Inhibition of Hydroxyl Radical Reaction with Aromatics by Dissolved

Mehlman, M. A.; Mumtaz, M. M.; Faroon, O.; De Rosa, C. T. J. Clean Technol. ...... Eileen F. Wheeler, Paul H. Heinemann, Kenneth B. Kephart, and Jerzy...
0 downloads 0 Views 134KB Size
Environ. Sci. Technol. 2000, 34, 444-449

Inhibition of Hydroxyl Radical Reaction with Aromatics by Dissolved Natural Organic Matter MICHELE E. LINDSEY† AND MATTHEW A. TARR* Department of Chemistry, University of New Orleans, New Orleans, Louisiana 70148

Reaction of aromatic compounds with hydroxyl radical is inhibited by dissolved natural organic matter (NOM). The degree of inhibition is significantly greater than that expected based on a simple model in which aromatic compound molecules bound to NOM are considered to be unreactive. In this study, hydroxyl radical was produced at steady-state concentrations using Fenton chemistry (H2O2 + Fe2+ f Fe3+ + HO- + HO‚). Suwannee River fulvic acid and humic acid were used as NOM. The most likely mechanism for the observed inhibition is that hydroxyl radical formation occurs in microenvironmental sites remote from the aromatic compounds. In addition to changes in kinetics, pyrene hydroxyl radical reaction also exhibited a mechanistic change in the presence of fulvic acid. The mechanism changed from a reaction that was apparently firstorder in pyrene to one that was apparently secondorder in pyrene, indicating that pyrene self-reaction may have become the dominant mechanism in the presence of fulvic acid. Dissolved NOM causes significant changes in the rate and mechanism of hydroxyl radical degradation of aromatic compounds. Consequently, literature rate constants measured in pure water will not be useful for predicting the degradation of pollutants in environmental systems. The kinetic and mechanistic information in this study will be useful for developing improved degradation methods involving Fenton chemistry.

Introduction Remediation of hydrophobic pollutants is complicated by sorption of these compounds to hydrophobic sites of dissolved natural organic matter (NOM), suspended particulates, soil, and sediment. This sorption causes the pollutants to be less easily degraded by remediation techniques (1). Although partitioning to these microenvironments has been studied with respect to pollutant transport (2-6) and bioavailability (1, 7, 8), little effort has been made to understand how partitioning affects chemical reactivity of the pollutants (1, 9-11). Both physical and chemical changes upon sorption can alter pollutant reactivity. For example, a pollutant in the interior of a particle may be physically isolated from reactants in bulk solution, or changes in rate constants or mechanisms may occur upon partitioning to a chemically distinct microenvironment. Understanding these changes and their effects on degradation processes is necessary in * Corresponding author phone: (504)280-6323; fax: (504) 2806860; e-mail: [email protected]. † Current address: U.S. Geological Survey, 4821 Quail Crest Place, Lawrence, KS 66049-3839. 444

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

order to devise effective and economically feasible remediation strategies. Polycyclic aromatic hydrocarbons (PAHs) are well studied pollutants with known carcinogenic effects (12). PAHs are ubiquitous pollutants (12, 13) with both natural (14) and anthropogenic sources (15-17). In addition, PAHs are hydrophobic compounds which partition into hydrophobic microenvironments, including hydrophobic sites of natural organic matter such as fulvic and humic acids (4, 18-23). Partition coefficients and binding kinetics have been investigated for these compounds (4, 18, 20-26), but the effects of binding on their chemical reactivity has not been thoroughly investigated. Numerous studies have, however, demonstrated the effects of binding to NOM on bioavailability (1, 7, 8). In general, in the presence of dissolved or colloidal NOM, hydrophobic compounds exhibit higher effective total aqueous solubility (freely dissolved plus bound to NOM). Bioavailability of compounds in the presence of dissolved or colloidal NOM is often decreased (27-30), although reports of increased (31) and unchanged (29, 32) bioavailability are also available. Fulvic acid (FA) and humic acid (HA) are classes of NOM that have been shown to bind hydrophobic compounds (4, 19-23). Humic acid molecules are typically larger and generally contain more aromatic character, while fulvic acid molecules often contain more carbonyl and aliphatic regions (33, 34). The structure of these molecules affects their ability to bind a molecule or sequester it from bulk solution. For example, HA molecules with highly aromatic regions tend to form micelle-like structures which are capable of sequestering hydrophobic molecules such as PAHs (1, 19). Schulten (19) determined through three-dimensional modeling that the hydrophobic regions of two humic molecules can form pockets that can trap molecules such as pentachlorophenol. In another study, HA was shown to be more effective than FA at sorbing pyrene, presumably due to the higher aromatic character of the HA (34). Humic and fulvic acids are also capable of binding metals, and several studies have reported on the role of humics in the formation of hydroxyl radical (HO‚) via iron-hydrogen peroxide reaction (35-37). Formation of hydroxyl radical by reaction of Fe2+ with H2O2, the Fenton reaction (38-43), has been utilized for the degradation of hydrophobic pollutants (44-48). Formation of hydroxyl radical occurs through the following reaction (38):

H2O2 + Fe2+ f Fe3+ + HO- + HO‚

(1)

The ferrous ion is regenerated through additional reactions and therefore acts as a catalyst. The hydroxyl radical reacts with a wide variety of compounds. Reaction with alkenes and aromatics is very fast, with second-order rate constants in the range of 1091010 M-1 s-1 (49). Since hydroxyl radical is nonselective, it may be useful for degrading a broad range of pollutants. However, reaction of hydroxyl radical with nonpollutant species can be a significant sink of the radical (9), resulting in decreased degradation efficiency. Furthermore, the presence of iron binding compounds, such as NOM, can alter the rate constant for reaction 1 or can alter the redox cycling of iron and thereby change the formation rate of hydroxyl radical (9). In this study, the effects of binding to dissolved NOM on the reaction of aromatic compounds with hydroxyl radical was investigated. Fenton chemistry was used with continuous addition of peroxide to yield steady-state hydroxyl radical 10.1021/es990457c CCC: $19.00

 2000 American Chemical Society Published on Web 12/31/1999

concentrations. The effect of FA and HA on the reaction kinetics was studied in order to better understand how binding to NOM can affect chemical reactions of hydrophobic pollutants in aqueous systems.

Experimental Section High purity water (NP) was obtained from a NanopureUV (Barnstead) water treatment system using a distilled water feed. Suwannee River humic acid (SRHA) and fulvic acid (SRFA) were purchased from the International Humic Substances Society (http://www.ihss.gatech.edu). Hydrogen peroxide (EM science, ∼30%) was standardized using iodometric titration (50). Iron(II) perchlorate (99+%) was purchased from Alfa. Benzoic acid (BA) (99.5+%), p-hydroxybenzoic acid (p-HBA) (99+%), o-cresol (99+%), phenanthrene (99+%), and pyrene (99%) were purchased from Aldrich. Phenol (99+%) was purchased from Fisher, and 1-propanol (PrOH) (99+%) was purchased from Mallinckrodt. All reagents were used as received. Hydroxyl radical was produced using Fenton chemistry (43-48). Prior to degradation, individual compounds were dissolved in water, and the pH was adjusted to 2.5 with hydrochloric acid. Just prior to the use of each solution, an aliquot of Fe(ClO4)2 (aqueous pH 2.5) was added to yield an initial Fe2+ concentration of 50 µM. Hydrogen peroxide was then added continuously with a syringe pump (KD Scientific). Low flow rates in the range of 0.15 mL h-1 were used so that volume changes were negligible over the course of an experiment (initial reaction volume was typically 3-10 mL). Continuous addition of H2O2 was used to establish a steadystate hydroxyl radical concentration in each degradation reaction. All reactions were performed at 20 °C with constant stirring. Hydroxyl radical concentration was measured using benzoic acid as a probe (9, 51). Formation of p-HBA was quantitated by high performance liquid chromatography using a Hewlett-Packard 1090 liquid chromatograph. A Spherisorb ODS-2 column (5 µm particle size, 25 cm length × 4.6 mm id) was used for all separations. Benzoic acid and the hydroxybenzoic acids were separated using the following procedure. Samples were loaded onto a 1.5 mL loop. After injection, the analytes were preconcentrated on-column during the initial 3 min and then were eluted by increasing the solvent strength. The elution gradient was water at pH ∼2.5 (A) and acetonitrile (B); 0-3 min 10% B, 3-12 min linear to 44% B, 12-13 min linear to 100% B. The mobile phase flow rate was 1 mL min-1. BA was detected by absorbance at 230 nm, and p-HBA was detected at 254 nm. Initial BA concentration was 50 µM. Benzoic acid was chosen as a hydroxyl radical trap because it has a known rate constant, exhibits a characteristic product (p-HBA) upon reaction, and has a low partition coefficient for binding to FA or HA. Consequently, this probe was expected to yield hydroxyl radical concentrations that were representative of the bulk aqueous phase. Phenol and o-cresol were also determined by HPLC using a Spherisorb ODS-2 column. These compounds were injected using a 100 µL loop, were separated isocratically with a mobile phase of 50/50 water/acetonitrile at 1 mL min-1, and were detected by absorbance at 254 nm. Each of these compounds was allowed to react at a steady-state hydroxyl radical concentration for predetermined time intervals. At these times, the reactions were quenched with 0.5 mL of PrOH per 10 mL of reaction solution, and the analytes were then quantitated. Each reaction represented one time point. Taken together, the individual points were used to reconstruct kinetic plots of analyte concentration verses time. The degradation of pyrene and phenanthrene at steadystate hydroxyl radical concentration was monitored continuously using fluorescence detection. Pyrene was excited

at a wavelength of 318 nm, and the decrease in fluorescence intensity at 370 nm was monitored during the reaction. Phenanthrene was excited at 292 nm, and the loss of fluorescence emission at 344 nm was monitored. Data were collected at 1 s intervals for periods of up to 40 min. Dark control experiments were carried out to ensure that the excitation radiation did not affect the reaction. All degradations were carried out at pH ) 2.5 (adjusted with HCl). The effects of FA and HA were evaluated by carrying out degradations in pure water and in the presence of up to 30 mg L-1 of SRFA or SRHA. Solid SRFA or SRHA was added directly to previously prepared solutions of the compound to be degraded. Solutions were then stirred vigorously to dissolve the added fulvic or humic acid and were allowed to equilibrate for several hours before use. Since no precipitate was observed after extended periods (several days), and since the solutions showed no visible opacity, it was assumed that the humics were completely dissolved or colloidal.

Results and Discussion To simplify kinetics, experiments were conducted under conditions of steady-state hydroxyl radical concentration. Hydrogen peroxide usually follows first-order disappearance in the presence of Fe2+ (9). The rate of disappearance of hydrogen peroxide will increase steadily as hydrogen peroxide is added until the rate of disappearance exactly balances the rate of addition, yielding a steady-state hydrogen peroxide concentration. Consequently, a constant formation rate of hydroxyl radical will result. Since hydroxyl radical loss also follows first-order kinetics (with respect to [HO‚]), loss and formation rates of HO‚ will also become equal, yielding a steady-state hydroxyl radical concentration. Indeed, the observance of pseudo-first-order reaction of aromatic hydrocarbons in this study verifies that steady-state hydroxyl radical concentration was achieved. Under the conditions used in this study, the major sink for hydroxyl radical was chloride. The chloride concentration was 3.2 mM, and its second-order rate constant for reaction with HO‚ is 4.2 × 109 M-1 s-1 (49). Fulvic acid was added at a maximum concentration of 30 mg L-1 (∼16 mg C L-1). Although a rate constant is not available for Suwannee River fulvic acid reaction with hydroxyl radical, an average value for several aquatic humics of (1.7 ( 0.7) × 104 (mg of C L-1)-1 s-1 has been reported (52). Based on this average value, at the maximum FA concentration used in this study (∼16 mg C L-1), reaction with chloride was ∼50 times more rapid than reaction with FA. Therefore it was assumed that FA was a minor sink for hydroxyl radical and that the FA did not significantly alter the [HO‚] through scavenging. Humic acid was also used at a maximum concentration of ∼16 mg C L-1, and therefore was also considered to be a minor sink for HO‚ scavenging. Although previous measurements have indicated that both FA and HA can alter the HO‚ formation rate, these effects are minimal at pH 2.5 (9, 35). Therefore, based on both scavenging and formation rate, it is reasonable to expect relatively constant hydroxyl radical concentrations upon addition of FA or HA under the conditions used here. Indeed, measured hydroxyl radical concentration showed no significant change upon addition of FA (Figure 1). HA showed similar results. Furthermore, pseudo-first-order kinetics were observed in the HO‚ quantitation measurements, and measured rate constants were in agreement with literature values (49) (see Table 1). These observations indicate that steady-state hydroxyl radical concentration was reached and maintained throughout the measurement period and that measured hydroxyl radical concentrations were reliable. Reaction of pyrene and o-cresol with hydroxyl radical in pure water showed first-order kinetics with pseudo-firstorder rate constants in agreement with literature values (49). Figure 2a illustrates the first-order kinetics observed for VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

445

FIGURE 1. Measured hydroxyl radical concentration as a function of FA concentration. Initial [Fe2+] ) 50 µM, 50 mM H2O2 (aqueous) added at 0.15 mL h-1, pH ) 2.5 with HCl. Benzoic acid (50 µM) used as hydroxyl radical probe.

TABLE 1. Second-Order Rate Constants for Reaction of Selected Aromatics with Hydroxyl Radical compound

o-cresol phenol benzene fluorene phenanthrene pyrene a

From ref 46.

b

lit. ka (M-1 s-1)

measd k (M-1 s-1)

1.1 × 1010 6.6 × 109 7.8 × 109 b b b

(9.2 ( 0.5) × 109 (6.4 ( 0.3) × 109 c (1.9 ( 0.3) × 1010 (2.3 ( 0.2) × 1010 (1.5 ( 0.6) × 1010

No literature value available. c Not measured.

pyrene degradation by hydroxyl radical. Table 1 presents rate constants measured in this study along with literature values. Our values are in agreement with previous reports using different methods. Pseudo-first-order kinetics were maintained for o-cresol upon addition of FA or HA. However, pyrene did not follow first-order behavior in the presence of FA. Instead, apparent second-order kinetics (with respect to pyrene) were observed (Figure 2b). At higher steady-state concentrations of HO‚ (1.2 × 10-13 M), apparent second-order kinetics for pyrene were also observed. These observations imply a mechanism that involves pyrene-pyrene reaction, probably involving a pyrene radical intermediate. Even at [HO‚] low enough (4.1 × 10-14 M) to yield pseudofirst-order kinetics for pyrene loss, addition of FA resulted in apparent second-order kinetics. The most likely explanation for this behavior is that pyrene-FA interaction involved aggregates containing more than one pyrene molecule. Pyrene molecules held in close proximity to each other in such aggregates would favor pyrene-pyrene reactions. This explanation, however, is not consistent with the expected extent of pyrene binding to FA. At 5 mg L-1 of FA, pyrene is expected to be only 4.9% bound to FA. Such a low fraction of bound pyrene should not be sufficient to shift the mechanism to second-order behavior. Since in the presence of FA no conditions could be found where pyrene exhibited pseudo-first-order kinetics, pyrene degradation experiments with FA were carried out under conditions of pseudo-secondorder kinetics. Pseudo-first-order kinetics were observed for pyrene in the presence of HA. Chin et al. (21, 24) have proposed that the degree of binding of a hydrophobic compound to dissolved NOM can be predicted by the following equation

f) 446

9

[OC]KDOC 1 + [OC]KDOC

(2)

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

FIGURE 2. (a) First-order degradation of pyrene in pure water. Initial [Fe2+] ) 50 µM, 2 mM H2O2 (aqueous) added at 0.15 mL h-1, pH ) 2.5 with HCl. (b) Second-order plot for degradation of pyrene in pure water with 5 mg L-1 FA, all other conditions same as (a). where f is the fraction of the compound bound to NOM, KDOC is the partition coefficient for binding to the NOM (normalized to dissolved organic carbon, DOC), and [OC] is the concentration of organic carbon in the NOM. Table 2 lists partition coefficients and percent bound to FA and HA for several compounds. Where measured values for KDOC were not available, values were calculated from eq 3 as proposed by Chin and Weber (34):

log KDOC ) 0.82 × log Kow + 0.1923

(3)

where Kow is the octanol water partition coefficient. Since this study focused on the effect of binding to NOM on the reactivity of pollutants, the maximum expected effect of binding was calculated in the following manner. The fraction bound was calculated using eq 2. It was assumed that the bound material did not react with hydroxyl radical (rate constant ) 0). The resulting apparent rate constant was then calculated by multiplying the rate constant in pure water by the fraction of compound freely dissolved [(1 - f) for pseudofirst-order, (1 - f)2 for pseudo-second-order]. This treatment yielded the maximum possible inhibition in reaction rate due to binding alone. Such inactivation could be due to reduced access of HO‚ to the site of the bound compound and/or a lower rate constant of the bound compound. To compare the calculated maximum inhibition to experimental observations, rate constants were measured under steady-state hydroxyl radical concentration as a function of added FA or HA concentration. This yielded pseudo-first-order rate constants for all compounds except for pyrene in the presence of FA, which yielded pseudosecond-order rate constants. Overall rate constants were then calculated by dividing by the steady state [HO‚], providing overall second-order rate constants (overall third-order for pyrene with FA).

TABLE 2. Partition Coefficients for Binding of Selected Aromatics to Natural Organic Matter

a

compound

log Kow

phenol o-cresol fluorene phenanthrene pyrene

1.47 (55) 1.94 (55) 4.18 (56) 4.52 (56) 5.00 (56)

Calculated from Kow using eq 3.

b

calcd KDOCa (L kg C-1) 24.99 60.69 4168 7920 19 602

exptl KDOC (L kg C-1)

8913 (34)b 10 000 (34)b 15 400 (57)c 31 500 (57)d

Commercial humic acid. c Suwannee River fulvic acid.

Calculated and measured second-order rate constants for o-cresol as a function of FA and HA concentration are presented in Figures 3a and 4a. The observed second-order rate constant is, in all cases, significantly smaller than the values predicted from a simple binding model as discussed above. In fact, o-cresol is predicted to be