Environ. Sci. Technol. 2006, 40, 5333-5339
Polychlorinated Naphthalenes in Great Lakes Air: Assessing Spatial Trends and Combustion Inputs Using PUF Disk Passive Air Samplers T O M H A R N E R , * ,† M A H I B A S H O E I B , † T O D D G O U I N , †,‡ A N D PIERRETTE BLANCHARD† Science & Technology Branch, Environment Canada, 4905 Dufferin Street, Toronto, Ontario, Canada, M3H 5T4, University of Toronto at Scarborough, Department of Physical and Environmental Sciences, 1265 Military Trail, Toronto, Ontario, Canada, M1C 1A4
Passive air samplers made from polyurethane foam (PUF) disks housed in stainless steel chambers were deployed over four seasons during 2002-2003, at 15 sites in the Laurentian Great lakes, to assess spatial and temporal trends of polychlorinated naphthalenes (PCNs). Sampling rates, determined using depuration compounds pre-spiked into the PUF disk prior to exposure, were, on average, 2.9 ( 1.1 m3 d-1, consistent with previous studies employing these samplers. PCN air concentrations exhibited strong urbanrural differencesstypically a few pg m-3 at rural sites and an order of magnitude higher at urban sites (Toronto, 1231 pg m-3 and Chicago,13-52 pg m-3). The high concentrations at urban sites were attributed to continued emissions of historically used technical PCN. Contributions from combustion-derived PCNs seemed to be more important at rural locations where congeners 24 and 50, associated with wood and coal burning, were elevated. Congener 66/67, associated with incineration and other industrial thermal processes, was elevated at two sites and explained by nearby and/or upwind sources. Probability density maps were constructed for each site and for every integration period were shown to be a useful complement to seasonally integrated passive sampling data to resolve source-receptor relationship for PCNs and other pollutants.
Introduction Polychlorinated naphthalenes are a group of chemicals that are no longer commercially produced. Historically, they were used for their thermal stability in dielectric fluids and insulators. PCNs were produced as technical formulations of varying composition, with a total of 75 congeners possible. The most common formulations were Halowaxes, Seekay waxes, and Nibreen waxes (1). Peak usage of PCNs occurred during the 1960s at levels of about 10% of the use of polychlorinated biphenyls (PCBs), which later replaced them * Corresponding author phone: 416-739-4837; fax: 416-739-5708; e-mail:
[email protected]. † Environment Canada. ‡ University of Toronto at Scarborough. 10.1021/es060872m CCC: $33.50 Published on Web 07/27/2006
2006 American Chemical Society
(1). Although PCNs have not been commercially produced for several decades, they are still routinely observed in air. For instance, Lee et al. (2) have demonstrated that PCN air concentrations are declining more slowly than PCB concentrations, and that PCNs now outweigh PCBs in many parts of the UK. While PCN air concentrations are typically elevated in urban/industrial regions (2-4), atmospheric concentrations observed in the Arctic (5, 6) suggest that these compounds are subject to long-range atmospheric transport. Evidence indicating that PCNs are persistent and are capable of entering remote ecosystems as a result of long-range transport is of concern, especially given the dioxin-like toxicity exhibited by PCNs (7). There are at least two main contemporary sources of PCNs to the atmosphere (1). The first is through evaporative emissions from sources related to historical usage (e.g., leakage from contaminated equipment, either discarded in landfills or still in use). The second is unintentional production through combustion. Several studies have shown that relative to technical PCN formulations, combustion-derived PCNs exhibit subtle differences in congener composition and are enriched in what are referred to as “combustion marker” PCNs (1, 8, 9). Thus the presence of combustion marker PCNs can be used as evidence of this input to the observed PCN burden. For instance, Helm et al. (6) show important longrange transport contributions of combustion derived PCNs to the air burden in the Canadian arctic, especially during the winter period. PCNs have also been measured in the Norwegian arctic and have been measured in the snowpack (10). Meijer et al. have shown that the contribution of combustion derived PCNs in soils have increased over the past several decades (8). Understanding the relative contribution of these two sources is important in implementing effective strategies aimed at reducing the environmental and human exposure of PCNs. PCNs have been identified as persistent, toxic, and bioaccumulative substances, and as such have been targeted by international bodies (e.g., United Nations Economic Commission for Europe, UNECE and the United Nations Environment Program, UNEP) that identify and regulate the use and production of persistent organic pollutants (POPs) (11). A main objective of these international bodies is to establish monitoring programssboth regional and global in scales that would provide information relating to the spatial and temporal distribution of POPs. Passive air samplers have been promulgated as a complement to active high volume air sampling as a feasible means to achieve spatially resolved data for persistent organic pollutants (POPs), at a modest cost (11). In the Great Lakes basin, a regional, binational air sampling strategy, the Integrated Atmospheric Deposition Network (IADN), monitors organic pollutants to assess spatial and temporal trends and to conduct loadings estimates to Lakes (12). Although numerous studies over the past decade have demonstrated the abundance of PCNs in all environmental compartments of the Great lakes region (13), so far, PCNs have not been monitored in the atmosphere as part of a concerted effort. In this study, polyurethane foam (PUF) disk passive air samplers (PAS) (Figure 1) (5, 14, 15) were deployed at 15 sites (including 13 sites operated under IADN) (12) over four consecutive, 3-month periods. A previous paper reported air concentrations of polychlorinated biphenyls (PCBs) and selected organochlorine pesticides (OCPs) in the samples (15). This study investigates the spatial and temporal trends of PCNs, including combustion marker congeners. VOL. 40, NO. 17, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5333
FIGURE 1. Schematic and photograph of PUF disk passive air sampler.
Materials and Methods The study design, including sampler preparation, extraction, and theory on PUF disk PAS is presented thoroughly in a recent publication (15). Briefly, previous calibration of the samplers for PCBs and PCNs (14) showed that PUF disks sample atmospheric POPs in air at a rate of ∼3m3 d-1. This is referred to as the linear sampling rate. Chemicals with low octanol-air partition coefficients (KOA), will have a low PUFdisk limited capacity in the PUF disk. After some time, they will reach equilibrium between the PUF disk and air and the linear sampling rate will decline until it finally reaches zero. At this point, the air and PUF disks are exchanging the chemical back and forth at the same rate and there is no further accumulation of the chemical. This is referred to as the plateau phase. PUF-disk air partition coefficients (KPUF-A) can be calculated from KOA values which are known for PCNs (16). The KPUF-A is included in the uptake expression (14, 15) for PUF-disk samplers to work out effective air sample volumes for each congener. The limited capacity of the PUF disk samplers is advantageous because it allows depuration compounds (DCs) to be used to evaluate site-to-site differences in sampling rates. Although the PUF disks are housed in chambers designed to protect the sampling media from the elements (Figure 1) including wind speed, some wind-dependency may occur over the range of air velocities experienced in the chamber (17, 18). The effect of wind speed on uptake and loss rates (which are both directly related to the air-side mass transfer coefficient) can be assessed using DCs. Unlike performance reference compounds (PRCs), which are used to account for variability in sample media performance relating to several factors, including assessing the kinetics of mass transfer, fouling of the media membrane, and leakage of triolein, which are associated with the use of semipermeable membrane devices (SPMDs) (19), the sole purpose of DCs is to assess only the air side mass transfer coefficient by observing the depuration rate of a uniformly applied chemical. DCs are chemicals that do not exist in the atmosphere, hence “exchange” is in one direction, i.e., out of the sample media. The following DCs were used in this study: 2,4,6-trichlorobiphenyl (PCB-30), deuterated γ-hexachlorocyclohexane (d6γ-HCH), 2,3,3′,4,5-pentachlorobiphenyl (PCB-107), and 2,2′,3,3′,4,5,5′,6-octachlorobiphenyl (PCB-198). PAS were deployed throughout the Laurentian Great Lakes (Figure 2) on a seasonal basis from July-October 2002 (period 1, i.e., summer), October-December 2002 (period 2, i.e., autumn), January-March 2003 (period 3, i.e., winter) and March-June 2003 (period 4, i.e., spring). Of the 15 sites, two sites are considered heavy urban/industrial (Toronto and Chicago), two sites are impacted by urban/industrial regions by their proximity (Downsview and St. Clair), and the 5334
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 17, 2006
remaining sites range from regional background to remote. The population densities in Figure 2 help to make this distinction, and details are summarized in Table 1. Field blanks (n ) 22) and replicate samplers were also deployed at selected sites. Details regarding the extraction method and sample preparation are presented elsewhere (15). A Hewlett-Packard 6890 gas chromatograph (GC)-5973 mass spectrometer (MS) was used to quantify 27 PCN congeners using external standard solutions. Analysis was similar to a published method (3, 4, 7), which performed well in a recent international intercalibration exercise involving several laboratories (20).
Results and Discussion Quality Control/Quality Assurance. Most PCN congeners that were detected in samples were not detected in the 22 field blanks. Only congeners 24, 42, 33/34/37, 47, and 75 were detected. Limit of detection values (pg m-3), defined as mean blank + 3SD, based on an air volume of 300m3 were: 0.10, 0.060, 0.16, 0.06, and 0.12, respectively. These were well below concentrations detected in samples, and samples were not corrected for blank values. Instrument detection limits (IDL, pg) for PCN congeners ranged from 0.002 (PCN 75) to 0.18 (PCN 71/72) with a mean for all congeners of 0.04 pg. Method recoveries for PCNs are congener dependent and range from ∼50% for the 3-4 Cl congeners and 65% and higher for the higher molecular weight congeners (3). Samples were not corrected for recovery. Recoveries for depuration compounds were used to calculate sampling rates for each individual sample. These were converted to sample volumes based on the duration of the deployment period for the sample. Details of this calculation are presented elsewhere (15, 21). Sampling rates ( SD were on average 2.9 ( 1.1 m3 d-1 for all sites and sampling periods (Figure 3). This agrees very well with the result of ∼3m3 d-1 from the initial calibration study for PUF disk samplers (14). Seasonal averages for all sites during the summer, fall, winter and spring deployment periods were 3.3 ( 1.0, 3.1 ( 1.0, 2.2 ( 1.2, and 3.0(0.9 m3 d-1 respectively. Deployment times were typically 80-100 days which is equivalent to sample volumes of ∼240-300 m3 (based on linear sampling). As discussed previously, for the more volatile chemicals with log KOA values less than ∼8.5-9 (and for deployment times of ∼3months), uptake by the PUF disk can begin to approach its equilibrium capacity. This results in effective air sample volumes that are less than those based on linear sampling over the entire period. In this study, this situation applies to the tri-chloro PCN congeners during the summer, autumn, and spring periods. This effect can easily be
FIGURE 2. Location of sampling sites relative to population density. (BNT, Burnt Island; BUR, Burlington; CHI, Chicago; DOW, Downsview; EGB, Egbert; EGH, Eagle Harbor; TOR, Toronto; GDB, Grand Bend; PPL, Pt. Pelee; PPT, Pt. Petre; RPT, Rock Point; SBD, Sleeping Bear Dunes; STC, St. Clair; STP, Sturgeon Point; TNT, Trent University Field Site). See Table 1 for site coordinates.
TABLE 1. Site Information and Summary of Passive Sampler Derived ΣPCN Air Concentrations ΣPCN (pg m-3) site
description
latitude/longitude
summer, 2002
fall, 2002
winter, 2003
spring, 2003
BNT, Burnt Island BUR, Burlington CHI, Chicago DOW, Downsview EGB, Egbert EGH, Eagle Harbor TOR, Toronto GDB, Grand Bend PPT, Pt. Petre PPL, Pt. Pelee RPT, Rock Point SBD, Sleeping Bear Dunes STC, St. Clair STP, Sturgeon Point Trent University Field Site
background semi-urban urban/industrial urban/industrial rural background urban/industrial background rural rural rural rural rural rural rural
45° 49′ 42" N/82° 56′ 53" W 43° 22′ 08" N/79° 52′ 13" W 41° 50′ 04" N/87° 37′ 29" W 43° 47′ N/79° 28′ W 44° 13′ 57" N/79° 46′ 53" W 47° 27′ 47" N/88° 08′ 59" W 43° 41′ N/79° 25′ W 43° 19′ 53" N/81° 44′ 30" W 43° 50′ 34" N/77° 09′ 13" W 41° 58′ 00" N/82° 31′ 58" W 42° 51′ 11" N/79° 33′ 30" W 44° 45′ 40" N/86° 03′ 31" W 42° 22′ 50" N/82° 24′ 15" W 42°41′ 35" N/79° 03′ 18" W 44° 33′N/78° 30′ W
0.8 2.1 12.8 2.5 1.0 0.7 12.3 1.4 0.5 4.7 1.6 0.7 6.4 2.3 not available
0.7 1.4 25.2 10.8 N/A 1.0 19.3 2.5 3.3 4.6 3.7 0.8 28.1 3.9 3.9
1.9 3.4 52.1 2.7 2.4 3.6 15.3 3.1 3.5 6.3 3.9 3.5 4.4 3.6 6.0
0.6 1.9 29.3 4.2 1.2 1.1 30.8 3.2 11.6 5.7 2.5 1.0 0.3 3.4 0.8
accounted for using the PUF disk uptake expression (15). For the most extreme cases in this study, sample volumes for tri-chloro congeners were calculated to be approximately half that for the high molecular weight, high KOA, PCNs. Agreement between replicate samples was good and consistent with previous analysis for PCBs, PBDEs, and OC pesticides (15) where coefficient of variance (COV) values were less than 35% for 81% of replicates and
