11) 10.1 ((7.36)
0.019 ((0.007) 0.0 ((0.0) 0.018 ((0.010) 0.206 ((0.013) 0.0 ((0.0) 0.098 ((0.002) 0.186 ((0.001) 0.0 ((0.0) 0.065 ((0.001) 0.085 ((0.062)
0.012 0.000 0.011 0.068 0.000 0.040 0.077 0.000 0.062 0.078
Triclosan-Containing Samples
VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
II IV V VII IX XII XIII XIV XVI XVII
1.90 ((0.08) 1.64 ((0.07) 115.0 ((11.8) 1.60 ((0.06) 1.38 ((0.05) 114.2 ((34.2) 1.88 ((0.00) 1.62 ((0.00) 151.8 ((33.9) 3.48 ((0.00) 3.01 ((0.00) 104.2 ((4.3) 1.74 ((0.03) 1.50 ((0.03) 24.7 ((12.0) 2.80 ((0.11) 2.42 ((0.09) 121.0 ((27.0) 2.80 ((0.11) 2.42 ((0.09) 121.3 ((27.0) 1.14 ((0.14) 0.098 ((0.12) 40.3 ((19.6) 1.22 ((0.03) 1.05 ((0.03) 76.1 ((17.1) 1.56 ((0.23) 1.09 ((0.20) 101.8 ((23.8)
III VI VIII X XI XV
0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00)
89.7 ((3.9) 56.8 ((1.4) 98.7 ((14.4) 80.5 ((12.7) 18.1 ((8.8) 94.5 ((16.3) 92.4 ((14.0) 39.0 ((18.4) 63.9 ((14.5) 80.2 ((16.3)
0.751 ((0.032) 0.476 ((0.012) 0.827 ((0.121) 0.674 ((0.107) 0.152 ((0.074) 0.792 ((0.137) 0.774 ((0.117) 0.327 ((0.154) 0.535 ((0.121) 0.672 ((0.137)
0.458 0.345 0.510 0.250 0.101 0.327 0.320 0.332 0.508 0.499
Triclosan-Free Controls 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00) 0.00 ((0.00)
0.0 ((0.0) 0.0 ((0.0) 39.3 ((18.0) 0.329 ((0.150) 5.6 ((3.9) 0.047 ((0.033) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
0.0 ((0.0) 23.0 ((7.8) 2.7 ((0.2) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
0.0 ((0.0) 0.192 ((0.065) 0.022 ((0.002) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
0.0 ((0.0) 0.0 ((0.0) 15.8 ((0.986) 0.132 ((0.008) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0) 0.0 ((0.0)
a Initial free chlorine concentration and temperature were varied. Conditions: [NaHCO ] ) 2 mM, pH ) 7.0. Free chlorine concentrations are given in mg/L as Cl . The standard deviation of triplicate samples is 3 2 given in parentheses.
9
2389
FIGURE 2. Triclosan concentration prior to chlorine addition and after 1 min of reaction time. No additional triclosan was consumed after the first minute. Conditions: [free chlorine]i ) 2.0 mg/L as Cl2 (28 µM), [NaHCO3] ) 2 mM, [soap] ) 0.25 g/L, pH ) 7.0, T ) 40 °C. Error bars reflect the standard deviation of triplicate samples. manufacturers of those products (1.2-3.0 mg triclosan/g product). To examine the potential for the triclosan present in these products to react with free chlorine, triclosan consumption was predicted using the second-order-model developed by Rule et al. (14). This model predicts that when a 0.25 g/L soap solution prepared with 3.48 mg triclosan/g soap is exposed to 2.0 mg/L as Cl2 (28 µM) at pH 7 and a temperature of 25 °C, that 65% of the initial triclosan would be removed after 1 min (see Supporting Information text and Figure S4). This model result suggests that rapid triclosan consumption is expected at room temperature, and by extension it is anticipated that even higher triclosan consumption rates will be observed at elevated temperatures common to household activities. To examine how the free chlorine-triclosan reaction rates are affected by elevated temperatures, triclosan consumption was quantified for seven randomly chosen samples (solutions: II, III (control), IV, VI (control), VII, IX, XII) at 40 °C. For the five soaps that contained triclosan, over half of the initial triclosan was consumed within 1 min of reaction time when the initial chlorine concentration was 2.0 mg/L as Cl2 (28 µM; Figure 2). For three of the five triclosan-containing soaps (II, VII, XII), all of the triclosan was consumed within 1 min. This result indicates that triclosan readily reacts with free chlorine, even in the presence of other soap components. These components exert a chlorine demand, as demonstrated by chlorine consumption experiments for both the triclosancontaining and the non-triclosan soaps; however, the measured chlorine demand tended to be more rapidly exerted for the triclosan-containing soaps than for the non-triclosan soaps (Figure S5). The more rapid loss of free chlorine in the triclosan-containing samples suggests that the reaction rates for triclosan are high enough that this chemical outcompetes many other potential reactants. Interestingly, the two soaps where residual triclosan was detected contained either the largest number of ingredients (soap IV) or contained known free chlorine scavengers such as sodium bisulfite and ammonium (soap IX). This latter observation suggests that personal care product manufacturers may be able to stabilize triclosan by judiciously adding free chlorine scavengers to their product formulations. Product Formation in Laboratory Experiments. Prior studies have shown that (chlorophenoxy)phenols, chlorophenols, and chloroform are produced when triclosan reacts with excess free chlorine (11-14). The formation of these products was monitored for solutions II, III (control), 2390
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
FIGURE 3. Experimental yields of chloroform, 2,4-dichlorophenol, and 2,4,6-trichlorophenol after exposure of a given soap to free chlorine for 1 min. Conditions: [free chlorine]i ) 2.0 mg/L as Cl2 (28 µM), [NaHCO3] ) 2 mM, [soap] ) 0.25 g/L, pH ) 7.0, T ) 40 °C. IV, VI (control), VII, IX, and XII at pH 7, 40 °C, and with an initial free chlorine concentration of 2 mg/L as Cl2 (28 µM). Of the triclosan-containing solutions, only soap IX produced detectable quantities of (5,6-dichloro-2-(2,4-dichlorophenoxy)phenol, 4,5-dichloro-2-(2,4-dichlorophenoxy)phenol, and 4,5,6-trichloro-2-(2,4-dichlorophenoxy)phenol) (data not shown). The molar yield (defined as moles of compound X produced/triclosan consumption in moles) for 2,4-dichlorophenol was zero for soaps IV and XII but ranged between 0.01 and 0.25 mol/mol for soaps II, VII, and IX (Figure 3). 2,4,6-Trichlorophenol was only detected for soaps VII and IX with yields of 0.012 and 0.027 mol/mol, respectively. Chloroform molar yields ranged from 0 to 0.36 mol chloroform/mol triclosan, with chloroform detected for all triclosan-containing soaps other than soap IX. In general, the chlorophenol yields inversely correlate with the chloroform yields, with the samples exhibiting the highest chlorophenol yields also having the lowest chloroform yields. In addition, the only solution (soap IX) that failed to produce detectable amounts of chloroform was also the only solution for which intermediate (chlorophenoxy)phenols were detected. This result may reflect the formation of chloramines in the solution due to the presence of an ammonium salt in soap IX. Prior studies have shown that chloramines react with triclosan to produce chlorophenols and (chlorophenoxy)phenols, but have insufficient oxidizing power to cleave the phenol ring of triclosan and produce chloroform (16). Collectively, these observations suggest that chloroform formation only occurs under conditions where a sufficient excess of free chlorine is present to cleave the phenol ring of triclosan. The amount of free chlorine required for ring cleavage is affected by the soap composition and the initial free chlorine to triclosan ratio. Rule et al. (14) had suggested that the majority of the chloroform produced upon chlorination of triclosan was produced via oxidation and ring cleavage of the phenol moiety of triclosan and not from reactions involving the product 2,4-dichlorophenol. To better understand how product yield varies with soap identity, an additional set of experiments was conducted to quantify chloroform formation for all 16 soap samples at pH 7 with initial free chlorine concentrations of 4.0, 2.0, and 1.0 mg/L as Cl2 (56, 28, and 14 µM, respectively) and temperatures of 40 and 30 °C (Table 1). For this entire set of conditions, chloroform concentrations ranged from 0 to 152 µg/L (0-1.3 µM) with an average of 60 ( 44 µg/L for the soaps containing triclosan and from 0 to 39.3 µg/L (0-0.33 µM) with an average of 2.3 ( 5.6 µg/L for the triclosan-free controls, after each
sample was contacted with free chlorine for 1 min. With the exception of soap IX at 40 °C and 2 mg/L Cl2 (28 µM), every sample that initially contained triclosan produced chloroform when contacted with 2.0 mg/L or more of free chlorine. At 40 °C and 4 mg/L (56 µM) free chlorine, the maximum contaminant level (MCL) of 80 µg/L (0.67 µM (20)) for total trihalomethanes was exceeded by chloroform formation alone in 7 of the 10 triclosan containing samples. The MCL was never exceeded for the control samples under any condition. No brominated THMs were detected under any condition. For an initial free chlorine concentration of 4 mg/L as Cl2 and a temperature of 40 °C, the average yield of chloroform for the soaps that contained triclosan was 0.50 mol chloroform/mol triclosan. Lowering the initial free chlorine concentration to 2.0 mg/L as Cl2 (average chloroform yield ) 0.29) or to 1 mg/L (average chloroform yield < 0.1) or reducing the reaction temperature to 30 °C (average chloroform yield ) 0.37) caused the chloroform yield to decrease. The decrease in the chloroform yield with a decrease in free chlorine concentration, for any given soap, suggests that variations in the free chlorine to triclosan ratio can significantly affect product yields. Furthermore, the chloroform yield cannot be readily predicted based solely on the triclosan content. Overall, the variability in the product yields is likely due in large part to variations in the free chlorine to triclosan ratio as well as the chlorine demand exerted by other ingredients in the soaps. A reagent-spike experiment was conducted to examine how a variation in the free chlorine to triclosan ratio alters product yields. For this purpose, a 0.25 g/L solution of soap VII was produced and split into three subsamples. One of these subsamples was left unmodified, whereas 0.25 mg/L (0.86 µM) and 2.0 mg/L (6.91 µM) of pure triclosan were added to the other subsamples, respectively. By adding pure triclosan to these subsamples, it was possible to vary the triclosan content, while keeping the total soap concentration constant. The initial total triclosan concentrations in the three samples were 0.78, 1.03, and 2.78 mg/L (2.78, 3.56, and 9.60 µM), respectively. As illustrated in Figure S6, for solutions treated with 2 mg/L as Cl2 free chlorine at 40 °C, all of the triclosan present in the 0.78 and 1.03 mg/L samples was consumed within 1 min, whereas 0.37 mg/L (2.28 µM) of triclosan remained after 1 min for the 2.78 mg/L solution. Examining the product yield as a function of the free chlorine to triclosan ratio, it was determined that the chloroform yield was the smallest at low free chlorine to triclosan ratios. Conversely, the chlorophenol yields increase with a decrease in the free chlorine to triclosan ratio (Figure 4). This result supports the hypothesis that the increased chlorine demand exerted by the additional triclosan affects the triclosan chlorination product yields. When ample free chlorine is present, triclosan is readily degraded to produce chloroform, and chlorophenol accumulation is diminished; however, when free chlorine is limiting, intermediate byproducts, such as the chlorophenols, are detected. This latter conclusion is supported by experiments with pure triclosan solutions, wherein chlorophenol accumulation was not observed at a high free chlorine to triclosan ratio ()28.66 µM free chlorine/ µM triclosan), but was observed at a lower ratio ()10.47 µM free chlorine/µM triclosan; Figure S7). Field Experiments. The observed variations in the product yields suggest that a priori predictions of product formation that are based solely on the triclosan content of a given soap may be difficult if not impossible. Furthermore, differences in the composition of the soap and the water in which the soap is used are also expected to affect the product yields. To examine product formation under real world conditions, field experiments were performed by augmenting Atlanta, GA and Danville, VA tap waters with several of the antibac-
FIGURE 4. Product yields versus the free chlorine to triclosan ratio for soap VII containing 0.87 mg/L (0.31 µM) triclosan spiked with varying amounts of triclosan. Soap VII with no added triclosan is represented by [free chlorine]i/[triclosan]i ) 10.5; soap VII with addition of 0.25 mg/L (0.86 µM) triclosan is represented by [free chlorine]i/[triclosan]i ) 7.9; soap vii with a 2.0 mg/L (6.9 µM) spike is denoted by [free chlorine]i/[triclosan]i ) 2.9. Conditions: [free chlorine]i ) 2.0 mg/L as Cl2 (28 µM), [NaHCO3] ) 2 mM, [soap VII] ) 0.25 g/L, T ) 40 °C, reaction time ) 1 min. terial hygiene products included in this study. For these experiments, water was obtained from distribution system taps after running the water for more than a half-hour to flush the pipes and reduce fluctuations in water quality. Atlanta tap water had an average free chlorine concentration of 1.0 mg/L as Cl2 (14 µM), a pH of 6.35, and a temperature of 33 °C over the experimental time period. Danville water maintained a 1.6 mg/L as Cl2 (23 µM) free chlorine residual with an average pH of 7.22 and a temperature of 38 °C. Baseline chloroform concentrations in the Atlanta and Danville tap waters at the onset of these experiments were 37.6 and 72.5 µg/L (0.315 and 0.607 µM), respectively. Experiments were conducted immediately after obtaining the tap water with products VI (control), VII, IX, and XII. Chloroform and phenol product yields for each field site are illustrated in Figure 5. In the water from the Atlanta distribution system there was minimal loss of triclosan from soaps IX and XII but complete consumption for soap VII (Figure S8). In contrast, triclosan was completely consumed from all of the soaps in the experiments with Danville water. While all three (chlorophenoxy)phenol intermediates were detected following chlorination of the Atlanta water, only one dichlorinated intermediate was detected with soap IX in Danville water. In contrast, 2,4-dichlorophenol and 2,4,6-trichlorophenol yields were considerably higher in the Danville water than in the Atlanta water, and significant quantities of chloroform above the baseline level were produced in the Danville water. The experiments in Atlanta water resulted in little chloroform formation above the baseline level of 37.6 µg/L (0.31 µM) present in the water coming from the tap. Relative to the results in laboratory water, the Atlanta and Danville waters generally produced lower levels of chloroform but higher levels of (chlorophenoxy)phenol intermediates. For the Danville water, although chloroform production was lower than observed in the laboratory, it was still produced at high yield for two of the soaps (0.07-0.17 mol chloroform/ mol triclosan; Figure 5). As evident from ICP analysis (Table S3), the Danville water contains many inorganic constituents that were not included in the laboratory experiments. These constituents as well as the slightly elevated pH of the Danville water are expected to affect the chloroform yield. The lower VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2391
FIGURE 5. Product yield comparisons between Atlanta and Danville field experiments after 1 min of reaction time. Atlanta conditions: [free chlorine]i ) 1.0 mg/L as Cl2 (14 µM), [soap] ) 0.25 g/L, T ) 33 °C, pH ) 6.35. Danville conditions: [free chlorine]i ) 1.6 mg/L as Cl2 (23 µM), [soap] ) 0.25 g/L, T ) 38 °C, pH ) 7.22. chloroform yields and the lack of triclosan consumption for two of the soaps in the Atlanta water may be attributed to the low pH of the water and to the low chlorine residual of 1 mg/L as Cl2. Prior studies (14) have shown that the free chlorine-triclosan reaction is fastest at slightly alkaline pH values, and thus it is not surprising that product formation is reduced at pH 6.35. In addition, a pH of 6.35 is very low for tap water and may have resulted in the higher concentrations of copper and zinc, relative to the Danville water, due to the corrosion of pipe materials (Table S3). These dissolved constituents may have affected the reactivity of free chlorine and triclosan. For instance, elevated copper levels can exert a chlorine demand (21) that could reduce the amount of free chlorine available to react with triclosan. Potential Health Significance. The present study has shown that under some circumstances chloroform levels in excess of the U.S. EPA MCL for drinking water can be detected shortly after triclosan comes into contact with chlorinated water. The ramifications of this previously unidentified chloroform exposure pathway are not known but require careful evaluation. In an effort to provide insight for future epidemiological studies, we used a simple exposure model to estimate the potential increase in a person’s chloroform exposure that could occur through the use of antibacterial soap. The exposure model was developed by modification of the Soap and Detergent Association (SDA) model typically used to estimate exposures to volatile contaminants present within soap products (18). The standard SDA model was modified to incorporate the molar yield of chloroform determined from the experiments described herein. For this purpose, the soap weight was modified by the yield of chloroform (YCF) to obtain an estimated inhalational exposure
inhalational exposure
) (mg yr )
A × FQ × TC × MWCF × YCF × IR × F × ED × T (1) MWTric × V where A ) amount of soap used (12.0 g body wash and 1.7 g liquid soap per use), FQ ) frequency (1.07 and 8 uses per day for body wash and liquid soap, respectively), IR ) inhalation rate (546 L/h), F ) respirable fraction (0.257), ED ) exposure duration (10 min for showering, 1 min for washing hands), T ) time correlation factor (365 days/year), V ) effective breathing air space (2000 L), and MWCF and MWTric 2392
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
are the molecular weights of chloroform and triclosan, respectively. Each of the model parameters is a typical value as specified by SDA. Using chloroform yields (YCF) ranging from 0.07 to 0.29 (a yield of 0.07 was the smallest yield observed experimentally when chloroform was detected in reactions at 40 °C with [free chlorine]i ) 2.0 mg/L (28 µM); 0.29 was the average yield for all triclosan-containing soaps for the same conditions) and a triclosan mass concentration (TC) of 2.0 mg triclosan/g soap, the calculated inhalational chloroform exposure was estimated to be 5.8-24.2 mg/yr. Neglecting ingestional exposures (typically 27% of an individual’s total exposure (20)), but accounting for the expected dermal exposure (typically 10% of an individual’s total exposure (22)), a person’s total exposure to chloroform produced by triclosan decomposition is estimated to be 6.828 mg chloroform/yr (i.e., 5.8 mg/yr × (63% + 10%)/63% ) 6.8 mg/yr). It was possible to estimate how the triclosan-mediated pathway contributes to an individual’s overall exposure by assuming an individual’s only other chloroform exposure results from the chloroform present in disinfected tap water. For this purpose, the magnitude of the tap water mediated exposure pathway was estimated using the comprehensive model of Kim et al. (23). This model incorporates established values for a variety of input parameters (such as the chemical properties of chloroform, shower size, house air exchange rates, etc.) to estimate annual chloroform exposures. Using this model, it was predicted that an individual’s yearly exposure to chloroform from normal water use (e.g., in the absence of triclosan-containing products) is 0.65 mg/yr per µg/L of chloroform present in the water at the tap. Using this value and the 6.8-28 mg chloroform/yr range calculated for the triclosan-mediated pathway, it was possible to construct a graph examining the percentage of an individual’s exposure that can be attributed to the use of triclosan-containing antibacterial products. As shown in Figure 6, a person’s overall exposure is predicted to be significantly enhanced by their use of antibacterial products that contain triclosan. For tap waters with a chloroform concentration at the MCL of 80 µg/L, the use of triclosan containing products could increase an individual’s overall exposure by 15-40% if the chloroform yield is within the simulated range. For waters below the MCL, the percentage of an individual’s exposure due to triclosan use is even higher. The simulations depicted in Figure 6 indicate that for those conditions where triclosan reacts to produce chloroform, that the resulting exposure
and ICP analyses. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited
FIGURE 6. Estimated consumer exposure to chloroform produced by the degradation of antibacterial soap as predicted using a modified exposure model. The solid lines in the figure illustrate the percentage of an individual’s exposure that results from the use of antimicrobial soaps under average- and low-chloroform yield conditions, shown by the red and black lines, respectively, as compared to tap water with chloroform concentrations between 0 and 150 µg/L. The dashed lines represent the total chloroform exposure of an individual (including both exposures due to soap use and due to normal water use). Exposures due to soap use were determined using a model established by Kim et al. (23) and exposures due to soap use were estimated using a model established by the Soap and Detergent Association (18). may be significant. We stress, however, that this is a model simulation based on a limited number of laboratory studies and needs to be verified at full scale as can only be done through well-designed epidemiologic studies. In a household environment, it is not clear that the use of triclosan-containing products provides enhanced antibacterial protection relative to triclosan-free products, and accordingly some have argued for the use of plain soap and water in lieu of triclosan-containing soaps (24-27). The laboratory and modeling studies described herein indicate that formation of chloroform and other chlorinated daughter products can occur when triclosan-containing antimicrobial products react with free chlorine and that these reactions can potentially lead to enhanced chloroform exposures. Even under conditions where chloroform formation is limited, other products of potential health concern are produced. This observation when coupled with the general lack of evidence clearly illustrating the efficacy of triclosan-containing products in the home suggests that a full risk-benefit analysis of these products should be conducted.
Acknowledgments This work was supported by a research grant from the American Water Works Association Research Foundation (AwwaRF). We thank Jody Smiley and Julie Petruska for their help with instrumentation and analytical methods. We also thank Dr. John Little for his help and input with the exposure modeling. The insightful comments of three anonymous reviewers significantly improved the content and flow of this manuscript, and we thank them for taking the time to carefully review this submission.
Supporting Information Available Figures of control experiments, model predictions of triclosan loss rates, chlorine consumption, product formation in the laboratory and in the field, and tables of product composition
(1) Minutes of the FDA Nonprescription Drugs Advisory Committee Meeting. October 20, 2005. http://www.fda.gov/ohrms/dockets/ ac/cder05.html#NonprescriptionDrugs (accessed August 1, 2006). (2) Jones, R. D.; Jampani, H. B.; Newman, J. L.; Lee, A. S. Triclosan: A review of effectiveness and safety in health care settings. Am. J. Infect. Control 2000, 28, 184-196. (3) McAvoy, D. C.; Schatowitz, B.; Jacob, M.; Hauk, A.; Eckhoff, W. S. Measurement of triclosan in wastewater treatment systems. Environ. Toxicol. Chem. 2002, 21, 1323-1329. (4) Singer, H.; Muller, S.; Tixier, C.; Pillonel, L. Triclosan: Occurrence and fate of a widely used biocide in the aquatic environment: Field measurements in wastewater treatment plants, surface waters, and lake sediments. Environ. Sci. Technol. 2002, 36, 4998-5004. (5) Kanda, R.; Griffin, P.; James, H. A.; Fothergill, J. Pharmaceutical and personal care products in sewage treatment works. J. Environ. Monit. 2003, 5, 823-830. (6) Federle, T. W.; Kaiser, S. K.; Nuck, B. A. Fate and effects of triclosan in activated sludge. Environ. Toxicol. Chem. 2001, 21, 1330-1337. (7) Lindstrom, A.; Buerge, I. J.; Poiger, T.; Bergqvist, P.-A.; Muller, M. D.; Buser, H.-R. Occurrence and environmental behavior of the bactericide triclosan and its methyl derivative in surface waters and in wastewater. Environ. Sci. Technol. 2002, 36, 23222329. (8) Tancrede, M.; Yanagisawa, Y.; Wilson, R. Volatilization of volatile organic-compounds from showers. 1. Analytical method and quantitative assessment. Atmos. Environ. Part A 1992, 26, 11031111. (9) Howard, C.; Corsi, R. L. Volatilization of chemicals from drinking water to indoor air: The role of residential washing machines. J. Air Waste Manage. Assoc. 1998, 48, 907-914. (10) American Water Works Association Drinking Water Utility Database; American Water Works Association Water Quality Section, 1997. (11) Onodera, S.; Ogawa, M.; Suzuki, S. Chemical-changes of organiccompounds in chlorinated water. 13. Gas-chromatographic mass-spectrometric studies of the reactions of Irgasan DP300 [5-chloro-2-(2,4-dichlorophenoxy)phenol] with chlorine in dilute aqueous-solution. J. Chromatogr. 1987, 392, 267-275. (12) Kanetoshi, A.; Ogawa, H.; Katsura, E.; Kaneshima, H. Chlorination of Irgasan DP300 and formation of dioxins from its chlorinated derivatives. J. Chromatogr. 1987, 389, 139-153. (13) Canosa, P.; Morales, S.; Rodriguez, I.; Rubi, E.; Cela, R.; Gomez, M. Aquatic degradation of triclosan and formation of toxic chlorophenols in presence of low concentration of free chlorine. Anal. Bioanal. Chem. 2005, 383, 1119-1126. (14) Rule, K. L.; Ebbett, V. R.; Vikesland, P. J. Formation of chloroform and chlorinated organics by free-chlorine-mediated oxidation of triclosan. Environ. Sci. Technol. 2005, 39, 3176-3185. (15) EPA. U.S. Environmental Protection Agency-National Center for Environmental Assessment, IRIS Database. http:// www.epa.gov/iris (accessed December 20, 2006). (16) Greyshock, A. E.; Vikesland, P. J. Triclosan reactivity in chloraminated waters. Environ. Sci. Technol. 2006, 40, 2615-2622. (17) Standard Methods for the Examination of Water and Wastewater; 18th ed.; Greenberg, A. E., Eaton, A. D., Clesceri, L. S., Eds.; APHA: Washington, DC, 1992. (18) Soap and Detergent Association. Exposure and risk screening methods for consumer product ingredients; 2005. (19) American Water Works Association Stats on Tap; Denver, CO. http://www.awwa.org/advocacy/pressroom/stats.cfm (accessed December 19, 2006). (20) EPA. U.S. Environmental Protection Agency: 1998; Vol. EPA 815-F-98-010. (21) Hong, P. K. A.; Macauley, Y.-Y. Corrosion and leaching of copper tubing exposed to chlorinated drinking water. Water, Air, Soil Pollut. 1998, 108, 457-471. (22) Kim, E.; Little, J. C. Exposure to chemical contaminants in drinking water; Report to the U.S. EPA; 2004. (23) Kim, E.; Little, J. C.; Chiu, N. Estimating exposure to chemical contaminants in drinking water. Environ. Sci. Technol. 2004, 38, 1799-1806. (24) Bendig, J. W. A. Surgical hand disinfection - comparison of 4-percent chlorhexidine detergent solution and 2-percent triclosan detergent solution. J. Hosp. Inf. 1990, 15, 143-148. VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2393
(25) Kampf, G.; Kramer, A. Epidemiologic background of hand hygiene and evaluation of the most important agents for scrubs and rubs. Clin. Microbiol. Rev. 2004, 17, 863-+. (26) Larson, E. L.; Aiello, A.; Lee, L. V.; Della-Latta, P.; Gomez-Duarte, C.; Lin, S. Short- and long-term effects of handwashing with antimicrobial or plain soap in the community. J. Community Health 2003, 28, 139-150. (27) Larson, E. L.; Lin, S. X.; Gomez-Pichardo, C.; Della-Latta, P. Effect of antibacterial home cleaning and handwashing products
2394
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
on infectious disease symptoms. Ann. Int. Med. 2004, 140, 321329.
Received for review September 18, 2006. Revised manuscript received January 4, 2007. Accepted January 22, 2007. ES062227L
