Environ. Sci. Technol. 2003, 37, 4403-4409
Role of Dissolved Organic Matter Composition on the Photoreduction of Cr(VI) to Cr(III) in the Presence of Iron M E G A N G A B E R E L L , † Y U - P I N G C H I N , * ,† STEPHAN J. HUG,‡ AND BARBARA SULZBERGER‡ Department of Geological Sciences, The Ohio State University, Columbus, Ohio 43210, and Swiss Federal Institute for Environmental Science and Technology (EAWAG), Duebendorf, Switzerland CH8600
The photochemical reduction of Cr(VI) by iron and aquatic dissolved organic matter (DOM) was investigated. DOM sampled from a number of surface waters (a eutrophic wetland, a blackwater stream, and river water from a mixuse watershed) was used in this study. Moreover, a fulvic acid from Lake Fryxell, Antarctica, was also used to represent a DOM derived from a strictly autochthonous source. Cr(VI) reduction to Cr(III) at pH 5.5 was observed for all target DOMs used in this study, but rates varied widely. In general, photoreduction rates increased with increasing iron concentrations, but the type of DOM appeared to influence the kinetics to a larger degree. The rate of reduction was significantly greater for DOM derived from terrestrial systems than from predominantly autochthonous materials even if additional iron was added to the later. A positive correlation was observed between rates of Cr(VI) photoreduction and properties of the isolated DOM samples whereby faster reduction was observed for larger more aromatic substrates. On the basis of the fast rates reported for the dark reduction of Cr(VI) to Cr(III) by Fe(II)organic ligands, we hypothesize that the rate-limiting step in these reactions is the photoreduction of Fe(III) to Fe(II) by a ligand-to-metal charge-transfer pathway after absorption of light by Fe(III)-DOM complexes or by reduction of Fe(III) by superoxide or other intermediates formed after light absorption by DOM. Thus, the rate of Cr(VI) photoreduction to Cr(III) in natural sunlit waters is dependent upon both the amount of iron present and the nature of the dissolved organic matter substrate.
Introduction Hexavalent chromium, Cr(VI), is a known carcinogen and mutagen and can be acutely toxic. Cr(VI) is the thermodynamically stable oxidation state of chromium under oxidized conditions (EH° value for the Cr(VI)/Cr(III) couple is approximately +1.2 V) and is present predominantly in the anion forms HCrO4- and CrO42- in the pH range 3-10 (1). Consequently, Cr(VI) is mobile in both aqueous and sub* Corresponding author phone: (614)292-6953; fax: (614)292-7688; e-mail:
[email protected]. † The Ohio State University. ‡ Swiss Federal Institute for Environmental Science and Technology. 10.1021/es034261v CCC: $25.00 Published on Web 09/05/2003
2003 American Chemical Society
surface environments. In contrast to Cr(VI), trivalent chromium Cr(III) is significantly less toxic and is considered a micronutrient for humans and animals (2, 3). In addition to its lower toxicity, Cr(III) is highly “particle reactive” and much less mobile in the environment when compared to hexavalent chromium (4). While Cr(VI) will predominate under oxidizing conditions in many surface waters, Cr(III) has been detected as a kinetically stable species (5-7). Two principally different mechanisms have been postulated for the existence of Cr(III) in natural waters: (i) reduction of Cr(VI) to Cr(III) by Fe(II) (existing under anoxic conditions or formed through light-induced reactions) (5, 8) and (ii) photolysis of Cr(VI)organic matter complexes (6). To date, little work has been done investigating the role of dissolved natural organic matter (DOM) on the photochemical reduction of Cr(VI) in natural waters. Past research has demonstrated that Cr(VI/III) cycling occurs in natural waters (5, 9). Kieber and Helz (5) observed an increase in Cr(III) concentrations from approximately 10 to 40 nM in river water over the course of a sunlit day. They surmised that Cr(VI) was being reduced by Fe(II) produced by the photoinduced dissolution of ferric iron oxyhydroxide particles in the river water with natural organic matter as the electron donor. Their control experiments demonstrated no observable Cr(VI) reduction in river water filtered to remove particles, leading them to conclude that particles are essential to Cr(VI) photoreduction. However, the short 1-h irradiation time used by the investigators may not have allowed for any measurable Cr(VI) reduction by DOM to occur. Nonetheless, the identification of DOM as an electron donor in the photochemical reduction of Cr(VI) by Fe(II) in natural waters is important. In a related study, Cr(VI) was observed to complex with DOM in natural waters, and irradiation of the water resulted in the release of Cr(III). Thus, the photodegradation of Cr(VI)-organic complexes may be an alternate pathway for the presence of Cr(III) in natural waters (6). Hug and co-workers (8) investigated the role of organic ligands on the Fe(III)-catalyzed photochemical reduction of Cr(VI) using oxalate and citrate as DOM analogues. They found the rate of Cr(VI) photoreduction in the presence of these two organic ligands and Fe(III) to be extremely fast, with over 95% Cr(VI) reduction occurring in 20-40 min. A kinetic model was postulated where Fe(II), HO2, and O2•were listed as the most likely reductants of Cr(VI). Precipitation of chromium(III) hydroxides was not observed, but the formation of a complex between the photoproduced Cr(III) and an oxalate ligand was confirmed. The importance of this work can be translated to natural aquatic systems, where the existence of low molecular weight organic acids in the DOM pool may be active participants in Cr(VI) photoreduction. However, there is little information available on the ironcatalyzed photoreduction of Cr(VI) using DOM. The photoreduction of Cr(VI) to Cr(III) has been demonstrated for simple Fe(III)-organic acid complexes. Given that DOM is an important component of environmental systems, we wish to investigate its role in the photoreduction of Cr(VI) in the presence of iron. The specific objectives of this study are to (i) isolate and characterize DOM from autochthonous and allocthonous sources, (ii) conduct Cr(VI) photoreduction experiments using the DOM isolates at various FeT concentrations in synthetic light at wavelengths present in sunlight, and (iii) correlate Cr(VI) photoreduction reaction kinetics to DOM composition. VOL. 37, NO. 19, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4403
Materials and Methods DOM Collection, Isolation, and Characterization. Natural water samples used in this research were collected from four areas: MacDonald’s Branch in the New Jersey Pine Barrens; the Hammonassett River in Connecticut; Old Woman Creek adjacent to Lake Erie in Ohio; and Lake Fryxell, a microbialdominated lake in Antarctica. The first two sites have predominantly terrestrial sources of DOM, while the latter two sites are dominated by autochthonous DOM (10-12). Water samples from the New Jersey Pine Barrens and Old Woman Creek were filtered through 0.45- and 0.22-µm membrane filters. The Hammonassett River DOM was reconstituted from lyophilized materials isolated using the tangential-flow ultrafiltration (TFUF) method of Everett and co-workers (13). TFUF was also used to isolate DOM from the other two sites for 13C NMR analysis (see below). Lake Fryxell fulvic acid fractions (isolated by XAD-8 chromatography) were provided by Dr. George Aiken at the USGS in Boulder, CO. All isolates were treated with a proton ionexchange resin (Bio-Rad Dowex AG 50W-X8) to remove trace metals. Dissolved organic carbon (DOC) for both raw water and reconstituted DOM isolates were quantified using a Shimadzu 5000 carbon analyzer and potassium hydrogen pthalate standards. DOM isolates were characterized by solid-state cross polarization magic angle spin (CPMAS) 13C NMR using a Bruker DSX 300 mHz instrument measured at a frequency of 75.5 MHz. The samples were placed in a 5-mm rotor and spun at 13 kHz at a magic angle of 54.7°, contact time of 2 ms, and a recycle delay of 1s. A Cary 1 UV-vis spectrophotometer was used to measure the absorbance of the DOM solutions. Scans were taken between 200 and 600 nm in a 1-cm cuvette and the molar absorptivity at 280 nm determined in conjunction with the DOC data (11). High-pressure size exclusion chromatography (HPSEC) was used to measure the weight-average molecular weight of the natural organic matter before and after irradiation according to the method described in Chin et al. (11). Briefly, the system was comprised of a Waters 717plus Autosampler, a Waters 486 tunable absorbance detector (λ ) 230 nm), a Waters 510 HPLC pump, and a Waters Protein-Pak 125 (7.8 × 300 mm) 10-µm modified silica column (Waters Corp.). Calibration of the HPSEC system was done using aqueous soluble polystyrene sulfonate (PSS) standards (Polysciences, Inc.) in 18 000, 8000, 5400, and 1800 Da molecular masses. Acetone was chosen to determine the total volume of the column and was used as the lowest molecular mass standard (58 Da). Details regarding column and standard selection and mobile-phase composition can be found elsewhere (11, 14). Photochemistry Experimental Protocol. Old Woman Creek and Pine Barrens natural water samples were processed as described above and adjusted to pH 5.5 ( 0.1. DOM isolates from the Hammonassett River and Lake Fryxell were first dissolved in water (pH 11) and then added to an aqueous solution of 1 mM MES buffer (pH 5.4-5.8). A spike from a stock solution of K2CrO4 (2 mM) was added to reach a concentration range of 16.6-24.1 µM. Experiments were conducted at high and low iron concentrations. For experiments involving low concentrations of iron, a spike of FeCl3‚ 6H2O (0.6-0.7 mM) was added to reach a concentration range of 0.5-0.7 µM. For experiments involving high concentrations of iron, a spike of FeCl3‚6H2O (1.9-2.7 mM) was added to reach a concentration range of 9.5-23.5 µM. The iron concentration for both Pine Barrens and Hammonassett River DOM (both raw water and isolates) that were either slightly adjusted or not adjusted after ICP analysis indicated elevated levels of natural iron (9.5-16 µM). This was true even after exhaustive cleaning using a proton ion-exchange resin as 4404
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 19, 2003
stated previously. Control experiments consisted of 16.518.5 µM Cr(VI) and 8.4-12.4 µM Fe(III) in MilliQ water or 1 mM MES buffer. All solutions were made in duplicate. Samples were placed in precleaned borosilicate culture tubes (Corning) and irradiated for 6-8 h in artificial light. Artificial-light experiments were performed using a merrygo-round reactor outfitted with a medium-pressure mercury arc lamp housed inside a borosilicate immersion well (Ace Glass, Vineland, NJ). Borosilicate filters out λ e 290 nm. Filters were placed around the well to isolate λ ) 366 nm, a wavelength in natural sunlight found to be an important contributor to photochemical reactions (15). Cold water (10 °C) was circulated through the immersion well to dissipate heat from the lamp. A fan was also placed above the reactor unit, keeping temperatures at roughly 27 ( 3 °C. Dark control samples were wrapped in aluminum foil and stored at room temperature, which was similar to the temperature range in the reactor. After irradiation, bichromate in the samples was converted to CrO42- by raising the pH to 11. The samples were then scanned from 500 to 250 nm at a scan rate of 90-150 nm/ min using a Cary 1 UV-vis spectrophotometer. Dark samples were handled similarly. Controls indicated that Cr(VI) reduction did not occur significantly at pH 11 during the course of analysis. Hexavalent chromium concentrations were calculated according to the method developed by Hug et al. (8). The UV-vis absorbance scan for each sample point was manipulated to remove the effect of iron addition by spectrally subtracting the control/blank scan taken of the DOM and iron solutions before any chromium addition. The spectral subtraction was done using Galactic software GRAMS/32 Version 5.22. After the subtraction, the Cr(VI) concentration was calculated by taking the absorbance value at 372 nm and dividing by the extinction coefficient for chromium at that wavelength (i.e., 4926 M-1 cm-1). The 372-nm wavelength was chosen from the absorbance scans as the peak maximum for chromate. Actinometry was conducted concurrently with Cr(VI) photoreduction experiments in order to normalize the results to differences in light intensity. The p-nitroanisole (PNA)/ pyridine actinometer was used because the kinetics of this reaction can be varied to last for the duration of the experiment by adjusting the concentration of pyridine (16). Within 1 h prior to the experiment, a reaction mixture consisting of 1.0 × 10-5 M PNA and 1.0-5.0 × 10-3 M pyridine was made. Samples of this reaction mixture were irradiated concurrently with the natural water samples. The PNA concentrations of the standards and samples were measured using reverse-phase HPLC (Waters Corp.). The observed quantum yield of photolysis of the actinometer (φobs) at 366 nm can be calculated using the following equation (16):
φobs ) 0.437[pyridine] + 0.00028
(1)
Correction Factors for Variability in Lamp Intensity and DOC. To account for daily differences in lamp intensity, a light intensity correction factor for each experiment was necessary. First, the observed actinometer rate constant (kobs) was divided by the actinometer quantum yield (Φobs) to account for differences in pyridine concentration. Next, an average value (kobs/Φobs)ave was determined for all the experiments. The light intensity correction factor was calculated according to the following equation:
light intensity correction factor )
(kobs/Φobs)av (kobs/Φobs)
) Ioav/Io (2)
FIGURE 1. Change in the DOM and Fe-corrected UV-vis absorbance spectra for observed Cr(VI) photoreduction as a function of time. Spectral subtraction was accomplished using GRAMS/32. Not all time points are shown for clarity.
TABLE 1. Apparent Rate Coefficients for Cr(VI) Photoreduction for Each Type of Dissolved Natural Organic Mattera
a
sample
exptl iron concn (µM)
exptl DOC concn (mg of C/L)
DOC-normalized rate coeff for Cr(VI) photoreduction min-1 (mg of C/L)-1
Pine Barrens Hammonassett River Old Woman Creek Lake Fryxell Old Woman Creek Lake Fryxell
high (23.5) high (9.5) high (15.5) high (16.6) low (0.5) low (0.7)
24.5 ( 0.5 5.2 6.2 ( 0.1 10.7 5.9 ( 0.1 13.5
2.8 × 10-4 ( 3.4 × 10-5 1.8 × 10-4 ( 3.4 × 10-5 6.7 × 10-5 ( 1.3 × 10-5 6.1 × 10-5 ( 1.1 × 10-5 2.7 × 10-5 ( 8.7 × 10-6 3.0 × 10-5 ( 5.1 × 10-6
The units reflect normalization to the carbon content of each sample and as such are not truly second order.
where Ioav is the average incident light intensity and Io is the light intensity for a given experiment. The reaction rate coefficient for each Cr(VI) photoreduction experiment was then multiplied by the appropriate light intensity correction factor. To account for differences in DOC concentrations for each experiment, the reaction rate coefficient was divided by the DOC for a specific sample. Errors associated with actinometry and DOC measurements were propagated with the error of the photoreduction experiment itself. The resulting units of the “apparent” reaction rate coefficient are min-1 (mg C/L)-1, but it should be noted that these units result from the normalization of the pseudofirst-order rate coefficient to DOC concentration and are not true second-order rate coefficients.
Results and Discussion Photoreduction of Cr(VI). The mercury arc vapor pressure lamp used in the merry-go-round reactor produced an average light intensity of 2.38 × 10-4 ( 1.69 × 10-5 einstein s-1 L-1 (n ) 15) in the reaction vials based upon our actinometry results. Hexavalent chromium photoreduction was observed for all the DOM samples in the presence of iron over the course of the 6-8-h experiments (Figures 1 and 2). UV-vis spectroscopy was used to measure the decrease in CrO42- absorbance over time for each DOM (Figure 1). Analysis of the dark control samples indicated that Cr(VI) reduction was not occurring on the 8-h time scale of these experiments. Lack of a discernible dark reaction on this time scale is supported by the literature (5, 6, 8). We observed Cr(VI) photoreduction in the MilliQ and MES buffer control spiked with iron; however, the pseudo-firstorder rate coefficients (not normalized to DOC) were smaller than in the samples with DOM. The one exception to this observation was the OWC “low iron” sample, which possessed
FIGURE 2. Kinetic plot of observed Cr(VI) photoreduction in the presence of iron and Hammonassett River DOM: irradiated samples (b) and dark controls (2). iron levels approximately a factor of 20 less than the controls (Figure 3). This observation demonstrates that Cr(VI) photoreduction by iron is enhanced by the presence of the DOM used in this study in the presence of iron. Light-normalized pseudo-first-order rate coefficients of Cr(VI) photoreduction varied for the different DOM types and at different iron concentrations (Figure 3). These differences are more noticeable when the rate coefficients are also normalized to DOC concentration (Table 1). The Pine Barrens and Hammonassett River DOM samples were the most efficient in photoreducing Cr(VI). Photoreduction of Cr(VI) by Lake Fryxell and OWC DOM were much slower and were almost indistinguishable from one another for a VOL. 37, NO. 19, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4405
FIGURE 3. Observed rate coefficients for Cr(VI) photoreduction in the presence of iron. FeT in the samples ranged from 0.5-0.7 µM (low) to 9.5-23.5 µM (high). CrT ) 16.6-24.1 µM. These pseudo-first-order rate coefficients have been adjusted for differences in lamp intensity but have not been normalized to DOC concentrations.
TABLE 2. DOM Aromaticity and Aliphatic Dataa
DOM
total aromaticity (%)
weightaverage mol wt
molar absorptivity at 280 nm ((mol of C/L)-1 cm-1)
Pine Barrens Hammonassett River OWC Lake Fryxella
24.0 25.8 19.0 15.9
1550 ( 130 1510 ( 30 1030 ( 10 1000 ( 10
506 ( 7 480 ( 7 247 ( 12 172 ( 5
a Total aromaticity for Lake Fryxell is based on 13C NMR data from the literature (31).
specific iron concentration. In experiments conducted at higher iron concentrations (10 µM or higher), Cr(VI) photoreduction was more rapid than at the lower iron concentrations (
