Triclosan Reactivity in Chloraminated Waters - Environmental Science

Mar 16, 2006 - Department of Civil and Environmental Engineering, Virginia ... treatment plant effluents and in source waters used for drinking water ...
1 downloads 0 Views 281KB Size
Environ. Sci. Technol. 2006, 40, 2615-2622

Triclosan Reactivity in Chloraminated Waters AIMEE E. GREYSHOCK AND PETER J. VIKESLAND* Department of Civil and Environmental Engineering, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24060

Triclosan, widely employed as an antimicrobial additive in many household personal care products, has recently been detected in wastewater treatment plant effluents and in source waters used for drinking water supplies. Chloramines used either as alternative disinfectants in drinking water treatment or formed during chlorination of nonnitrified wastewater effluents have the potential to react with triclosan. This study examined triclosan reactivity in chloraminated waters over the pH range of 6.5-10.5. Experimental and modeling results show that monochloramine directly reacts with the phenolate form of triclosan; however, the reaction is relatively slow as evinced by the second-order rate constant kArO-NH2Cl ) 0.025 M-1 s-1. Kinetic modeling indicates that for pH values less than 9.5, reactions between triclosan and two monochloramine autodecomposition intermediates, hypochlorous acid (kArO-HOCl ) 5.4 × 103 M-1 s-1) and dichloramine (kArO-NHCl2 ) 60 M-1 s-1), are responsible for a significant percentage of the observed triclosan decay. The products of these reactions include three chlorinated triclosan byproducts as well as 2,4dichlorophenol and 2,4,6-trichlorophenol. Low levels of chloroform were detected after 1 week at pH values of 6.5 and 7.5. The slow reactivity of triclosan in the presence of chloramines explains the recalcitrance of this species in nonnitrified wastewater effluents.

Introduction Triclosan (5-chloro-2-(2,4-dichlorophenoxy)phenol; Supporting Information Figure S1) is an antibacterial agent commonly added to personal care products such as soaps, deodorants, toothpastes, and mouthwash. Because of its widespread industrial and domestic use, estimated at 3-5 mg/capita/day (1, 2), triclosan is often detected in untreated wastewaters. Removal efficiencies for triclosan within activated sludge wastewater treatment plants (WWTPs) typically exceed 95% (3). Because of its incomplete removal, however, triclosan at concentrations of 0.24-37.8 µg/L has been detected in WWTP effluents (1, 2, 4, 5). Within the U.S., these effluents are often treated with free chlorine (e.g., molecular chlorine, Cl2; hypochlorous acid, HOCl; and hypochlorite ion, OCl-), at doses ranging from 0.05 to 0.3 mM (6, 7) for a contact time of 30-120 min (8) prior to discharge. In nonnitrified wastewater effluents, which contain ammonia at concentrations in the range of 0.7-2.3 mM (9-11), this applied chlorine is rapidly converted to inorganic (e.g., monochloramine and dichloramine) and organic chloram* Corresponding author phone: (540)231-3568; fax: (540)231-7916; e-mail: [email protected]. 10.1021/es051952d CCC: $33.50 Published on Web 03/16/2006

 2006 American Chemical Society

ines as the result of reactions between free chlorine and ammonia or free chlorine and organic nitrogen, respectively. In drinking water disinfection, monochloramine is increasingly being used as an alternative to free chlorine because it is less reactive with organic material and produces lower levels of regulated disinfection byproducts (12). Studies have shown that monochloramine is not stable and undergoes a series of autodecomposition reactions in which monochloramine reacts with itself, leading to the oxidation of ammonia and the reduction of chlorine (13-16). A model describing monochloramine autodecomposition in the absence of other reactive species was developed by Valentine and co-workers (14, 16) and is given by eqs 1.1-1.14 in Table 1. Autodecomposition is dependent on factors such as pH and the molar ratio of chlorine to ammonia nitrogen (Cl/N). As the solution pH decreases, monochloramine decays more rapidly due to the pH dependency of general acid-catalyzed monochloramine disproportionation (reaction 1.5; ref 13) and the pH-dependent speciation of chlorine and ammonia. A larger Cl/N ratio means that less free ammonia is present for a fixed monochloramine concentration, resulting in faster oxidation of ammonia and more rapid monochloramine decay. In general, a decrease in the free ammonia concentration enhances chlorine and dichloramine production. Prior studies have shown that the reactivity of pharmaceutical and personal care products (PPCPs) is typically lower in chloraminated waters than in chlorinated waters. In a direct comparison, Pinkston and Sedlak obtained half-lives for acetaminophen and naproxen that were ∼104 longer in chloraminated waters than in the presence of free chlorine (7). Similar differences in reactivity were also reported for sulfamethoxazole (17) and enrofloxacin (18). From these studies, it was concluded that the reactions between chloramines and PPCPs are too slow to significantly affect PPCP concentrations in nonnitrified wastewaters. The detection of triclosan in chlorinated wastewater effluents, even though triclosan and free chlorine react rapidly (19), would support this conclusion. In a recent study, the triclosan concentration decreased only 12-57% upon chlorination of activated sludge effluents (20). When reactive substances are present in chloraminated waters, their reactivity must be considered while simultaneously accounting for autodecomposition and the formation of chlorine and dichloramine. This approach has been proven to be necessary in studies in which monochloramine reacts with natural organic matter (21, 22) and pharmaceuticals (7). The results of the present study show that the formation and reactivity of hypochlorous acid and dichloramine dictates triclosan loss rates at low pH values but that the direct reaction between triclosan and monochloramine dominates at higher pH values.

Experimental Procedures General Laboratory Procedures. Deionized water was produced using an Aries water purification system. pH measurements were obtained using a Fisher Scientific Model 60 pH meter with a ThermoOrion Ross PerpHect combination electrode. Laboratory glassware was cleaned by soaking in 10% nitric acid and in free chlorine (∼5000 mg/L) for at least 24 h in each bath. Preformed monochloramine stock solutions were prepared using established procedures (16, 23) in deionized water containing 2 mM sodium bicarbonate. Monochloramine-Triclosan Experiments. Deionized water containing 2 mM bicarbonate was adjusted to the desired reaction pH via acid (HNO3) or base (NaOH) addition. To this solution, a specific volume of monochloramine stock VOL. 40, NO. 8, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2615

TABLE 1. Model Describing Monochloramine Autodecomposition and Triclosan Loss reaction 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 1.12 1.13 1.14 1.15 1.16 1.17 1.18

HOCl + NH3 f NH2Cl + H2O NH2Cl + H2O f HOCl + NH3 HOCl + NH2Cl f NHCl2 + H2O NHCl2 + H2O f HOCl + NH2Cl NH2Cl + NH2Cl f NHCl2 + NH3 NHCl2 + NH3 f NH2Cl + NH2Cl NHCl2 + H2O f Ib I + NHCl2 f HOCl + products I + NH2Cl f products NH2Cl + NHCl2 f products HOCl S OCl- + H+ NH4+ S NH3 + H+ H2CO3 S HCO3- + H+ HCO3- S CO32- + H+ triclosan S phenolate-triclosan + H+ phenolate-triclosan + NH2Cl f products phenolate-triclosan + HOCl f products phenolate-triclosan + NHCl2 f products

rate coefficient/equilibrium constant (25 °C)

a k ) k +[H+] + k -2 s-1, k -2 s-1, and k + ) 6.9 × 103 M-2 s-1. d H H2CO3[H2CO3] + kHCO3[HCO3 ] where kH2CO3 ) 11 M HCO3 ) 0.22 M H monochloramine autodecomposition intermediate.

was added to reach the desired initial concentration. Additional ammonium chloride stock was added to the reaction solution at this time for those experiments employing excess ammonia. The DPD colorimetric method was used to determine the monochloramine concentration of the stock and diluted monochloramine solutions (24). For this analysis, a Beckman DU 640 spectrophotometer was employed. Once the initial monochloramine concentration of the samples was determined, an experiment was initiated by adding a known volume of triclosan stock. Triclosan was obtained from SigmaAldrich, and a primary stock solution (∼2000 mg/L) was prepared in 100% methanol. The concentration of methanol in the samples was 0.05% for reactions in which monochloramine was in excess and 0.40% for reactions in which triclosan was in excess. For methanol levels below 1.0%, as employed in these experiments, cosolvent effects do not need to be considered (25). After the addition of triclosan, the reaction solution was transferred to amber vials with screw top lids and stored headspace free in the dark at 22 °C. Periodic samples were taken for monochloramine and triclosan. Triclosan and its nonvolatile products were measured via solid-phase extraction, derivatization with pentafluorobenzyl bromide, and GC/MS detection (19). The solution pH was measured each time a sample was analyzed. Over the course of a typical experiment, the pH varied by e0.4 pH units in both controls and test solutions. Samples were analyzed for chloroform using EPA method 502.2 with purge and trap GC, as described elsewhere (19). Immediately prior to analysis via either GC/MS or purge and trap GC, the samples were dechlorinated with a 3× molar excess of sodium sulfite to monochloramine. Kinetic Model Formulation. A kinetic model for monochloramine autodecomposition incorporating reactions 1.11.14 in Table 1 was developed. For this purpose, differential equations describing each species in Table 1 were derived, and the extended Debye-Hu ¨ ckel equation was used to calculate activity coefficients. Temperature corrections were included for the rate constants for reactions 1.1-1.3 and 1.5 and for the equilibrium constants (16). For this study, the autodecomposition model was run using MicroMath Scientist version 2.01 (Salt Lake City, UT) and a base model code supplied by R. Valentine (University of Iowa). Inputs to the model included solution pH, Cl/N ratio, temperature, initial reactant concentrations, ionic strength, and carbonate buffer capacity. The program uses 2616

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 8, 2006

ref

k1.1 ) 4.2 × 106 M-1 s-1 k1.2 ) 2.1 × 10-5 s-1 k1.3 ) 2.8 × 102 M-1 s-1 k1.4 ) 6.4 × 10-7 s-1 kda k1.6 ) 6.1 × 104 M-1 s-1 k1.7 ) 1.1 × 102 M-1 s-1 k1.8 ) 2.8 × 104 M-1 s-1 k1.9 ) 8.3 × 103 M-1 s-1 k1.10 ) 1.5 × 10-2 M-1 s-1 pKa ) 7.5 pKa ) 9.3 pKa ) 6.3 pKa ) 10.3 pKa ) 7.9 kArO-NH2Cl ) 0.025 M-1 s-1 kArO-HOCl ) 5.4 × 103 M-1 s-1 kArO-NHCl2 ) 60 M-1 s-1

35 35 36 36 16 37 38 39 39 39 40 40 40 40 28 this paper 19 this paper b

I is an unidentified

the Episode numerical integration package to solve the set of simultaneous differential equations over time for the species of interest.

Results and Discussion Over the pH range 6.5-10.5, triclosan readily decayed in the presence of an 8.5-fold excess of monochloramine, while in its absence triclosan was stable. Under these conditions, triclosan loss can be described using pseudo-first-order kinetics, and appropriate rate constants (kobs; s-1) were determined using the method of initial rates. These kobs values can be related to an apparent second-order rate constant (kapp; M-1 s-1)

d[triclosan] ) -kapp[triclosan]T[oxidant]T dt

(1)

by the following:

kobs ) kapp[oxidant]T,t)0

(2)

where [triclosan]T represents the total concentration of triclosan, [oxidant]T is the total concentration of oxidant in the system (i.e., any combination of HOCl, OCl-, NH2Cl, and NHCl2), and [oxidant]T,t)0 is the initial excess oxidant concentration. Using this approach, kapp values were calculated and plotted as a function of the solution pH (Figure 1). As shown, the kapp values for monochloramine increase from a low of 0.008 ((0.004 at the 95% confidence level) M-1 s-1 at pH 9.5 to a high of 0.07 ((0.17) M-1 s-1 at pH 6.5. Under the conditions used to obtain the kinetic data shown in Figure 1, the monochloramine demand of the controls was the same as that measured in the presence of triclosan for pH values between 10.5 and 7.5 and was only slightly less at pH 6.5 (Figure 2; Supporting Information Figure S2). In contrast to the results with excess monochloramine, a significant monochloramine demand was exhibited relative to controls without triclosan when a >10× excess triclosan concentration was employed (Figure 3; Figure S3). This difference was observed because under excess triclosan conditions, reactions with triclosan are a comparable monochloramine loss pathway to autodecomposition. The results depicted in Figures 1-3 and S2 and S3 indicate that triclosan reacts in the presence of monochloramine, but that the reactions occur over a longer time frame than observed with free chlorine (19). For comparative purposes,

FIGURE 1. Effect of pH on the apparent second-order rate constants for the electrophilic substitution (Cl+ transfer) of triclosan in the presence of monochloramine (filled circles) and free chlorine (open squares). For the monochloramine experiments: [NH2Cl]0 ) 42.3 µM, [triclosan]0 ) 3.8-7.9 µM, Cl/N ≈ 0.72, and [NaHCO3] ) 2 mM. For the free chlorine experiments: [HOCl]0 ) 14.3 µM, [triclosan]0 ) 5.4 µM, and [NaHCO3] ) 2 mM. (The previously unpublished free chlorine data was supplied by K. Rule, Virginia Tech.)

FIGURE 2. Monochloramine and triclosan decay at pH 8.5. Cl/N ≈ 0.72, T ) 22 °C, [NH2Cl]0 ) 42.3 µM, [triclosan]0 ) 5 µM, and [NaHCO3] ) 2 mM. Error bars depict 95% confidence interval about the mean (duplicates for monochloramine control and triplicates for monochloramine-triclosan samples). kapp values for triclosan loss in the presence of free chlorine are plotted in Figure 1. As shown, triclosan loss in chloramine solutions is 2-4 orders of magnitude slower than in free chlorine solutions. This difference in relative reaction rates helps explain the recalcitrance of triclosan in nonnitrified wastewater effluents in which chloramines are the dominant form of active chlorine. The observed increase in the triclosan loss rate with a decrease in pH parallels results reported for free chlorine with triclosan (19) as well as other phenols such as acetaminophen (7) and chlorophenols (26). In each of those cases, it was suggested that the rate enhancing effect of a change

FIGURE 3. Monochloramine decomposition in the presence and absence of triclosan, triclosan decay, and corresponding model predictions at pH ) 6.5. Cl/N ≈ 0.74, T ) 22 °C, [NH2Cl]0 ≈ 2.5 µM, [triclosan]0 ≈ 27.0 µM, and [NaHCO3] ) 2 mM. Error bars depict 95% confidence interval about the mean (duplicates for the monochloramine control and triplicates for the monochloramine-triclosan samples). in pH from pH 10 to 7 was the result of the increased concentration of reactive HOCl at lower pH. Similar pH trends were also observed by Pinkston and Sedlak for the reaction of acetaminophen with monochloramine (7); however, given that monochloramine speciation does not appreciably change over the pH 7-10 range, the authors noted that this pH effect would be unexpected if monochloramine was the only reactive oxidant in solution. They suggested that the formation of dichloramine and hypochlorous acid via acidcatalyzed monochloramine autodecomposition could explain this phenomenon, but they never mechanistically evaluated their hypothesis. Elevated Ammonia Conditions. The potential for hypochlorous acid to act as a reactive species was evaluated by conducting chloramination experiments under elevated ammonia conditions. At high ammonia concentrations, the production of hypochlorous acid is significantly repressed (Figure S4), thus making it possible to isolate the reactivity of monochloramine and dichloramine (23, 27). For this purpose, a set of experiments was run at pH 6.5 under excess monochloramine conditions and three chlorine to ammonia molar ratios (Cl/N ) 0.72, 0.072, and 0.0072). Figure 4A shows that triclosan decays much slower in the two experiments with elevated ammonia than at a Cl/N ratio of 0.72. This observation supports the hypothesis that hypochlorous acid formation enhances triclosan loss under these reaction conditions. Furthermore, the slow but significant decay of triclosan under excess ammonia conditions suggests that monochloramine and dichloramine also react with triclosan. Reaction Modeling. The monochloramine autodecomposition model described by eqs 1.1-1.14 in Table 1 was employed to mechanistically evaluate how the production of hypochlorous acid and dichloramine during autodecomposition affects triclosan decay. Model results predicting autodecomposition in the absence of triclosan are given in Figures 2, 3, and S2 and S3. The model predictions correlate VOL. 40, NO. 8, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2617

FIGURE 4. Triclosan decay and product formation at three Cl/N ratios. (A) Triclosan, (B) 5,6-dichloro-2-(2,4-dichlorophenoxy) phenol, (C) 2,4-dichlorophenol, and (D) 2,4,6-trichlorophenol. pH ) 6.5, T ) 22 °C, [NH2Cl]0 ) 42.3 µM, [triclosan]0 ≈ 3.2 µM, and [NaHCO3] ) 2 mM. The formation of 4,5-dichloro-2-(2,4-dichlorophenoxy)phenol and 4,5,6-trichloro-2-(2,4-dichlorophenoxy)phenol exhibited similar trends as shown in panel B for 5,6-dichloro-2-(2,4-dichlorophenoxy) phenol. Error bars depict 95% confidence interval about the mean of triplicate samples. well with the experimental data, thus corroborating previous results (16). One minor limitation of the model is the prediction of autodecomposition at high pH values and low initial monochloramine concentrations. As shown for pH 10.5 in Figure S3a, the autodecomposition model underpredicts the decay experimentally observed for an initial monochloramine concentration of 2.5 µM. For a higher initial monochloramine concentration of 43 µM, however, the autodecomposition model fits the experimental data readily (Figure S2a). The autodecomposition model was originally developed over a pH range of 6.5-8.3 (21), and therefore, additional reactions may need to be considered to correctly 2618

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 8, 2006

predict autodecomposition at alkaline pH values and low initial chloramine concentrations. Determination of these parameters was outside the scope of this project. In aqueous solutions, triclosan exists either as an uncharged neutral species or as a phenolate anion (phenolatetriclosan). The phenolate form predominates when the pH is above a pKa of 7.9 (28). Prior studies have shown that the phenolate form of triclosan, and phenols in general, is considerably more reactive toward electrophilic substitution (Cl+ transfer) than the neutral form (ref 19 and references therein), and thus, phenolate-triclosan was considered the only reactive form in the modeling exercise. The reactions

of phenolate-triclosan with monochloramine (eq 1.16), hypochlorous acid (eq 1.17), and dichloramine (eq 1.18) were added to the main autodecomposition model (Table 1) to predict both monochloramine loss and triclosan decay. kArO-NH

2Cl

NH2Cl + phenolate-triclosan 98 products kArO-HOCl

HOCl + phenolate-triclosan 98 products kArO-NHCl

2

NHCl2 + phenolate-triclosan 98 products

(1.16) (1.17) (1.18)

The rate constant for reaction 1.17, the reaction of phenolatetriclosan with HOCl, was previously determined to be 5.4 × 103 M-1 s-1 (19). To obtain the rate constant (kArO-NH2Cl) for the direct reaction between triclosan and monochloramine, the rate constant for the reaction between triclosan and dichloramine (kArO-NHCl2) was initially set to zero, and the excess triclosan data sets at pH 8.5 and 10.5 were fit using Scientist. This approach is reasonable because at high pH values and low initial monochloramine concentrations, dichloramine production is insignificant (