Comparison of Four Active and Passive Sampling Techniques for

Apr 6, 2010 - Department of Physical and Environmental Sciences,. University of Toronto Scarborough, 1265 Military Trail,. Toronto, Ontario, Canada M1...
0 downloads 0 Views 536KB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 3410–3416

Comparison of Four Active and Passive Sampling Techniques for Pesticides in Air S T E P H E N J . H A Y W A R D , †,‡ T O D D G O U I N , †,§ A N D F R A N K W A N I A * ,†,‡ Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, Ontario, Canada M1C 1A4, and Department of Chemical Engineering & Applied Chemistry, University of Toronto, 200 College St. W, Toronto, Ontario, Canada M5S 3E5

Received August 20, 2009. Revised manuscript received March 14, 2010. Accepted March 26, 2010.

Four sampling systems were evaluated for their ability to determine the concentrations of pesticides in the atmosphere of rural southern Ontario. Two active air samplers (AAS, highvolume and low-volume pumps) and two passive air samplers (PAS, polyurethane foam disks and XAD-resin) were deployed between March 2006 and September 2007 using different sampling frequencies (biweekly to annually) and durations (24 h to 1 yr). Concentrations of nine pesticides in air determined by the different systems were compared at time scales of two weeks, two months, and one year. Agreement in the average concentrations obtained by different techniques improved with increasing length of the comparison period, especially for pesticides with high short-term temporal concentration variability. Such variability was high for the most volatile and reactive pesticides (trifluralin and pendimethalin). Except for these two pesticides, the annually averaged air concentrations determined by the different systems are within a factor of 2.5 for all pesticides and are not statistically different. Even though the PUF-PAS may have approached equilibrium with the atmosphere during deployment, the air concentrations are not statistically significantly different from those determined by AAS when averaged over longer time scales. Two month XADPAS deployments during the second summer resulted in sufficient sampling volumes to reliably establish air concentrations. If the sole purpose of collecting air samples is the assessment of long-term air concentration trends, this can be achieved most cost-effectively, i.e., with the least number of samples with year-long XAD-PAS.

Introduction Atmospheric monitoring programs have been established to assess levels of organic pollutants in air, their spatial variability, and temporal trends at various geographic scales (1-3). Currently, the majority of atmospheric monitoring programs rely on the use of active air samplers (AAS), * Corresponding author phone: +1-416-287-7225; e-mail: [email protected]. † University of Toronto Scarborough. ‡ University of Toronto. § Current address: Safety and Environmental Assurance Centre, Unilever, Colworth Science Park, Sharnbrook, Bedfordshire, United Kingdom MK44 1LQ. 3410

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010

deployed for short periods but in high frequency. Advantages of using AAS include the ability to provide reliable quantitative concentration data (4), information on gas/particle partitioning (5), and high temporal resolution (6). The latter is useful for assessing the influence of factors controlling shortterm concentration variability such as temperature and air mass origin (6). Disadvantages are relatively high maintenance and costs of operation and the need for a stable source of electricity. Conversely, passive air samplers (PAS) are inexpensive and do not require electricity, making them better suited for providing high spatial resolution data, especially when sampling is done in remote areas. PAS have been used to study seasonal and spatial trends of semivolatile organics in the atmosphere (7, 8). Because PAS have lower air sampling rates than AAS, and therefore lower temporal resolution, they are often used in studies where long sampling periods are desired (9). Although the usage of PAS is increasing, to date there have been only a limited number of comparisons between different types of AAS and PAS (8, 10-12), and often the comparisons are between only two types of samplers, for example when an AAS is used to calibrate the data from a PAS. The aim of this study is to provide a quantitative performance comparison among different types of air sampling techniques, including active and passive sampling systems in an agricultural source region in southern Ontario. Four different types of air samplers are assessed in terms of quantifying concentrations of pesticides in air. Specifically, XAD-resin and polyurethane foam (PUF) are evaluated as PAS media, and combinations of both sampling media are used in high- and low-volume AAS. There are a number of studies that have recently investigated levels of pesticides in air at several sites located throughout the Great Lakes region (8, 13-16). With the exception of one study deploying PAS throughout southern Ontario (8), most studies relied on high-volume (HV) AAS. Currently, however, there is a lack of understanding of how different sampling methods perform with respect to quantifying the long-term average concentrations of pesticides and their temporal variability. Pesticides should be particularly suited for assessing the comparability among different sampling techniques because their episodic application results in strong temporal air concentration variability on a seasonal and day-to-day scale. This variability presents a particular challenge to achieving consistent results from different techniques.

Methods Air was sampled at the Centre for Atmospheric Research Experiments (CARE), in Egbert, Ontario, Canada (44°13′52′′ N, 79°46′59′′ W) from March 2006 to September 2007 using two AAS and two PAS techniques. The sampling period was chosen to include two pesticide application seasons and one winter. The overall deployment strategy was designed to take advantage of the characteristics of each system. Active Air Sampling. Two different types of AAS were deployed in two week intervals for 18 months, resulting in 31 sampling events. Specifically, a 24 h HV-AAS sample was collected once every two weeks, whereas a low-volume (LV) AAS was operated continuously for each two week period. The HV-AAS utilized a PS1 sampler module (Tisch Environmental); air was aspirated through a glass fiber filter (GFF) placed in front of a PUF-XAD-PUF sandwich (10 g XAD, between two 3 cm × 6 cm PUF) at a calibrated sampling rate of 372 ( 6 m3 per 24 h. Additionally, a BGI-400S LV-AAS (BGI, 10.1021/es902512h

 2010 American Chemical Society

Published on Web 04/06/2010

Inc.) was used to aspirate air through a PUF-XAD-PUF sandwich (5 g of XAD, between 2 cm × 3 cm PUF), but without GFF to separately capture particles. No analytes were found consistently on the HV-AAS GFF, indicating that the majority of the pesticides were in the gas phase. The LV-AAS was calibrated to sample 2.9 ( 0.2 m3 d-1 resulting in a sample volume of 40.6 m3 for each two week sample. Passive Air Sampling. Both PAS techniques used in this work operate in the linear uptake phase for semivolatile organic compounds (17) and aim to provide time-averaged air concentrations. If a PAS is retrieved before chemical equilibrium between the atmospheric gas phase and the sampling medium is approached, the average air concentration (CAir) over the entire sampling deployment (t) may be derived from the sampling rate (R) of the PAS and the amount of chemical sorbed by the PAS (mPAS). CAir )

mPAS Rt

(1)

Equation 1 assumes that during the linear uptake phase the rate of analyte desorption from the sampling medium is very slow when compared to the rate of uptake, which means that the concentration close to the sampling medium is assumed to be zero. The time to reach equilibrium depends on the capacity of the sampling medium for the target chemicals, which in turn depends on temperature-dependent sorption coefficients of the analyte to the sampling medium. PAS utilizing XAD-resin and PUF disks have equilibration times for semivolatile organic contaminants on the order of years and months, respectively. The larger the sorption capacity of the PAS, the less chance that any of the analytes will reach equilibrium, even if the equilibration times differ between analyte chemicals (17). The low sampling rates of PAS, however, necessitate much longer sampling periods than for the AAS. Duplicate PUF-disk PAS (PUF-PAS) were deployed in sequential two month intervals throughout the 18 month study. These PUF-PAS consist of a precleaned PUF disk (diameter 14 cm, thickness 1.35 cm; Pacwill Environmental, Stoney Creek, ON) placed in a shelter consisting of two hemispherical stainless steel containers, providing shelter from the elements but allowing air circulation (18). Although the shelter greatly reduces the effective wind speed near the PUF disk, the sampling rate R of the PUF-PAS is somewhat dependent on wind speed (19). The use of depuration compounds (DCs) allows for the determination of R values that apply to a specific location and sampling period (20); prior to deployment PUF disks were spiked with a suite of DCs. Details of the PUF-PAS sampling rate determination can be found in the Supporting Information. In the current study, R for the PUF-PAS averaged 5.6 ( 1.0 m3 d-1 and ranged from 4.3 to 6.8 m3 d-1. This is slightly higher than the R previously reported for PUF-PAS at Egbert (16) but is consistent in that the R value reported for this site is typically higher than averages reported for the PUF-PAS, which may be related to high wind speeds at this site (9). One pair of XAD-PAS were deployed in March 2006 and retrieved in February 2007. In the second year (February 27 to September 15, 2007), XAD-PAS samples were deployed in three consecutive two month intervals, just like the PUFPAS. The XAD-PAS, based on the design described in ref 17, consists of a stainless steel mesh cylinder containing precleaned XAD-2 resin, which is protected from precipitation by a stainless steel housing designed to minimize the effect of wind speed on the sampling rate. The sampler housing and sorbent resin chamber are approximately 8 cm in height, shorter than the original design (17). The shorter design utilizes approximately half as much resin (approximately 10 g), which can now be extracted in a Soxhlet apparatus without removal from the mesh container.

Recent evidence suggests that the sampling rate R for the XAD-PAS is dependent on the chemical and on the temperature at the sampling site (12). The original calibration which determined an R of 0.52 m3 d-1 was performed in the Arctic, while temperate locations showed much more variable sampling rates (17). A recent calibration experiment (12), along with the original calibration data (17), show that R increases with decreasing molecular size and increasing temperatures. Both studies used 20 cm long XAD-PAS filled with approximately 20 g of XAD-2 resin. In this study, by only using half as much XAD-2, with half the total surface area available for chemical sorption, our sampling rates should be half of those in the previous studies. Sampling rates for the XAD-PAS in this study were obtained by dividing by two the average of the sampling rates derived for a specific compound in previous studies (12, 17). If no such R was available for a compound, the average for all compounds was used. The full data set can be found in Table S1 of the Supporting Information. Time to Equilibrium of PAS. To demonstrate that equilibrium has not been attained, it is possible to estimate a pesticide’s partition coefficient (K′, L g-1) between XAD-2 and air (K′XAD-2/air) (description in the Supporting Information), which in turn can be used to estimate each pesticide’s time to equilibrium between PAS and the atmospheric gas phase (t95) (21) (Table S9 and Figure S2b of the Supporting Information). HCB has the lowest K′XAD-2/air of the analytes in this study, and the shortest t95 (1.6 × 106 d). As this time is much longer than our longest deployment of one year, this confirms that HCB is not reaching equilibrium with the XAD-resin and that the other pesticides would be even less likely to approach equilibrium during the deployment. Furthermore, we can estimate the period of linear uptake for the pesticides (t25) (21) on XAD-2, which is >105 d for all pesticides, confirming that during a one year deployment the XAD-PAS remain in the linear uptake range. It is also possible to estimate sorption to PUF (K′PUF/air, L g-1) (description in the Supporting Information). The PUF has a significantly lower K′PUF/air for all of the pesticides than the XAD-2 (Figure S3a of the Supporting Information). The estimated time until equilibration with the atmosphere for the PUF (t95) is as short as 53d for HCB and less than 100 d for trifluralin. With deployments of approximately 60 d in Egbert, the PUF-PAS may not be operating exclusively in the linear uptake region for such chemicals. It is possible to estimate a reduced, effective PUF-PAS sampling volume when equilibrium is approached (18), but this estimation introduces significant uncertainty because of discrepancies in the maximum uptake capacity of a PUF estimated by different methods (calculations shown in Table S5 of the Supporting Information). For this reason, the sampling volume calculated from the loss of the DCs during deployment was not reduced to account for the possibility of nonlinear uptake.

Results Air Concentrations Determined by Different Sampling Techniques. Nine pesticides were consistently detected in the samples taken by all four sampling systems and are therefore used in the assessment of the comparability of the techniques. They include the insecticides endosulfan and disulfoton, the herbicides trifluralin, pendimethalin, atrazine, alachlor, and metolachlor, and the fungicides chlorothalonil and hexachlorobenzene (HCB, a banned pesticide, which continues to be found in the atmosphere for reasons possibly other than its use as pesticide). The amounts of these nine pesticides in the extracts of the XAD-PAS and PUF-PAS are given in Tables S2 and S4 of the Supporting Information, respectively. Tables S6 and S7 of the Supporting Information give the volumetric air concentrations that have been derived from those data on the basis of estimated sampling rates R. VOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3411

FIGURE 1. Bland-Altman plots of active air sampling systems for two week sampling periods at Egbert, Ontario during March 2006 to September 2007. Solid horizontal line indicates the mean difference between AAS samples, while dashed horizontal lines are 95% confidence intervals ((1.96 std. dev. of AAS differences). Tables S8 and S9 of the Supporting Information list the volumetric air concentrations obtained with the LV-AAS and HV-AAS techniques, respectively. Figure S2 of the Supporting Information displays the air concentration data as a function of time. Concentrations increase strongly during the warm growing season and fall below detection limits during the winter for a number of the pesticides. The timing of the concentration peaks is not the same for all pesticides, however, which is likely due to the different application periods associated with each compound. This is consistent with previous observations of pesticide concentrations in the same region (8). Comparison of the Two Active Sampling Systems at High Temporal Resolution. To compare the air concentrations determined by the two AAS during 2006-2007, we show the data for each two week sampling period in a Bland-Altman plot (22) in Figure 1. The average of the air concentrations determined by both AAS for each sampling period is plotted on the x-axis, while the difference between the two air concentrations (∆CAAS) determined by each AAS is plotted on the y-axis. Caverage ) (CHV-AAS + CLV-AAS)/2

(2)

∆CAAS ) CHV-AAS - CLV-AAS

(3)

The arithmetic mean average difference between paired air concentrations ∆CAAS (mean) and the 95% confidence interval (1.96 times the standard deviation of ∆CAAS) may also be calculated, which estimate the average agreement between AAS and the variability between all paired samples, 3412

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010

respectively (Figure 1, Table S11 of the Supporting Information). To compare among the pesticides, these values are also normalized to the mean air concentrations during the growing period. Although the air concentrations measured by both AAS show large variation for all chemicals, ∆CAAS (mean) is only significantly different from zero for pendimethalin and alachlor (two-tailed t test, p ) 0.05). Such differences may be attributed to the systems’ sampling duration. For instance, if an air concentration spike occurs, most likely in response to the application of a pesticide, on a day when no HV-AAS sample is collected, the LV-AAS concentration is likely to be higher than the HV-AAS. If, however, such a spike happens to occur on the day of HV-AAS sampling, it is likely to give higher concentrations than the LV-AAS. Although the data in Figure 1 show that both situations may occur, whenever ∆CAAS (mean) is different from zero (pendimethalin and alachlor), it is less than zero, indicating a higher probability that the HV-AAS misses the application of these pesticides. This is particularly true for pendimethalin, where all data points have ∆CAAS < 0. Air Concentration Variation between Subsequent Active Air Samples. The temporal profile of the air concentrations determined by the two AAS often show significant variability from one sampling period to the next (Figure S2 of the Supporting Information). This variability or “spikiness” may reveal important aspects of a pesticide’s behavior in the atmosphere. In particular, the extent of spikiness should be related to the frequency and variability of a pesticide’s application during the study period as well as its atmospheric residence time, which in turn is governed by its physical-

chemical properties. The spikiness in an AAS data series can be quantified by the Active Sampling Variation (ASV)

Caverage )

n

∑ |C i

ASV )

∑n ∑ν

AAS

(5)

AAS

Ci+1 |

i)0

where n is the amount of pesticide detected in each AAS, and v is the volume of air sampled by each AAS over the averaged time period. In Figure 2, the air concentrations determined by PUFPAS, XAD-PAS (summer 2007), and HV-AAS (2-month averages) are compared with the LV-AAS two month average air concentrations in Bland-Altman plots. When averaging the AAS data over two month intervals, the HV-AAS air concentrations show much better agreement with the LVAAS air concentrations for all pesticides except chlorothalonil and disufoton, where the agreement is consistent at both time scales. (Table S11 of the Supporting Information). For all other pesticides, the confidence intervals decrease by at least 30%. In general, with increasing ASVHV/ASVLV, we see a larger improvement in the confidence intervals (i.e., larger ∆C.I., Table S11 of the Supporting Information), calculated using eq 6

(4)

nCavg

where Ci is the air concentration during a given sampling period, Ci+1 is the air concentration during the following sampling period, and n is the number of sampling periods in the calculation (May-September, where both Ci and Ci+1 are greater than detection limits). Cavg is the average air concentration during the n sampling periods. As we attempt to compare the ASV between two systems (HV- and LV-AAS), Ci and Ci+1 must be greater than detection limits for both sampling systems. The data for disulfoton did not meet these criteria and thus was excluded from this comparison. Table 1 lists the ASV of the HV- and LV-AAS data series, along with the pesticides’ vapor pressures (PL, Pa) and atmospheric degradation half-lives (t1/2, hr). Although the ASV values for different pesticides cover a wide range (0.21-0.84 for the HV-AAS data series; 0.21-0.63 for LV-AAS data), they correlate strongly with neither PL nor t1/2, partially because the two properties have opposite effects on atmospheric residence time but also because the frequency of pesticide application and the variability in application rate are unquantifiable confounding factors. As PL increases, a larger portion of a chemical will remain in the atmosphere and be less susceptible to wet and dry deposition or particle sorption and fallout. At the same time, as t1/2 decreases, a chemical is removed more quickly from the atmosphere by reaction with OH radicals. However, different combinations of t1/2 and PL may lead to similar ASV ratios. The ratio between the ASVs of the two active sampling systems, ASVHV/ASVLV (Table 1), helps to identify chemicals that have higher variability in one AAS. Only trifluralin and pendimethalin have an ASVHV/ASVLV ratio significantly different from unity. Both have short atmospheric residence times and high PL, meaning that there may be high air concentrations immediately after application events, which quickly decrease in the following days. The ASV ratio of pendimethalin lends further weight to the theory that the HV-AAS was not always deployed during application of this pesticide. For all other pesticides, the ASVHV/ASVLV ratio is within 30% of 1.0 because these chemicals have either short t1/2 or high PL but not both. Comparison at the Scale of Two Months. The air concentrations determined by the AAS were averaged over two month periods using the equation

∆C.I. ) (3.92σtwo week)-(3.92σtwo month)

(6)

where σtwo week and σtwo month are the standard deviations of ∆CAAS at the respective time scales. Conceptually this makes sense as short-term variations in air concentrations are reduced as the data are averaged over longer time periods, so pesticides with greater ASVHV/ASVLV will see the most improvement in agreement between the two AAS. Although the correlation between AAS systems improved for pendimethalin, the HV-AAS still significantly underestimates the air concentration (the only chemical where all ∆CAAS < 0). This may be due in part to the sampling strategy during the first year, where the HV-AAS was deployed at the end of the work week. This systematic deployment may have led to the HV-AAS “missing” the presence of this herbicide in the atmosphere, as it has a short atmospheric residence time and could have been applied systematically earlier in the week, leading to lower air concentrations determined by the HV-AAS. The air concentrations determined by the two PAS show relatively good agreement with the air concentrations determined by the LV-AAS. For the XAD PAS, the normalized ∆CXAD-LVAAS for the summer 2007 samples are less than (0.5, i.e., the average difference between the two systems is less than 50% of the summer time average air concentration. For the PUF-PAS, the normalized ∆CPUF-LVAAS > -0.5 (except for pendimethalin and trifluralin), but the values are always

TABLE 1. Average Air Concentrations for Pesticides in Active Sampling Systems, during the 2006 and 2007 Growing Seasons (May-September) at Egbert, Ontarioa average, both systems (pg m-3) pendimethalin trifluralin endosulfan HCB metolachlor alachlor chlorothalonil atrazine

34.4 ( 12.7 42.3 ( 14.1 94.7 ( 39.1 58.5 ( 17.8 561 ( 78 468 ( 99 944 ( 427 597 ( 48

range, both systems (pg m-3) 3.6-91.3 4.8-120.8 1.8-321 0.07-133 145-822 125-900 8.5-2845 352-823

PL(Pa) -2 b

2.7 × 10 9.8 × 10-3 b 8.9 × 10-3 c 9.4 × 10-3 c 1.7 × 10-3 b 2.9 × 10-3 b 1.3 × 10-2 b 1.3 × 10-3 b

t1/2 (h) d

2.9 0.35e 12g 3.8 × 108 h 2.5f 2.25f 528h 2.6h

ASVHV (n samples)

ASVLV (n samples)

ASVHV/ASVLV

0.52 (12) 0.84 (17) 0.74 (17) 0.37 (12) 0.40 (11) 0.52 (11) 0.43 (17) 0.21 (12)

0.21 (12) 0.40 (17) 0.63 (17) 0.32 (12) 0.37 (11) 0.51 (11) 0.54 (17) 0.29 (12)

2.4 2.1 1.2 1.2 1.1 1.0 0.80 0.72

a Vapor pressure (PL) atmospheric degradation half-life (t1/2), active sampling variation (ASV) for HV-AAS and LV-AAS systems, and the ratio of ASVHV/ASVLV. b Literature-derived values (LDV) compiled by Muir et al. (23). c Final adjusted values (FAV) compiled by Shen et al. (24). d Moza et al., 1992, estimated from photolysis half-life in aqueous solution. (25). e Woodrow et al., 1978, calculated in air. (26). f Tanaka et al., 1981, estimated from photolysis half-life in aqueous solution. (27). g Muir, 2004, estimated from OH radical reaction rate. (23). h Atkinson, 1987, estimated from calculated OH radical reaction rate. (28). h Millet et al., 1998, estimated from photolysis half-life in aqueous solution. (29).

VOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3413

FIGURE 2. Bland-Altman plots of air sampling systems for two month sampling periods at Egbert, Ontario, during March 2006-September 2007. For each axis, i, AS refers to either (() HV-AAS, (O) PUF PAS, or (×) XAD PAS.

FIGURE 3. Annual average air concentrations using four types of air sampling systems for trifluralin, pendimethalin, endosulfan I, hexachlorobenzene (HCB), and disulfoton (left axis), and chlorothalonil, atrazine, alachlor, and metolachlor (right axis), during March 2006-February 2007 at Egbert, Ontario. negative, reflecting the fact that the PUF-PAS tends to underestimate the air concentrations relative to the LV-AAS. The relative short time to equilibrium may help explain the relatively poor agreement between the PUF-PAS and the other sampling techniques for trifluralin and pendimethalin, whereby the PUF-PAS may be approaching equilibrium and no longer be operating in the linear uptake region, unlike the results shown for the XAD-PAS. Comparison at the Time Scale of One Year. Figure 3 compares the annually averaged air concentrations obtained with the different sampling techniques. For the PUF-PAS, LV-AAS, and HV-AAS techniques, the confidence intervals in Figure 3 are an indication of temporal variability, whereas in the case of the XAD-PAS, it is the range obtained by duplicates. The annual averages agree within a factor of 2.5, except for trifluralin and pendimethalin, which vary by a factor of 3.5 and 3.1, respectively. On an annual time scale, the differences in concentrations among sampling systems 3414

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010

are further reduced, as the data are averaged over much longer periods. Because the data cannot be assumed to distribute normally, a Kruskal-Wallis one-way (independent) analysis of variance test was performed, which does not require a normal distribution. At the 95% confidence level (i.e., p g 0.05), it was determined that neither the HV-AAS, LV-AAS or PUF-PAS air concentrations were significantly different from the mean for each pesticide (average concentration of these three systems). Because the annually averaged XAD-PAS air concentrations in Figure 3 are based on a single pair of replicate samples, it is not possible to statistically determine if the air concentrations are different from those determined by the other systems. However, except for trifluralin and pendimethalin, the XAD-PAS air concentrations appear to fall within the range of confidence intervals for the other sampling systems. The lower concentrations determined by the XAD-PAS systems may be due to degradation loss of both pesticides from the XAD-PAS (12).

Recommendations. This study has helped demonstrate the strengths and weaknesses of a variety of atmospheric sampling techniques. In determining long-term averaged air concentrations, there appears to be no systematic bias introduced by any one of the sampling techniques (Figure 3), with results being consistent among each of the methods. This suggests that when the purpose of collecting air concentration data is limited to determining long-term trends, such as assessing the efficacy of international regulatory instruments at reducing environmental levels, it should be achievable with minimal effort using XAD-PAS deployed for a year. The number of samples that are used in the derivation of the annual averages displayed in Figure 3 range from 2 (XAD-PAS) to 21 (HV- and LV-AAS). For information about environmental levels on shorter than annual time scales, such as seasonal trends, both PAS were typically able to sequester quantifiable amounts of pesticides from the atmosphere within two months, particularly during the application period, and are thus able to provide temporal data related to usage patterns. For some of the pesticides, the PUF-PAS appeared less reliable on a bimonthly basis, possibly because the PUF approached equilibrium with the atmospheric gas phase for some analytes and was therefore not sampling only in the linear uptake range. To compensate for problems associated with the lower capacity of the PUF, it is suggested that either shorter deployment periods (i.e., monthly) be considered when sampling at high temperature or when targeting analytes that may approach equilibrium quickly, or an effective reduced air sampling volume should be estimated, assuming the maximum uptake capacity of the PUF for a particular compound can be reliably determined at the temperature of deployment. Where atmospheric levels are below detection limits for a given deployment period, more sensitive analytical methods will need to be employed. Where temporal resolution shorter than one month is desired, the best choice is an AAS system. Overall there is good agreement between both types of AAS, but if information on pesticide air concentration at higher than seasonal resolution is sought, continuous sampling with an AAS may be a better choice than episodic sampling with a HV-AAS. This is because episodic sampling may be missing periods of high air concentrations, especially if a schedule of sampling on fixed weekdays is adopted and the target analyte is lost from the atmosphere relatively quickly. Surprisingly, continuous AAS appears to be the technique least commonly deployed in North America. However, where a study aims to focus on short episodic events, air concentration maxima, or the mechanisms of atmospheric transport, short-deployment HV-AAS remains the most suitable system.

Acknowledgments We are grateful to the Natural Sciences and Engineering Research Council of Canada and the Canadian Foundation for Climate and Atmospheric Sciences for funding.

Supporting Information Available Tables showing the amounts of pesticides sampled by each technique, the volumetric air concentrations determined by each technique, the compound-specific R values for the XADPAS, compilation of pesticide solvation parameters and K′ for XAD and PUF, specific partition coefficients and t95 values for select pesticides on XAD-PAS and PUF-PAS, average air concentration difference ∆CAAS between HV-AAS and LVAAS for samples at two week intervals and for averages of samples during two month intervals, and improvement in confidence interval ∆C.I. Figures showing the air concentration for selected pesticides at Egbert and the vapor pressure of select pesticides plotted against the ratio of ASVHV/ASVLV. Details on the determination of specific partition coefficients,

sample preparation and analysis, PUF-PAS sampling rate determination. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Meijer, S. N.; Sweetman, A. J.; Halsall, C. J.; Jones, K. C. Temporal trends of polycyclic aromatic hydrocarbons in the UK atmosphere: 1991-2005. Environ. Sci. Technol. 2008, 42, 3213–3218. (2) Hung, H.; Blanchard, P.; Halsall, C. J.; Bidleman, T. F.; Stern, G. A.; Fellin, P.; Muir, D. C. G.; Barrie, L. A.; Jantunen, L. M.; Helm, P. A.; Ma, J.; Konoplev, A. Temporal and spatial variabilities of atmospheric polychlorinated biphenyls (PCBs), organochlorine (OC) pesticides, and polycyclic aromatic hydrocarbons (PAHs) in the Canadian Arctic: Results from a decade of monitoring. Sci. Total Environ. 2005, 342, 119–144. (3) Buehler, S. S.; Basu, I.; Hites, R. A. Gas-phase polychlorinated biphenyl and hexachlorocyclohexane concentrations near the Great Lakes: A historical perspective. Environ. Sci. Technol. 2002, 36, 5051–5056. (4) Dobson, R.; Scheyer, A.; Rizet, A. L.; Mirabel, P.; Millet, M. Comparison of the efficiencies of different types of adsorbents at trapping currently used pesticides in the gaseous phase using the technique of high-volume sampling. Anal. Bioanal. Chem. 2006, 386, 1781–1789. (5) Scheyer, A.; Morville, S.; Mirabel, P.; Millet, M. Gas/particle partitioning of lindane and current-used pesticides and their relationship with temperature in urban and rural air in Alsace region (east of France). Atmos. Environ. 2008, 42, 7695–7705. (6) Hoff, R. M.; Muir, D. C. G.; Grift, N. P. Annual cycle of polychlorinated-biphenyls and organohalogen pesticides in air in southern Ontario. 1. Air concentration data. Environ. Sci. Technol. 1992, 26, 266–275. (7) Motelay-Massei, A.; Harner, T.; Shoeib, M.; Diamond, M.; Stern, G.; Rosenberg, B. Using passive air samplers to assess urbanrural trends for persistent organic pollutants and polycyclic aromatic hydrocarbons. 2. Seasonal trends for PAHs, PCBs, and organochlorine pesticides. Environ. Sci. Technol. 2005, 39, 5763– 5773. (8) Gouin, T.; Shoeib, M.; Harner, T. Atmospheric concentrations of current-use pesticides across south-central Ontario using monthly-resolved passive air samplers. Atmos. Environ. 2008, 42, 8096–8104. (9) Harner, T.; Pozo, K.; Gouin, T.; Macdonald, A. M.; Hung, H.; Cainey, J.; Peters, A. Global pilot study for persistent organic pollutants (POPs) using PUF disk passive air samplers. Environ. Pollut. 2006, 144, 445–452. (10) Gouin, T.; Harner, T.; Blanchard, P.; Mackay, D. Passive and active air samplers as complementary methods for investigating persistent organic pollutants in the Great Lakes basin. Environ. Sci. Technol. 2005, 39, 9115–9122. (11) Waite, D. T.; Droneck, N. R.; Tuduri, L.; Sproull, J. F.; Chau, D. F.; Quiring, D. V. Comparison of active versus passive atmospheric samplers for some current-use pesticides. Bull. Environ. Contam. Toxicol. 2005, 74, 1011–1018. (12) Gouin, T.; Wania, F.; Ruepert, C.; Castillo, L. E. Field testing passive air samplers for current use pesticides in a tropical environment. Environ. Sci. Technol. 2008, 42, 6625–6630. (13) James, R. R.; Hites, R. A. Chlorothalonil and Dacthal in Great Lakes air and precipitation samples. J. Great Lakes Res. 1999, 25, 406–411. (14) Peck, A. M.; Hornbuckle, K. C. Gas-phase concentrations of current-use pesticides in Iowa. Environ. Sci. Technol. 2005, 39, 2952–2959. (15) Yao, Y.; Harner, T.; Blanchard, P.; Tuduri, L.; Waite, D.; Poissant, L.; Murphy, C.; Belzer, W.; Aulagnier, F.; Sverko, E. Pesticides in the atmosphere across Canadian agricultural regions. Environ. Sci. Technol. 2008, 42, 5931–5937. (16) Yao, Y.; Tuduri, L.; Harner, T.; Blanchard, P.; Waite, D.; Poissant, L.; Murphy, C.; Belzer, W.; Aulagnier, F.; Li, Y. F.; Sverko, E. Spatial and temporal distribution of pesticide air concentrations in Canadian agricultural regions. Atmos. Environ. 2006, 40, 4339– 4351. (17) Wania, F.; Shen, L.; Lei, Y. D.; Teixeira, C.; Muir, D. C. G. Development and calibration of a resin-based passive sampling system for monitoring persistent organic pollutants in the atmosphere. Environ. Sci. Technol. 2003, 37, 1352–1359. (18) Harner, T.; Shoeib, M.; Diamond, M.; Stern, G.; Rosenberg, B. Using passive air samplers to assess urban-rural trends for persistent organic pollutants. 1. Polychlorinated biphenyls and organochlorine pesticides. Environ. Sci. Technol. 2004, 38, 4474– 4483. VOL. 44, NO. 9, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3415

(19) Tuduri, L.; Harner, T.; Hung, H. Polyurethane foam (PUF) disks passive air samplers: Wind effect on sampling rates. Environ. Pollut. 2006, 144, 377–383. (20) Moeckel, C.; Harner, T.; Nizzetto, L.; Strandberg, B.; Lindroth, A.; Jones, K. C. Use of depuration compounds in passive air samplers: Results from active sampling-supported field deployment, potential uses, and recommendations. Environ. Sci. Technol. 2009, 43, 3227–3232. (21) Shoeib, M.; Harner, T. Characterization and comparison of three passive air samplers for persistent organic pollutants. Environ. Sci. Technol. 2002, 36, 4142–4151. (22) Bland, J. M.; Altman, D. G. Measuring agreement in method comparison studies. Stat. Methods Med. Res. 1999, 8, 135–160. (23) Muir, D. C. G.; Teixeira, C.; Wania, F. Empirical and modeling evidence of regional atmospheric transport of current-use pesticides. Environ. Toxicol. Chem. 2004, 23, 2421–2432. (24) Shen, L.; Wania, F. Compilation, evaluation, and selection of physical-chemical property data for organochlorine pesticides. J. Chem. Eng. Data 2005, 50, 742–768.

3416

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 9, 2010

(25) Moza, P. N.; Hustert, K.; Pal, S.; Sukul, P. Photocatalytic decomposition of pendimethalin and alachlor. Chemosphere 1992, 25, 1675–1682. (26) Woodrow, J. E.; Crosby, D. G.; Mast, T.; Moilanen, K. W.; Seiber, J. N. Rates of transformation of trifluralin and parathion vapors in air. J. Agric. Food Chem. 1978, 26, 1312–1316. (27) Tanaka, F. S.; Wien, R. G.; Mansager, E. R. Survey for surfactant effects on the photodegradation of herbicides in aqueous-media. J. Agric. Food Chem. 1981, 29, 227–230. (28) Atkinson, R. A structure-activity relationship for the estimation of rate constants for the gas-phase reactions of OH radicals with organic compounds. Int. J. Chem. Kinet. 1987, 19, 799– 828. (29) Millet, M.; Palm, W. U.; Zetzsch, C. Abiotic degradation of halobenzonitriles: Investigation of the photolysis in solution. Ecotoxicol. Environ. Saf. 1998, 41, 44–50.

ES902512H