Environ. Sci. Technol. 2005, 39, 9471-9477
Photodegradation of Dimethyl Sulfide (DMS) in Natural Waters: Laboratory Assessment of the Nitrate-Photolysis-Induced DMS Oxidation R E N EÄ - C H R I S T I A N B O U I L L O N * , † A N D WILLIAM L. MILLER‡ Dalhousie University, Department of Oceanography, Halifax, Nova Scotia, Canada, B3H 4J1
The interaction of sunlight and dissolved chromophoric matter produces reactive chemical species that are significant in the removal of dimethyl sulfide (DMS) in the surface ocean. Using artificial solar radiation, we examined the role of several inorganic components of seawater on the kinetics of NO3--photolysis-induced DMS removal in aqueous solution. This study strongly suggests that NO3- photolysis products react significantly with DMS in aqueous solution possibly via an electrophilic attack on the electron-rich sulfur atom. This supports previous field observations that indicate that NO3- photolysis has a substantial control on DMS photochemistry in nutrient-rich waters. A key finding of this research is that the oxidation rate of DMS induced by NO3- photolysis is dramatically enhanced in the presence of bromide ion. Moreover, our results suggest that bicarbonate/carbonate ions are involved in free radical production/scavenging processes important for DMS photochemistry. These reactions are pH dependent. We propose that DMS removal by some selective free radicals derived from bromide and bicarbonate/carbonate ion oxidation is a potentially important and previously unrecognized pathway for DMS photodegradation in marine waters.
Introduction DMS biogeochemistry has received considerable attention because of the potential of DMS in the atmosphere to partially counteract the warming effect of greenhouse gases (1). Biogenic DMS from the ocean is the most important source of reduced sulfur compounds to the atmosphere in remote ocean locations (2). Atmospheric oxidation of DMS is a major source of sulfate aerosols and cloud condensation nuclei (CCN) that are thought to be involved in Earth’s climate regulation by influencing scattering, absorption, and reflection of solar radiation (1). To assess the DMS flux to the atmosphere, it is crucial to understand the processes controlling DMS distribution in the surface ocean. However, DMS production and removal processes are only partially characterized. Dimethylsulfoniopropionate (DMSP), which * Corresponding author phone: (910)962-3458; fax: (910)962-3013; e-mail:
[email protected]. † Current address: Department of Chemistry and Biochemistry, University of North Carolina Wilmington, Wilmington, NC, 28403. ‡ Current address: Department of Marine Science, Room 211, Marine Sciences Bldg., University of Georgia, Athens, GA, 30602. 10.1021/es048022z CCC: $30.25 Published on Web 11/08/2005
2005 American Chemical Society
is produced by some marine phytoplankton species, is the major source of oceanic DMS (3). Removal of DMS from surface waters involves mainly three processes: sea-air ventilation, bacterial transformation, and photodegradation (2-5). Until recently, DMS photochemical degradation has received little quantitative attention even though it may account for between 7% and 40% of the total loss of DMS in the surface ocean (5-8). Photodegradation of DMS in aqueous solutions proceeds via an indirect pathway that is initiated by the photochemical production of oxidants (Ox) (5, 9, 10). Previous studies have shown that the photoexcitation of chromophoric dissolved organic matter (CDOM) in natural waters is a major source of oxidants, such as singlet oxygen (1O2), hydroxyl radical (OH•), hydrogen peroxide (H2O2), and photoactivated CDOM, and these have been suggested as potential reactants in the photosensitized removal of DMS (11-13). On the basis of field investigations, an alternative source of photooxidants for DMS has recently been proposed (14, 15). A strong positive correlation between the quantum efficiency of DMS photochemical removal and NO3- concentrations in the northeast Pacific Ocean waters suggests that NO3- photolysis also induces DMS degradation in natural waters (14). In addition, Toole et al. (15) observed that the photodegradation rates of DMS increased linearly with addition of NO3- to seawater samples collected from the Southern Ocean. The purpose of this study was to investigate the kinetics and the possible mechanisms of DMS oxidation initiated by NO3- photolysis in aqueous solution, a process that we previously suggested can efficiently remove DMS from seawater. In previous studies, free radical reactivity toward organic sulfur compounds has been mainly determined via flash photolysis and pulse radiolysis experiments (16-18). In this study, however, free radicals were generated using irradiation of an aqueous NO3- solution with simulated solar radiation, an approach that closely replicates photochemical processes in the natural environment.
Background Photolysis of NO3- in seawater generates a series of free radicals (19). In UV-irradiated aqueous solution NO3- is photodecomposed into a variety of intermediates which include NO2-, NO2•, O(3P), and O- (20).
NO3- + hν f NO2- + O(3P)
(1)
NO3- + hν f NO2• + O-
(2)
Warneck and Wurzinger (21) reported that pathway 2 is about 10 times more efficient than pathway 1 during irradiation at 305 nm. O- radicals are readily protonated to form OH•. In aqueous solution, OH• is a highly reactive short-lived oxidant (E ) +1.83 V) that reacts with most inorganic and organic compounds at rates close to diffusion-controlled (i.e., with rate constants for a bimolecular reaction on the order of 109 to 1010 L mol-1 s-1; (22)). O(3P) is believed to react with O2 producing O3 or to react with NO3- forming NO2- and O2 (23). In seawater, Zafiriou et al. (24) determined that OH• reacts almost exclusively with bromide ion (97%) and, to a lesser extent, with the carbonate system (1.8%) and dissolved organic matter (DOM) (0.2%). These secondary photochemical reactions produce some longer-lived but less reactive free radicals such as Br2-•, and CO3-•. Br2-• is produced via VOL. 39, NO. 24, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
9471
OH• addition to the bromide ion as shown in the following simplified equations (25):
Br- + OH• f BrOH-•
(3)
BrOH-• f Br• + OH-
(4)
BrOH-• + Br- f Br2-• + OH-
(5)
Br• + Br- f Br2-•
(6)
The bimolecular rate constant of eq 3 is very large (kBr-,OH• ) 1.1 × 1010 L mol-1 s-1). Different pathways have been proposed for the production of CO3-• (26, 27). First, OH• can directly oxidize bicarbonate and carbonate anions (26) via electron-transfer reactions:
OH• + HCO3- f H2O + CO3-•
(7)
OH• + CO3-2 f OH- + CO3-•
(8)
Bimolecular rate constants reported in the literature (28) indicate that the reaction of OH• with CO3-2 is moderately fast (3.9 × 108 L mol-1 s-1) and the corresponding reaction with HCO3- is considerably slower (8.5 × 106 L mol-1 s-1). Moreover, the reaction between Br2-• and CO3-2 has been also proposed as an alternative source of CO3-• (27):
Br2-• + CO3-2 f 2 Br- + CO3-•
(9)
True and Zafiriou (27) calculated a rate constant of 6.5 × 106 L mol-1 s-1 for this reaction (eq 9). In the absence of any other scavengers, O3 directly reacts with NO2- (29) or decomposes to OH• (30). O3 may also directly react with Br- (to form OBr-/HOBr), HCO3-, and CO3-2. However, the bimolecular rate constants for these reactions are very low, i.e., less than 300 L mol-1 s-1 (29, 31) and likely not significant in seawater. Since organic sulfur compounds, including sulfides, are known to be readily oxidized by OH•, Br2-•, and CO3-• radicals (16-18), we have previously speculated that these free radicals could significantly affect DMS biogeochemical cycles in the surface ocean (14, 32).
Experimental Section Chemicals. DMS (>99.0%), sodium bicarbonate (>99.5%), sodium bromide (>99.0%), sodium hydroxide (>98.0%), and sodium nitrate (>99.0%) were obtained from Sigma-Aldrich Canada Ltd. Dimethylsulfoniopropionate hydrochloride (DMSP-HCl) was purchased from Research Plus Ltd. All chemicals were reagent grade and were used without further purification. All solutions were prepared using purified distilled NANOpure-UV water (Barnstead) having a final resistivity >18 MΩ cm-1. Seawater Sample Collection. Seawater samples were collected on May 13, 2003 in the northwest Atlantic Ocean (43°24′ N; 57°43′ W) from CCGS Hudson. Seawater from 10-m depth was collected using a rosette system equipped with 24 10-L Niskin sample bottles. Water samples were gravityfiltered directly from the Niskin bottles through an acidcleaned Whatman POLYCAP AS 0.2-µm (820-cm2) filter capsule prerinsed with seawater into a fluorinated highdensity polyethylene Nalgene 25-L bottle. Prior to sample collection, filter capsules were extensively precleaned with HCl solution and then with NANOpure-UV water. Samples were stored at 4 °C (dark) until used in irradiation experiments (about 7 months). Just before each experiment, samples were 9472
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005
refiltered through a 0.2-µm nylon membrane (Whatman no. 7402-004) prerinsed with NANOpure water and then with seawater. Irradiation Experiments. A SUNTEST CPS solar simulator (DSET Laboratories) equipped with a 1500 W xenon arc lamp (Heraeus), an infrared glass mirror, and a glass filter that eliminates the transmission of wavelengths below 290 nm was used for all polychromatic irradiation experiments. Spectral irradiance (E(λ): moles photons m-2 s-1 nm-1) intensity at the surface of the quartz sample cells was measured at 1-nm intervals from 250 to 700 nm using an Optronics OL754 spectroradiometer fitted with a quartz fiberoptic cable. The spectroradiometer was calibrated using a NIST traceable Optronic OL-752-10E irradiance standard. Repetitive scans showed less than 1% variation in the irradiance measured at any wavelength. Experimental Details. A detailed description of each irradiation solution used in this study is presented in Table 1. NaBr and NaHCO3 were added to solutions so that bromide and bicarbonate concentrations were similar to those found in seawater. However, NaNO3 was added at concentration somewhat higher than those generally found in surface oceans (260 nm) (9), and volatilization does not explain this loss since the cells are gastight and no loss was observed in dark controls. One feasible explanation for this, given the fact that we did not use trace metal clean techniques, is the possible contamination by trace amounts of a photoreactive species such as Fe. Fe photoredox reactions, in oxygenated aqueous solution, are known to produce reactive oxygen species, in particular hydroxyl radicals that may react with DMS. The kobs values presented in Table 1 were corrected for photochemical DMS loss not due to NO3- photolysis. kobs values for each treatment were calculated by subtracting the measured DMS concentration removed in the “no NO3added” experiments from the measured DMS concentration removed in the “NO3- added” experiments. The pseudo-first-order rate coefficient for DMS degradation determined in solutions containing 25 µmol L-1 of NO3was 0.12 ( 0.01 h-1 (Figure 1A). Among known products of NO3- photolysis, OH• is considered as the most important in inducing oxidation of organic compounds (41). It has been previously shown that OH• reacts with DMS in aqueous solution with a very large bimolecular rate constant (kDMS,OH• ) 1.9 × 1010 L mol-1 s-1; 18). Thus, we assume that, in this system, the photodegradation rate of DMS could be describe by eq 13:
-
d[DMS] ) kDMS,OH•[DMS][OH•]ss dt
(13)
Because OH• is a highly reactive, nonselective, free radical, as will be discussed later, it is not likely that all of the OH• radical generated by NO3- photolysis reacts with DMS. A possible loss pathway for OH• radical is the “cage recombination” with NO2• radicals produced from NO3- photolysis (44). Moreover, impurities present in the irradiated solution and even the quartz cell walls may also consume OH•. Addition of NaBr (0.8 mmol L-1) to irradiated NO3solutions drastically enhanced photochemical DMS loss rates (kobs ) 1.1 ( 0.1 h-1) (Figure 1B). OH• produced by NO3photolysis can react with bromide anion to form Br2-• (eqs 3-6). It has been shown previously that Br2-• also efficiently oxidizes DMS with a measured bimolecular rate constant of 3.2 × 109 L mol-1 s-1 (45). Consequently, in the presence of both OH• and Br2-• radicals, the photodegradation rate of DMS is defined by eq 14:
-
d[DMS] ) kDMS,OH•[DMS][OH•]ss + dt kDMS,Br2-•[DMS][Br2-•]ss (14)
On the basis of published second-order rate constant values, the reaction between OH• and 0.8 mmol L-1 Br- is approximately 25 000 times greater than the reaction between OH• and 20 nmol L-1 DMS. Therefore, in this system, the direct degradation of DMS by OH• is negligible, since almost all OH• will react with bromide. According to literature values, kDMS,OH• is approximately 6 times greater than kDMS,Br2-•. Taken alone, this would suggest that the addition of NaBr to NO3solutions should have lowered DMS degradation rates, a result contrary to our observations. This discrepancy can be explained in the following way. Br2-• radical has been demonstrated to be relatively more selective than OH• in its reaction with certain organic compounds (16, 17). Adams et al. (16) reported a study of Br2-• reactions with a wide range of amino acids and found that only sulfur-containing and aromatic amino acids have an appreciable reactivity with Br2-•. Non-sulfur-containing aliphatic amino acids gave no 9474
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 24, 2005
detectable reaction with Br2-•. Br2-• is also removed from the system via a bimolecular self-decay reaction (24). It appears likely that higher DMS photochemical loss rate coefficients observed with the addition of NaBr result from relatively higher steady-state concentrations of Br2-• in the irradiated solution. The photochemical DMS loss rate coefficient also increases with the addition of sodium bicarbonate (NaHCO3, 2 mmol L-1) to NO3- solutions (Figure 1C), giving a measured rate coefficient of 0.20 ( 0.02 h-1. In aqueous solution, reaction of OH• with both HCO3- and CO3-2 anions produces CO3-• radical (eqs 7 and 8). CO3-• is a good one-electron oxidant that reacts efficiently with reduced organo-sulfur compounds via an electrophilic attack (16, 17). For example, methionine which has a similar structure to DMS (R2S), reacts with CO3-• at a rate constant of 3.6 × 107 L mol-1 s-1. In this system, the photodegradation of DMS by OH• and CO3-• can be described by the following equation:
-
d[DMS] ) kDMS,OH•[DMS][OH•]ss + dt kDMS,CO3-•[DMS][CO3-•]ss (15)
In a solution containing 2 mmol L-1 NaHCO3, 20 nmol L-1 DMS, and at pH 8, literature values for kHCO3-, OH•, kCO3-2, OH•, and kDMS,OH• indicate that about 99.6% of OH• will react with HCO3-/CO3-2. Under these conditions, we suggest that CO3-• is the main oxidant for DMS and that the direct reaction of OH• with DMS plays only a minor role in observed DMS loss. Similarly to Br2-•, CO3-• selectively reacts with sulfurcontaining amino acids and enzymes (16, 17). Chen and Hoffman (17) observed that the bimolecular rates constants are low (k e 105 L mol-1 s-1) for the reaction of CO3-• with non-sulfur-containing aliphatic compounds but considerably larger (k ) 106 to 107 L mol-1 s-1) for the reaction of CO3-• with sulfur-containing compounds. The reaction of CO3-• radical with another CO3-• radical is believed to be an important loss process (26). Higher pseudo-first-order rate coefficients for the reactions of CO3-• with DMS observed in the present study also imply that the steady-state concentration of CO3-• was higher (i.e., longer lived species) than OH•, potentially due to its lower scavenging rates with components other than DMS. When both NaBr and NaHCO3 were added together and irradiated with NO3- (Figure 1D), the measured kobs was 0.42 ( 0.03 h-1, a significant decrease compared to the NO3- + NaBr solution without NaHCO3. The equation describing this photodegradation rate is as follows:
d[DMS] ) kDMS,OH•[DMS][OH•]ss + dt kDMS,Br2-•[DMS][Br2-•]ss + kDMS,CO3-•[DMS][CO3-•]ss (16)
-
Using literature values for the second-order rate constants for relevant reactions, we attempted to elucidate the dominant pathway for DMS degradation in an irradiated solution containing NO3-, NaBr, and NaHCO3. As mentioned above, the direct oxidation of DMS by OH• is negligible in a solution containing NaBr. In this system, we calculated that more than 99% of OH• reacts with Br-. It is likely that carbonate species and DMS compete for Br2-•, since Zafiriou and colleagues (24, 27) suggested that Br2-• radical is efficiently scavenged by reactions with the carbonate system. Indeed, based on the published second-order rate constant for the reactions of Br2-• with HCO3-/CO3-2 (27) and Br2-• with DMS (45), the rate of reaction of Br2-• with 2 mmol L-1 HCO3-/ CO3-2 (at pH 8) is approximately 25 times greater than its reaction with 20 nmol L-1 DMS. As indicated in eq 9, Br2-•
can oxidize CO3-2 and possibly HCO3- to form CO3-•, which can then react with DMS. A decrease in kobs values with the addition of NaHCO3 to a NaBr + NO3- solution is consequently expected since CO3-• is a less reactive oxidant for DMS than Br2-•. A fraction of Br2-• that is not scavenged by reaction with HCO3-/CO3-2 remains to react with DMS. As suggested by True and Zafiriou (27), in a solution containing both NaHCO3 and NaBr, bromide might just act as an intermediate in converting OH• to CO3-•. In summary, we propose that in an irradiated aqueous solution containing NO3-, NaBr, and NaHCO3, the dominant pathway for the photodegradation of DMS could be simplified to the following: (1) production of OH• via NO3- photolysis; (2) production of Br2-• via reaction of OH• and bromide anion; (3) production of CO3-• via reaction of Br2-• and HCO3-/ CO3-2; (4) degradation of DMS via reactions with both CO3-• and Br2-• species. Role of DIC in DMS Photochemistry in Natural Waters. From our experimental work with solutions prepared using NANOpure-UV water, it appears that DIC could play a significant role in DMS photochemical removal in marine waters. To determine the net effect of DIC in marine DMS photochemistry, two separate photochemical experiments were performed with seawater (Table 1). In both experiments, NO3- was added to samples collected from the Atlantic Ocean to a final concentration of 25 µmol L-1. In the first experiment, the sample pH was adjusted to 8. In the second experiment, DIC was stripped from solution according to the procedures explained above and the pH was then also adjusted to 8. kobs was about 2 times greater in the sample where DIC was removed (Table 1), consistent with results from experiments carried out with NANOpure-UV water. It is possible that the DIC removal procedure (i.e., acidification of the seawater samples) influenced the photochemical reactivity of CDOM present in the sample. Johannessen and Miller (34) noted
