Environ. Sci. Technol. 2010, 44, 5824–5829
Correlations between Dissolved Organic Matter Optical Properties and Quantum Yields of Singlet Oxygen and Hydrogen Peroxide ´ E M. DALRYMPLE,† RENE AMY K. CARFAGNO,‡ AND C H A R L E S M . S H A R P L E S S * ,§ Department of Chemistry, University of Wisconsin, Madison, Wisconsin 53706, SpecPro Environmental Services, Lorton, Virginia 22079, and Department of Chemistry, University of Mary Washington, Fredericksburg, Virginia 22401
Received March 29, 2010. Revised manuscript received June 9, 2010. Accepted June 15, 2010.
Various aquatic dissolved organic matter (DOM) samples produce singlet oxygen (1O2) and hydrogen peroxide (H2O2) with quantum yields of 0.59 to 4.5% (1O2 at 365 nm) and 0.017 to 0.053% (H2O2, 300-400 nm integrated). The two species’ yields have opposite pH dependencies and strong, but opposite, correlations with the E2/E3 ratio (A254 divided by A365). Linear regressions allow prediction of both quantum yields from E2/ E3 in natural water samples with errors ranging from -3% to 60%. Experimental evidence and kinetic calculations indicate that less than six percent of the H2O2 is produced by reaction between 1O2 and DOM. The inverse relationship between the 1 O2 and H2O2 yields is thus best explained by a model in which precursors to these species are populated competitively. A model is presented, which proposes that important precursors to H2O2 may be either charge-transfer or triplet states of DOM.
Introduction Natural dissolved organic matter (DOM) is a heterogeneous mixture produced by decay of plant and plankton biomass. Its brown color derives from a moderately high molecular weight (approximately 1 to 20 kD) aromatic fraction containing the so-called hydrophobic acids (humic and fulvic) (1). These participate in many environmentally significant reactions, among which is a diverse array of photochemical reactions. Upon absorbing light, DOM produces several reactive intermediates, including many reactive oxygen species (ROS) (2) such as singlet oxygen (1O2), superoxide (O2-), and hydrogen peroxide (H2O2). Excited state DOM triplets (3DOM*) are also important oxidants (3, 4). The photochemistry of DOM is significant for several reasons. First, ROS and 3DOM* can oxidize organic water contaminants (4-7). Second, the ROS can influence the redox cycling of biogeochemically significant metals, such as Fe and Hg (8, 9). Finally, DOM photooxidation is an important part of oceanic carbon cycling, although the links between DOM oxidation and ROS and 3DOM* production rates are * Corresponding author phone:
[email protected]. † University of Wisconsin. ‡ SpecPro Environmental Services. § University of Mary Washington. 5824
9
(540)654-1405;
e-mail:
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010
not well established (10-12). Because of their general significance to environmental redox kinetics, ROS and 3DOM* must be a part of models attempting to quantify these processes. Although photochemical models for individual ROS can be complex (13), at their heart is a calculation of the reactive species’ production rate, which depends on its quantum yield. In the case of DOM, quantum yields vary with DOM source, season, and water chemistry. This makes it difficult to predict reactive species’ quantum yields in a general fashion. Researchers and regulators must instead assess them experimentally on a case-by-case basis. An alternative is to use empirical models of how DOM affects contaminant photolysis rates (14, 15), but these models lack a mechanistic underpinning and thus generality. While the complexity of DOM hinders modeling its molecular-scale photochemistry, observed relationships between chemical, optical, and photochemical properties of DOM hint at the possibility of predicting photochemical reaction rates from spectroscopic data. For instance, several fluorescence properties (e.g., quantum yield) correlate negatively with DOM average molecular weight, and absorption coefficients depend strongly on DOM lignin phenol content (16). Also, low molecular weight HA fractions photodegrade 2,4,6-trimethylphenol more efficiently than high molecular weight fractions (17), and there is a positive correlation between photoreactivity and carbon normalized absorption coefficients (18). Furthermore, there is an inverse relationship between DOM molecular weight and the E2/E3 ratio, the absorbance at 254 nm divided by that at 365 nm (19). To explain these phenomena, a charge-transfer model has been proposed for DOM optical properties, which may also explain some aspects of DOM photochemistry (20). This model posits that tailing of DOM absorbance into the visible region reflects charge-transfer (CT) absorbance and is evidence for the existence of CT states. Understood this way, low E2/E3 ratios suggest that DOM photochemistry will be influenced by CT interactions, while high E2/E3 ratios suggest more reaction via non-CT pathways. This paper reports quantitative correlations between the yields of two ROS produced by DOM (1O2 and H2O2) and the E2/E3 ratio. Both ROS result from reaction of molecular oxygen with excited state DOM, but 1O2 is the product of energy transfer while H2O2 is the product of electron transfer. Energy transfer may be expected when non-CT photochemistry dominates DOM + hν f 1DOM∗ f 3DOM∗
(1)
DOM∗ + O2 f 1DOM + 1O2
(2)
1
3
Reaction 1 populates 3DOM* states by intersystem crossing from 1DOM* after excitation. Other fates of 1DOM* such as fluorescence and internal conversion have been omitted for notational convenience. Reaction 2 shows energy transfer from 3DOM* to produce 1O2. Conversely, when CT photochemistry is significant, electron transfer reactions occur, potentially populating CT states (DOM · +/ · -), which may serve as precursors to H2O2. DOM + hν f DOM·+/·-
1
10.1021/es101005u
(3)
2010 American Chemical Society
Published on Web 07/01/2010
DOM∗ f DOM·+/·-
(4)
DOM∗ f DOM·+/·-
(5)
1 3
DOM
·+/·-
+
+ O2 f ·DOM +
O2
+ 2O2 + 2H f H2O2 + O2
DOM∗ + O2 f ·DOM+ + O2
Ra )
∑ (E
0 p,λ(1
(11)
- 10-aλz)/z)
(12)
λ
(6) (7)
Here, DOM · +/ · - represents an intramolecular CT complex with diradical character. In reaction 3, the DOM · +/ · - state is populated by direct CT absorbance. Reactions 4 and 5 populate it by intramolecular electron transfer involving excited states. Reaction 6 shows electron transfer from DOM · +/ · - to O2 to produce O2-, which dismutates to form H2O2 (reaction 7). An alternative route to H2O2 not involving DOM · +/ · - is reaction 8, where 3DOM* is oxidized by O2. 3
Φ1O2 ) Rp /Ra
(8)
The correlations we present here are consistent with this scheme, and they have practical utility for estimating DOM photochemical reaction rates. Our results and discussion also demonstrate that H2O2 production does not involve 1O2 to any significant extent.
Here, E0p,λ is the spectral photon irradiance (millieinsteins cm-2 s-1) at the sample surface as determined radiometrically (23), λ indicates wavelength dependence, z is the optical path length (cm), and aλ is the solution absorption coefficient (cm-1). The summation in eq 12 was carried out below 400 nm. Photochemistry in this region is dominated by the 365 nm line of the Hg lamp, so reported Φ1O2 values approximate those at 365 nm. More detail concerning this approach can be found in the Supporting Information. In some cases, quantum yields were determined relative to SDOM at pH 7, which was repeatedly measured to be 0.018 mol einstein-1 using the approach described above. These experiments used a merry-go-round photoreactor (Ace Glass) with a 450 W Hg lamp and 365 nm band-pass filters. The experiments employed optically matched DOM samples containing FFA. Relative quantum yields were calculated using eq 13, where kobs,i and kobs,SDOM are rate constants for FFA loss in sample i and SDOM, respectively. This equation holds only when the DOM solutions are optically matched. Φrel,i ) (kobs,i /kobs,SDOM) × 0.018
Materials and Methods DOM Samples. Several DOM samples were used, the vast majority being from aquatic sources. A full list is provided in the Supporting Information. Some were purchased from the International Humic Substances Society (IHSS, http:// www.ihss.gatech.edu), while others were isolated using ultrafiltration or adsorption chromatography (see the Supporting Information). Stock and Experimental Solutions. All solutions were prepared in either 5 mM phosphate or borate buffered 0.1 M KCl. When using powdered DOM, 1 g L-1 stock solutions were prepared in pH 7 buffer followed by filtration (0.45 µm) and storage at 4 °C. Experimental solutions were prepared by diluting stock DOM or XAD isolates in buffer until the absorption at 365 nm was approximately 0.3 cm-1. This gave final concentrations of approximately 25 to 125 mg L-1 depending on the sample. Prior to experiments, samples were pH adjusted with 0.1 M HCl or NaOH, and absorption spectra were acquired from 200 to 600 nm at 1 nm intervals. 1 O2 Quantum Yields. Furfuryl alcohol (FFA, Aldrich, 99%) was used as a probe for 1O2. It is highly selective for 1O2 and does not undergo reactions with free radicals or other oxidants in irradiated DOM solutions (3, 21). Samples were spiked to approximately 5 µM FFA and sparged with air for 15 min. In most experiments, 25 mL of sample was magnetically stirred in a Petri dish beneath a 450 W Hg lamp and 3 mm Pyrex to filter out wavelengths below 300 nm. At selected intervals 0.2 mL were collected and analyzed for FFA by HPLC (C-18 column, 18% MeOH mobile phase, UV detection at 220 nm). Pseudo-first-order rate constants for FFA loss (kobs) were used to calculate steady-state 1O2 concentrations via eq 9, where kFFA is the rate constant for reaction between 1O2 and FFA, 1.2 × 108 M-1 s-1 (21). These were converted into 1O2 production rates (Rp, M s-1) by assuming that deactivation by water is the dominant 1O2 decay path, where kd ) 2.4 × 105 s-1 (22). [1O2]ss ) kobs /kFFA
(9)
Rp ) [1O2]ss × kd
(10)
Quantum yields (Φ1O2) were found using eqs 11 and 12, where Ra is the rate of light absorption (einsteins L-1 s-1).
(13)
The Supporting Information lists all the samples, E2/E3 ratios, and Φ1O2 values. H2O2 Quantum Yields. Rates of H2O2 production were measured in air-saturated DOM solutions. All experiments employed a 300 or 450 W Xe lamp whose emission was filtered through 3 mm Pyrex and a 5.8 cm path length water filter. Samples were irradiated in borosilicate test tubes and stirred magnetically while open to the air. Small portions of sample were regularly removed for H2O2 analysis (details provided as Supporting Information). Production rates of H2O2 were often constant with time, although in several cases they decreased when H2O2 concentrations exceeded approximately 2 or 3 µM. Therefore, initial production rates approximated as the change in concentration with time below 2 µM - were used in the calculations. Quantum yields (ΦH2O2) were then determined using eqs 12 and 14. ΦH2O2 ) (∆[H2O2]/∆t)/Ra
(14)
Because the Xe lamp is polychromatic and Ra was calculated using wavelengths below 400 nm, our ΦH2O2 values are a weighted average at all wavelengths between 300 and 400 nm. More detail concerning this approach can be found in the Supporting Information, which also contains a list of all the samples, E2/E3 ratios, and ΦH2O2 values.
Results and Discussion E2/E3 as a Predictor of Φ1O2. It is well-known that Φ1O2 is sensitive to solution conditions such as pH (24), where high pH leads to low Φ1O2 (Figure 1(a)). While these data might be taken to suggest that pH is a useful predictor of Φ1O2 in natural waters, this is not the case, as Figure 1(b) comprising data from diverse DOM sources demonstrates. Concomitant with a change in pH, a notable and universal change occurs in DOM absorbance spectra. As pH increases, the long wavelength tailing of the spectrum increases, which decreases E2/E3, the absorbance at 254 nm divided by that at 365 nm (Supporting Information). This ratio is a simple way to quantify the spectral tailing into the visible range. Another widely used parameter for this purpose is the spectral slope, which is determined by fitting the spectrum as an exponential curve (25). To our knowledge, there is no evidence VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5825
FIGURE 2. 1O2 quantum yields (as percents) versus E2/E3 (A254/ A365) for diverse DOM samples with regression line and equation and 95% confidence interval for regression (red lines).
TABLE 1. Predicted and Measured Φ1O2 for Whole Water Samples and an XAD Isolate sample
E2/E3
predicted Φ1O2 (%)a
measured Φ1O2 (%)
% error
Mt. Pleasant Tappahannock Waterview XAD
4.51 4.59 5.32
2.4 2.5 3.1
1.5 2.1 3.2
60% 19% -3%
a Predicted values equation in Figure 2.
FIGURE 1. 1O2 quantum yields (as percents) versus pH (a) for Suwannee River DOM and (b) for a collection of diverse DOM samples. to suggest that either E2/E3 or the spectral slope is to be strongly preferred for characterizing the absorption tailing. When the same Φ1O2 data shown in Figure 1(b) is plotted versus E2/E3, a reasonably linear correlation between the parameters appears. Figure 2 displays these data along with the regression line, the regression equation, and the 95% confidence interval for the regression. The data shown in Figures 2 and 1(b) represent single experiments, so error bars are not included. A list of all samples, E2/E3 ratios, and measured quantum yields is provided in the Supporting Information. Despite the scatter, there is a clear positive correlation between E2/E3 and Φ1O2. Importantly, this relationship holds regardless of pH for diverse samples, including whole water, unfractionated DOM obtained by reverse osmosis, and hydrophobic acid fractions obtained by adsorption chromatography. This trend is not unexpected; several researchers have shown that photosensitization efficiency increases as DOM molecular weight decreases (17, 18), and it is also known that E2/E3 increases as DOM molecular weight decreases (19). Thus, a correlation between E2/E3 and photosensitization efficiency makes sense. This correlation may be used to predict Φ1O2 from E2/E3. In our data set Φ1O2 ranges from less than 1% to almost 5%, and similar variability may be expected in natural waters. Predicted quantum yields could thus be used to improve the accuracy with which DOM photochemical reaction rates can 5826
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010
calculated
using
the
regression
be estimated or modeled. Table 1 shows predicted and measured values of Φ1O2 for three DOM samples not included in the data used for regression analysis (samples described in the Supporting Information). The predicted quantum yields have errors ranging from -3 to 60% (Table 1), and all of the measured values are well within the prediction interval at 95% confidence, (1.5% at the E2/E3 ratios shown here (26). This correlation thus provides a way to predict steady-state 1O2 concentrations in surface waters and estimate degradation rates of environmental contaminants that react with 1O2. 1
O2 + C f products kobs ) kC[1O2]ss
(15)
Here, C is a contaminant, and kC is the rate constant for reaction between 1O2 and C. In sunlight, the value of [1O2]ss is calculated via eq 16, in which Ra,DOM is the rate of light absorption by DOM, and the subscript λ indicates wavelength dependence. [1O2]ss )
∑R
a,DOM,λΦ1O2,λ
(16)
λ
In many cases, it may be possible to simplify eq 16 by using a solar spectrum weighted average of Φ1O2. The correlation here relates E2/E3 to Φ1O2 at 365 nm, so it may not apply directly to solar spectrum weighted values of Φ1O2. Refining it to do so would require further experimentation, but we suspect that the 365 nm quantum yield is a good reflection of DOM photochemical reactivity and that E2/E3 would also correlate with solar spectrum weighted yields. Limited data in our lab (not presented here) suggest that 365 nm quantum yields are linearly related to solar spectrum weighted values,
Fe3+) may scavenge O2- (30, 31) and alter the absolute production rate, or they may catalyze H2O2 decomposition and alter the observed production rate. Therefore, water chemistry data may be needed to refine predictions based on spectroscopic correlations.
Proposed Basis for the Observed Relationships The results presented above suggest that the ability of DOM to form CT excited states, as indicated by the E2/E3 ratio, correlates negatively with 1O2 yields and positively with H2O2 yields. Two different mechanisms could explain this. In one, 1 O2 reacts with DOM more rapidly as E2/E3 decreases, producing H2O2 in the process and lowering [1O2]ss, which would be interpreted using our experimental approach as a lower Φ1O2. In the other mechanism, H2O2 precursor states form in competition with 1O2 precursor states. If 1O2 reacts more rapidly with DOM as E2/E3 decreases, faster H2O2 production would be observed if O2- were produced (reaction 17). FIGURE 3. H2O2 quantum yields (as percents) versus E2/E3 (A254/ A365) for various IHSS DOM samples with regression line and equation and 95% confidence interval for regression (red lines). Error bars represent the standard deviation of duplicate experiments.
TABLE 2. Predicted and Measured ΦH2O2 for a Whole Water Sample and Two XAD Isolates sample
E2/E3
predicted ΦH2O2 (%)a
measured ΦH2O2 (%)
% error
Tappahannock2 Tapp2 XAD Siders Pond XAD
4.38 4.12 4.42
0.019 0.024 0.018
0.025 0.020 0.019
-25% 19% -6%
a
Predicted values equation in Figure 3.
calculated
using
the
regression
so it may be possible to use the present correlation with a linear adjustment. E2/E3 as a Predictor of ΦH2O2. We also examined whether there was a relationship between E2/E3 and the H2O2 quantum yield, ΦH2O2. For various reasons, only IHSS samples were used in this portion of the work (see the Supporting Information). As observed by other researchers, we find ΦH2O2 to be approximately 100 times smaller than Φ1O2 (27-29). Raising the pH increases ΦH2O2, which is the opposite of the trend for Φ1O2. As with 1O2, there did not seem to be a strong correlation between pH and ΦH2O2 when data from multiple samples were combined. However, ΦH2O2 shows a negative correlation with E2/E3 for the IHSS samples (Figure 3) in direct contrast to the positive correlation for Φ1O2. It is also somewhat stronger, but this may be because the DOM samples are less diverse than in the 1O2 experiments. The utility of this correlation was tested on three samples, river water and two XAD isolates (Table 2, samples described in the Supporting Information). The predicted H2O2 quantum yields have similar percent errors to predicted 1O2 quantum yields and are also well within the prediction interval at 95% confidence, (0.016% at these E2/E3 ratios. This suggests that the regression may also be useful for environmental modeling. However, it must be applied judiciously because our ΦH2O2 values are expressed on the basis of all light absorbed below 400 nm, not as solar spectrum weighted values. Also, predicting field concentrations of H2O2 from ΦH2O2 is more difficult than predicting steady-state 1O2 concentrations from Φ1O2. In the case of 1O2, its deactivation by water (kd ) 2.4 × 105 s-1) dominates the decay kinetics in the majority of environmentally significant situations, thus simplifying the calculations. For H2O2, however, dissolved metals (e.g., Cu2+,
+ O2 + DOM f O2 + ·DOM
1
(17)
Analogous reactions occur with various reductants in water (32, 33). With DOM, reaction 17 could occur two different ways: by encounter between freely diffusing 1O2 and DOM or by reaction of 1O2 within the DOM microenvironment before it escapes into solution (34). To test for reaction 17, we conducted two experiments (for details, see the Supporting Information). First, H2O2 production rates were measured in D2O, extending the 1O2 lifetime ten times relative to water (35). If reaction 17 occurs, its rate would thus increase 10-fold with correspondingly faster H2O2 production. We observed approximately 20% faster H2O2 production. Although this is similar to the typical deviation in duplicate ΦH2O2 measurements, it is based on the average of duplicate experiments in both solvents. Also, it is in general agreement with a recent report that H2O2 production by Suwannee River fulvic acid increases in D2O compared to H2O (36). Thus, we feel that the difference is real and interpret the additional H2O2 as arising from enhanced reaction between 1O2 and DOM. Assuming that the only difference in the two solvents is the 10-fold higher concentration of 1O2, the 20% of H2O2 production attributable to 1O2 in D2O would decrease to 2% of the H2O2 formed in H2O. However, slower O2- dismutation in D2O, which increases [O2-]ss, introduces uncertainties that complicate quantifying the effect in H2O. As an upper bound estimate, the 3-fold faster dismutation of O2- in H2O compared to D2O (37) could increase the H2O2 forming through reaction 17 to approximately 6% of the total. A second experiment, conducted in H2O, employed the hydrophobic 1 O2 scavenger β-carotene at a bulk concentration of 200 nM to test whether 1O2 reacts within the DOM microenvironment before escaping into solution (34). No effect was observed, consistent with a very low percentage of H2O2 being formed through reactions of 1O2 with DOM. Rate calculations support this conclusion. For example, Cory et al. report the rate constant for reaction 18 as approximately 25 L mg-C-1 s-1 (12). 1
O2 + DOM f products
(18)
Under our experimental conditions with 25 mg C L-1 at pH 7, we typically achieve [1O2]ss ∼ 10-12 M, giving 6 × 10-10 M s-1 as the rate of 18. If O2- were the only product (i.e., reaction 17), this represents the maximum rate of its formation from 1 O2. From this, one can calculate the rate at which the reaction produces H2O2 to be 3 × 10-10 M s-1 (see the Supporting Information), which is approximately 10 times smaller than VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5827
SCHEME 1. Simplified Model of Excited State DOM Reactions with O2 Illustrating the Proposed Role of CT Complexes
our experimental H2O2 production rates. Thus, 10% or less of the H2O2 production can be ascribed to reaction between 1 O2 and DOM. Furthermore, it has been estimated that less than 14% of 1O2 undergoing reaction 18 produces H2O2 (36), which implies an H2O2 production rate of 8 × 10-11 M s-1 if the rate of reaction 18 is 6 × 10-10 M s-1. This is approximately 3% of our observed rates, further supporting the estimate of two to six percent of the H2O2 forming via reaction 17. Thus, our results indicate that 1O2 is not a significant route to H2O2, making it unlikely that the inverse relationship between Φ1O2 and ΦH2O2 is due to increased 1O2 scavenging by DOM with lower E2/E3. An alternative explanation is that as E2/E3 decreases, the production rate of H2O2 precursor states increases at the expense of 1O2 precursors. It appears that the whole 3DOM* pool, rather than a subfraction of the triplets, is capable of forming 1O2 (38). In question then is the nature of the H2O2 precursors, and either 3DOM* or the proposed DOM · +/ · species seem likely candidates (reactions 8 and 6). If 3DOM* is the H2O2 precursor, then the opposite trends in Φ1O2 and ΦH2O2 suggest that as E2/E3 decreases, the rate of reaction 8 increases. If reaction 8 is the source of H2O2, it would need to proceed at about 3 × 10-9 M s-1 in our experiments. Assuming that we achieve [3DOM*]ss ∼ 5 × 10-13 M (38), then the rate constant for reaction 8 would need to be 2.4 × 107 M-1 s-1 in aerated solution ([O2] ) 250 µM), which is reasonable, but somewhat low for electron transfer. If the rate constant were on the order of 1.2 × 1010 M-1 s-1 (39), then only a small fraction of triplets (∼ 0.002) must be reacting to form H2O2. These are plausible scenarios, so reaction 8 is a reasonable hypothesis. It is also reasonable to hypothesize that CT states are H2O2 precursors. The anion-radical portion of DOM · +/ · should efficiently reduce O2 with a rate constant for reaction 6 of near 1010 M-1 s-1. At such high rates, ΦH2O2 on the order of 5 × 10-4 could only occur if DOM · +/ · - were long-lived with a low quantum yield or if it were a very short-lived species. Lifetimes of different DOM radical pools have been reported to lie in the range 1 to 250 µs with an average of about 100 µs (39). This value combined with our experimental conditions and previously estimated rates of DOM radical production (39) allows us to estimate the rate constant for reaction 6 as at least 4 × 109 M-1 s-1 (calculations in the Supporting Information). This is a reasonable value for reaction between radicals and O2, suggesting that DOM radicals produced by CT reactions could be H2O2 precursors. These considerations lead us to propose the tentative model in Scheme 1. The model applies for pristine DOM solutions that do not contain exogenous electron donors (e.g., methylphenols) that could reduce DOM, which would introduce additional routes to O2-. In this model, the 1O2 precursor is 3DOM*, and both 3DOM* and CT states, 5828
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010
DOM · +/ · -, are shown as H2O2 precursors. The model omits production of O2- by reaction of O2 with e-(aq) on the basis of previous research indicating that this is a negligible source of H2O2 (40). The CT states are populated either by direct CT absorbance or deactivation of DOM excited states. Both 1 DOM* and 3DOM* may produce DOM · +/ · -, but we note that the inability of O2 to quench radical precursors in DOM (39) leads us to suggest that short-lived singlets are the main precursors to DOM · +/ · - rather than the longer lived triplets. This model thus explains the inverse relationship between Φ1O2 and ΦH2O2 as arising from competitive formation of the 1 O2 and H2O2 precursors. If 3DOM* is the H2O2 precursor, the competition is between the formation of triplets that either produce 1O2 (reaction 2) or H2O2 (reaction 8). If DOM · +/ · - is the H2O2 precursor, the competition could be loss of 3DOM* to DOM · +/ · - via reaction 5 or formation of DOM · +/ · - from 1 DOM* (reaction 4) at the expense of 3DOM* formed by intersystem crossing (reaction 1). Future tests of the model could include measurements and comparisons of the 1O2 and H2O2 precursor lifetimes and studies of how triplet quenchers affect H2O2 production rates.
Acknowledgments This research was supported by grants from the Research Corporation (Cottrell College Science Award #6839) and the Virginia Academy of Science (Small Project Research Funds). We also acknowledge the support of the University of Mary Washington’s Summer Science Research Program and Undergraduate Research Grants. We also thank Dr. Ryan Fimmen for providing the Pony Lake and Lake Toolik samples.
Note Added after ASAP Publication A reaction arrow was omitted from equation 6 in the version of this paper published ASAP on July 1, 2010. The correct version published July 6, 2010.
Supporting Information Available DOM samples used in the research, hydrophobic acid isolation procedures, lamp spectra, spectral distribution of sample light absorption rates, absorption spectra and E2/E3 ratios for SDOM, numerical presentation of quantum yields for all samples, H2O2 analytical details, details of experiments to test for 1O2 intermediacy in H2O2 production, and calculation of the maximum rate of H2O2 production by reaction between 1O2 and DOM. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Thurman, E. M.; Malcolm, R. L. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 1981, 15, 463–466. (2) Blough, N. V.; Zepp, R. G. Reactive oxygen species in natural waters. In Active Oxygen in Chemistry; Foote, C. S., Valentine, J. S., Greenberg, A., Liebman, J. F., Eds.; Chapman & Hall: NY, 1995. (3) Halladja, S.; ter Halle, A.; Aguer, J.-P.; Boulkamh, A.; Richard, C. Inhibition of humic substances mediated photooxygenation of furfuryl alcohol by 2,4,6-trimethylphenol. Evidence for reactivity of the phenol with humic triplet excited states. Environ. Sci. Technol. 2007, 41, 6066–6073. (4) Chen, Y.; Hu, C.; Hu, X.; Qu, J. Indirect photodegradation of amine drugs in aqueous solution under simulated sunlight. Environ. Sci. Technol. 2009, 43, 2760–2765. (5) Guerard, J. J.; Chin, Y.-P.; Mash, H.; Hadad, C. M. Photochemical fate of sulfadimethoxine in aquaculture waters. Environ. Sci. Technol. 2009, 43, 8587–8592. (6) Boreen, A. L.; Edhlund, B. L.; Cotner, J. B.; McNeill, K. Indirect photodegradation of dissolved free amino acids: the contribution of singlet oxygen and the differential reactivity of DOM from various sources. Environ. Sci. Technol. 2008, 42, 5492–5498. (7) Gerecke, A. C.; Canonica, S.; Mu ¨ ller, S. R.; Scha¨rer, M.; Schwarzenbach, R. P. Quantification of dissolved natural organic matter (DOM) mediated phototransformation of phenylurea herbicides in lakes. Environ. Sci. Technol. 2001, 35, 3915–3923.
(8) Rose, A. L.; Waite, T. D. Role of superoxide in the photochemical reduction of iron in seawater. Geochim. Cosmochim. Acta 2006, 70, 3869–3882. (9) Costa, M.; Liss, P. S. Photoreduction of mercury in sea water and its possible implications for Hg0 air-sea fluxes. Mar. Chem. 1999, 68, 87–95. (10) Wang, W.; Johnson, C. G.; Takeda, K.; Zafiriou, O. C. Measuring the photochemical production of carbon dioxide from marine dissolved organic matter by pool isotope exchange. Environ. Sci. Technol. 2009, 43, 8604–8609. (11) Clark, C. D.; Hiscock, W. T.; Millero, F. J.; Hitchcock, G.; Brand, L.; Miller, W. L.; Ziolkowski, L.; Chen, R. F.; Zika, R. G. CDOM distribution and CO2 production on the Southwest Florida Shelf. Mar. Chem. 2004, 89, 145–167. (12) Cory, R. M.; Cotner, J. B.; McNeill, K. Quantifying interactions between singlet oxygen and aquatic fulvic acids. Environ. Sci. Technol. 2009, 43, 718–723. (13) Vione, D.; Maurino, V.; Minero, C.; Carlotti, M. E.; Chiron, S.; Barbati, S. Modelling the occurrence and reactivity of the carbonate radical in surface freshwater. C. R. Chim. 2009, 12, 865–871. (14) Lam, M. W.; Tantuco, K.; Mabury, S. A. PhotoFate: a new approach in accounting for the contribution of indirect photolysis of pesticides and pharmaceuticals in surface waters. Environ. Sci. Technol. 2003, 37, 899–907. (15) Linden, K. G.; Sharpless, C. M.; Andrews, S. A.; Atasi, K. Z.; Korategere, V.; Stefan, M.; Suffet, I. H. (Mel). Innovative UV Technologies to Oxidize Organic and Organoleptic Chemicals. AwwaRF Report 91033F; Water Research Foundation Report Series, IWA Publishing: London, 2005. (16) Boyle, E. B.; Guerriero, N.; Thiallet, A.; Del Vecchio, R.; Blough, N. V. Optical properties of humic substances and CDOM: relation to structure. Environ. Sci. Technol. 2009, 43, 2262–2268. (17) Richard, C.; Trubetskaya, O.; Trubetskoj, O.; Reznikova, O.; Afanas’Eva, G.; Aguer, J.-P.; Guyot, G. Key role of the low molecular size fraction of soil humic acids for fluorescence and photoinductive activity. Environ. Sci. Technol. 2004, 38, 2052– 2057. (18) Cavani, L.; Halladja, S.; ter Halle, A.; Guyot, G.; Corrado, G.; Ciavatta, C.; Boulkamh, A.; Richard, C. Relationship between Photosensitizing and Emission Properties of Peat Humic Acid Fractions Obtained by Tangential Ultrafiltration. Environ. Sci. Technol. 2009, 43, 4348–4354. (19) Peuravuori, J.; Pihlaja, K. Molecular size distribution and spectroscopic properties of aquatic humic substances. Anal. Chim. Acta 1997, 337, 133–149. (20) Del Vecchio, R.; Blough, N. V. On the origin of the optical properties of humic substances. Environ. Sci. Technol. 2004, 38, 3885–3891. (21) Haag, W. R.; Hoigne, J.; Gassman, E.; Braun, A. M. Singlet oxygen in surface waters - part I: furfuryl alcohol as a trapping agent. Chemosphere 1984, 13, 631–640. (22) Average of values compiled at the Notre Dame Radiation Laboratory, Radiation Chemistry Data Center. http://www.rcdc. nd.edu/compilations/SingOx/table1/SOX_1140.HTM (accessed 2/17/10). (23) Bolton, J. R.; Linden, K. G. Standardization of methods for fluence (UV dose) determination in bench-scale UV experiments. ASCE J. Env. Eng. 2003, 129, 209–215.
(24) Frimmel, F. H.; Bauer, H.; Putzlen, J.; Murasecco, P.; Braun, A. M. Laser flash photolysis of dissolved aquatic humic material and the sensitized production of singlet oxygen. Environ. Sci. Technol. 1987, 21, 541–545. (25) Helms, J. R.; Stubbins, A.; Ritchie, J. D.; Minor, E. C.; Kieber, D. J.; Mopper, K. Absorption spectral slopes and slope ratios as indicators of molecular weight, source, and photobleaching of chromophoric dissolved organic matter. Limnol. Oceanogr. 2008, 53, 955–969. (26) Larsen, R. J.; Marx, M. L. An Introduction to Mathematical Statistics and its Applications, 2nd ed; Prentice Hall: NJ, 1986; Chapter 10. (27) O’Sullivan, D. W.; Neale, P. J.; Coffin, R. B.; Boyd, T. J.; Osburn, C. L. Photochemical production of hydrogen peroxide and methylhydroperoxide in coastal waters. Mar. Chem. 2005, 97, 14–33. (28) Andrews, S. S.; Caron, S.; Zafiriou, O. C. Photochemical oxygen consumption in marine waters: A major sink for colored dissolved organic matter. Limnol. Oceanogr. 2000, 45, 267–277. (29) Cooper, W. J.; Zika, R. G.; Petasne, R. G.; Plance, J. M. C. Photochemical formation of H2O2 in natural waters exposed to sunlight. Environ. Sci. Technol. 1988, 22, 1156–1160. (30) Kwan, W. P.; Voelker, B. M. Decomposition of hydrogen peroxide and organic compounds in the presence of dissolved iron and ferrihydrite. Environ. Sci. Technol. 2002, 36, 1467–1476. (31) Voelker, B. M.; Sedlak, D. L.; Zafiriou, O. C. Chemistry of superoxide radical in seawater: reactions with organic Cu complexes. Environ. Sci. Technol. 2000, 34, 1036–1042. (32) Fukuzumi, S.; Fujita, S.; Suenobu, T.; Yamada, H.; Imahori, H.; Araki, Y.; Ito, O. Electron transfer properties of singlet oxygen and promoting effects of scandium ion. J. Phys. Chem. A 2002, 106, 1241–1247. (33) Harbour, J. R.; Issler, S. L. Involvement of the azide radical in the quenching of singlet oxygen by azide anion in water. J. Am. Chem. Soc. 1982, 104, 903–905. (34) Latch, D. E.; McNeill, K. Microheterogeneity of singlet oxygen distributions in irradiated humic acid solutions. Science 2006, 311, 1743–1747. (35) Turro, N. J. Singlet oxygen and chemiluminescent organic reactions. In Modern Molecular Photochemistry; University Science Books: Sausalito, CA, 1991. (36) Cory, R. M.; McNeill, K.; Cotner, J. P.; Amado, A.; Purcell, J. M.; Marshall, A. G. Singlet oxygen in the coupled photochemical and biochemical oxidation of dissolved organic matter. Environ. Sci. Technol. 2010, 44, 3683–3689. (37) Riley, D. P.; Rivers, W. J.; Weiss, R. H. Stopped-flow kinetic analysis for monitoring superoxide decay in aqueous systems. Anal. Biochem. 1991, 196, 344–349. (38) Zepp, R. G.; Schlotzhauer, P. F.; Sink, R. M. Photosensitized transformations involving electronic energy transfer in natural waters: role of humic substances. Environ. Sci. Technol. 1985, 19, 74–81. (39) Blough, N. V. Electron paramagnetic resonance measurements of photochemical radical production in humic substances. 1. Effects of O2 and charge on radical scavenging by nitroxides. Environ. Sci. Technol. 1988, 22, 77–82. (40) Thomas-Smith, T. E.; Blough, N. V. Photoproduction of hydrated electron from constituents of natural waters. Environ. Sci. Technol. 2001, 35, 2721–2726.
ES101005U
VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5829
