Environ. Sci. Technol. 2004, 38, 4113-4119
Photooxidation and Its Effects on the Carboxyl Content of Dissolved Organic Matter in Two Coastal Rivers in the Southeastern United States H U I X I A N G X I E , * ,†,§ OLIVER C. ZAFIRIOU,‡ WEI-JUN CAI,| RICHARD G. ZEPP,§ AND YONGCHEN WANG| Institut des Sciences de la Mer de Rimouski, Universite´ du Que´bec a` Rimouski, Que´bec, Canada G5L 3A1, Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, Department of Marine Sciences, University of Georgia, Athens, Georgia 30602, and Ecosystems Research Division, National Exposure Research Laboratory, The United States Environmental Protection Agency, Athens, Georgia 30605
Photodecarboxylation (often stoichiometrically expressed as RCOOH + 1/2O2 f ROH + CO2) has long been postulated to be principally responsible for generating CO2 from photooxidation of dissolved organic matter (DOM). In this study, the quantitative relationships were investigated among O2 consumption, CO2 production, and variation of carboxyl content resulting from photooxidation of DOM in natural water samples obtained from the freshwater reaches of the Satilla River and Altamaha River in the southeastern United States. In terms of loss of dissolved organic carbon (DOC), loss of optical absorbance, and production of CO2, the rate of photooxidation of DOM was increased in the presence of Fe redox chemistry and with increasing O2 content. The ratio of photochemical O2 consumption to CO2 photoproduction ranged from ∼0.8 to 2.5, depending on the O2 content, the extent of involvement of Fe, and probably the initial oxidation state of DOM as well. The absolute concentration of carboxyl groups ([-COOH]) on DOM only slightly decreased or increased over the course of irradiation, possibly depending on the stages of photooxidation, while the DOC-normalized carboxyl content substantially increased in the presence of Fe redox chemistry and sufficient O2. Both the initial [-COOH] and the apparent loss of this quantity over the course of irradiation was too small to account for the much larger production of CO2, suggesting that carboxyl groups were photochemically regenerated or that the major production pathway for CO2 did not involve photodecarboxylation. The results from this study can be chemically rationalized by a reaction scheme of (a) photodecarboxylation/ regeneration of carboxyl: CxHyOz(COOH)m + aO2 + (metals, hv) f bCO2 + cH2O2 + Cx-bHy′Oz ′(COOH)m-b(COOH)b or * Corresponding author email:
[email protected]; phone: (418)724-1767; fax: (418)724-1842. † Universite ´ du Que´bec a` Rimouski. ‡ U.S. Environmental Protection Agency. § Woods Hole Oceanographic Institution. | The University of Georgia. 10.1021/es035407t CCC: $27.50 Published on Web 06/24/2004
2004 American Chemical Society
of (b) nondecarboxylation photooxidation: CxHyOz(COOH)m + aO2 + (metals, hv) f bCO2 + cH2O2 + Cx-bHy′Oz ′(COOH)m.
Introduction The absorption of solar radiation, particularly in the ultraviolet (UV) region, by dissolved organic matter (DOM) in natural waters initiates an array of physical and chemical processes, which profoundly influence the chemical, physical, and biological characteristics of aquatic ecosystems. DOMinvolved photochemical processes produce a suite of important transient reactive species (e.g., singlet oxygen, hydroxyl radicals, superoxide radicals) (1-3), relatively longlived small molecules (e.g., carbon monoxide, hydrogen peroxide) (4-6), and the bulk of photochemically altered DOM. Photoprocessing of DOM also plays an important role in the hydrospheric carbon cycle through remineralization of dissolved organic carbon (DOC) to CO2 (7) and by production of low molecular weight (LMW) organic compounds, which are generally more labile to microbial processes than is the parent DOM (8). In addition, photodegradation of DOM results in loss of its absorbance (photobleaching), mainly in the UV and visible regions, allowing sunlight to penetrate into deeper depths. Although it is well-recognized that in aqueous solutions some simple carboxyl-containing organic compounds can undergo photodecarboxylation, which produces CO2, consumes O2, and can be catalyzed by metals (e.g., iron, copper) (9-12), little is known of the mechanisms for the CO2 production from DOM photochemistry. Miles and Brezonik (13) speculated that decarboxylation might also occur during photooxidation of DOM and further proposed that the overall reaction can be expressed as RCOOH + 1/2O2 f ROH + CO2. This equation implies a massive loss of carboxyl groups if DOM is extensively photooxidized. The loss of carboxyl groups may have important environmental and ecological impacts since the carboxyl group is the major contributor to the aqueous solubility and acidity of dissolved organic matter (DOM), thus controlling the buffering capacity of high-DOM freshwater systems (14). Moreover, the concentration and acidity of carboxyl groups also profoundly influence DOM’s physicochemical properties, such as aggregation and disaggregation, hydrophobic binding, and trace metal binding (15). In this study, we investigated the kinetics of O2 consumption, CO2 production, DOC loss, photobleaching, and iron redox chemistry during the photooxidation of DOM from two coastal rivers in the southeastern United States. More importantly, we assessed the effects of aquatic photochemistry on the carboxyl content of DOM, examined the roles of iron and oxygen in this process, and discussed the possible processes responsible for photochemical CO2 production.
Experimental Procedures Sample Collection and Treatment. Freshwater samples were taken in March 2001 from two coastal rivers in Georgia: the blackwater Satilla River and the whitewater Altamaha River. The Satilla River originates in the coastal plain of Georgia and is characterized by a high level of DOM (DOC: ∼20 mg L-1) and low pH (14, 16); the origin of its DOM is considered to be mainly allochthonous. The Altamaha River originates in the Piedmont region and has a relatively low level of DOM (DOC: ∼6 mg L-1) and high pH (14); DOM in this river is thought to be derived from both allochthonous and auVOL. 38, NO. 15, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4113
tochthonous sources (17). The samples were sequentially passed through Whatman GF/F glass fiber filters and 0.22 µm Millipore polycarbonate filters upon collection. To improve the sensitivity of determining the carboxyl content, the Altamaha River water (ARW) was concentrated by a factor of 3 under vacuum-evaporation (60 °C) using a SC210A Savant SpeedVac Concentration System. Part (∼10 L) of the filtered Satilla River water (SRW) or of the concentrated ARW was treated with 25 µM deferoxamine mesylate (DFOM) (Novartis). DFOM is a strong Fe(III)-complexing ligand that forms nearly photo-inert complexes (18). Two weeks in the dark at 4 °C allowed the completion of the DFOM-Fe(III) complexing process. Equilibrium calculations indicate that 25 µM DFOM was sufficient to complex all the Fe in SRW and ARW. All samples were kept at 4 °C in the dark until being irradiated. Immediately before irradiation, the samples were refiltered (0.22 µm) and purged with air (medical grade), pure O2, or pure N2 to obtain three widely different initial O2 levels: air-, O2-, or N2-saturated. For simplicity, the initially air-, O2-, and N2-saturated samples with no DFOM added will be designated as air-sat, O2-sat, and N2-sat samples, respectively, and the initially air- and O2-saturated samples with DFOM added as air-sat & DFOM and O2-sat & DFOM samples, respectively. Irradiation. Samples were irradiated using an Oriel 1000W xenon arc lamp source whose light beam was turned upward with a Beam Turning Assembly (Oriel). The irradiation vessel, a cylindrical quartz cell measuring 8.5 cm in length and 3.95 cm in i.d., was placed vertically in the light beam. The cell had two forked ports near the upper window for sampling. The vertical orientation facilitated sample mixing by creating heat-induced convection within the cell. Prior to reaching the cell window, the IR and short UV radiation were reduced by pure water and a Schott long band-pass cutoff filter (50%T at 320 nm), respectively. The spectrum of the incident light after the IR and short UV filtration was similar to natural sunlight, with the light intensity at 320 nm being ∼11 times higher than that of the clear-sky natural sunlight measured at midafternoon in midspring in Athens, GA. The number of photons (320 nm) absorbed per unit time by the sample with the cell vertically oriented in the artificial light are estimated to be 3.5 times the number of photons absorbed as if the cell were placed horizontally in the natural sunlight. Analyses. Measurements were made of the carboxyl content, total CO2, dissolved organic carbon (DOC), dissolved O2, UV-visible absorption spectra, pH, and iron. CO2 was stripped out with pure N2 from acidified samples and quantified using an infrared CO2 analyzer (Li-Cor 6252) (19). Analysis of DOC was performed with a Shimadzu TOC-5000 carbon analyzer calibrated with potassium biphthalate. O2 was measured with a WTW Oxi340 DO meter. An Accumet pH meter (model 20) fitted with a Ross Orion combination electrode was used to determine pH; the system was standardized with three NIST buffers at pH 4, 7, and 10. The absorption spectra, referenced to pure water (Nanopure), were determined using an HP8453 spectrophotometer fitted with a 1 cm quartz cell. Absorption coefficients were calculated as 2.303 times the absorbance, ratioed to the light path-length of the cell in meters. Previous studies (18, 20) showed that formation of particles could occur through coagulation of DOM under extensive irradiation. For the present study, visible brown particles formed only in the O2-sat SRW irradiated for 46 h. To minimize the effect of particle formation, this sample was filtered with Millipore polycarbonate membrane filters (0.22 µm pore size) prior to the determinations of absorbance, DOC, and carboxyl content. Total iron and [Fe(II)] were measured with the ferrozine method (21). The concentration of carboxyl groups ([-COOH]) was quantified using potentiometric titration (14). 4114
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 15, 2004
Briefly, samples were titrated with HCl to pH ∼2.7, bubbled with ultra-pure N2 for 30 min, and finally NaOH-titrated to pH ∼10.5 in small increments under a N2 atmosphere. The HCl remaining in the solution after the forward (HCl) titration (i.e., the excess strong acid) was obtained as the difference between the total HCl added and the total alkalinity of the sample determined with the Gran protocol (22). The endpoint for the carboxylic titration was determined by a doubledifferential plot (15). The [-COOH] was derived by subtracting the excess strong acid from the amount of NaOH added to the endpoint. Taking the initial (t ) 0) values in Table 1, we estimated the analytical uncertainty (relative standard deviation) to be ∼3% for O2 and ∼2% for DOC and absorbance (at 320 nm). The analytical precision for the measurement of CO2 was shown to be 1% by analyzing five replicates with a mean concentration of 12.2 µM. The analytical precision for the determination of the carboxyl concentration was evaluated by titrating replicate (unirradiated) samples, arriving at 102 ( 12 µM (n ) 16) for SRW and 112 ( 13 µM (n ) 8) for ARW or 12% for both of them. The accuracy of the titration was assessed by determining the recovery rates of a synthetic reference prepared from a combination of various model organic acids (23). The mean recovery rate was 93% (standard deviation: 0.6%, n ) 3).
Results and Discussion Iron Kinetics, O2 Consumption, DOC Loss, and pH Change. Total iron was 13.6 µM in SRW and 8.3 µM in the concentrated ARW. The kinetics of Fe(II) (Fe2+ and its complexes) production during iron-catalyzed photooxidation of DOM (13, 18, 24, 25) was monitored only for the SRW samples (Figure 1). The [Fe(II)] was generally undetectable in the DFOM-added samples but reached a steady-state concentration of 1.2 µM after 1 h irradiation in the air-sat samples. Hence, the interaction of DOM with iron was nearly completely eliminated by DFOM. For the samples without DFOM, the [Fe(II)] in the O2-sat samples was slightly but consistently lower than that in the air-sat samples, probably due to faster oxidation of Fe(II) by H2O2 in the O2-sat samples (unfortunately, experimental constraints prohibited simultaneous measurements of H2O2). Exposure of SRW and ARW to the solar-simulated light led to an approximately exponential decrease in dissolved oxygen in all treatments (Tables 1 and 2). The exponentially decreasing trend, however, was followed by an abrupt leveloff in the air-sat SRW after 24 h irradiation due to the almost total depletion of O2. Addition of DFOM roughly halved the rate of O2 consumption in both SRW and ARW. The average rate of O2 consumption during the first 4 h of exposure was nearly 2.5 times higher in the O2-sat samples than in the air-sat samples, although the initial O2 concentration was about 5 times higher in the O2-sat samples. These results indicated that photochemical O2 consumption strongly depended on, but was less than proportional to, the O2 content in the solution. Irradiation of DOM resulted in a notable loss of DOC in the presence of O2: 17-46% for SRW and 25-33% for ARW, depending on treatment (Table 1). The DOC loss was, however, extremely small in the N2-sat samples (∼2% after 46 h irradiation), suggesting a pivotal role of O2 in DOM photooxidation. DOC loss was more or less linear with respect to the irradiation time except in the air-sat SRW samples in which loss of DOC after 24 h exposure dramatically slowed, apparently as a result of O2 depletion. In contrast to O2 consumption, the difference in the loss of DOC was not substantial between the air- and O2-sat samples until the development of O2 deficiency in the air-sat samples near the end of the irradiation (Table 1). Addition of DFOM reduced,
TABLE 1. Time Series of Variables during Irradiationa irradiation time (h) O2 (µM)
DOC (µM)
a320 (m-1)
CO2 (µM)
-COOH (µM)
O2 (µM)
DOC (µM)
a320 (m-1)
CO2 (µM)
-COOH (µM)
a
0
2
4
8
Satilla River water (SRW) 240.6 210.6 154.7 1062.5 1012.5 937.5 220.6 1078.1 1519 1502 1471 1570 1535 1451 1519 1515
air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM N2-sat air-sat O2-sat air-sat & DFOM O2-sat & DFOM N2-sat air-sat O2-sat air-sat & DFOM O2-sat & DFOM N2-sat air-sat O2-sat air-sat & DFOM O2-sat & DFOM N2-sat
276.9 1187.5 284.1 1193.8 1570 1620 1588 1545 1574 81.5 81.9 78.7 78.3 82.5 10.5 9.2 13.8 10.1 10.5 109.1 104.4 91.7 94.0 103.6
air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM
Altamaha River water (ARW) 298.8 237.2 1225.0 1075.0 281.3 1215.6 1442 1376 1430 1404 1441 1406 54.9 44.1 54.4 41.8 54.7 55.1 19.1 71.7 14.0 87.3 15.0 11.0 111.5 98.0 113.0 145.8 127.0 98.4
13
18
24
46
100.6 862.5
58.8 790.6 176.3 990.6 1343 1271 1468 1437
37.8 709.4 141.3 915.6 1207 1158 1426 1432 1538 40.6 26.5 47.5 40.5 76.4 334.8 449.5 150.3 144.0 56.4 128.9 105.7 88.5 109.0
37.2 477.5 63.4 806.3 1131 875 1323 1279 1548 39.6 2.7 38.0 30.4 72.7 394.7 739.2 254.4 260.2 87.0 114.6 144.4 94.1 70.2 88.0
94.7 862.5
42.2 746.9
1391 1376
74.5 73.1
69.8 67.3
61.4 56.6 60.6 58.3
52.5 46.3
44.7 36.2 52.0 48.5
56.4 62.1
98.5 97.9
167.9 181.9 61.4 67.3
233.9 262.9
296.7 350.7 109.0 116.6
97.5 104.4
101.0 56.5
114.5 64.5 105.9 59.3
142.3 101.0
167.4 123.7 99.3 104.4
174.4 987.5 179.4 1025.0 1296 1240
1180 1132
1082 962
1302 1300 35.3 31.2
29.1 22.7
25.0 13.7
252.0 290.8
340.3 455.5
78.2 98.4
89.0 66.5
32.6 29.1 166.1 178.8 112.7 112.2 127.0 119.2 117.4 142.9
Blanks indicate that variables were not determined.
FIGURE 1. Concentration of Fe(II) as a function of irradiation time under air-saturation, O2-saturation, and with and without addition of DFOM. but did not stop the loss of DOC, implying the existence of photodegradation processes without the involvement of iron. The initial pH was ∼7.4 for the air-sat samples and ∼7.7 for the O2- and N2-sat samples. For both SRW and ARW and for all the treatments, the pH decreased (faster initially) with
increasing irradiation time. The reduction of pH during a 46 h period of irradiation was higher in the O2-sat samples (2.3 unit in SRW and 1.9 unit in ARW) than in the air-sat samples (1.8 unit in SRW and 1.5 unit in ARW). Addition of DFOM reduced the reduction of pH from 2.3 to 1.9 unit in the O2-sat SRW and from 1.8 to 1.7 unit in the air-sat SRW. The N2-sat samples had the smallest decrease in pH (0.7 unit). The reduction of pH was due principally to the formation of CO2 and organic acids as previously reported (26, 27). Photobleaching. DOM photooxidation gave rise to a considerable decrease in the absorption coefficients throughout the UV and visible wavelengths with the maximum relative loss in the range of 320-370 nm. Photobleaching was faster with higher O2 content (almost no photobleaching in the N2-sat samples) and with Fe involved and was approximately first-order with respect to the irradiation time until the occurrence of O2 depletion and the formation of particles in the air-sat and O2-sat SRW, respectively (Tables 1 and 2). The spectral slope coefficient (S), defined as the slope of the linear regression of the semilog plot of absorption coefficient versus wavelength, breaks around 400 nm (not shown), with its value in the UV wavelengths (300-400 nm) VOL. 38, NO. 15, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4115
TABLE 2. Regression Analysis between Pairs of Variablesa x
y
t
O2
t
DOC
t
a320
t
CO2
∆DOC
∆CO2
t
O2
t
DOC
t
a320
t
CO2
∆DOC
∆CO2
treatment
eq
Satilla River water (SRW) y ) 289.6-0.085x y ) 1114.8-0.0187x y ) 294.9-0.0324x y ) 1161.5-0.0084x y ) -13.9x + 1566 y ) -16.3x + 1592 y ) -5.6x + 1573 y ) -5.8x + 1554 y ) 78.9-0.0296x y ) 81.5-0.0458x y ) 74.3-0.0142x y ) 74.8-0.0184x y ) 15.7x + 26.5 y ) 18.1x + 23.8 y ) 5.2x + 17.6 y ) 5.3x + 17.5 y ) 0.80x + 52.7 y ) 0.96x + 7.9 y ) 0.99x - 22.6 y ) 0.90x + 16.9
air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM air-sat O2-sat air-sat & DFOM O2-sat & DFOM
n
R2
7b 8 5 5 7b 8 5 5 7b 7b 5 5 6c 7b 5 5 7 7 4 4
0.996 0.990 0.992 0.971 0.978 0.989 0.986 0.985 0.988 0.998 0.932 0.95 0.987 0.997 0.997 0.996 0.921 0.999 1.000 0.977
5 5 5 5 5 5 4b 4b 4 4
0.994 0.944 0.953 0.970 0.879 0.977 0.990 0.989 0.993 0.982
Altamaha River water (ARW) air-sat O2-sat air-sat O2-sat air-sat O2-sat air-sat O2-sat air-sat O2-sat
y ) 288.9-0.0426x y ) 1150.6-0.0101x y ) -7.7x + 1409 y ) -10.4x + 1414 y ) 47.8-0.016x y ) 48.6-0.0288x y ) 9.6x + 28.6 y ) 11.1x + 28.4 y ) 0.89x + 3.0 y ) 0.84x + 31.7
a t: irradiation time; ∆DOC: loss of DOC; and ∆CO : production of 2 CO2. b The last data point (t ) 46 h) is excluded. c The last two data points (t ) 24, 46 h) are excluded.
being considerably higher than in the visible wavelengths (400-500 nm). The S value for the original ARW is 12% higher than that of the original SRW (0.0172 vs 0.0154) in the UV region but comparable (0.0136 vs 0.0130) in the visible region. During the irradiation, S in the UV range remained fairly stable (air-sat SRW and ARW) or increased (O2-sat SRW and ARW) in the presence of sufficient O2 but decreased when O2 deficiency developed in the air-sat SRW and ARW. S in the visible region steadily decreased with increasing irradiation time except at the end of exposure where an increase in S occurred in the air- and O2-sat SRW due supposedly to the formation of anoxia and brown particles, respectively, in these two treatments (Table 1 in Supporting Information). The effect of DOM photooxidation on S in the UV range observed in this study is different from that of Gao and Zepp (18), who reported a substantial decrease in S for DOM from the Satilla River. This difference could result from a number of factors, including the differences in the wavelength range used to derive S (Gao and Zepp (18): 290-450 nm), in the extent to which the photo-Fenton chemistry is involved (28), and possibly in the extent and duration of anoxia as well. Production of CO2 and Carboxyl Content. The linear regression analysis between CO2 production and DOC loss (Table 2) indicates that the former could account for more than 80-90% of the latter in air-saturated conditions, confirming that CO2 is the largest carbon-containing product of DOM photochemistry (7, 29). The production of CO2 displayed similar patterns in both the SRW and ARW; the production rate in descending order is O2-sat, air-sat, O2-sat & DFOM, and air-sat & DFOM. The CO2 production rate in the N2-sat SRW was only ca. 10% and 20% of those in the O2-sat and air-sat samples, respectively (Table 1). Closer examination of the data shows that the difference in CO2 production between the O2-sat and the air-sat treatments 4116
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 15, 2004
FIGURE 2. DOC-normalized carboxyl content of the Satilla River water as a function of irradiation time. was rather small until the O2 concentration in the air-sat water became very low ( 2.7 (i.e., upper limits: 137 µM for the SRW and 162 µM for the ARW) and that all these strong-acid groups had been photochemically degraded to CO2, the unaccounted photoproduction of CO2 at the end of the time series irradiation were still 65% in the air-sat SRW, 81% in the O2-sat SRW, 52% in the air-sat ARW, and 64% in the O2-sat ARW. Furthermore,
FIGURE 4. Molar ratio of photochemical oxygen consumption to photoproduction of CO2 as a function of irradiation time. The ratio is cumulative with respective to irradiation time. (A) Satilla River and (B) Altamaha River. the molar ratio of O2 consumption to CO2 production (RO2/CO2) observed in our study (Figure 4) is different and more variable than the value (∼0.5) obtained by Miles and Brezonik (13). Our main observations are (1) RO2/CO2 ranges from ∼0.8 up to ∼2.5, depending on sample treatment and irradiation time, (2) RO2/CO2 tends to become ∼0.8 after long irradiation of air-sat samples, (3) initially RO2/CO2 is higher, and (4) more O2 or less Fe chemistry makes it higher, and both together makes it even higher. Obviously, the Miles and Brezonik’s equation is incompletely characterizes the more complex behavior revealed by our data. Our results strongly suggest that, if photodecarboxylation was the primary mechanism involved, carboxyl groups must have been photochemically regenerated at a rate high enough to compensate for their destruction leading to CO2 formation. The lack of a substantial increase in the absolute [-COOH] in the irradiated DFOM-added samples implies that photoregeneration of carboxyl groups, like photodecarboxylation, should also be dependent on Fe chemistry. Here, we propose the following reaction scheme to chemically rationalize the relationships among O2 consumption, CO2 production, decarboxylation, and carboxyl regeneration: metals, hv
CxHyOz(COOH)m + aO2 98 bCO2 + cHOOH + Cx-bHy′Oz′ (COOH)m-b(COOH)b For clarity in expressing the chemical formula of DOM, all minor elements (e.g., N, S, P) are ignored, and all C, H, and O are lumped together except -COOH. Transition metals, such as Fe and Cu, can catalyze DOM photooxidation (3, 9, 13, 36). Hydrogen peroxide (HOOH) is included in the products list since this species is one of the major photoproducts of DOM and plays a key role in the Fenton and photo-Fenton reactions that regulate the redox chemistry of Fe and O2 consumption (25, 28, 37). This reaction scheme shows that b mol of -COOH are destroyed to produce b mol VOL. 38, NO. 15, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4117
of CO2, while an equivalent number of moles of new -COOH are regenerated (as pointed out previously, the majority of our data did not show significant change of [-COOH] over the course of irradiation). Photoproduction of carboxylic acids from scission of aromatic rings in lignin model compounds was reported in wood chemistry (38). Similar processes might also be responsible for part of the photoformation of carboxylic acids in natural waters (27). Although lignin photochemistry would be a poor model for SRW because of the very small contribution of lignin phenols to the overall composition of SRW DOM (39), the study in the wood chemistry does demonstrate the possibility of photoconversion of non-carboxyl carbon to carboxylic acids. Ample evidence indicates that photodecarboxylation of simple Fe(III)-polycarboxylate (e.g., oxalate) complexes involves ligand-to-metal charge transfer (LMCT) during which an electron is transferred from the polycarboxylate to Fe(III), resulting in the oxidation of the former and reduction of the latter (3, 9, 10). In view of the relatively high abundance of carboxyl groups on aquatic DOM, it has been suggested that Fe(III) may form complexes with DOM and that the LMCT process may also be involved in DOM photooxidation (13). Analogous to the mechanistic scheme proposed by Faust (24) for simple Fe(III)-polycarboxylate complexes, the photoinduced LMCT within Fe(III)-DOM complexes presumably produces Fe(II) and a carboxylate radical, RCOO•. The fate of RCOO• is determined by several competing processes: (a) back reaction to reform Fe(III)-DOM; (b) decarboxylation to form CO2 and a carbon-centered radical, R•; (c) reaction with O2 to produce the superoxide radical,O2- ; and (d) reduction of another Fe(III). The R• there derived can also react with O2 to produce O2- or with another Fe(III) to form Fe(II). The O2- formed from the reactions of RCOO• and R• with O2 can undergo dismutation to generate HOOH; the Fe(II) and HOOH there evolved lead to the Fenton reaction, which produces HO• and Fe(III), thus completing the Fe-catalyzed photooxidation of DOM. It is also likely that the RCOO• may be produced, although probably at lower rates, by LMCT involving metals other than Fe or through non-LMCT processes (9, 10). However, removal of the Fe redox chemistry (by addition of DFOM in our case) leads to relatively more RCOO• and R• to react with O2, thereby increasing the ratio of O2 consumption to CO2 production (RO2/CO2). The higher initial RO2/ CO2 observed in our study could result from two factors. First, the oxidation state of DOM might increase with irradiation time (40), requiring more O2 to oxidize it initially than afterward. As the degree of the transformation of DOM became very high (under O2-saturation), RO2/CO2 should approach that required to oxidize DOM to CO2. Second, HOOH (its formation consumes O2) builds up initially until a steady state sets in (28). The relatively high light intensity of our irradiation system might have enhanced the role of reactive oxygen species in alteration of DOM and RO2/CO2 as compared to photoreactions initiated by natural sunlight or artificial radiation with lower light intensity. Nevertheless, the values of RO2/CO2 from this study were comparable to those obtained from a previous preliminary study on the Satilla River using light intensity only 30% higher than the natural sunlight (41), suggesting that the effect of the light source we used on RO2/CO2 was probably unimportant. Microbial involvement in O2 consumption and CO2 production during irradiation was assumed to be insignificant, considering that all the samples were 0.22 µm refiltered immediately before irradiation and that bacteria, if there were any remaining, would be killed quickly by the intense UV radiation. In addition to the decarboxylation/carboxyl-regeneration reaction scheme as described previously, an alternative, but basically speculative, reaction scheme that can explain our observations is that the primary pathway responsible for CO2 4118
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 15, 2004
production does not entail photodecarboxylation. This scheme can be depicted as metals, hv
CxHyOz(COOH)m + a O2 98 bCO2 + cHOOH + Cx-bHy′Oz′ (COOH)m Va¨ha¨talo et al. (42) demonstrated that solar radiation can convert the carbon atoms in the aromatic rings of a synthetic lignin to CO2, thus ascertaining that carbon atoms photochemically transformed to CO2 are not necessarily from original carboxyl groups. However, it is unknown whether the aromatic carbon must be first oxidized to carboxyl groups that are subsequently photodecarboxylated to CO2 or whether the aromatic carbon is oxidized to CO2 via other processes. Despite the ambiguity remaining in the mechanisms for photoconversion of aromatic carbon to CO2, the study by Va¨ha¨talo et al. (42) does strongly suggest that the cleavage of aromatic ring structures may be a major event in DOM photomineralization in organic-rich waters. Indeed, the aromatic carbon content of our samples, as indicated by the DOC-normalized absorption coefficient at 270 nm (a270), steadily decreased with irradiation time, with the maximum abatement (83%) occurring in the O2-sat SRW. It is likely that the oxidative cleavage of the aromatic rings also contributed prolifically to the formation of CO2 observed in our study. As expected, the fragmentation of the high molecular weight aromatic ring structures caused a reduction of the average molecular size of DOM, as inferred by an increase in the ratio of the absorption coefficient at 250 nm to that at 365 nm (43).
Acknowledgments Part of this study was conducted when H.X. held a U.S. NRC associateship. This paper is based upon work supported by grants from NSF (EAR-0087695) and ONR (N00014-98-F0202). We thank H. Ji and F. Chen for assisting with the titrations and M. A. Moran for use of the water filtration system. This paper has been reviewed in accordance with the U.S. Environmental Protection Agency’s peer and administrative review policies and approved for publication. Mention of trade names or commercial products does not constitute an endorsement or recommendation for use by the U.S. EPA. This is Contribution 11152 of the Woods Hole Oceanographic Institution.
Supporting Information Available One table and three figures depicting the spectral irradiance of the light source, the changes of the spectral slope coefficient, aromatic carbon content, and molecular size with irradiation time. This material is available free of charge via the Internet at http://acs.pubs.org.
Literature Cited (1) Zafiriou, O. C.; Joussot-Dubien, J.; Zepp, R. G.; Zika, R. G. Environ. Sci. Technol. 1984, 18, 358A-371A. (2) Zafiriou, O. C.; Blough, N. V.; Micinski, E.; Dister, B.; Kieber, D.; Moffett, J. Mar. Chem. 1990, 30, 45-70. (3) Blough, N. V.; Zepp, R. G. In Active Oxygen in Chemistry; Foote, C. S., Valentine, J. S., Greenberg, A., Liebman, J. F., Eds; Chapman and Hall, New York, 1995; pp 280-333. (4) Valentine, R.; Zepp, R. Environ. Sci. Technol. 1993, 27, 409-412. (5) Zafiriou, O. C.; Andrews, S. S.; Wang, W. Global Biogeochemical Cycles; 17 (1), 1015, doi: 10.1029/2001GB001638, 2003. (6) Cooper, W. J.; Zika, R. G.; Petasne, R. G.; Plane, J. M. C. Environ. Sci. Technol. 1987, 22, 1156-1160. (7) Miller, W. L.; Zepp, R. G. Geophys. Res. Lett. 1995, 22, 417-420. (8) Moran, M. A.; Zepp, R. G. Limnol. Oceanogr. 1997, 42, 13071316. (9) Langford, C. H.; Wingham, M.; Sastri, V. S. Environ. Sci. Technol. 1973, 7, 820-822. (10) Budac, D.; Wan, P. J. Photochem. Photobiol. A 1992, 67, 135166.
(11) Faust, B. C.; Zepp, R. G. Environ. Sci. Technol. 1993, 27, 25172522. (12) Andrianirinaharivelo, S. L.; Pilichowski, J.-F.; Bolte, M. Transition Met. Chem. 1993, 18, 37-41. (13) Miles, C. J.; Brezonik, P. L. Environ. Sci. Technol. 1981, 15, 10891095. (14) Cai, W.-J.; Wang, Y.; Hodson, R. E. Geochim. Cosmochim. Acta 1998, 62, 473-483. (15) Thurman, E. M. Organic Geochemistry of Natural Waters; Martinus Nijhoff/W. Junk: Boston, 1985; p 497. (16) Beck, K. C.; Reuter, J. H.; Perdue, E. M. Geochim. Cosmochim. Acta 1974, 38, 341-364. (17) Moran, M. A.; Sheldon, W. M., Jr.; Sheldon, J. E. Estuaries 1999, 22, 55-64. (18) Gao, H.; Zepp, R. G. Environ. Sci. Technol. 1998, 32, 2940-2946. (19) Cai, W.-J.; Wang, Y. Limnol. Oceanogr. 1998, 62, 473-483. (20) Kulovaara, M. Intern. J. Anal. Chem. 1996, 62, 85-95. (21) Stookey, L. L. Anal. Chem. 1970, 42, 779-781. (22) Gran, G. Analyst 1970, 77, 661-671. (23) Paxe´us, N.; Wedborg, M. Anal. Chim. Acta 1985, 169, 87-98. (24) Faust, B. C. In Aquatic and Surface Photochemistry; Helz, G. H., Zepp, R. G., Crosby, D. G., Eds; Lewis Publishers: Boca Raton, FL, 1994; pp 3-38. (25) Voelker, B. M.; Morel, F. M. M.; Sulzberger, B. Environ. Sci. Technol. 1997, 31, 1004-1011. (26) Corin, N.; Backlund, P.; Kulovaara, M. Chemosphere 1996, 33, 245-255. (27) Bertilsson, S.; Tranvik, L. J. Limnol. Oceanogr. 1998, 43, 885895. (28) White, E. M.; Vaughan, P. P.; Zepp, R. G. Aquat. Sci. 2003, 65, 399-411. (29) Kieber, D.; Mopper, K.; Qian, J. Abstracts Pacifichem 2000, 1, ENVR-28. (30) Perdue, E. M.; Reuter, J. H.; Ghosal, M. Geochim. Cosmochim. Acta 1980, 44, 1841-1851.
(31) Oliver, B. G.; Thurman, E. M.; Malcolm, R. L. Geochim. Cosmochim. Acta 1983, 47, 2031-2035. (32) Ephraim, J. H.; Reddy, M. M.; Marinsky, J. A. In Humic Substances in the Aquatic and Terrestrial Environment; Allard, B., Bore´n, H., Grimvall, A., Eds.; Springer-Verlag: New York, 1991; pp 263276. (33) Leenheer, J. A.; Wershaw, R. L.; Reddy, M. M. Environ. Sci. Technol. 1995, 29, 393-398. (34) Leenheer, J. A.; Wershaw, R. L.; Reddy, M. M. Environ. Sci. Technol. 1995, 29, 399-405. (35) Leenheer, J. A.; Wershaw, R. L.; Brown, G. K.; Reddy, M. M. Appl. Geochem. 2003, 18, 471-482. (36) Voelker, B. M.; Sedlak, D. L.; Zafiriou, O. C. Environ. Sci. Technol. 2000, 34, 1036-1042. (37) Zepp, R. G.; Faust, B. C.; Holgne, J. Environ. Sci. Technol. 1992, 26, 313-319. (38) Sun, Y.-P.; Nguyen, K. L.; Wallis, A. F. A. Holzforschung 1998, 52, 61-66. (39) Opsahl, S. P.; Zepp, R. G. Geophys. Res. Lett. 2001, 28, 24172420. (40) Schmitt-Kopplin, P.; Hertkorn, N.; Schulten, H.-R.; Ketrup, A. Environ. Sci. Technol. 1998, 32, 2531-2541. (41) Xie, H.; Wang, Y.; Cai, W.-J.; Zafiriou, O. C.; Zepp, R. G. IHSS 10 Proc. 2000, 1, 253-256. (42) Va¨ha¨talo, A. V.; Salonen, K.; Salkinoja-Salonen, M.; Hatakka, A. Biodegradation 1999, 10, 415-420. (43) Strome, D. J.; Miller, M. C. Inter. Er. Theor. Angew. Limnol. Verh. 1978, 20, 1248-1254.
Received for review December 16, 2003. Revised manuscript received April 25, 2004. Accepted May 6, 2004. ES035407T
VOL. 38, NO. 15, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4119
