Environ. Sci. Technol. 2004, 38, 2701-2706
Characterizing Uptake Kinetics of PAHs from the Air Using Polyethylene-Based Passive Air Samplers of Multiple Surface Area-to-Volume Ratios M I C H A E L E . B A R T K O W , * ,† DARRYL W. HAWKER,‡ KAREN E. KENNEDY,† AND JOCHEN F. MU ¨ LLER† National Research Centre for Environmental Toxicology (ENTOX), University of Queensland, 39 Kessels Road, Coopers Plains, 4108, Australia, and Faculty of Environmental Science, Griffith University, Nathan, 4111, Australia
Polyethylene passive sampling devices (PSDs) were deployed to investigate how passive samplers of multiple surface area-to-volume ratios could be used to characterize uptake kinetics for polyaromatic hydrocarbons (PAHs). Theoretically, uptake profiles for different thickness PSDs of the same surface area should show the following: where uptake is linear, the amount of compound accumulated in the different PSDs will be the same and where equilibrium is approached, the amount accumulated by the different PSDs will be proportional to sampler thickness. Polyethylene sheets of the same surface area and approximately 100 and 200 µm thickness were collected after 30, 60, and 90 days of exposure along with samples from a codeployed high volume sampler. Twelve priority pollutant PAHs could be routinely quantified in replicate PSDs. Overall, reproducibility between replicate PSDs was satisfactory, with normalized differences rarely exceeding 25%. The smallest analytes quantified, fluorene, phenanthrene, and anthracene, were shown to approach equilibrium during the deployment period, whereas uptake for fluoranthene and pyrene moved into the curvilinear stage. For most of the larger molecular weight PAHs such as indeno[1,2,3-cd]pyrene, uptake could be described using a linear uptake model. Preliminary sampling rates for the compounds which remained in the linear stage of uptake ranged between 0.5 and 1.5 m3 d-1 dm-2. Sampler to air partition coefficients were estimated for PAHs which approached equilibrium and predicted for some of the other compounds. Results suggest that a single deployment of PSDs with multiple surface area-to-volume ratios can be sufficient to determine whether uptake was linear or approaching equilibrium for a range of PAHs.
Introduction Atmospheric concentrations of semivolatile organic compounds (SOCs) such as PAHs and PCBs are typically measured with a high volume sampling system incorporating a pump * Corresponding author phone: 0061 7 3274 9147; fax: 0061 7 3274 9003; e-mail:
[email protected]. † University of Queensland. ‡ Griffith University. 10.1021/es0348849 CCC: $27.50 Published on Web 03/24/2004
2004 American Chemical Society
together with filters and/or sorbent. High volume or “active” sampling systems are relatively expensive to prepare and maintain, vulnerable to vandalism, and require power to operate; hence deployment on a large scale and in remote areas may be problematic and is limited. In this instance passive samplers are useful as they do not require electricity to sample SOCs. Also they do not require a high level of maintenance while deployed in the field. A range of passive air samplers, including semipermeable membrane devices (SPMDs), polyurethane foam (PUF), XAD resin based samplers, tristearin-coated fiberglass sheets, and polymer-coated glass samplers (POGs) (1-7), have been used for passive sampling of SOCs from air. Typically, uptake of chemicals into passive samplers is initially linear over time, then moves into a curvilinear stage, and eventually can approach equilibrium. For simplicity, the estimation of atmospheric concentrations from passive sampling data is usually based on a linear (kinetic) model or an equilibrium model, assuming constant air concentrations. Semivolatile organic compounds such as PAHs exhibit a wide range of physical and chemical properties. Thus a sampler may be an equilibrium sampler for some chemicals and a kinetic sampler for others. To apply the appropriate model when estimating air concentrations for specific compounds it is relevant to understand the position of a sampler with respect to the linear or equilibrium stage. There are three alternate approaches by which the stages of uptake into a passive sampler may be experimentally characterized (8). The common approach is where samples are collected at various time points during an exposure period and the concentration of the analyte in the sampler plotted as a function of time (9). In the second approach equilibrium is confirmed when concentrations in both uncontaminated and contaminated samplers converge over subsequent time periods. In effect, the loss of compounds from samplers loaded at concentrations higher than the expected equilibrium concentration can be compared with the accumulation of the same or similar compounds in uncontaminated samplers (3). The third approach to characterizing the uptake kinetics of a passive sampler is obtained by manipulating the surface area-to-volume ratio of codeployed samplers, which are equivalent in all other respects. A convenient means of achieving this is to codeploy samplers with negligible differences in surface area but varying volume or thickness. During the initial, linear uptake stage, the samplers of different thickness will theoretically accumulate analytes at the same rate as determined by the interfacial surface area of the sampler regardless of sampler thickness. However as the samplers attain equilibrium, which takes different time periods, the amount of analyte in the samplers would be proportional to sampler thickness. To test the third approach we deployed low-density polyethylene sheets (PSDs) of two different thicknesses but with the same cross-sectional area. The multiple surface area/ volume ratio system was used to show whether uptake for particular PAHs was in the linear stage or approaching equilibrium. On that basis, sampling rates were calculated for compounds in the linear stage of uptake and approximate sampler to air partition coefficients were determined where compounds approached equilibrium. Passive Sampler Theory: Influence of Sampler Surface Area and Volume. The following section illustrates in mathematical terms how multiple surface area-to-volume passive samplers can be used to characterize uptake kinetics. The diffusive exchange of a chemical between the air and VOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2701
sampler can be represented by the following flux equation (3):
VS
(
CS dCS ) A SK C V dt KSV
)
(1)
where CS and CV are the concentration in the sampler and vapor phase, respectively (ng m-3), VS is the volume of the sampler (m3), AS is the area of the sampler (m2), K is the overall mass transfer coefficient (m s-1), KSV is the sampler to vapor phase partition coefficient, and t represents time of exposure (days). In the early stages of uptake, when CS is very small, loss from the sampler is relatively insignificant, and the change in concentration in the sampler could be considered to be due only to uptake:
dCS VS ) ASKCV dt
(2)
If the analyte concentration in the surrounding vapor phase is considered to remain constant, then this equation can be integrated and expressed on a mass basis:
NS ) ASKCVt
(3)
where NS represents the amount accumulated in the sampler (ng). This equation describes an initial, linear uptake phase of a chemical into the sampler and shows that the amount in the sampler, on a mass basis, is directly proportional to the area of the sampler. Therefore while uptake is in the linear phase, the amount of analyte accumulated in two PSDs of similar surface area but different volume should be the same. As the concentration in the sampler increases, loss from the sampler becomes more important until at equilibrium eq 1 can be rearranged to the following:
NS(eq) ) KSVCVVS
(4)
If the PSDs are constructed of the same material (hence sampler to air partition coefficients are the same) and exposed to the same vapor phase concentrations, then the amount in each sampler at equilibrium (NS(eq)) is directly proportional to the volume of the sampler. Thus, if the surface area of each sampler is the same, the amount in each sampler at equilibrium is then proportional to the thickness of the sampler. Figure 1 shows how the uptake curves for codeployed samplers with the same surface area would theoretically differ according to the previously described equations if one sampler was 200 µm thick and the other sampler was 100 µm thick. Where uptake is linear, the amount accumulated in either PSD is the same because surface area governs how much of the analyte accumulates. Uptake for the 100 µm PSD then begins to approach equilibrium before the 200 µm PSD until finally both samplers have attained equilibrium and the amount in the 200 µm PSD (2 NS) is twice the amount in the 100 µm PSD (NS). Therefore, by simultaneously deploying these PSDs, linear uptake or equilibrium can be inferred based on the ratio of the amount of a particular analyte in the different PSDs. Furthermore, if the atmospheric concentration of the respective analyte is measured and can be assumed to be constant with time, then key parameters can be estimated. If uptake was linear, then sampling rates can be calculated according to the following:
RS )
NS CVt
(5)
where RS is the sampling rate (m3 day-1). 2702
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004
FIGURE 1. Theoretical uptake profiles for 100 µm and 200 µm thickness PSDs. When uptake is initially linear, the amount accumulated by both PSDs remains the same. The 100 µm PSD then approaches equilibrium sooner than the 200 µm PSD. Finally, at 99% of equilibrium, the 200 µm PSD has accumulated approximately twice as much analyte (2 NS) than the 100 µm PSD (NS). If equilibrium has been attained the sampler to vapor phase partition coefficient can be calculated by:
KSV )
CS(eq) CV
(6)
Materials and Methods Air Sampling. Each PSD was comprised of approximately 30 cm by 30 cm of low-density polyethylene film. Sheets were cut from 2 different rolls of polyethylene of mean thickness 99 ( 1 µm (n ) 15) and 196 ( 2 µm (n ) 15) and mean mass, 7.3 ( 0.6 g and 16.1 ( 0.5 g, respectively. Each PSD was pre-extracted in approximately 800 mL of redistilled hexane for 24 h followed by pre-extraction for a further 24 h in fresh redistilled hexane. This was done not only to ensure the samplers were clean but also to remove compounds such as lower molecular weight polyethylene oligomers and other compounds (e.g. slip additives, antistatic agents, and antioxidants) which could potentially interfere with the analysis of samples (10). PSDs (including field blanks) were wrapped in aluminum foil immediately after preparation and stored at -17 °C. The codeployed filter-sorbent high volume air sampling system consisted of a glass fiber filter paper (Whatman GFF, 9 cm diameter) to collect particle-associated PAHs, two XAD filled cartridges in series to capture the vapor phase compounds (XAD-2, Sigma), a pressure gauge, gas meter, and vacuum pump. The XAD resin was prepared by rinsing with 500 mL each of H2O, acetone (Merck, residue analysis grade), and methanol (Unichrom, AR grade), loaded into the sampler cartridge and then Soxhlet extracted for 4 h, first using n-hexane (Merck, GC grade) and then toluene (Merck, residue analysis grade). The cartridge containing the resin was then removed from the Soxhlet and rinsed using 500 mL of dichloromethane (DCM) (Merck, residue analysis grade) before being dried with purified N2, sealed, and stored in the dark at -17 °C. GFFs were rinsed with acetone, heated in an oven (130 °C) for 24 h, and wrapped in aluminum foil prior to deployment. The sampler was run during deployment at about 3.6 m3 h-1 (pressure corrected capacity) using a dry vane vacuum pump. Previous testing for sampling artifacts associated with the adsorption of vapor phase pollutants on the GFF and the breakthrough of compounds through the XAD-2 cartridge has shown the sampling system to be reliable (11). Ideally samplers would be deployed under conditions of constant air concentrations; however, given the difficulties
involved in creating a constant atmosphere for PAHs indoors, a field site was chosen, where variation in atmospheric PAHs was expected to be low. The experiment commenced in April 2002 at a Queensland Environmental Protection Agency (QEPA) Ambient Air Monitoring Station situated at Springwood in metropolitan Brisbane, Queensland, Australia. Springwood is a predominantly residential suburb located 20 km south east of Brisbane CBD. This suburb is situated in the rapidly expanding corridor between Brisbane and the Gold Coast. The M1 motorway which is several kilometers from the monitoring station bisects Springwood. The station itself is located on the grounds of a school, approximately 20 m from a 4-lane roadway that services residential areas. Passive samplers were deployed 1.5 m above the ground in galvanized iron chambers with an open bottom and louvers on all sides. The chambers were designed to ensure adequate air flow while minimizing the effect of direct sunlight, rain and wind on sampler performance and were similar to those used by Ockenden et al. (2). Three chambers were deployed; with each chamber containing 2 sheets of 100 µm and 200 µm thick PSDs. PSD replicates (two 100 µm and two 200 µm PSDs) and XAD cartridges were collected after 30, 60, and 90 days. Mean temperatures during each 30 day period were 22 °C, 18 °C, and 16 °C, respectively, while mean wind speeds remained relatively constant (0.8 m s-1, 0.8 m s-1, 0.9 m s-1, respectively) (12). Filters were regularly inspected and collected every 3 or 4 days, depending on the amount of accumulated particulate material. Passive samplers and filter papers were wrapped in foil on-site, and the XAD cartridges were sealed with ground glass stoppers. Samples were transported to the laboratory and stored at -17 °C. Analysis. PSDs were extracted in redistilled hexane at room temperature (22 °C) without any precleaning. Each PSD was extracted in the dark using 800 mL of solvent, and after 24 h the solvent was replaced and the samplers were extracted for a further 24 h. For quantification the PSD extract was spiked with an internal standard mix containing known amounts of 10 deuterated PAHs (2D10-acenaphthene, 2D10fluorene, 2D10-phenanthrene, 2D10-fluoranthene, 2D12-benz[a]anthracene, 2D12-chrysene, 2D12-benzo[b]fluoranthene, 2D -indeno[1,2,3-c,d]pyrene, and 2D -benzo[g,h,i]perylene). 12 12 The combined extracts from each sampler were reduced to about 1 mL using rotary evaporators. XAD and filter papers were spiked with the PAH internal standard and then Soxhlet extracted separately for 10 h in toluene. Both active and passive samples were then cleaned up using adsorption chromatography in a glass column (0.8 cm i.d.) filled with 2 g of activated silica. The column was first rinsed with DCM and then cyclohexane before the sample was added to the column. The first fraction was eluted with 10 mL of cyclohexane and discarded. The PAH fraction was eluted using of a DCM/cyclohexane mixture and concentrated to approximately 50 µL. The separation and quantification of the PAHs was performed using a Varian 3400 GC equipped with a Finnigan A200S liquid autosampler (splitless; injector temperature 295 °C; GC column: J&W DB-1, originally 20 m, 0.2 mm i.d., 0.33 µm film thickness; temperature program: 65 °C [isothermal 2 min], 20 °C min-1 to 295 °C [isothermal 10 min]) and coupled to a Finnigan SSQ 710 single stage quadrupole mass selective detector. The following PAHs were routinely quantified: naphthalene (Nap), acenaphthylene (Acy), acenaphthene (Ace), fluorene (Flo), phenanthrene (Phe), anthracene (Ant), fluoranthene (Flu), pyrene (Pyr), benz[a]anthracene (B[a]A), chrysene (Chr), benzo[b+k]fluoranthene (B[b,k]F), benzo[e]pyrene (B[e]P), benzo[a]pyrene (B[a]P), perylene (Per), indeno[1,2,3-c,d]pyrene (I[c,d]P), dibenz[a,h]anthracene (D[a,h]A), and benzo[g,h,i]perylene (B[g,h,i]P). Passive sam-
pler results were reported as the total amount of each PAH in a PSD (ng). Vapor phase air concentrations (ng m-3) were determined from high volume sampler results and reported as an average over 30, 60, and 90 days. Specific vapor phase and particle phase air concentrations are reported in Supporting Information (Table A). In some cases, the vapor phase concentration of the 5- and 6-ring PAHs were very close to the detection limits (see below), which increased the uncertainty associated with the values. QA/QC. The quantification criteria included confirmation of the retention times of the labeled standard and respective analyte. Routinely, the mass fragment with the highest intensity (base peak) was used for quantification. About 10% of the samples consisted of QC samples (i.e. field blanks, solvent blanks, and spikes). Field blanks were taken into the field when samplers were deployed and collected and then extracted and analyzed in the same manner as the deployed samples. The sample detection limits for individual compounds in a high volume air sample were defined as three times the analyte concentration measured in the field blank. Detection limits for the PSD samples were defined by the mean amount in the field blanks plus three times the standard deviation. These detection limits are reported in the Supporting Information (Table B). Where a compound could not be identified in the blank, the detection limit was set as three times the average noise peak area. Recoveries of the surrogate standards ranged from 32 to 88% in PSDs and between 80 and 120% in high volume air samples. This is within the acceptable criteria set by the U.S. EPA for trace analytical techniques using isotope dilution methods (13). It is however noteworthy that in contrast to many other trace analytical methods, a decrease in recovery with increasing molecular weight was observed.
Results and Discussion Reproducibility. Reproducibility was measured by calculating the mean normalized difference between PSD replicates (n ) 2). The % normalized differences were calculated for replicates A and B according to [(valueA - valueB)/((valueA + valueB)/2)]100. Reproducibility between PSD samples for compounds larger than Ace was very good, with the normalized difference rarely exceeding 25%. Due to the consistently poor reproducibility for Nap, Acy, and Ace in all samplers, the results for these analytes were excluded from further interpretation. For the high volume air sampler data, normalized differences were calculated from the previous deployment of two air sampling systems. The normalized differences between two replicate active air samples during this sampling replication were generally low and only exceeded 25% for Flu (27%). Characterization of Uptake Kinetics. The amount of individual PAHs that accumulated in the PSDs and also the vapor phase air concentration of these compounds are presented in Table 1. More than a microgram of Phe and Flu accumulated in the 200 µm PSDs that were exposed for 90 days, while between 100 and 500 ng of Flo, Pyr, B[a]A, and Chr accumulated. Polyaromatic hydrocarbons with 5 rings were dominated by B[b]F and B[k]F (not separately quantified thus reported as B[b,k]F) with up to almost 60 ng, whereas the other PAHs were present at smaller quantities. Perylene and D[a,h]A were not detected in any PSDs. Overall the PAH profiles in the PSDs and the active air samples appear to be similar. The compounds Flo through to Pyr were predominantly in the vapor phase (i.e. collected on the XAD), while the compounds B[b,k]F through to B[g,h,i]P were mostly associated with particles (i.e. collected on the GFF) (refer Supporting Information, Table A). Although variation in air concentrations was evident we have not attempted to develop more complex models to account for the effect of this variability on sampler behavior. At this stage we use the simple VOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2703
TABLE 1. Mean Vapor Phase Air Concentration (ng m-3) of PAHs Measured Using High Volume Sampling System and Amount of PAHs (ng PSD-1) Accumulated by PSDsa amount of PAH accumulated in passive sampler (ng PSD-1) 0-30 daysc vapor phase air concn (ng PAH
0-30
Flo Phe Ant Flu Pyr B[a]A Chr B[b,k]F B[e]P B[a]P Per I[c,d]P D[a,h]A B[g,h,i]P
daysc
1.5 2.5 0.3 0.9 0.8 0.04 0.06 0.02 0.01 nd nd nd nd 0.002
0-60
daysc
2.0 2.1 0.3 1.0 0.9 0.07 0.1 0.04 0.02 0.009 nd 0.005 nd 0.01
m-3)
0-90
daysc
1.9 2.5 0.3 1.0 0.9 0.06 0.1 0.03 0.01 0.008 nd 0.004 nd 0.01
100d PAH
1e
0-60 daysc
200d 2e
1e
Flo 71 42 120 Phe 390 310 740 Ant 59 49 88 Flu 340 330 420 Pyr 230 210 220 B[a]A 31 29 34 Chr 36 30 33 B[b,k]F 20 17 17 B[e]P 6 6 4 B[a]P 6 5 6 Per nd nd nd I[c,d]P 3 3 3 D[a,h]A nd nd nd B[g,h,i]P 10 12 10
100d 2e
1e
120 740 92 440 250 35 31 18 5 7 nd 4 nd 11
65 740 110 620 320 65 48 38 5 9 nd 8 nd 13
0-90 daysc
200d 2e
1e
100d 2e
1e
200d 2e
1e
2e
57 170 170 210 220 360 350 810 1600 1600 1300 1500 2300 2100 110 210 220 140 190 320 270 530 870 790 700 740 1300 1100 280 400 380 300 330 550 440 58 88 70 84 83 110 100 45 50 59 62 76 96 95 34 42 38 36 53 58 54 5 6 6 8 5 8 13 8 12 12 11 13 18 18 nd nd nd nd nd nd nd 8 7 8 9 10 8 10 nd nd nd nd nd nd nd 14 17 18 16 15 22 24
a Values for each replicate PSD are provided to show the variation between replicates. b nd: not detected (detection limits for perylene and indeno[1,2,3-cd]pyrene were 0.001 ng m-3 in high volume air samples and 1 ng in PSDs; refer to Supporting Information for a full list of detection limits). c Exposure. d Thickness (mm). e Replicates.
modeling approach already discussed to show whether this model approximates sampler behavior in the field. To evaluate the uptake kinetics using the multiple surface area-to-volume approach it is useful to select three compounds, Phe, Pyr, and I[c,d]P, which can represent the range of partitioning behavior of semivolatile organic compounds such as PAHs in the PSDs. The data demonstrates that the quantity of Phe accumulated in the 200 µm PSD at 30 and 60 days was approximately twice the amount sampled by the 100 µm PSD (Figure 2a and Table 1). This was also the case for Flo (at 30 days) and Ant (at 60 days). The results suggest that these PAHs have approached equilibrium in both the 100 and 200 µm PSDs within the deployment time. Thus the PSD is acting as an equilibrium sampler. This result is not surprising as these PAHs have low molecular weight and high vapor pressure relative to the larger PAHs and are therefore expected to approach equilibrium more quickly. Another interesting aspect shown in Figure 2a is the increase in the mean amount of Phe in the PSDs from 30 to 60 days. The data shows that the PSDs approached equilibrium at 30 days and also at 60 days. An increase in the amount of Phe in the PSDs would suggest that there was a corresponding increase in the atmospheric concentration of Phe. However the atmospheric concentration of Phe dropped slightly from 30 to 60 days. In this case it is likely that the decrease in mean air temperature (mean 30 day air temperature decreased from 22 °C to 18 °C) resulted in an increase in the sampler to air partition coefficient. In other words, the PSDs had a high “capacity” for these compounds at equilibrium. In contrast, PAHs with more than 3 rings did not approach equilibrium at any stage during the deployment. For example, the amount of Pyr that accumulated in the 100 µm and 200 µm PSDs was similar after 30 days. After 60 and 90 days we observe that the 200 µm PSD has accumulated about 30% and 60% more Pyr than the 100 µm PSDs (Figure 2b and Table 1). Hence the data indicates that for the first 30 days the uptake of Pyr could be described satisfactorily by a linear model; however, after 30 days the uptake of Pyr is entering a nonlinear phase. For I[c,d]P, one of the larger PAHs quantified in this study, the quantity in 100 µm and 200 µm PSDs is similar throughout the entire 90 day exposure period (Figure 2c). In accordance 2704
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004
with eq 3, we can conclude that for the entire deployment, uptake in both types of PSDs can be approximated by a linear uptake model. In other words, the PSD is acting as a kinetic passive sampler. Having characterized the stage of uptake for each of the target compounds, we can use air concentrations measured with the codeployed high volume air sampler to estimate sampling rates and equilibrium partition coefficients, with the following caveat. We have assumed that uptake from the atmosphere to the passive samplers was predominantly via the vapor phase, in accordance with modeling theory. This clearly applies to the lighter molecular weight PAHs which occur predominantly in the vapor phase. However larger PAHs are mostly bound to particles in the atmosphere, hence particle deposition to the passive samplers could form a significant component of uptake if particles adhere to the sampler surface. The following sampling rates for the PAHs, B[a]A through to B[g,h,i]P, would be significantly lower if calculated using the total air concentration (CV +CP). Further work is required to determine the significance of this process, hence the sampling rates reported in this study are only preliminary. Sampling rates were calculated for the PAHs, Flu through to B[g,h,i]P, using eq 5 and are reported in Table 2. Uptake for Flu and Pyr moved into the curvilinear phase of uptake during the deployment, so only PSDs collected at 30 days and the 0 to 30 day vapor phase air concentration was used. For B[a]A through to B[g,h,i]P (all of which were assumed to be undergoing linear uptake), PSDs deployed for 90 days and the mean vapor phase air concentration for 0 to 90 days was used. In all cases, data from the 200 µm PSDs were used in case uptake in the 100 µm PSD was moving into the curvilinear stage for any compounds (e.g. Flu and Pyr). Sampling rates ranged between 9 and 27 m3 d-1 and showed no distinct trend with molecular weight. An analysis of the sampling rates can indicate whether uptake was airside or sampler-side limited. If air-side resistance dominates uptake, then sampling rates are not directly related to KSV, whereas in the case of sampler-side resistance, sampling rates would increase, proportionally, with KSV (5). While sampling rates demonstrated little variation, KSV values for the corresponding compounds ranged over 3 orders of magnitude (see Table 3). The absence of a direct relationship between the sampling rates and KSV indicates that the resistance to
TABLE 2. Sampling Rates for 200 µm PSDs and Area Normalized Sampling Rates PAH
RS 200 (m3 d-1)
area normalized RS 200 (m3 d-1 dm-2)
Flu Pyr B[a]A Chr B[b,k]F B[e]P B[a]P I[c,d]P B[g,h,i]P
17a 11a 19b 11b 19b 9b 25b 27b 23b
0.9a 0.6a 1.1b 0.6b 1.1b 0.5b 1.4b 1.5b 1.3b
a Sampling rates determined over 30 days only. calculated over 90 days.
b
Remaining rates
TABLE 3. Log KOA Values Determined by Beyer et al. (14) for Certain PAHs and Corresponding log KSV Values Determined in This Study
a
FIGURE 2. a. Plot showing the mean amount of phenanthrene accumulated in 100 µm and 200 µm PSDs at 30, 60, and 90 days. At 30 days and 60 days, the mean amount accumulated by the 200 µm PSDs was approximately twice the mean amount accumulated by the 100 µm PSDs, suggesting equilibrium had been approached. b. Plot showing the mean amount of pyrene accumulated in 100 µm and 200 µm PSDs at 30, 60, and 90 days. After 30 days, the mean amount of pyrene accumulating in the 100 µm PSD is similar to the mean amount accumulated in the 200 µm PSD. The difference between the mean amount accumulating in the different PSDs increases with time suggesting that uptake is moving into the curvilinear stage. c. Plot showing the mean amount of indeno[1,2,3-cd]pyrene accumulated in 100 µm and 200 µm PSDs at 30, 60, and 90 days. The mean amount of indeno[1,2,3-cd]pyrene is similar in both PSDs for the duration of the deployment, suggesting that uptake remained in the linear phase. diffusion into the PSDs was air-side controlled for most of the PAHs. Although there are no other directly measured sampling rates for PAHs reported in the literature, rates for PCBs using SPMDs or PUF passive samplers are available for comparison (4). Rates reported by Shoeib and Harner (4) are, on the whole, significantly lower (2-9.9 m3 d-1). However, the
PAH
log KOA (from ref 14)
adjusted or predicted log KSV (25 °C)
Flo Phe Ant Flu Pyr B[a]A Chr B[a]P
6.76 7.58 7.70 8.63 8.70 9.85 10.21 11.15
6.3a 6.9a 7.2a 7.9b 8.0b 9.0b 9.3b 10.1b
Temperature adjusted to 25 °C.
b
Predicted from Figure 3.
surface areas of the SPMDs and PUF samplers used by Shoeib and Harner were 495 cm2 and 365 cm2, respectively, while the PSDs were approximately 1800 cm2. Once normalized for surface area, sampling rates for PSDs (0.5-1.5 m3 d-1 dm-2) compared closely to the rates reported by Shoeib and Harner (0.53-2.2 m3 d-1 dm-2) for PCBs with similar KOA values. Results for the compounds which approached equilibrium were used to estimate sampler to vapor phase partition coefficients (KSV) using eq 6. For calculating the KSV for these compounds, analyte concentration in the PSD (CS) was based on a mean concentration of all replicates of 100 µm and 200 µm samplers. Corrections were made for temperature, using the modified van’t Hoff equation presented by Beyer et al. (14). The adjusted log KSV values (25 °C) for Flo, Phe, and Ant (Table 3) were plotted against corresponding log KOA values (25 °C) calculated by Beyer et al. (14) (refer to Figure A, Supporting Information). A linear log-log relationship was obtained with
log KSV ) 0.890 log KOA + 0.216 (r2 ) 0.95)
(7)
Although this relationship was determined using 3 compounds and must be regarded as preliminary, eq 7 then allows for an estimation of log KSV values for compounds which did not approach equilibrium during the deployment and for which Beyer et al. (14) provides KOA values (Flu, Pyr, B[a]A, Chr, and B[a]P). These results are reported in Table 3 and show that for the range of compounds considered (3-5 ring PAHs) log KSV values ranged between 6.3 for Flo and 10.1 for B[a]P.
Future Developments The results from this study show that the use of passive samplers of multiple surface area-to-volume ratios can assist in the characterization of uptake kinetics for atmospheric VOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2705
PAHs and potentially other semivolatile organic compounds. Polyethylene sheets proved to be a simple, accessible and cost-effective passive sampling medium, easily adaptable to different surface area-to-volume ratios. In future work it should be possible to deploy two PSDs of low volume as equilibrium samplers for smaller, more volatile PAHs, while two thicker PSDs could also be deployed as kinetic samplers for larger PAHs. Such an approach could potentially reduce the need for lengthy deployments and also reduce the large amounts of samples associated with conventional passive sampling strategies by providing kinetic and equilibrium data at single time points.
Acknowledgments The authors thank Michael McLachlan for comments on the manuscript, James Huckins for discussions on passive sampling, and Neil Holling for training on GC-MS. We are also grateful to our reviewers for their constructive comments. This work and Ph.D. program was funded by an ARC SPIRT Linkage Grant, with industry support from Queensland EPA, Queensland Health Scientific Services, and ERGO. M.B. receives an APAI scholarship. Queensland Health provides funding for The National Research Centre for Environmental Toxicology.
Supporting Information Available Mean vapor phase and particulate phase air concentration of PAHs measured using high volume sampling system and the percent in the vapor phase (Table A), limits of detection (determined using field blanks) for the target PAHs in both high volume air samples and PSD samples (Table B), and linear regression of experimentally determined log KSV vs log KOA (Figure A). This material is available free of charge via the Internet at http://pubs.acs.org.
2706
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004
Literature Cited (1) Petty, J. D.; Huckins, J. N.; Zajicek, J. L. Chemosphere 1993, 27, 1609-1624. (2) Ockenden, W. A.; Sweetman, A. J.; Prest, H. F.; Steinnes, E.; Jones, K. C. Environ. Sci. Technol. 1998, 32, 2795-2803. (3) Mu ¨ller, J. F.; Hawker, D. W.; Connell, D. W.; Komp, P.; McLachlan, M. S. Atmos. Environ. 2000, 34, 3525-3534. (4) Shoeib, M.; Harner, T. Environ. Sci. Technol. 2002, 36, 41424151. (5) Harner, T.; Farrar, N. J.; Shoeib, M.; Jones, K. C.; Gobas, F. A. P. C. Environ. Sci. Technol. 2003, 37, 2486-2493. (6) Soderstrom, H. S.; Bergqvist, P. A. Environ. Sci. Technol. 2003, 37, 47-52. (7) Wania, F.; Shen, L.; Lei, Y. D.; Teixeira, C.; Muir, D. C. G. Environ. Sci. Technol. 2003, 37, 1352-1359. (8) Mayer, P.; Tolls, J.; Hermens, J. L. M.; Mackay, D. Environ. Sci. Technol. 2003, 37, 184A-191A. (9) Ockenden, W. A.; Corrigan, B. P.; Howsam, M.; Jones, K. C. Environ. Sci. Technol. 2001, 35, 4536-4543. (10) Huckins, J. N.; Tubergen, M. W.; Lebo, J. A.; Gale, R. W.; Schwartz, T. R. J. Assoc. Off. Anal. Chem. 1990, 73, 290-293. (11) Mu ¨ ller, J. F.; Hawker, D. W.; Wermuth, U. D.; Connell, D. W. Clean Air 1996, 30, 37-41. (12) South East Queensland Ambient Air Quality Monitoring Program; Air Services, Environmental Sciences Division, Queensland Environmental Protection Agency: 2003; electronic database. (13) Method 1625C Semivolatile Organic Compounds by Isotope Dilution GCMS; Office of Science and Technology Engineering and Analysis Division, U.S. EPA: 1989. (14) Beyer, A.; Wania, F.; Gouin, T.; Mackay, D.; Matthies, M. Environ. Toxicol. Chem. 2002, 21, 941-953.
Received for review August 10, 2003. Revised manuscript received December 4, 2003. Accepted February 11, 2004. ES0348849