Environ. Sci. Technol. 2007, 41, 2165-2171
Coupling Passive Air Sampling with Emission Estimates and Chemical Fate Modeling for Persistent Organic Pollutants (POPs): A Feasibility Study for Northern Europe ROSALINDA GIOIA, ANDY J. SWEETMAN, AND KEVIN C. JONES* Centre for Chemicals Management and Department of Environmental Science, Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ, UK
Passive air samplers (polyurethane foam disks) were deployed at 23 background locations along a broadly westeast transect in 8 northern European countries and analyzed for PCBs, PBDEs, PAHs, and a range of organochlorine pesticides (HCB, DDTs, and DDEs). PCBs and PAHs were highest at the center of the transect (Denmark) and lowest in northern Norway. HCB was relatively uniformly distributed, reflecting its persistence and high degree of mixing in air. Higher DDE and DDT levels occurred in Eastern Europe and at several sites in Central Europe. PBDE levels were generally similar at all sites, but lower for some locations in Eastern Europe and Ireland. Emissions information for PCBs, HCB, and PBDEs was used as input for a multi-media chemical fate model, to generate predicted air concentrations and compare with these measured values. Different scenarios were highlighted by this exercise: (i) country and compound combinations where the national inventory gave predicted air concentrations in close agreement with those measured (e.g., PCBs in the UK); (ii) country and compound combinations where predicted concentrations were well below those measured, but where advection of emissions from elsewhere is likely to be important (e.g., PCBs in Norway); (iii) consistent underestimation of compound concentrations by the emissions modeling (i.e., HCB); and (iv) general overestimation of ambient concentrations (i.e., PBDEs). Air mass trajectory analysis showed the likely role of long-range atmospheric transport (LRAT) on national levels. In general, advection from the south and west of Europe appeared to contribute to ambient POPs levels for countries in the center and northeast of the transect. Guidelines are presented as to how countries that want to assess their POPs source inventories can do so with this relatively cheap initial screening approach.
Introduction As international efforts are made to ban and reduce sources of persistent organic pollutants (POPs) (1, 2), scientists and regulators need more sophisticated approaches to monitor * Corresponding author e-mail:
[email protected]; phone: + 44 1524 593972; fax: + 44 1524 593985. 10.1021/es0626739 CCC: $37.00 Published on Web 03/07/2007
2007 American Chemical Society
for changing levels in the environment, to identify the contribution of more diffuse sources, and to better understand compound environmental fate and source-sink relationships. Essentially 3 sets of tools are available to achieve this. (i) Source inventory estimates, usually at the national or regional scale (1, 2). These can be subject to large uncertainties, and need to be validated. (ii) Passive air sampling (PAS) techniques, which are cheap and easy to use, and can therefore be deployed over large areas. Knowledge of the way PAS respond to ambient conditions and reflect them is continually improving (3-5), such that they can be used to compare levels in different places at the same time, or trends in ambient levels (6-9). PAS have been used to study POPs at the local, regional and continental scale in recent years, and to detect differences in the concentrations and composition of atmospheric POPs along environmental gradients (e.g., urban-rural; latitudinal; altitudinal; chiral signatures) (e.g., 3, 4, 6, 7); and (iii) Chemical fate models (10, 11), which describe the receiving environment and the chemical’s properties, and can be used to derive predicted concentrations which can be compared to ambient measurements (12, 13). This allows the testing of available knowledge and hypotheses. To date, these 3 tools have largely been used separately by researchers and regulators. However, they can become very powerful when used together, to help identify uncertainties over inventory estimates, model descriptions and parameterization, and knowledge of chemical fate processes. This paper presents a “feasibility study” in which these three tools are combined and the power of the approach is demonstrated. Northern Europe was selected as the study area, because (i) reasonably comprehensive POPs emissions data has been assembled for European countries (e.g., 14, 15, 16); (ii) Northern Europe contains a range of contrasting countries, with high and low population density, “source” and “sink” regions for POPs, and a high level of information on meteorology; and (iii) under the auspices of EMEP (the Programme for Monitoring and Evaluation of the Long-Range Transmission of Air Pollutants in Europe), a network of well managed air monitoring stations has been established and co-ordinated, with a focus on background/remote locations (17). A general west-east transect was established, such thats in simple termssthe changing composition of European background air arriving from the west (i.e., from the Atlantic) could be monitored as it passed over Ireland, the United Kingdom, central Europe/Scandinavia (Denmark, Norway, Sweden, Finland), and into eastern Europe (Estonia, Russia). Figure 1 shows the study area, countries, sites, and broad air mass movements. PAS were deployed at 23 sites in 8 countries for 8-11 weeks during August-October 2005. Polyurethane foam (PUF) disks were the samplers of choice. These have been well studied and used in the past (3-5, 7, 9, 18), and are ideal for deployments over several weeks. Sampling times of several weeks/a few months provide a good time integration of meteorological and air mass back trajectory conditions, from which to make inferences about source-sink relationships. Data were obtained for a range of POPss namely PCBs, HCB, DDTs, PBDEs, and PAHssin the European background air. These represent compounds with a range of industrial, agricultural, combustion-derived, and legacy or recent uses.
Materials and Methods Sampling and Site Characterization. Details of the sites and their locations are given in Supporting Information Table VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2165
FIGURE 1. Location map for the sampling sites. Wind roses derived by air mass back trajectories during the period of deployment are also shown at some of the sites. The angular variable (θ) indicates the direction from where the air mass is coming and the radial variable (R) indicates the number of days during the sampling period from which the air mass was coming. SI-1. The samplers have been described elsewhere (18). They were transferred to the sampling locations in airtight containers to avoid contamination from ambient air. A courier was used to send samplers deployed outside the UK. The length of the exposure time differed slightly among locations, depending on availability of local volunteers for collection. Deployment was as follows: Ireland (2 samplers), UK (2), Denmark (6), Norway (4), Sweden (1), Finland (5), Estonia (1), and Russia (2) (see Table S1). Local volunteers were given guidance on choice of deployment locations: 3 were on islands, 4 were in coastal/marine locations, and 11 were in inland rural/remote areas. At the end of the deployment period, the samplers were retrieved by volunteers, re-sealed in their original transport containers, and returned by courier to Lancaster University. Upon receipt, the PUF disks were removed from the samplers and stored in sealed, solventrinsed, glass jars at -20 °C until extraction. Sample Processing. All PUF disks were pre-extracted with dichloromethane (DCM) using a Dionex accelerated solvent extractor. Upon return to Lancaster, samples were handled and extracted in a dedicated clean laboratory, which has filtered, charcoal-stripped air and positive pressure conditions. Each sample was spiked with a recovery standard of 13C -labeled PCB congeners (13C -PCB 28, 52, 101, 138, 153, 12 12 180, 209) and 1 PAH compound (dibenzo(ah)anthracened10) to monitor PCBs and PAHs respectively. They were then individually extracted in a Buchi extraction unit for 18 h with DCM. The extracts were concentrated using rotoevaporation and nitrogen-evaporation. Each sample was eluted on a 9 mm i.d. column with 1 g alumina (BDH neutral Alumina) column, 2 g of silica gel (Merck Silica 60), and 1 cm of sodium sulfate (all baked at 450 °C overnight). The extracts were then eluted through gel permeation columns containing 6 g of Biobeads SX 3. A portion (20%) of each sample was reduced in volume under N2 gas and solvent exchanged to 500 µL of acetonitrile for PAHs analysis. The rest of each sample was reduced in volume to about 100 µL under N2 gas 2166
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
and solvent exchanged to 25 µL of dodecane containing PCB 30, [13C12] PCB 141, and [13C12] PCB 208 as internal standard for PCBs, organochlorine pesticides, and PBDEs. The samples were analyzed by gas chromatography-mass spectrometry (GC-MS) with an EI+ source operating in selected ion mode (SIM) for PCBs and organochlorine pesticides. PBDEs were analyzed separately with a Thermo Trace GC-MS system operated in negative chemical ionization in SIM mode using ammonia as reagent gas. Details of the instruments, temperature program, and monitored ions are given elsewhere (19-21). The following compounds were monitored in the PUF disks: tri-PCBs 18, 22, 28, and 31; tetra-PCBs 44, 49, 53, 70, and 74; penta-PCBs 87, 90/101, 95, 99, 105, 110, 118, and 123; hexa-PCBs 138, 141, 149, 151, 153/132, and 158; heptaPCBs 170, 174, 180, 183, and 187; octa-PCBs 194, 199, and 203; HCB; o,p′-DDT, o,p′-DDE, p,p′-DDT, and p,p′-DDE; and PBDEs 28, 47, 49, 99, 100, 153, 154, and 183. PAHs were analyzed with a Perkin-Elmer HPLC system with LC250 binary pump, LS40 fluorescence detector, and ISS200 autosampler. The analytical column was a PAH Spherisorb column 15 cm × 4.6 mm i.d. (thermostatically controlled) and the mobile phase was an acetonitrile/water gradient. Twelve PAHs (fluorene, phenanthrene, anthracene, 1-methylphenanthrene, fluoranthene, pyrene, benzo[a]anthracene, chrysene, benz[e]pyrene, benzo[a]pyrene, indeno[123-cd]pyrene, and benzo[ghi]perylene) routinely detected in samples were quantified. QA/QC. Laboratory and solvent blanks were included at a rate of 20% and 10%, respectively. No field blanks were taken during this campaign, so laboratory blanks were used to blank correct, subtracting the mean of them from the estimated concentration of the samples. Jaward et al. (2, 8) found no difference between concentration of analytes in laboratory and fields blanks in a previous study throughout Europe, indicating contamination was minimal during transport, storage, and analysis. The limits of detection (LOD) were calculated as 3 times the standard deviation of the mean
FIGURE 2. Levels in ng/sample for North European Atmospheric Transect survey 2004: (a) PCBs; (b) HCB. laboratory blanks, and the data below the limit of detection were excluded. However, most of the PCB and PBDE congeners and OCs were not detected in the laboratory blanks. In this case the concentration of the lowest calibration standard was taken as the detection limit, namely 0.03 ng/ sample. LOD ranged from 0.03 to 0.05 ng/sample for PCBs (depending on congeners), 0.03 ng/sample for all PBDE congeners and OCs, and 0.6-7.8 ng/sample for PAHs. The solvent blanks were used as a method check. Recoveries were routinely monitored using the 13C12 PCBs as surrogate standards for PCBs and dibenzo(ah)anthracene-d10 for PAHs. Recoveries ranged from 60 to 107% for PCB and OCs, 75 to 100% for PBDEs, and about 100% for PAH recovery standards. Reported values are not recovery corrected. Extraction and cleanup method efficiencies were monitored by spiking cleaned PUF disks with validation standard for PCBs, PBDEs, OCs, and PAHs (matrix spikes) and extracting and analyzing those PUFs in the same way as samples. Recoveries for matrix spikes ranged from 75 to 110% for all compounds.
Results and Discussion General Comments on the Spatial Distribution Maps. PUF disks typically sample air at a rate of between 2.5 and 5 m3 d-1 (18). Information on sampler performance and methods to derive air concentrations were published previously (18). However, the initial discussions on compound spatial distribution uses data presented as absolute amounts (i.e., ng/sample). The data are expressed in this form in the Supporting Information Table S2. PCBs. Lowest levels were in Ireland, Norway, Finland, Sweden, and Estonia, and the highest were in Russia, Denmark, and the UK (Figure 2a). The data are consistent with other studies of European background air, and lower than those for urban centers (3, 7). The observation of relatively high levels in the UK and Eastern Europe is consistent with expectations, based on their past use and manufacture (14, 15). Manufacture and use of PCBs may have continued in Russia after bans were imposed in western VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2167
countries. The two Russian sites were also quite close (within 30 km) to Moscow, the largest urban area in the country. Diffusive atmospheric emissions from urban locations continue to contaminate the regional atmosphere (e.g., 3). HCB. Differences between the lowest and highest sample were only a factor of 3, less than that for any other compound (Figure 2b). HCB is extremely stable in the atmosphere and as primary sources reduce, it has become very well mixed in the atmosphere (22). The data indicate the highest sequestered levels generally in the most northerly sites, where temperatures were lowest, indicating that HCB has approached equilibrium in the samplers during the deployment time. These results are all consistent with previous observations (3, 7, 23). DDTs. In general, the DDT/DDE ratio was
