Effects of Fulvic Acid on Fe(II) Oxidation by Hydrogen Peroxide


Vishal Verma, Roberto Rico-Martinez, Neel Kotra, Laura King, Jiumeng Liu, Terry W. Snell, .... J. D. Willey,, R. F. Whitehead,, R. J. Kieber, and, D. ...
1 downloads 0 Views

Environ. Sci. Technol. 1996, 30, 1106-1114

Effects of Fulvic Acid on Fe(II) Oxidation by Hydrogen Peroxide BETTINA M. VOELKER* AND BARBARA SULZBERGER Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, CH-8600 Duebendorf, Switzerland

Iron redox cycling can catalyze the oxidation of humic substances and increase the rate of oxygen consumption in surface waters rich in iron and organic carbon. This study examines the role of Fenton’s reaction [oxidation of Fe(II) by hydrogen peroxide] in this catalytic cycle. A number of competing processes were observed in model systems containing dissolved Fe, hydrogen peroxide, and Suwannee River fulvic acid. First, the effective rate constant of Fenton’s reaction increased with increasing fulvic acid concentration, indicating the formation of Fe(II)-fulvate complexes that react more rapidly with hydrogen peroxide than Fe(II)-aquo complexes. This effect was significant at pH 5 but negligible at pH 3. A second effect was scavenging of the HO• radical produced in Fenton’s reaction by fulvic acid, forming an organic radical. The organic radical reduced oxygen to HO2•/O2•-, which then regenerated hydrogen peroxide by reaction with Fe(II). Finally, Fe(III) was reduced by a dark reaction with fulvic acid, characterized by an initially fast reduction followed by slower processes. The behavior of Fe(II) and hydrogen peroxide over time in the presence of fulvic acid and oxygen could be described by a kinetic model taking all of these reactions into account. The net result was an iron redox cycle in which hydrogen peroxide as well as oxygen were consumed (even though direct oxidation of Fe(II) by oxygen was not significant), and the oxidation of fulvic acid was accelerated.

Introduction Both Fe(II) and hydrogen peroxide are common constituents of oxygenated natural waters. In sunlit surface waters containing natural organic matter, photochemical reactions can result in the rapid formation of both Fe(II) through ligand-to-metal charge transfer reactions of Fe(III)-organo complexes (either in solution or on the surface of iron oxides) (1-7) and hydrogen peroxide mainly through the * Corresponding author present address: Fye Laboratory, Woods Hole Oceanographic Institution, Woods Hole, MA 02543; e-mail address: [email protected]; telephone: (508) 289-2641; fax: (508) 457-2164.

1106

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996

reduction of oxygen by photo-excited organic substances (8-10). Photochemical reactions are not the only source of Fe(II) and hydrogen peroxide. Humic substances have been shown to reduce Fe(III) to Fe(II) in the absence of light (11-13), and microbial processes are also potential sources of both Fe(II) (14-16) and hydrogen peroxide (17). If hydrogen peroxide is present at sufficient concentrations, it will oxidize Fe(II) faster than oxygen does. Hydrogen peroxide is likely to be a dominant oxidant of Fe(II) not only in acidic natural waters, where the rate of oxidation of Fe(II) by oxygen is very slow, but also in marine waters at pH 8 (18, 19). Conversely, Fe(II) can be a dominant reductant of hydrogen peroxide in some surface waters (20). The reaction of Fe(II) with hydrogen peroxide (also called Fenton’s reaction) is therefore of interest in a variety of natural water systems. Organic substances can affect both the rate and the products of Fenton’s reaction. Carboxylate ligands are known to accelerate the effective rate of the reaction by forming complexes with Fe(II) that react faster than the aquo complexes (21, 22). Although it is often assumed that the HO• radicals produced by Fenton’s reaction oxidize another Fe(II), HO• is also rapidly scavenged by organic matter. The organic intermediate product formed by the latter reaction can reduce oxygen to HO2•/O2•-, whose reaction with Fe(II) results in regeneration of hydrogen peroxide and oxidation of Fe(II). Finally, if organic reducing agents are present, Fe(II) may be regenerated from Fe(III). Humic substances contain a high density of carboxylate functional groups that complex iron (23) and are therefore likely to affect the rate of Fenton’s reaction. They are also known to react rapidly with HO• radicals (24) and to reduce Fe(III) (11-13). Each of the latter two processes results in the oxidation of humic acid. In an aquatic system containing humic substances then, the reaction of Fe(II) with hydrogen peroxide can play a significant part in both the redox cycling of iron and in the oxidation of otherwise refractory organic matter. In this study, we examine Fenton’s reaction in the absence of light in systems containing hydrogen peroxide, a model fulvic acid, and iron at concentrations similar to those found in acidic surface waters. We show that fulvic acid simultaneously plays the role of a carboxylate ligand, a radical scavenger, and a reductant of Fe(III). By examining each of these roles in detail, we can construct a kinetic model describing the behavior of Fe(II) and hydrogen peroxide in our experimental systems. This model can then be used to understand fulvic acid’s competing roles in Fe oxidation and reduction and to make a qualitative assessment of the overall effect of Fenton’s reaction in the redox cycling of iron and the oxidation of organic matter in natural surface waters.

Background Fulvic Acid as a Ligand and Its Effect on the Rate of Fenton’s Reaction. Hydrogen peroxide is degraded and Fe(II) is oxidized by Fenton’s reaction:

Fe(II) + H2O2 f Fe(III) + HO• + OH-

(1)

The overall (apparent) rate constant of this reaction is the sum of the rate constants k of the reactions of each Fe(II)

0013-936X/96/0930-1106$12.00/0

 1996 American Chemical Society

species multiplied by the fraction R of the total Fe(II) present as this species:

kapp )

∑kR

1

∫ [Fe(II)] dt ≈ ∑2(t t

0

t

i+1

- ti)([Fe(II)]i+1 + [Fe(II)]i)

i

(6)

(2)

x x

x

Even a minor species (Rx , 1) can have a large effect on kapp if the reaction rate constant kx of this species is much higher than that of the major species. For example, the pH dependence of kapp is a result of the fact that, in the absence of other ligands, the kFeOH+RFeOH+ term dominates the sum above even though RFeOH+ is much smaller than RFe2+ within the pH range normal for natural waters (18, 19). The same effect has been observed in the presence of some organic ligands: Sedlak and Hoigne´ (22) demonstrated that oxalate strongly accelerates the apparent rate of reaction 1 even when only a small portion of the Fe(II) is complexed by the ligand. Other carboxylate ligands behave similarly (21). Stabilization of Fe(II) by the formation of complexes reacting with hydrogen peroxide more slowly than the Fe(II)-aquo complexes is also a possibility. Note that while even minor species can have an accelerating effect on Fe(II) oxidation, stabilization of Fe(II) through complexation can occur only if an Fe(II)-organic complex is a major species. Fulvic acid contains numerous carboxylate binding sites (6 µg/mg for Suwannee River fulvic acid; 23), which probably dominate its metal complexation behavior. (Because phenolate sites are less abundant and generally have much higher pKa’s than carboxylate sites, they are unlikely to play a dominant role at the acidic pH values considered here.) We would therefore expect an increase in the rate of Fe(II) oxidation by hydrogen peroxide in the presence of fulvic acid. In this case, kapp is given by

kapp ) kFe2+RFe2+ + kFeOH+RFeOH+ +

∑k

orgRorg

(3)

where ∑korgRorg refers to the complexes of Fe(II) with the carboxylate sites present in fulvic acid. The decrease of hydrogen peroxide concentration over time due to reaction 1 (in the absence of oxygen, so that no regeneration of hydrogen peroxide can take placessee below) is described by the second-order rate equation:

d[H2O2] ) -kapp[Fe(II)][H2O2] dt

(4)

where kapp is given by eq 3. If Fe(II) were present in large excess (i.e., invariable with time) or present in constant proportion to hydrogen peroxide (in the absence of reactions other than reaction 1), this equation would have a simple solution. However, because of the reduction of iron by fulvic acid in our systems, the concentration of Fe(II) is not a simple function of time. To allow direct comparison of the results of experiments when initial Fe(II) concentrations and the behavior of Fe(II) over time vary from one experiment to the next, the data can be linearized as follows. Integrating eq 4 gives

-ln

(

[H2O2]t

[H2O2]o

)

) kapp

∫ [Fe(II)] dt t

0

t

(5)

We estimate the value of integral using the Fe(II) measurements made during the course of the experiments:

Plotting ln[H2O2]t/[H2O2]o versus this estimate of the integral (which could be described as a time-weighted average Fe(II) concentration, multiplied by time) should yield a linear function with a slope equal to kapp. Because this linearization is based on measured Fe(II) concentrations, it is valid no matter by which reaction Fe(II) is produced or consumed. For example, comparisons of kapp in the presence and the absence of fulvic acid are possible as long as reaction 1 is the only sink of hydrogen peroxide. The possibility of Fe(III) reduction by H2O2 or its conjugate base HO2- is classically invoked to explain the catalytic effect of Fe(III) on H2O2 decomposition in the absence of Fe(II) and at high H2O2 to Fe(III) ratios (25, 26). Because this reaction is at least several orders of magnitude slower than reaction 1 (26), we need not consider its effect in our experimental systems, which contain an excess of Fe(II) over Fe(III). Fulvic Acid as a Scavenger of HO•sProduction of HO2•/ O2•-. One product of Fenton’s reaction (reaction 1) is HO•, which can rapidly oxidize organic substances as well as Fe(II). The rate constant of the reaction of HO• with humic substances is on the order of 3 × 104 (mg/L of DOC)-1 s-1 (24). In our systems [5 mg/L DOC and 0-5 µM Fe(II)], the reaction of HO• with fulvic acid should therefore easily outcompete the reaction of HO• with Fe(II) (k ) 3 × 108 M-1 s-1; 27). In the presence of oxygen, HO2•/O2•- (pKa 4.8; 28) is a common product of the reaction of HO• with organic compounds, such as benzene, formate, glyoxylate, ethanol, methanol, and humic acid (29). For example, the addition of HO• to benzene results in phenol and HO2•/O2•- (30). Alternatively, HO• may abstract a hydrogen atom from the organic molecule:

R-H + HO• f R• + H2O

(7)

Oxygen addition then leads to a peroxy radical, which may eliminate HO2•/O2•-:

R• + O2 f RO2• f Rox + HO2•/O2•-

(8)

Formation of HO2•/O2•- from HO• in the presence of fulvic acid and oxygen could occur either via HO• addition or via hydrogen atom abstraction. It has been suggested (21, 31, 32) that the transient oxidant generated by Fenton’s reaction may not be HO• but possibly a ferryl (FedO2+) species, even in the case of the “classical” Fenton reagent, aqueous Fe2+ and hydrogen peroxide in acidic solution (32). Since this species can also rapidly oxidize organic substances, for example, by hydrogen atom abstraction (32), HO2•/O2•remains a plausible product in the presence of oxygen. HO2•/O2•- generally degrades by dismutation, forming hydrogen peroxide by the overall reaction: H+

2HO2•/O2•- 98 H2O2 + O2

(9)

HO2•/O2•- may also act either as a reductant or as an oxidant of iron: H+

Fe(II) + HO2•/O2•- 98 Fe(III) + H2O2

VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

(10)

9

1107

-H+

Fe(III) + HO2•/O2•- 98 Fe(II) + O2

(11)

Thus, the HO2•/O2•- generated in the presence of oxygen not only regenerates hydrogen peroxide but also participates in the iron redox cycle. The ratio of hydrogen peroxide formed per HO2•/O2•- could vary from 0 to 1 depending on which of the three reactions above are the dominant reaction pathways.

Experimental Section Materials. All reagents used were reagent grade unless otherwise mentioned. All glassware and other vessels were soaked in 0.1 N HCl for at least 12 h before use. All solutions were prepared in distilled, deionized water (Barnstead NANOpure system). Suwannee River fulvic acid (SRFA) was isolated by J. Leenheer according to the method described in Leenheer (33). Stock solutions of SRFA in water were prepared fresh every month and stored in the refrigerator. Concentrations of Fe present initially in the SRFA were found to be insignificant (2 µM), ferrozine was used. Ferrozine measurements agreed well with the measurements carried out using the luminol method. Solutions were kept in a water bath at 25 °C for temperature control during the experiments. Optimization of Unknown Kinetic Parameters. To model Fe reduction, unknown kinetic parameters were determined using an iterative simplex routine combining the mathematics capabilities of MATLAB (a mathematics software/programming package sold by The Math Works, Inc., Natick, MA 01760) with the kinetic problem-solving

kapp ) kapp,0 +

FIGURE 1. Comparison of the rates of hydrogen peroxide degradation at various fulvic acid concentrations in de-aerated solutions. The x-axis represents a time-average Fe(II) concentration multiplied by time, given by the integral of [Fe(II)]t dt. See eqs 5 and 6 in the text for a further explanation of this linearization of the data. Experiments were conducted on time scales of several hours. (a) pH 3, [H2O2]o ≈1 µM, [Fe(II)]o ≈5 µM; [SRFA]: 0 (O), 10 (B, 3), and 30 mg/L (1). (b) pH 5, [H2O2]o ≈1 µM, [Fe(II)]o ≈2 µM; [SRFA]: 0 (O), 3 (9), 10 (3), and 30 mg/L (1).

program ACUCHEM (44). This procedure is described in more detail in Voelker (7).

Results and Discussion Fulvic Acid as a Carboxylate Ligand. To determine whether the ∑korgRorg term in eq 3 is significant, we examined the effect of fulvic acid concentration on the apparent rate constant of Fenton’s reaction in de-oxygenated systems at pH 3 and pH 5. At pH 3, the effect of fulvic acid concentration was negligible (Figure 1a), while at pH 5, an increase in concentration of fulvic acid resulted in significant acceleration of the degradation of hydrogen peroxide (Figure 1b). This behavior is consistent with a pHdependent complexation of Fe(II) by fulvic acid. If we assume that only a small portion of the total Fe(II) is organically complexed in these systems, which is likely because the binding of Fe(II) by carboxylate ligands is generally observed to be weak, then RFe2+ and RFeOH+ should not change as a function of fulvic acid concentration. In this case, the observed kapp is given by

∑k

orgRorg

(12)

where kapp,0 is the apparent reaction rate constant in the absence of fulvic acid. If the iron binding sites in the fulvic acid are present in excess, then each Rorg should be a linear function of fulvic acid concentration, and a plot of kapp versus fulvic acid concentration will yield a straight line, as is the case at pH 5 (Figure 2). At pH 3, kapp at high fulvic acid concentrations is still approximately equal to kapp,0 (Figure 2), suggesting that less Fe(II) is complexed by fulvic acid at pH 3 than at pH 5 (so that at pH 3, ∑korgRorg is much smaller than kapp,0 at the fulvic acid concentrations studied), presumably because of competition of Fe2+ and H+ for the metal-binding sites. Our measurement of kapp,0 at pH 3 (43.8 ( 1.6 M-1 s-1) is comparable to previously observed values, which range from 40 to 80 M-1 s-1 in acidic media containing various concentrations of a background electrolyte (19, 45-49). The kapp,0 value obtained at pH 5 is difficult to compare with other studies because of the ionic medium effects observed by Millero et al. (19, 45), but the higher value is consistent with the effect of increasing [FeOH+] that is generally observed. We conclude that with regard to the effect of fulvic acid on the rate of Fenton’s reaction, the substance behaves similarly to a small carboxylic acid like oxalate. In this study, we find no evidence of sufficient amounts of Fe(II)-stabilizing binding sites to affect Fe(II) oxidation by hydrogen peroxide. Fulvic Acid as a Radical Scavenger. We observed a significant effect of oxygen on the apparent degradation rate of hydrogen peroxide in the presence of fulvic acid and Fe(II) (Figure 3). To show that this effect is due to a regeneration of hydrogen peroxide in the presence of oxygen via HO2•/O2•- (see reactions 8-10), we conducted an experiment in the presence of NO•(aq). NO•(aq) is an efficient HO2•/O2•- scavenger (kO2- ) 4.3 × 109 M-1 s-1, kHO2 ) 3.2 × 109 M-1 s-1; 50), forming NO3- by rearrangement of the peroxonitrite intermediate (51, 52):

NO•(aq) + HO2•/O2•- f OONO- f f NO3- (13) Although the reaction of NO•(aq) with HO• is also rapid (k ≈ 1010 M-1 s-1; 53), at the concentrations of fulvic acid present in our systems, at most 11% of the HO• produced will react with NO•(aq) (at the equilibrium concentration of NO•(aq), see Experimental Section), and the rest will react with fulvic acid. NO•(aq) may also react with the organic radical intermediates resulting from oxidation of fulvic acid by HO•, but because reaction rate constants of both O2 and NO•(aq) with these intermediates are expected to be near the diffusion-controlled limit (54), reaction with O2 will outcompete reaction with NO•(aq) by at least a factor of 100 in aerated solutions. In a solution containing both NO•(aq) and oxygen, the rate of hydrogen peroxide degradation by Fe(II) was the same as the rate observed in the deoxygenated system (Figure 3a), confirming the role of HO2•/O2•- in the regeneration of hydrogen peroxide. [NO•(aq) in the deoxygenated system had no effect on the rate of hydrogen peroxide degradation.] From the difference in slopes between the aerated and de-aerated experiments in Figure 3 (linear regressions not shown), we calculate that 47.6 ((2.7)% (at pH 3) and 36.3 ((2.9)% (at pH 5) of the hydrogen peroxide degraded by reaction 1 is regenerated by the HO2•/O2•- mechanism

VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1109

FIGURE 2. Apparent rate constant of Fenton’s reaction (reaction 1 in text) as a function of fulvic acid concentration at pH 3 (O) and pH 5 (b). kapp was calculated from linear regressions of the data in Figure 1. Error bars represent 95% confidence limits on the slopes of the regression lines. When no error bars are shown, errors are smaller than 5%.

(errors given are 95% confidence limits). If HO2•/O2•- reacts mainly by reaction 10, so that each HO2•/O2•- formed results in one molecule of hydrogen peroxide, we can conclude that the efficiency of HO2•/O2•- formation from the reaction of HO• with fulvic acid is approximately 48% and 36% at pH 3 and pH 5, respectively. If reaction 11 also plays a role, this stoichiometry cannot be determined so easily. However, we expect that most of the Fe(III) in this system is complexed by fulvic acid (see below). In general, Fe(III)organo complexes are much less reactive with HO2•/O2•than inorganic Fe(III) complexes (28, 22), so we would expect reaction 11 to be insignificant in this system. Fulvic Acid as a Reductant of Fe(III). To gain a better understanding of the kinetics of Fe(III) reduction in our systems, we have studied this process in the absence of hydrogen peroxide in aerated systems at pH 3 and pH 5 containing, initially, 10 mg/L SRFA and varying amounts of Fe(III). In all of the experiments at both pH 3 and pH 5, we observed a very fast initial reduction, followed by much slower processes (Figure 4). Significant formation of hydrogen peroxide did not occur during Fe(III) reduction, indicating that reaction of oxidized fulvic acid with O2 (leading to formation of HO2•/O2•-) was not important. Also, no formation of Fe(II) was observed in the absence of SRFA. The observed behavior seems to indicate reduction by two (or more) reducing sites: after a limited amount of fast-reducing sites have been oxidized, the reduction proceeds more slowly at other sites. The amount of Fe(III) reduced by the fast process should then depend only on the total number of sites available for this reaction. However, our data show that the total amount of Fe(III) reduced by the fast process increased when the concentration of initially added Fe(III) was increased, even though the concentration of fulvic acid (and therefore the total number of reducing sites) was kept constant. In fact, the fraction of Fe(III) reduced by the fast process remained nearly constant even when the iron-to-ligand ratios were varied by over an order of magnitude. One plausible explanation of these results is a competition between different functional groups present on the fulvic acid: an Fe(III) binding ligand L (such as a carboxylate

1110

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996

FIGURE 3. Comparison of the rates of hydrogen peroxide degradation in aerated (b) and de-aerated (O) systems. [SRFA] )10 mg/L. (a) pH 3, [H2O2]o ≈1 µM, [Fe(II)]o ≈5 µM. 1 and 3: experiments conducted in the presence of NO(aq) in aerated and de-aerated solutions, respectively. (b) pH 5, [H2O2]o ≈1 µM, [Fe(II)]o ≈2 µM.

functional group) and an Fe(III)-reducing site L′ (for example, a quinone-type structure). Dissolved Fe(III) can thus either be complexed by L:

Fe(III) + L f Fe(III)-L

(14)

or reduced by L′:

Fe(III) + L′ f Fe(II) + L′ox

(15)

(For simplicity, the net charges on the species have been omitted.) If the rates of both reactions are very fast, and both L and L′ are present in excess, the fraction of iron that is reduced quickly will be k15/(k14 + k15) independent of the iron-to-ligand ratio (where k14 and k15 are the pseudo-firstorder rate constants of reactions 14 and 15). According to this model, the rate of reduction then slows down because all of the remaining Fe(III) is present as the Fe(III)-L complex. If complexed Fe(III) is not reduced directly, the rate of the slower reduction process is limited by the dissociation of Fe(III)-L:

Fe(III)-L f Fe(III) + L

(16)

where the Fe(III) released by this reaction can again be reduced by reaction 15. The observed rate of reduction

FIGURE 4. Reduction of Fe(III) by fulvic acid in the dark in aerated solutions. [SRFA] ) 10 mg/L. (a) pH 3; total Fe(III) added: 150 nM ()), 500 nM (4), 1.50 µM (O), 5.00 µM (0), and 1.00 µM Fe(III) added to a solution already containing 4.9 µM Fe(II) (3). (b) pH 5; total Fe(III) added: 500 nM (4), 1.50 µM (O), 5.00 µM (0), and 1.00 µM added to a solution containing 2.0 µM Fe(II) (3). In the experiments in the presence of added Fe(II) (3), [Fe2+]/[FeT] on the y-axis refers to the fraction of the added Fe(III) that is reduced over time. Solid lines represent model fits (see text).

should then be proportional to the concentration of Fe(III)-L. However, at both pH 3 and pH 5, the reduction rate slowed down significantly in the 100 min following the initial fast reduction, even though the concentration of Fe(III)-L had not changed much in this time period (Figure 4). This suggests that the dissociation rate of the Fe(III)-L complex slows down over time. This effect was also observed by Choppin and Clark (55) for the dissociation of UO2-humate complexes and could indicate either a slow rearrangement of the fulvic acid, resulting in stronger binding of the metal, or simply a redistribution of the Fe(III) among different fulvic acid binding sites. We represent this process by the reaction:

Fe(III)-L f Fe(III)-L′′

(17)

where Fe(III)-L′′ represents an Fe(III) complex that dissociates more slowly than Fe(III)-L:

Fe(III)-L′′ f Fe(III) + L′′

(18)

The reduction kinetics observed in our data can be

approximated by a kinetic model consisting of reactions 14 to 18 (solid lines in Figure 4). Fitting parameters were the ratio k15/(k14 + k15) (the exact values of k14 and k15 are not relevant to the kinetic model since both reactions are assumed to be fast) and the first-order rate constants of reactions 16-18 (Table 1). The iterative fitting routine used to determine values of the fitting parameters converged on the same set of parameters independent of initial guesses, indicating that unique “best fits” could be obtained from this kinetic model. Fits were obtained by assuming the rate constants of reactions 16-18 to be independent of iron-to-ligand ratios (so that the same k16, k17, and k18 were used to model all the experiments at one pH). However, because the data show a slight decrease of the fraction of iron reduced in the fast initial step with increasing total iron concentration, the ratio k15/(k14 + k15) was assumed to vary as a function of the total iron in the experiments. Apparently, micromolar concentrations of Fe(II) [either present initially or produced by reduction of Fe(III)] affect the rate of reduction k15, possibly through complexation at the reducing site L′. At pH 5, the fraction of iron reduced by the fast reaction is much smaller than at pH 3, but there is little effect of total Fe(II) on this fraction. This model provides an adequate description of the kinetics of reduction of Fe(III) by fulvic acid for the purpose of this study, which is to understand the contribution of this process to the observed behavior of the Fe(II)-SRFAH2O2 system. Further studies are necessary before a more detailed and rigorous mechanistic interpretation of this complicated process can be made. Kinetic Model of Reactions in the Fe(II)-SRFA-H2O2 System. In an oxygenated solution containing Fe(II), SRFA, and H2O2 then, dissolved Fe(II) is oxidized by hydrogen peroxide and HO2•/O2•- and reduced by a dark reaction with fulvic acid (Figure 5). We can construct a kinetic model of this system using the rate constants (kapp) of the reaction of Fe(II) and hydrogen peroxide shown in Figure 2, the efficiency of HO2•/O2•- formation from HO• in the presence of oxygen calculated from Figure 3 (under the assumption that all HO2•/O2•- formed reacts with Fe(II) via reaction 10), and the description of Fe(III) reduction kinetics summarized in Table 1. While the experimental data only provide information on the net oxidation of Fe(II) and the net disappearance of hydrogen peroxide, model calculations let us compare the rates of competing processes occurring simultaneously (Figure 5). Our model accurately reproduces the observed concentrations of Fe(II) and hydrogen peroxide over time in de-aerated and aerated fulvic acid solutions at pH 3 and pH 5 (Figures 6 and 7). While the hydrogen peroxide data shown in Figures 6 and 7 (appearing in linearized form in Figures 2 and 3) were used to calibrate the model’s parameters, the Fe data were not. The model’s success in explaining the behavior of Fe(II) over time therefore indicates that we have correctly identified the most important processes taking place in our experimental systems. Including the reduction of Fe(III) by HO2•/O2•in the model would lead to predictions of somewhat higher Fe(II) concentrations than were observed, justifying our assumption that this reaction is insignificant. Other reactions, such as a direct reaction of organic radical intermediates with Fe(II) or Fe(III), may be occurring, but also do not seem to contribute significantly to the observed behavior of Fe(II) and hydrogen peroxide.

VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1111

TABLE 1

Reactions and Parameters Used in the Kinetic Model reactionsa dark reduction of Fe(III) by fulvic acid (14) Fe(III) + L f Fe(III)-L (15) Fe(III) + L′ f Fe(II) + L′ox k15/(k14 + k15)b

(16) Fe(III)-L f Fe(III) + L c (17) Fe(III)-L f Fe(III)-L′′ c (18) Fe(III)-L′′ f Fe(III) + L′′ c additional reactions used in kinetic model (1) Fe(II) + H2O2 f Fe(III) + HO• + OH- d % yield of HO2•/O2•- from HO• reaction w/SRFA (10) Fe(II) + HO2•/O2•- f Fe(III) + H2O2 f

total Fe(III) added (µM)

0.150 0.500 1.50 5.00 1.00 + Fe(II)e

model parameters pH 3

pH 5

fast fast 0.51 0.45 0.31 0.23 0.17 1.16 × 10-3 2.55 × 10-4 7.05 × 10-6

fast fast 0.22 0.20 0.19 0.19 9.15 × 10-4 6.22 × 10-4 1.96 × 10-5

49.8 ( 2.5 47.6 ( 2.7 fast

176.2 ( 7.0 36.3 ( 2.9 fast

a Number of reactions as used in the text. b k -1 14 and k15 are pseudo-first-order rate constants in s . Because the rates of reactions 14 and 15 are faster than the time scale of the measurements (minutes), only the ratio of these parameters could be determined from the data (see text). c Rate constants in s-1. d Apparent rate constant in M-1 s-1. e Initial concentrations of Fe(II): 4.9 µM (pH 3) and 2.0 µM (pH 5). f We assumed that all of the HO2•/O2•- produced reacted with Fe(II).

FIGURE 5. Summary of iron redox reactions in the experiments shown in Figures 6 and 7 (aerated solutions only). The numbers in italics indicate the amount of iron (in µM) oxidized or reduced by each process after 100-min reaction time (results of model calculations; model parameters are listed in Table 1) at pH 3 (boldface, first number) and at pH 5 (lightface, second number).

The effect of Fenton’s reaction on oxygen consumption and on the oxidation of Fe(II) and fulvic acid in our model system may be more easily visualized if we formulate an overall stoichiometry. This is derived by adding reactions 1 and 7 to reactions 8 and 10, where the latter two reactions are multiplied by a factor of 0.4, the approximate efficiency of HO2•/O2•- formation:

1.4Fe(II) + R-H + 0.6H2O2 + 0.4O2 + 1.8H+ f 1.4Fe(III) + 0.4Rox + 0.6R′ox + 2H2O (19)

FIGURE 6. Kinetics of degradation of hydrogen peroxide and oxidation of Fe(II) in aerated (b) and de-aerated (O) solutions at pH 3. [SRFA] ) 10 mg/L. Solid lines represent model fit.

(R′ox represents the products of reactions of HO• with fulvic acid that do not result in formation of HO2•/O2•-.) A total of 1.4 equivalents of electrons is produced by oxidation of Fe(II) to Fe(III), and another 1.4 equivalents by the oxidation of R-H to Rox• and R′ox (or the analogous oxidation products of HO• addition). Some equivalents (1.2) are consumed by the reduction of H2O2 to H2O and the rest (1.6 equivalents) by reduction of O2 to H2O. Much of the Fe(III) formed in this reaction is reduced back to Fe(II) within a

few hours (Figure 5), resulting in further oxidation of fulvic acid. The overall result of Fenton’s reaction is a catalytic oxidation of fulvic acid in the presence of iron, similar to that observed by Miles and Brezonik (56). Miles and Brezonik (56) assumed that the mechanism of oxygen consumption in the systems they examined was direct reaction with Fe(II). However, at acidic pH this reaction should be too slow to consume oxygen on a time scale of

1112

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996

Heather Lounsbury for introducing us to the chemiluminescence flow system and to Jerry Leenheer for providing us with a sample of Suwannee River fulvic acid.

Literature Cited

FIGURE 7. Same as in Figure 6, except at pH 5.

hours. Our study presents an alternative explanation: oxygen is reduced to HO2•/O2•- by an organic radical produced through Fenton chemistry, and the HO2•/O2•then oxidizes Fe(II). Extrapolating the rates of reactions observed in our model systems to natural waters is not a simple matter. The effective rate constant of Fenton’s reaction depends on the extent of organic complexation of Fe(II), which is likely to vary with pH, ionic strength, iron-to-ligand ratios, and the nature of the organic material. The efficiency of O2 reduction to HO2•/O2•- does not seem to be strongly affected by changes in pH, but could vary as a function of the type and concentration of humic material present. Finally, the dark reduction of Fe(III) was found to be a complicated process whose kinetics are affected by pH, Fe concentrations, and reaction time. While the results of this case study cannot be generalized in a quantitative way, the properties of Suwannee River fulvic acid relevant to Fenton’s reaction, such as its abilities to complex and reduce iron and act as a radical scavenger, are ones that are generally observed in humic material. A redox cycle comparable to the one depicted in Figure 5 should be a common phenomenon in natural waters, especially in the presence of light, since both Fe(II) and hydrogen peroxide are rapidly produced by photochemical reactions involving humic substances.

Acknowledgments We wish to thank Ju ¨ rg Hoigne´ and Oliver Zafiriou for their comments on this manuscript; Stephan Hug for his help with Matlab and the kinetic modeling; and Silvio Canonica, Franc¸ ois Morel, David Sedlak, and Yuegang Zuo for helpful discussions. We are also grateful to Whitney King and

(1) Waite, T. D.; Morel, F. M. M. J. Colloid Interface Sci. 1984, 102, 121-137. (2) Waite, T. D.; Morel, F. M. M. Environ. Sci. Technol. 1984, 18 , 860-868. (3) Siffert, C.; Sulzberger, B. Langmuir 1991, 7, 1627-1634. (4) Zepp, R. G.; Faust, B. C.; Hoigne´, J. Environ. Sci. Technol. 1992, 26, 313-319. (5) Zuo, Y.; Hoigne´, J. Environ. Sci. Technol. 1992, 26, 1014-1022. (6) Faust, B. C.; Zepp, R. G. Environ. Sci. Technol. 1993, 27, 25172522. (7) Voelker, B. M. Ph.D. Dissertation No. 10901, Eidgeno¨ssische Technische Hochschule, Zu ¨ rich, Switzerland, 1994. (8) Petasne, R. G.; Zika, R. G. Nature 1987, 325, 516-518. (9) Cooper, W. J.; Zika, R. G.; Petasne, R. G.; Fischer, A. M. In Aquatic Humic Substances: Influence on Fate and Treatment of Pollutants; Suffett, I. H., MacCarthy, P., Eds.; American Chemical Society: Washington, DC, 1989; pp 333-362. (10) Sturzenegger, V. T. Ph.D. Dissertation No. 9004, Eidgeno¨ssische Technische Hochschule, Zu ¨ rich, Switzerland, 1989. (11) Szila`gyi, M. Soil Sci. 1971, 111, 233-235. (12) Langford, C. H.; Kay, R.; Quance, G. W.; Khan, T. R. Anal. Lett. 1977, 10, 1249-1260. (13) Skogerboe, R. K.; Wilson, S. A. Anal. Chem. 1981, 53, 228232. (14) Fischer, W. R. In Iron in Soils and Clay Minerals; Stucki, J. W., Goodman, B. A., Schwertmann, U., Eds.; D Reidel Publishing Company: Dordrecht, 1988; pp 715-748. (15) Jones, G. J.; Morel, F. M. M. Plant Physiol. 1988, 87, 143147. (16) Price, N. M.; Maldonado, M. T. Abstract presented at Workshop on Iron Speciation and its Availability to Phytoplankton in Seawater, Bermuda Biological Station, May 1-5, 1994. (17) Palenik, B.; Zafiriou, O. C.; Morel, F. M. M. Limnol. Oceanogr. 1987, 32, 1365-1369. (18) Moffett, J. W.; Zika, R. G. Environ. Sci. Technol. 1987, 21, 804810. (19) Millero, F. J.; Sotolongo, S. Geochim. Cosmochim. Acta 1989, 53, 1867-1873. (20) Moffett, J. W.; Zafiriou, O. C. J. Geophys. Res. 1993, 98 (C2), 23072313. (21) Rush, J. D.; Maskos, Z.; Koppenol, W. H. Methods Enzymol. 1990, 186, 148-156. (22) Sedlak, D. L.; Hoigne´, J. Atmos. Environ. 1993, 27A, 2173-2185. (23) Aiken, G. R.; Brown, P. A.; Noyes, T. I; Pinckney, D. J. In Humic Substances in the Suwannee River, Georgia: Interactions, Properties, and Proposed Structures; Averett, R. C., Leenheer, J. A., McKnight, D. M., Thorn, K. A., Eds.; U.S. Geological Survey Open File Report 87-557; U.S. Geological Survey: Denver, CO, 1989; pp 163-178. (24) Hoigne´, J.; Faust, B. C.; Haag, W. R.; Scully, F. E.; Zepp, R. G. In Aquatic Humic Substances: Influence on Fate and Treatment of Pollutants; Suffett, I. H., MacCarthy, P., Eds.; American Chemical Society: Washington, DC, 1989; pp 363-381. (25) Haber, F.; Weiss, J. Proc. R. Soc. Ser. A 1934, 147, 332-351. (26) Barb, W. G.; Baxendale, J. H.; George, P; Hargrave, K. R. Trans. Faraday Soc. 1951, 47, 591-616. (27) Christensen, H.; Sehested, K. Radiat. Phys. Chem. 1981, 18, 723731. (28) Bielski, B. H. J.; Cabelli, D. E.; Arudi, R. L.; Ross, A. B. J. Phys. Chem. Ref. Data 1985, 14, 1041-1100. (29) Staehelin, J.; Hoigne´, J. Environ. Sci. Technol. 1985, 19, 12061213. (30) Dorfman, L. M.; Taub, I. A.; Bu ¨ hler, R. E. J. Chem. Phys. 1962, 36, 3051-3061. (31) Sawyer, D. T.; Kang, C.; Llobet, A.; Redman, C. J. Am. Chem. Soc. 1993, 115, 5817-5818. (32) Wink, D. A.; Nims, R. W.; Saavedra, J. E.; Utermahlen, W. E.; Ford, P. C. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 6604-6608. (33) Leenheer, J. A. Environ. Sci. Technol. 1981, 15, 578-587. (34) Bader, H.; Sturzenegger, V.; Hoigne´, J. Water Res. 1988, 22, 11091115. (35) Moss, M. L.; Mellon, M. G. Ind. Eng. Chem. (Anal. Ed.) 1942, 14, 862-865. (36) Stookey, L. L. Anal. Chem. 1970, 42, 779-781. (37) King, D. W.; Lounsbury, H. A.; Millero, F. J. Environ. Sci. Technol. 1995, 29, 818-824. (38) Seitz, W. R.; Hercules, D. M. Anal. Chem. 1972, 44, 2143-2149.

VOL. 30, NO. 4, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1113

(39) Klopf, L. L.; Nieman, T. A. Anal. Chem. 1983, 55, 1080-1083. (40) O’Sullivan, D. W.; Hanson, A. K.; Kester, D. R. Marine Chem., submitted for publication. (41) Skoog, D. A.; West, D. M. Fundamentals of Analytical Chemistry, 2nd ed.; Holt International: London, 1969. (42) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 73rd ed.; CRC Press: Boca Raton FL, 1992-1993, p 6-4. (43) Awad, H. H.; Stanbury, D. M. Int. J. Chem. Kinet. 1993, 25, 375381. (44) Braun, W.; Herron, J. T.; Kahaner, D. K. Int. J. Chem. Kinet. 1988, 20, 51-62. (45) Millero, F. J.; Sotolongo, S.; Stade, D. J.; Vega, C. A. J. Solution Chem. 1991, 20, 1079-1092. (46) Po, H. N.; Sutin, N. Inorg. Chem. 1968, 7, 621-624. (47) Wells, C. F.; Salam, M. A. J. Chem. Soc. A, 1968, 24-29. (48) Barb, W. G.; Baxendale, J. H.; George, P.; Hargrave, K. R. Trans. Faraday Soc. 1951, 47, 462-500. (49) Harwick, T. J. Can. J. Chem. 1957, 35, 428-436. (50) Goldstein, S.; Czapski, G. Free Radical Biol. Med. 1995, 19, 505510.

1114

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 4, 1996

(51) Blough, N. V.; Zafiriou, O. C. Inorg. Chem. 1985, 24, 35023504. (52) Micinski, E.; Ball, L. A.; Zafiriou, O. C. J. Geophys. Res. 1993, 98 (C2), 2299-2306. (53) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. J. Phys. Chem. Ref. Data 1988, 17, 513-886. (54) Zafiriou, O. C.; Blough, N. V.; Micinski, E.; Dister, B.; Kieber, D.; Moffett, J. Mar. Chem. 1990, 30, 45-70. (55) Choppin, G. R.; Clark, S. B. Mar. Chem. 1991, 36, 27-38. (56) Miles, C. J.; Brezonik, P. L. Environ. Sci. Technol. 1981, 15, 1089-1095.

Received for review March 28, 1995. Revised manuscript received October 23, 1995. Accepted November 21, 1995.X ES9502132 X

Abstract published in Advance ACS Abstracts, February 1, 1996.