Atmospheric Transport of Polybrominated Diphenyl Ethers and

Jan 8, 2004 - Atmospheric Transport of. Polybrominated Diphenyl Ethers and. Polychlorinated Biphenyls to the. Baltic Sea. ARNOUT F. H. TER SCHURE,* , ...
0 downloads 0 Views 201KB Size
Research Atmospheric Transport of Polybrominated Diphenyl Ethers and Polychlorinated Biphenyls to the Baltic Sea A R N O U T F . H . T E R S C H U R E , * ,† PER LARSSON,† CECILIA AGRELL,† AND JAN P. BOON‡ Chemical Ecology and Ecotoxicology, Department of Ecology, Lund University, Ecology Building, S-223 62 Lund, Sweden, and Department of Marine Biogeochemistry & Toxicology, Royal Netherlands Institute for Sea Research (NIOZ), P.O. Box 59, 1790 AB Den Burg, Texel, The Netherlands

The atmospheric transport of polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) was compared by measuring concentrations in air and deposition on an island located in the central basin of the Baltic Sea. Median ∑PBDE and ∑PCB concentrations (gaseous + particle) were 8.6 and 7.4 pg m-3, respectively. Airborne PCBs were mainly found in the gaseous phase, while most of the PBDEs were detected on particles, which agrees with predicted particle/gas distributions. ∑PBDE levels were dominated by the decabrominated BDE209 followed by the tetrabrominated BDE47 and pentabrominated BDE99. BDE209 is a marker for the environmental distribution of the commercial deca-BDE formulation (>99.5% BDE209), whereas BDE47 and BDE99 are markers for the commercial penta-BDE mixture. General correlations between PBDEs and PCBs suggested similarities in sources and transport mechanism, while more detailed examination of the data identified notable behaviors and exceptions. Differences in regression slopes among tetra-, penta-, and decabrominated PBDEs may reflect different transport processes and the change in usage pattern. Tetraand pentabrominated PBDEs may originate from secondary sources such as air surface exchange in a manner similar to that of the PCBs, while the deca-BDE209 formulation still has primary sources. The tribrominated BDE17 was also detected and is proposed to be a breakdown product due to atmospheric debromination processes. PBDEs had higher washout ratios than PCBs, explaining their higher concentrations compared to PCBs in precipitation (median of 6.0 and 0.5 ng L-1 for ∑BDE and ∑PCB concentrations (“dissolved” + particle), respectively) than in air. The calculated yearly deposition of PBDEs and PCBs indicated that the atmospheric input of PBDEs to the Baltic Proper is currently exceeding that of the PCBs by a factor of 40, while that of the PCBs is decreasing.

* Corresponding author phone: + 46 (0)46 2224598; fax: + 46 (0)46 2223790; e-mail: [email protected]. † Lund University. ‡ Royal Netherlands Institute for Sea Research. 1282

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 5, 2004

Introduction The Baltic Sea is a vulnerable ecosystem due to its semienclosed character and hydrology (1). It experiences high input of many anthropogenic contaminants, such as polychlorinated biphenyls (PCBs). As their usage ceased in most industrialized countries in the mid-1970s, levels of PCBs in Baltic biota have decreased (2). In contrast, polybrominated diphenyl ethers (PBDEs) are currently still applied as additive brominated flame retardants, and their concentrations in biota from the Baltic and other parts of the world are increasing (3-8). PCBs and PBDEs are structurally similar and have therefore similar physical-chemical properties. Consequently, PBDEs and PCBs may behave analogously in the environment (9-11). Concentration measurements in biota are, however, not the best way to directly estimate the load of pollutants to an ecosystem. Sources, flow, and environmental processes that the compounds undergo must be understood to quantify and predict their environmental fate and transport (1). The major transport route for many semivolatile organic compounds (SOCs) is the atmosphere (12). PCBs have been studied extensively in the atmosphere (13-16). It has been estimated that atmospheric deposition constitutes around 50% of the annual chlorinated contaminant input, including PCBs to the Baltic Proper, which is the central basin of the Baltic Sea (17, 18). Airborne PBDEs, however, have received less attention, and the input of PBDEs to the Baltic Sea by atmospheric deposition remains until now unknown. However, it seems useful to investigate this since air samples from urban and rural sites, near the U.S./ Canadian Great Lakes, and Arctic regions have been found to contain PBDEs (19-21). Thus, PBDEs are also subjected to long-range atmospheric transport, probably in a manner similar to that of PCBs. Understanding the atmospheric lifetime of PBDEs and PCBs and their removal from the atmosphere is essential for determining their air transport susceptibility to remote regions, and hence their environmental fate (22, 23). Contaminants in the vapor phase and associated with particles can be washed out via rain and/or snow. In rain, the fraction of particle-associated PCBs is generally less than 3% and that of PBDEs varies between 68% and 100% (24-27). The dry flux of chemicals depends on the vapor pressure of a chemical, type of surface, resistance to mass transfer in the deposition layer, particle size, air concentrations, and microand macrometeorology (12, 13, 28). Recently it has also been shown that wet and dry scavenging of PBDEs is equally important for their removal from the atmosphere (26). Airborne SOCs can also be removed by chemical reactions with OH and NO3 radicals, O3, and photolysis by UV light. Atmospheric breakdown involves mainly gas-phase substances, although it has been shown that contaminants sorbed to aerosols can react with OH radicals (22, 29, 30). Photolysis of the decabrominated PBDE congener BDE209 has been observed, whereas PCBs are known to react with OH radicals (31-33). Atmospheric degradation is however often partial, which can lead to the formation of breakdown products that are more persistent than their parent compounds. Both types of contaminants may therefore ultimately deposit in remote ecosystems and become bioavailable. To study the atmospheric long-range transport and the environmental fate of PBDEs and PCBs, air and deposition were sampled on a small island (Gotska Sando¨n) located in the central basin of the Baltic Sea, the Baltic Proper. An annual input budget by atmospheric deposition of the substances 10.1021/es0348086 CCC: $27.50

 2004 American Chemical Society Published on Web 01/08/2004

FIGURE 1. Sampling location in the Baltic Proper, including latitude. Given is the temperature range and the mean precipitation volume during the sampling campaign. The sub-basin boundaries of the Baltic Proper are drawn with dotted lines. to the area was calculated. Airborne PBDEs have to our knowledge hitherto not been extensively measured for the Baltic Sea, and the results were therefore compared with reported levels from other areas.

Experimental Section Sampling Location and Techniques. Air and atmospheric bulk deposition were sampled simultaneously from Sept 21, 2001, until Nov 11, 2001, on the island Gotska Sando¨n, situated in the Baltic Proper (Figure 1). The island was chosen because of its central position in the Baltic Sea, its small size, and its use as a meteorological station and since it has previously been found to be a representative site for the Baltic Proper as a whole, since there are no local point sources in its neighborhood (34, 35). During 19 periods each of 2 days, approximately 1600 m3 of air was drawn through a 150 mm diameter glass fiber filter (GF/A) and then subsequently through three polyurethane foam (PUF) plugs (size 150 × 20 mm) in series, using a high-volume sampler, as described elsewhere (36). A bulk deposition sampler (wet + dry) was deployed for 10 periods each of 4 days. Wet deposition percolated through a 47 mm diameter glass fiber filter (GF/ A) and then through two PUF plugs (size 27 × 40 mm) connected in series (27). At the beginning of the sampling campaign, the funnel was washed and scrubbed with a water/ soap solution and then cleaned with 95% ethanol. The sampler was closed with a stainless steel lid during the days when deposition was not collected. If it did not rain during a sampling period, i.e., dry deposition only, the funnel was washed and rinsed with 1.5 L of distilled water, which then percolated through the filter setup. Filters and PUFs for air and deposition were changed after each 2- and 4-day period, respectively. All samples were put in aluminum foil, placed in airtight bags, and kept frozen until preparation for analysis. Sample Preparation and Instrumentation. PBDE congeners are given different numbers according to the same IUPAC system used for PCB congeners (37). The methods for sample preparations were modified from Bremle et al. (38) and are described in detail elsewhere (27, 36). Briefly, PBDEs including tri (#17, #28), tetra (#47), penta (#85, #99, #100), hexa (#153, #154), octa (#183), and deca (#209) congeners and PCBs including tetra (#52), penta (#92, #95,#110), hexa (#128, #132, #144, #149, #151, #153), hepta (#177, #180, #183, #187), and octa (#194, #196,#201, #202) congeners sorbed on

the PUFs and GFs were Soxhlet extracted with a mixture of acetone and n-hexane (10:7 v/v) for 16 h. Octachloronaphthalene (OCN) was used as extraction standard (ES). Concentrated extracts were cleaned and fractionated on acid/ basic dual-layer silica gel columns (38). Samples were analyzed for PCBs and PBDEs by GC/ECD (Varian 3400) with an on-column injector and a 10 m methyl-deactivated precolumn in series with a 30 m DB5-HT capillary column (i.d. 0.25 mm, 0.1 µm film thickness). Pentachlorobenzene was used as a chromatographic standard. BDE209 was analyzed separately with the same technique using a similar precolumn in series with a 15 m DB5-HT capillary column (i.d. 0.25 mm, 0.1 µm film thickness). To validate the GC/ ECD results, a set of 18 randomly chosen samples were also analyzed by GC/MS for PBDEs at the Royal Netherlands Institute for Sea Research (NIOZ). Details of the analytical method are given elsewhere (39). The GC instrument was a Hewlett-Packard 6890, with a split-splitless injection and a 25 m CP Sil-8 capillary column (i.d. 0.25 mm, 0.25 µm film thickness). The mass-selective detector was a HewlettPackard 5973. Ionization gas CH4 and negative chemical ionization (NCI) in the SIM mode at the m/z ratios of both bromine isotopes (79 and 81) and m/z ) 487 (for BDE 209 only) were used. Quality Assurance. Extraction efficiency of the ES was 90 ( 20%, and samples were not corrected for recovery. Results for the mass spectrometry analysis were compared with the GC/ECD analysis, and for each PBDE congener their concentrations were not significantly different (ANOVA, P > 0.37, all cases), except for BDE100 (P < 0.001). Concentrations between methods were found to be highly correlated (Pearson correlation, r > 0.7, P < 0.002, all cases). Hence, PBDEs and PCBs were quantified using the GC/ ECD results. Blanks (field, n ) 12, and analytical, n ) 9) were examined every 14 samples and used for limits of quantification (LOQ). For each congener and type of filter analyzed (GF-air, GF-rain, PUF-air, and PUF-rain), no significant differences between field and analytical blanks were observed (paired t-test, P > 0.06, all cases). In case of interference, the LOQ was defined as 3 times the average blank sample levels. For BDE209 a 5 times threshold was used. The average amount of the PBDE congeners found in the blank GFs-air, GFs-rain, first and second PUFs-air, and PUFs-rain (expressed as the percentage of the average amount measured in each type of sampling matrix, (ng in the blank/ng in the sample) × 100) varied between 0% and 6%, 0% and 1%, 10% and 25%, and 7% and 20%, respectively, depending on the congener. To examine collection efficiency for the air sampler, for each sampling occasion the amount of PBDEs and PCBs on the third PUF was compared to that on all three PUFs. Breakthrough was on average 12% ( 22% for all congeners, whereas the breakthrough for the deposition sampler was assumed to be 0.12, all cases). An overview of reported atmospheric concentrations of several PBDEs and PCBs is given in Table 1. Those most relevant to our study are discussed here. The total PBDE (four congeners) levels of approximately 1 pg m-3 from air samples collected on the southern part of Gotland, the island in the Baltic Proper south of our sampling station (Figure 1), are somewhat lower than our results. No BDE209 was found, but the detection level for BDE209 was, however, reported to be high (19). PBDE concentrations measured at the rural site Stoke Ferry in the U.K. (19) were higher compared to the concentrations at Gotska Sando¨n, indicating that our station is situated at a less polluted site. 1284

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 5, 2004

In addition, the total concentrations found by Strandberg et al. (21) at the remote station Eagle Harbor near the Great Lakes are comparable with our results, although we detected higher concentrations of BDE209. During the early 1990s, Agrell et al. (34) measured total PCB concentrations in air at the same island (Gotska Sando¨n, Table 1). Although the authors analyzed a greater number of PCB congeners than those reported in this study, and over a 1-year time period, comparison indicates that PCB concentrations in air have decreased during the past decade at this site. Our median ∑PCB concentration is at the lower edge of the range found by Axelman et al. (41) in the mid 1990s (Table 1). Although the authors reported concentrations in air as the mean of five stations situated over the entire Baltic Sea area, they found that the PCB concentrations were not related to latitude and may thus be applied to the Baltic Proper area. We therefore suggest that PCB concentrations in the atmosphere above the Baltic Proper have decreased since the 1990s, which corresponds with declining atmospheric PCB concentrations in other parts of Europe and decreasing concentrations in Baltic Sea biota (2, 41). Relationships of PBDEs and PCBs in Air. All PCB concentrations were highly correlated with each other and with ∑PCB concentrations (r > 0.7, P < 0.003, all cases), except for PCB52 and PCB194. Hence, the PCBs in this study may have the same secondary sources, such as volatilization from soil surfaces (42-44). ∑BDE total concentrations were highly correlated with BDE47, BDE100, BDE209, and ∑PCB concentrations (Figure 4). Although a variety of factors are possible, such as similarity in sources and small- and largescale meteorological driving forces, the general correlations between PCBs and PBDEs suggest similarities in their atmospheric transport mechanisms that deliver the contaminants on a regional scale to the island. Regression analysis resulted in similar regression slopes between ∑BDE and BDE47, BDE100, and ∑PCB concentrations, whereas BDE209 had a different regression slope (04-0.5, 1.2, respectively), suggesting different underlying atmospheric transport processes and/or sources between BDE209 and

Given are the median or mean concentrations with the minimum and maximum values between parentheses. Re ) remote, SeRu ) semirural, Ru ) rural. c Median values. d Mean values. e Measured prior bud burst, directly after first snowmelt over a 3-day period. f Measured after bud burst over a 40-day period. g Not given. h Not detected. i Not analyzed. j BDE209 included. k Sum of 105 congeners. l Sum of 41 congeners. m Sum of 51 congeners. n Sum of 7 congeners.