Environ. Sci. Technol. 2002, 36, 1245-1256
Dynamic Oxidation of Gaseous Mercury in the Arctic Troposphere at Polar Sunrise STEVE E. LINDBERG* Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6038 STEVE BROOKS Oak Ridge Associated Universities, P.O. Box 117, Oak Ridge, Tennessee 37831-0117 C.-J. LIN Department of Civil Engineering, P. O. Box 10024, Lamar University, Beaumont, Texas 77710 KAREN J. SCOTT Department of Microbiology, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada MATTHEW S. LANDIS U.S. EPA, 79 TW Alexander Drive, Human Exposure Analysis Branch, MD-56, Research Triangle Park, North Carolina 27711 ROBERT K. STEVENS Florida Department of Environmental Protection, 2600 Blair Stone Road, Tallahassee, Florida 32399 MIKE GOODSITE National Environmental Research Institute of Denmark, Copenhagen, Denmark ANDREAS RICHTER Institute of Environmental Physics, University of Bremen, D-28359 Bremen, Germany
Gaseous elemental mercury (Hg0) is a globally distributed air toxin with a long atmospheric residence time. Any process that reduces its atmospheric lifetime increases its potential accumulation in the biosphere. Our data from Barrow, AK, at 71° N show that rapid, photochemically driven oxidation of boundary-layer Hg0 after polar sunrise, probably by reactive halogens, creates a rapidly depositing species of oxidized gaseous mercury in the remote Arctic troposphere at concentrations in excess of 900 pg m-3. This mercury accumulates in the snowpack during polar spring at an accelerated rate in a form that is bioavailable to bacteria and is released with snowmelt during the summer emergence of the Arctic ecosystem. Evidence suggests that this is a recent phenomenon that may be occurring throughout the earth’s polar regions.
Introduction Mercury has been targeted for global concern as a highly toxic contaminant. Exposures are thought to be increasing, * Corresponding author phone: (865)574-7857; fax: (865)576-8646; e-mail:
[email protected]. 10.1021/es0111941 CCC: $22.00 Published on Web 02/13/2002
2002 American Chemical Society
especially among indigenous populations who consume fish and piscivorus species contaminated with methylmercury (1, 2), a neurotoxin that biomagnifies in aquatic food chains. Environmental mercury levels are known to be elevated in the Arctic, to generally increase with latitude, and to have increased over time (3, 4). Extensive wetlands exist in the Arctic (5), and such areas are now recognized as important sources of methylmercury (6). In addition to Hg0, somewhat lesser amounts of oxidized reactive gaseous mercury (RGM) species are now known to be emitted from industrial sources (7). RGM species are watersoluble, exhibit a much shorter atmospheric lifetime than Hg0, and their potential contribution to atmospheric deposition is widely recognized (8, 9). Although RGM compounds represent only a few percent of the overall gaseous mercury in typical ambient air, their dry deposition velocities and scavenging ratios exceed those of Hg0 by more than an order of magnitude (8). Since industrial emissions of water-soluble RGM compounds are controllable to some extent (7), the demonstration of direct production of RGM in the atmosphere has profound implications on ecosystem and human exposure. The concept of global fugacity has been used to explain the condensation and accumulation of organic toxins in Arctic regions (10), but Hg0 does not effectively “condense out” even at -50 °C. However, other forms of airborne Hg, especially oxidized Hg, might strongly partition from gas to solid phases at low temperatures (11). Recent reports of ground-level ozone (O3) and Hg0 depletions at Alert in the Canadian High Arctic (∼82° N) provided the first evidence that conditions may exist in the upper Arctic following polar sunrise that promote depletion of airborne Hg0 (12). Unlike O3, which is chemically destroyed, Hg only changes its oxidation state, and depletion from the airmass implies accumulation elsewhere. While the original Alert study did not include measurements of the fate of the depleted Hg, a recent paper reports elevated levels of Hg in snow throughout the Canadian Arctic (13). The characterization of the associated Hg species and their ultimate fate in the Arctic ecosystem is critical to our understanding of mercury depletion events (MDEs). In 1998, we initiated Hg0 measurements as part of the Barrow Arctic Mercury Study (BAMS; 14) to determine the geographic extent and reaction mechanism of MDEs. In 1999, we added the first semi-continuous measurements of RGM in the Arctic and measured both species during and after polar sunrise through June 2001. In an intensive campaign during April-June 2001, we also collected particulate Hg (Hgp) samples at Barrow, aircraft measurements of RGM, and aerodynamic measurements of RGM fluxes over snow. We report here the first detailed analysis of the BAMS data and demonstrate that Hg0 is being “depleted” as a result of in-air oxidation reactions that produce some form of oxidized gaseous mercury that is rapidly deposited to the snow surface at Barrow.
Methods Study Site. All of the data reported here were collected at the NOAA Climate Monitoring and Diagnostic Laboratory (CMDL) in Barrow, AK. There are no Hg sources within the CMDL, no emission points on the roof where our intakes were located, nor any known major Hg point sources in the town of Barrow, which is located ∼10 km southwest of the CMDL. Prevailing winds are from the northeast across the Beaufort Sea. The CMDL is located near the peninsula at Point Barrow, ∼2 km from the shoreline, and is surrounded primarily by VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1245
FIGURE 1. Schematic diagram of the sample flow path in the automated Tekran model 2537A Hg0 analyzer with the 1130 gaseous speciation denuder module and 1135 particulate Hg pyrolysis unit. water to the north, east, and west. Barrow is geographically the northern-most point in Alaska, located at 71°19′ N, 156°37′ W, and the CMDL is ∼9 m above mean sea level. In latitude, Barrow is ∼1600 km south of Alert. We initiated atmospheric Hg sampling at Barrow in September 1998 and began additional parallel sampling of speciated Hg in air 1 yr later using the methods described below. These measurements will continue through at least 2003. We also collected fresh surface snow and snow cores for Hg analysis from the tundra in the CMDL clean air sector during 2000 and 2001. Particulate Hg, eddy flux, and aircraft Hg measurements were added during the intensive BAMS-2001 campaign in March-June 2001 at the Barrow lab. Ancillary data available at the CMDL include routine meteorological data and trace gases such as ozone (15). Sampling and Analytical Approaches. Atmospheric Hg0 was determined using a Tekran 2537A vapor-phase mercury analyzer. Descriptions of the Tekran and its operating parameters have been published (16), and it has been the subject of a number of intercomparison studies (17-19). The 2537A instrument utilizes two parallel solid gold traps to preconcentrate Hg0 that is subsequently thermally desorbed into a cold vapor atomic fluorescence spectrometer (20). The instrument was configured to sample at a 5-min time resolution using a heated Teflon inlet line mounted ∼5 m above the ground on a mast ∼1 m above the roof of the NOAA CMDL building. The insulated sampling line was maintained at 50 °C by a PID temperature controller. Atmospheric Hg speciation was determined by integrating a Tekran 1130 speciation unit with the Tekran 2537A. The model 1130 speciation unit consists of a heated denuder module, a pump module, and a controller module. The model 1130 controller module integrates the analytical capabilities of the Tekran model 2537A unit with the 1130 speciation module allowing for continuous measurement of both Hg0 1246
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002
and RGM at pg m-3 concentrations. The 1130 speciation unit was configured to collect 2-h RGM samples onto a KCl-coated quartz annular denuder at a 10 L/min flow rate. During the 2-h sampling period, 5-min Hg0 samples were continuously quantified by the 2537A analyzer. After the 2-h sampling period, the 1130 system was flushed with Hg-free air, and the annular denuder was heated to 500 °C. The RGM collected on the denuder was thermally decomposed into a Hg-free airstream and subsequently analyzed as Hg0. The denuders collect oxidized reactive gaseous mercury compounds with a diffusion coefficient >0.1 cm2/s that readily adhere to a KCl coating at 50 °C (21). The most probable candidate compounds are HgCl2 and HgBr2; HgO is less likely. In this configuration, the fine particulate phase mercury (Hg-p; e2.5 µm) that passes the impactor inlet was collected on a downstream quartz fiber filter but was not analyzed. For 3 months during the BAMS-2001 spring intensive campaign, our EPA colleagues deployed additional mercury monitoring equipment to allow separate quantification of Hg0, RGM, and also Hg-p. The equipment included a Tekran 2537A/1130 configuration described above with the addition of a prototype Tekran 1135 particulate pyrolysis unit. One of the greatest problems with conventional Hg-p measurement methods is that RGM has been shown to adsorb to filter media and previously collected aerosols. This can result in large and variable measurement artifacts (21). The Tekran 1130/1135 speciation system avoids this problem by collecting RGM prior to collecting the Hg-p onto a quartz fiber filter. The quartz filter and quartz denuder components were sequentially desorbed during the analysis phase at 500 and 650 °C, respectively. During the filter heating step, the pyrolyzer was maintained at 650 °C to ensure complete decomposition of all Hg-p compounds evolved during the filter temperature ramp-up to Hg0. Figure 1 shows a schematic of the flow path for the overall 1130/1135 system as integrated
with the 2537A. During desorption, the sampling system was flooded with zero air to eliminate background air and achieve good analysis blanks. This zero air also acts as the carrier gas during subsequent analysis steps. The pyrolyzer for the quartz regenerable filter was preheated to convert to elemental form any mercury compounds that are eluted during subsequent steps. As the regenerable particulate trap was heated, the particle-bound mercury captured on the trap was desorbed and quantified by the Tekran 2537A. The heating process also reconditioned the trap for subsequent cycles. After desorption was complete, the entire sampling train was cooled to 50 °C. After being cooled, the denuder and particulate trap were ready for another measurement cycle. Atmospheric Hg fluxes were quantified during the BAMS2001 intensive by deployment of a relaxed eddy accumulation (REA) system developed for RGM. The Danish REA system uses a METEK sonic anemometer coupled with three heated manual RGM denuders and filter packs (22). The basis of the REA logging and control system used is as described in the literature (23). The system was operated over the snow during April 2001 to collect 12 4-h samples at a 10 L/min flow rate and a switching frequency of 1 Hz to quantify the RGM concentrations in the upflow and downflow eddies and a deadband (we realize that there is decreased resolution of the turbulence at 1 Hz; however, we feel that the data represent a general trend and are in good agreement with fluxes calculated based on the constant sampling during the same time periods). The vertical flux was calculated as described in ref 24: F ) σW(C_up - C_down)β, where σW is SD (vertical wind speed), β is a proportionality constant (both measured in real time), and C_up - C_down is the direct difference in concentration between the upflow and downflow denuders for each period. The denuders were analyzed manually as described in refs 22 and 25, a method based on the automated approach described above. Development continues on the REA system to increase the sampling frequency and maintain a constant laminar flow in the denuders. Surface snow samples (upper 10 cm) were collected in acid-washed Teflon bottles and maintained frozen until analysis. Subsurface samples were similarly collected from the freshly exposed snowpack face. Snowmelt runoff was collected in prefired Pyrex bottles with Teflon caps. Total Hg and methylmercury analyses were performed at Flett Research Ltd. (Winnipeg, MB) and at ORNL using cold vapor atomic fluorescence spectrophotometry (26). Assays for bioavailable mercury (bioHg) were performed on a number of snow samples using mer-lux bioreporters, genetically engineered bacteria that produce light when divalent inorganic mercury enters their cells (27). Samples were melted in the dark and analyzed immediately. Sample preparation was done in a Class 100 laminar flow hood in a HEPA-filtered Hg clean lab at the Freshwater Institute (Winnipeg, MB). The general assay method employed was as described in ref 27 with the following modifications: Vibrio anguillarum was the host species, not Escherichia coli; cells were grown in Glucose Minimal Medium; the assay medium was modified to include 5 mM glucose and 18 mM (NH4)2SO4; in addition, 3 mM sodium/potassium phosphate buffer (pH 6), used to wash and resuspend the mer-lux cells, was added to the sample in the final cell suspension. The specific growth requirements and cell preparation of V. anguillarum are described in ref 28. The primary Hg standard used was a 1 µg mL-1 Hg(NO3)2 solution prepared by Flett Research Ltd. (Winnipeg, MB). Light production was measured with a Beckman LS 6500 scintillation counter in noncoincidental mode. Maps of BrO distribution in the total column were generated from GOME satellite data (Global Ozone Monitoring Experiment). GOME is a 4-channel UV/visible grating spectrometer on board the ESA satellite ERS-2. GOME
radiances and irradiances were used with a published algorithm (29) to derive total column BrO. GOME measurements do not yield profile information but reflect the BrO distribution throughout the column. Since stratospheric BrO (which mainly depends on photolysis of the reservoirs and total stratospheric BrY) is relatively constant with time and shows little spatial variation, as supported by model calculations and ground-based observations of stratospheric BrO (29), the BrO plots show mainly the boundary-layer variations plus a more or less constant offset. This idea is supported by enhanced boundary-layer BrO observed independently at other polar sites using ground-based instruments that match with enhanced BrO values in the GOME data (30). Quality Assurance Activities. At Barrow, the Tekran 2537A routinely undergoes automated periodic recalibrations daily using an internal permeation source. Two-point calibrations (zero and span) are performed separately for each cartridge. A Commercial permeation tube (VICI Metronics) provides approximately 1 pg/s at +50.0 °C. Since it is not practical to certify these low rates gravimetrically, manual injections were used in our home lab before shipment to Barrow to initially calibrate the tube against a saturated mercury vapor standard (16, 31). Since then, this procedure has been repeated once or twice annually. The adjustment for perm tube drift has been on the order of 1-2% per year. During routine operations at Barrow, the Tekran was also subject to periodic zero checks and spikes of ambient air with a known amount of Hg0. During 1999, we performed spikes and zero checks every 25 h using an automated system that externally controlled the perm source to deliver 16 pg of Hg0 into the Teflon sampling line. Over the year, spike recoveries ranged from 85 to 114% and averaged 102 ( 3% recovery. There was no significant difference in recovery during the MDE period as compared to the rest of the year. Zero checks were consistently at the detection limit (0.09 ( 0.01 ng/m3). Although the 1130 RGM speciation system does not yet include a direct calibration method with known amounts of RGM, all analyses are performed with the 2537A after thermal desorption and conversion of the RGM to Hg0, a process demonstrated to be quantitative with an efficiency of 100% (21). Field intercomparisons of paired 1130 instruments performed elsewhere over a range of RGM from ∼20 to 400 pg m-3 showed that the denuder method exhibits good precision (( 15%) and that no significant RGM breakthrough occurs for the 2-h denuder samples collected over the RGM concentration range seen at Barrow (21). We also performed an intercomparison for RGM in Barrow using the Tekran 1130 automated denuder operated adjacent to a manual quartz annular denuder of similar design (21). During May 1-6, 2000, these systems were operated simultaneously on the roof of the Barrow CMDL to collect six sets of replicate samples. The Tekran 1130 denuder was analyzed as described above, while the manual denuders were thermally desorbed in a tube furnace, and the exhaust gas was directed into a second 2537A for an independent analysis of desorbed Hg (21, 25). Five of the six replicates gave excellent results; the data were well correlated (r2 ) 0.91), and the means were within ∼5% (Tekran ) 108 ( 23 pg/m3, manual ) 115 ( 24 pg/m3). One set differed by a factor of 2 (manual lower), but RGM was increasing rapidly during this period, and the two samples were slightly offset in time. During the REA flux measurements, the total concentration of the three denuder tubes for each run compared favorably with a collocated manual denuder sampling system as a QA check. On the basis of parallel measurements in the lab and field study, the manual denuders exhibit a precision of 10%, which infers a sampling error of 40% (95% confidence level) for the RGM flux determined by the REA system, given an allowed 10% error in the micrometeorological determinations of β and σw (23). Flow in the denuders was set by a VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1247
mass flow controller to 10 L/min and controlled prior to and after each measurement. However, because of flow development during switching, a more cautious number for the REA system would be a denuder precision of 20%, thus giving a sampling error for flux as determined by the REA system of ∼60% (95% confidence level; 22). Approximately half of the Barrow snow samples were split and analyzed in duplicate by each analytical lab with good agreement ((15%). Assays for bioavailable mercury (bioHg) in snow samples were routinely run in replicate with a precision on the order of 5-10%.
Results and Discussion Dynamics and Chemistry of Depletion Events. The 1999 data (Figure 2) provide the first confirmation of MDEs at this more southerly Arctic site (Barrow is ∼1600 km south of Alert), and the 2000 data show that RGM is produced during MDEs (Figure 3). The early 2001 data exhibit comparable trends. This is the first evidence that the MDEs could be a widespread phenomenon of Arctic dawn and that RGM only appears at significant levels when Hg0 is being depleted. Although others have suggested that the depleted Hg0 at Alert accumulates in the aerosol-phase Hg (12), our data clearly indicate an important change in gaseous speciation during MDEs, producing levels of RGM unprecedented at remote and rural sites (8, 18, 25). Depletion events begin within a few days of polar sunrise (late January) and persist until snowmelt (early June, Figure 2), suggesting a role of both sunlight and frozen surfaces. During this period, Hg0 exhibits a strong correlation with O3 (e.g., r2 ) 0.76 for the period graphed) as also seen at Alert (12). Surface O3 destruction is a common feature at Barrow during Arctic spring (15). There is no correlation between O3 and Hg0 in the months before polar sunrise (r2 < 0.1). Gaseous and aerosol Br also exhibit strong seasonal cycles at Barrow and, like RGM, peak annually between January and June (32). During this period, aerosol Br increases nearly 20-fold over typical concentrations and can exceed 100 ng m-3. Hypotheses for the sources of this Br include aerosol enrichment by bubble bursting from the sea-surface microlayer, gaseous reactions resulting from organic Br emissions from marine algae (e.g., bromoform is thought to be emitted by ice algae), and/or other aerosol-related reactions. The most probable mechanism involves heterogeneous reactions at the interface of hygroscopic sea-salt aerosols (15), many of which are initiated in the surface microlayer of snowflakes or the snowpack (33). Several gaseous reactive halogen species (e.g., BrO, see Figure 2) may result with the potential to oxidize Hg0 to gaseous Hg(II) (RGM) compounds. Many of these halogen compounds exhibit a strong diel pattern (e.g., BrO, ClO, Br, and Cl), indicating the importance of sunlight and photochemical reactions (34), as reflected in the diel cycle of the airborne Hg species. Our data indicate that peak RGM production and Hg0 depletion generally occur at midday under maximum UV (Figure 4). We hypothesize that RGM is formed through a rapid, in-situ oxidation of Hg0 in the gaseous phase during MDEs. Production of RGM may be attributed to the same photochemically active halogen species involved in surface O3 destruction (15), suggesting that the overall process is heterogeneous. On the basis of reported halogen activation mechanisms in the remote marine boundary layer (34-39), we propose the physicochemical pathways conceptualized in Figure 5 for our observation of RGM during MDEs. In the reaction mechanism, bromine and chlorine radicals are produced autocatalytically from a heterogeneous photochemical mechanism involving sea-salt aerosol. The halogen radicals (Br/Cl) and halogen oxide radicals (BrO, ClO) produced from the ozone destruction reaction (reaction 1) exhibit strong diurnal patterns with solar radiation (34, 1248
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002
36) and serve as the primary oxidants that produce RGM (Figure 5):
Br/Cl + O3 f ClO/BrO + O2
(1)
BrO/ClO + Hg0 f HgO + Br•/Cl•
(2)
Hg0 + 2Br•/Cl• f HgBr2/HgCl2
(3)
and/or
These proposed mechanisms are plausible for the excellent correlation between Hg0 and O3 concentrations during MDEs. There are other reactive halogen species present in the Arctic that are thermodynamically favorable in oxidizing Hg0 to form RGM in the gaseous phase, including Cl2, Br2, HOCl, HOBr, and BrCl. However, we feel that molecular Cl2, Br2, and BrCl are not likely to cause the in-situ RGM formation seen here as they can be rapidly photolyzed under sunlit conditions (34-36). HOCl and HOBr are more resistant to solar irradiation. However, since they do not exhibit a strong diel cycle in the remote marine boundary layer (35-38), they may not account for the observed RGM at Barrow. Figure 5 identifies two RGM species as proposed products: mercury oxide (HgO) and mercury halides (HgBr2/HgCl2). Considering the extremely low concentration of all reacting species (both Hg0 and reactive halogen), the bimolecular oxidation of Hg0 by halogen oxide would be the favorable RGM formation pathway, yielding HgO as the product. However, since the denuders may preferentially collect HgBr2/HgCl2 over HgO (21), there may also be a significant contribution from halogen radicals in the trimolecular oxidation of Hg0. On the basis of published studies of reactive halogens in the Arctic (15, 3240), HgBr2 should be favored. Research to determine the predominant species of RGM at Barrow continues. The reaction scheme in Figure 5 explains the strong Hg/ O3 correlation and also the extreme seasonality of MDEs: the depletions of O3 and Hg0 require both sunlight and a frozen aerosol or snow surface. The reactive halogens are initiated by the production of HOBr (or HOCl) from hydroxyl radical reactions with Br- (or Cl-), which are highly concentrated in the surface layer of a frozen water droplet or snow crystal (40). The abrupt end of the MDEs precisely at snowmelt suggests that these reactions do not proceed when the droplets deliquesce, decreasing the surface Br- and Clconcentrations as the droplets become homogeneously mixed (Figure 5). Meteorological factors strongly influence the extreme levels of RGM measured at Barrow. We developed a simple predictive model for airborne Hg depletion and dry deposition to snowpack using local meteorological data and an inverse boundary-layer approach (41). Taken together, boundary-layer entrainment rates and deposition velocities (both modeled as a function of wind speed, UV-B, and air temperature) explained ∼70% and ∼80% of the measured variance in airborne Hg0 and RGM, respectively. The highest RGM concentrations consistently occurred during periods of reduced wind speed and maximum UV-B. Elevated RGM also coincided with periods characterized by increased levels of BrO in the vicinity of Barrow (e.g., Figure 2). These conditions often follow periods of elevated wave activity and sea-salt aerosol generation. For example, an extensive area of elevated BrO occurred on March 29, 2000, when the Beaufort gyre had opened a series of leads north of Barrow. Under these conditions, RGM reached 900-950 pg/m3 (e.g., Figure 3, days 91 and 138-148), levels that exceed those measured near industrial point sources (8, 42). Even the average daytime RGM of 180 pg m-3 at Barrow during MarchMay is several times higher than typical event levels at rural sites (8, 25).
FIGURE 2. Trends in Hg0 at Barrow during 1999, showing a strong correlation between Hg0 and O3 depletions, and development of a zone of elevated BrO near Barrow during this period (mean vertical column BrO is expressed as molecule cm-2, derived from GOME satellite data). VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1249
FIGURE 3. Trends in Hg0 (upper plot) and reactive gaseous mercury (lower) at Barrow during 2000, showing parallel behavior in total Hg in the surface snowpack (lower) and mean daily UV-B (lower). Fate of RGM and Dynamics of Hg in Snow. RGM is formed continuously as long as O3 and reactive Br are present during polar spring, and the observed air concentrations reflect the dynamic balance between formation and deposition. Integrating the data in Figure 4 suggests that ∼30-40% of the depleted Hg0 appears as airborne RGM and that the remainder is deposited to the snow surface directly as RGM and/or is scavenged by fine aerosols. Our inverse boundarylayer model of the local sink strength (41) predicted Hg deposition fluxes of ∼40 µg m-2 at Barrow during FebruaryMay 1999 and ∼55 µg m-2 during January-May 2000. These estimated 5-month fluxes are much higher than annual wet deposition rates measured in the northeastern United States (∼5-15 µg m-2 yr-1; 9) and can be supported only by direct deposition of RGM since submicron aerosols are subject to far less local deposition than is RGM (8). Our year 2000 snow chemistry data confirm these modeled rates of Hg accumulation (Figure 3). The measured Hg concentrations in the snowpack and calculated snowpack depth (86 cm) prior to melt in 2000 (assuming a 70% water 1250
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002
equivalency) yield an estimated average flux of ∼50-60 µg m-2. This flux is reflected in total Hg in snow, which increased steadily from 90 ng/L over this period (Figure 3). Such dramatic increases have not been seen in snow from lower latitudes (43), but 1998 data from a ship frozen in the Beaufort Sea 550 km north of Barrow showed similar temporal trends in surface snow (although at lower concentrations; 44), suggesting that this phenomenon should be investigated at higher latitudes. These concentrations of Hg in the May snowpack are not only unprecedented for non-Arctic remote sites but also exceed typical levels near industrialized regions (9, 43). The fraction of the total Hg pool in snow that was biologically available to bacteria (bioavailable mercury) was determined using the mer-lux bioreporter V. anguillarum pRB28 and its constitutive control V. anguillarum pRB27. The mer-lux bioreporters are genetically engineered bacteria that produce light when divalent inorganic mercury (Hg(II)) enters their cells (27), thereby distinguishing biologically labile from biologically inert Hg(II) species that do not enter the
FIGURE 4. Typical diel cycle of tropospheric gaseous Hg species and UV-B at Barrow (UV is measured in near-realtime, while Hg0 and RGM represent integrated samples of 5 min and 2 h, respectively, as described in the text).
FIGURE 5. Conceptual diagram of proposed Hg0 oxidation reaction sequences in the Arctic at Barrow. As explained in the text, there are several possible pathways and products. This schematic presents those that appear most favorable given the observations (dashed lines represent inter-phase transport; solid lines are reaction pathways). bacterial cell. This measurement is of interest because microorganisms play an important role in the transformation of Hg(II) in the environment (6). In January (2000), prior to polar sunrise, bioavailable Hg(II) was undetectable in Barrow snow. It then increased from 0.22 ng/L (∼1% of total Hg) in March to 8.8 ng/L (nearly 13% of the total Hg) in May. Prior to this study, bioavailable Hg(II) had never been measured in the Arctic. However, concentrations in snow and precipitation in a remote Boreal site in northern Canada were on average
