Environ. Sci. Technol. 2005, 39, 7631-7637
Atmospheric Size Distribution of PAHs: Evidence of a High-Volume Sampling Artifact ERIC G. SANDERSON* AND J.-P. FARANT Department of Occupational Health, Faculty of Medicine, McGill University, 1130 Pine Avenue West, Montre´al, Que´bec, Canada H3A 1A3
The purpose of this study was to characterize atmospheric levels of four- to six-ring polycyclic aromatic hydrocarbons (PAHs) in the vicinity of a horizontal stud So¨ derberg aluminum smelter in terms of the size distribution of particulate matter (September to December 2002). It was found that the vast majority of the PAHs was associated with particle diameters less than 1 and 3 µm. A profile comparison of the PAH mixturesusing benzo[a]pyrene (B[a]P) relative abundance ratios (PAH/B[a]P)sfor the cascade impactor filters indicated the formation of a sampling artifact. Overall, the PAH stability scale generated in this study agrees with those produced experimentally for ozone and nitrogen dioxides or developed using other in situ measurement techniques. Correlations of the four- to six-ring PAHs with other atmospheric variables suggested that smelter plume conditions and particle characteristics may play a potentially important role in the overall PAH reactivity. To our knowledge, this is the first study to report a sampling artifact for the four- to six-ring PAHs during in situ high volume sampling under real world conditions.
Introduction Within Canada, primary So¨derberg aluminum smelters are a large source of anthropogenic emissions for polycyclic aromatic hydrocarbons (PAHs) to the atmosphere (1, 2). These emissions of PAHs are in turn linked to the carbon anode, which undergoes incomplete combustion in the electrolytic cells during the reduction process (3). Prior to emission at the aluminum smelter considered in the current study, the flue gas travels through a wet scrubber where removal efficiencies are approximately 18 and 14% for B[a]P and total PAHs, respectively (4). After stack emission, the fate of the remaining PAHs associated with particulate matter (PM) will then depend on the particle properties and atmospheric conditions that drive the extent of dry and wet deposition, as well as any chemical transformation. During the last several decades, a multitude of studies have used size fractionating techniques to characterize PAH size distributions in a variety of locations. For example, studies have been performed in urban areas (including urban-traffic sites) (5, 6), industrial zones (7, 8), rural areas (5, 9, 10), and marine environments (10, 11). In general, the majority of the four- to seven-ring PAHs is found on the fine fraction of PM and displays a unimodal distribution. Several studies have, on occasion, found a second mode of PAHs associated with the larger coarse particles, often due to the relationship of * Corresponding author: phone +43 2236 807 ext. 369; fax +43 2236 71 313; e-mail
[email protected]. 10.1021/es0510111 CCC: $30.25 Published on Web 08/27/2005
2005 American Chemical Society
vehicle emissions and the resuspension of road dust (12, 13). Hence, the mechanisms that govern PAH association with a range of particle sizes are not solely limited to the growth of combustion-generated particles (aerosol aging) during atmospheric transport. To illustrate the fate of particle associated PAHs, Allen et al. proposed that mass transfer by vaporization and condensation helps explain the distribution of PAHs among particle sizes (5). Here, the preferential removal of individual PAHs is likely to occur by photooxidation and would affect the partitioning of PAHs on atmospheric particles (5). Other particle size distribution studies for PAHs in Germany (9, 14) indicate that coarse particles contribute most to dry deposition, while wet deposition of PAHs is dominated by fine particles. Overall, such examples from the literature illustrate the complex nature of the atmospheric fate of particle associated PAHs. Any changes in the size distributions of the PAHs downwind of a given source would be due to a combination of chemical transformation and size selective deposition during airborne transport. Historically, only the PAH content on total suspended particulates (TSP) and particulate matter with an aerodynamic diameter (dae) less than 10 µm (PM10) has been routinely characterized in Canada. For this reason, the primary objective of the current study was to investigate the relationships between the particle size distribution and the associated four- to six-ring PAH content in the atmosphere surrounding a So¨derberg aluminum reduction facility.
Materials and Methods Site Characterization and Sampling Strategy. The PAH content on the TSP, PM10, and cascade impactor samples was monitored between September and December 2002. The urban sampling station was located in Beauharnois in the Southwestern region of Que´bec, Canada (15). The station was situated in the prevailing wind direction (1.7 km) northeast from a horizontal stud So¨derberg (HSS) aluminum smelter. Detailed descriptions of the facility and production process are found in previous publications (16-19). Historical monitoring data in the vicinity of the facility indicate that the smelter’s emissions are the primary source of atmospheric PAHs in the area (17, 20, 21). TSP (n ) 15) and PM10 (n ) 15) samples were collected using conventional high-volume (hi-vol) samplers for a period of 24 h (00:00-24:00) from September 5 to November 28, 2005. Measurements for TSP and PM10 were performed every 6 days according to the national schedule defined by Environment Canada. A detailed description of the hi-vol sampling methodology is published elsewhere (17). Size fractionated samples (n ) 14) were collected in the interval between each sampling date of the national schedule using a hi-vol cascade impactor (Anderson model 235) that was attached to the PM10 sampler. These samples were collected for a period of 72 h (00:00-24:00) to ensure sufficient collection of particles to quantify individual PAHs on all impactor stages. For example, if PM10 samples were collected on October 17 and 23, the impactor sample was collected from October 19-21 when the PM10 sampling was not in progress. Note, three PM10 samples (October 17 and November 16 and 22) and two impactor samples (October 13-15 and November 18-20) were excluded since the PM10 sampler stopped during sampling due to excess brush wear in the motor. The particle sizes collected using the cascade impactor were separated into six ranges: dae < 0.49 µm (stage 6 or backup filter; these terms are used interchangeably throughVOL. 39, NO. 19, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7631
out the paper), 0.49-0.95 µm (stage 5), 0.95-1.5 µm (stage 4), 1.5-3 µm (stage 3), 3-7.2 µm (stage 2), and 7.2-10 µm (stage 1). All TSP, PM10, and impactor sampling was performed at a flow rate of 1.13 m3/min, and the flow accuracy was maintained using an automatic flow controller. TSP and PM10 samples were collected on glass fiber filters (20.3 cm × 25.4 cm, Gelman Type A/E). Impactor samples were collected on five separate fritted glass fiber filters (15.2 cm × 17.8 cm, Tisch Environmental Type A) with a backup glass fiber filter (20.3 cm × 25.4 cm, Gelman Type A/E). One field blank for every 10 TSP and 10 PM10 measurements and one field blank of the fritted filters per impactor sampling period was taken in this study. Directly following each sampling period, filters were recovered and stored in a desiccator for 24 h in the dark. Note, all filters (hi-vol and fritted) were treated in the same manner during sample preparation and transportation in the field, as well as sample weighing and workup needed for PAH extraction in the laboratory. After weighing, each filter was wrapped in aluminum foil and stored at 4 °C in a refrigerator until extraction and analysis. Because of the hivol sampling strategy (with no vapor phase PAH collection), only the four- to six-ring PAHs normally associated with particulate matter were considered in this study: benzo[a]anthracene (B[a]A), chrysene (Chrys), benzo[e]pyrene (B[e]P), benzo[b]fluoranthene (B[b]F), benzo[k]fluoranthene (B[k]F), benzo[a]pyrene (B[a]P), benzo[g,h,i]perylene (B[g,h,i]P), dibenzo[a,h]anthracene (D[a,h]A), indeno[1,2,3-cd]pyrene (Indeno), and benzo[j]fluoranthene (B[j]F). All sample concentrations were adjusted to standard atmospheric conditions (20 °C and 101.3 kPa). Atmospheric gases (ozone, O3; nitrogen oxides, NOx; nitrogen dioxide, NO2; and nitrogen oxide, NO) were continuously monitored at the urban sampling station. Meteorological conditions (temperature, T; relative humidity, RH; wind direction, WD; barometric pressure, BP; and wind speed, WS) were continuously monitored in a field directly adjacent to the HSS smelter. A complete summary of the atmospheric conditions during the TSP, PM10, and impactor sampling periods are presented in the Supporting Information. The warm weather period refers to all samples collected between September 5 and October 21 (Tavg ) 15 °C). The cold weather period refers to all samples collected between October 23 and December 8 (Tavg ) 0.5 °C). Collection Efficiency of the Cascade Impactor. Hi-vol cascade impactors can efficiently collect size fractionated samples (22, 23). Use of glass fiber filters also reduces particle bounce off and diffusional losses of the very fine particles in the upper stages (24). However, glass fiber filters experimentally shift the cutoff diameter for each stage downward (25). Then again, this effect is minor ( Indeno > D[a,h]A ∼ Chrys ∼ B[g,h,i]P > B[e]P ∼ B[b]F > B[k]F. This scale is consistent with several other reactivity scales that can be extracted from the literaturessee Supporting Information (43-48).
Discussion Reanalysis of the GMD for each PAH by adjusting for the loss of PAHs (EDMs) indicated that artifacts (particularly for B[a]P) were formed during the hi-vol sampling. To illustrate, GMDs of nonadjusted and adjusted PAHs (Table 2) did not differ appreciably with the exception of B[a]P, whose GMD decreased by 10% (0.66-0.6 µm). This downward shift of the GMD suggested that the adjusted PAH concentrations provided a more realistic indication of the particle size distribution of the individual PAH masses. The low GMD values (
