Temperature Dependence of Atmospheric PCB ... - ACS Publications


Temperature Dependence of Atmospheric PCB...

2 downloads 102 Views 237KB Size

Environ. Sci. Technol. 2005, 39, 740-747

Clapeyron slope (a1, in K) is reported instead of ∆HSA in order to avoid giving thermodynamic meaning to this parameter. For consistency, we will use this convention here. Clearly, the value of a1 for a given compound at a given site can be obtained from a linear regression of ln P versus 1/T.

Temperature Dependence of Atmospheric PCB Concentrations DANIEL L. CARLSON AND RONALD A. HITES* School of Public and Environmental Affairs, Indiana University, Bloomington, Indiana 47405

An analysis of data from the U.S. Integrated Atmospheric Deposition Network (IADN) sites near the Great Lakes and a review of the literature shows that the temperature dependence of atmospheric PCB concentrations cannot be used to distinguish sites dominated by long-distance transport from those with local sources. We observe that calculations based on data sets with only ∼25 measurements over a period of 1 year are unreliable indicators of the longterm temperature dependence at a given location, that temperature independence occurs at temperatures below freezing, and that low PCB concentrations can bias analyses toward a weaker temperature dependence. After accounting for these factors, a similar temperature dependence is observed at all IADN sites, and this temperature dependence increases for PCBs with increasing chlorination. Although previously described models do fit the data, the resulting calculated constants are not meaningful. Instead, a simple new model in which temperatures < 273 K are set equal to 273 K, and the Clausius-Clapeyron equation is used, can account for observed temperature dependence phenomena.

Introduction Over the past several decades, atmospheric concentrations of PCBs have been measured around the world in an attempt to understand the behavior of these persistent contaminants in the environment. In the Great Lakes region of North America, measurements have been made at several Integrated Atmospheric Deposition Network (IADN) sites every 12 days for the past 14 years (1). With this wealth of data now available, a more detailed picture is emerging of how PCBs cycle between air, water, and land. The atmospheric concentrations of PCBs and other semivolatile organic compounds have been shown to depend on atmospheric temperature, and this dependence can be modeled by the Clausius-Clapeyron (C.C.) equation

ln P ) a0 +

∆HSA a1 ) a0 + RT T

740

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005

Since these studies were published, more PCB data have become available, including nearly twice as much data from IADN (4-40). In this paper, we examine the temperature dependence of atmospheric PCB concentrations at IADN sites in detail and analyze the wealth of literature reports. The results show that the range of calculated C.C. slopes is not due to local vs long-range PCB sources, as suggested by the Hoff and Wania models. The IADN data alone contradicts this idea because all of the sites, both urban and remote, have essentially the same C.C. slope. Instead, we propose several alternate explanations for the observed variations in the literature C.C. slopes: First, much of the variability is due to small data sets; it appears that one needs > 100 atmospheric temperature-concentration pairs to get reasonable results. Second, C.C. slopes vary depending on the range of temperatures included, decreasing as more low-temperature data are included. Third, C.C. slopes are not reliable for congeners or PCB mixtures present at very low levels; random noise added to a PCB signal can reduce the C.C. slopes by a factor of ∼2. Fourth, we find that C.C. slopes for PCB congeners vary systematically with the number and positions of chlorines at both the urban Chicago location and at remote sites; thus, this behavior is a function of thermodynamic properties and not transport distance. Finally, although derived differently using different tools, Hoff’s and Wania’s models fit PCB data equally well, but neither produces logical results when applied to the IADN data; instead, a new model is proposed (which fits just as well) in which atmospheric temperatures < 273 K are set equal to 273 K (a frozen water model). We therefore conclude that Hoff’s and Wania’s models should only be applied to the limited situations for which they were derived, namely locations with constant PCB concentrations in the inflowing air. This means that the Clausius-Clapeyron slope alone cannot determine PCB transport distance.

Experimental Section (1)

where P is the partial pressure of the analyte (in atm), ∆HSA is a heat of surface-air partitioning (in kJ/mol), T is the local atmospheric temperature (in K), R is the gas constant (0.0083 kJ/K‚mol, in this case), and a0 is an intercept which resolves the units. In this case, “surface-air partitioning” refers to partitioning of a trace atmospheric component between the air and an unknown component of the earth’s surface; thus ∆HSA includes the effect of temperature on vaporization from soil, vegetation, and water. In many papers, a Clausius* Corresponding author e-mail: [email protected].

Two recent papers have discussed the relevance of the C.C. slope for semivolatile organic compounds, including PCBs. Both papers have stated in the abstract that the temperature dependence (expressed as the C.C. slope) indicates whether local sources or long-distance transport are more important for the compound in question (2, 3). In the text, however, both papers show that this conclusion is only applicable at sites with a constant pollutant concentration in the inflowing air.

Original data presented here are for PCBs only in the gas phase; these samples were collected from five U.S. IADN sites beginning in the following years through the end of 2002: Brule River (1996) and Eagle Harbor (1990), both near Lake Superior; Sleeping Bear Dunes (1991) and Chicago (1996), both near Lake Michigan; and Sturgeon Point (1991), near Lake Erie. Details of these sites can be found on the IADN Web site (www.smc-msc.ec.gc.ca/iadn/index.html). The data presented here are the sum of ∼105 PCB congeners or congener groups. Air samples are taken every 12 days for 24 h using a Graseby (General Metal Works, Cleaves, OH) high volume air sampler; gas-phase contaminants are collected by an Amberlite XAD-2 polymer resin (Supelco) 10.1021/es049081f CCC: $30.25

 2005 American Chemical Society Published on Web 12/29/2004

TABLE 1. Clausius-Clapeyron Slopes for Atmospheric PCB Concentrations from the Literature site

N

slope, a1

SE

r2

Eagle Harbor, Michigan Sturgeon Point, New York Sleeping Bear Dunes, Michigan Lista, Norway Brule River, Wisconsin Chicago, Illinois New Brunswick, New Jersey Lancaster, U.K. Egbert, Ontario Sandy Hook, New Jersey Camden, New Jersey Pinelands, New Jersey Madrid, Spain Jersey City, New Jersey Birmingham, U.K. Manchester, U.K. Bloomington, Indiana Marcell Bog, Minnesota Cardiff, U.K. Tuckerton, New Jersey Birmingham, U.K. Washington’s Crossing, NJ Northwest England, U.K. Delaware Bay, New Jersey Bloomington, Indiana Baltimore Harbor, Maryland Ro¨ rvik, Sweden Northwest England, U.K. Ansung, South Korea Northwest England, U.K. Chester, New Jersey Galveston Bay, Texas

319 300 295 200 183 181 166 161 143 83 75 72 71 68 62 55 52 50 48 43 41 40 37 36 36 35 32 28 28 27 25 25

-4950 -5640 -5420 -6610 -5370 -5130 -4730 -4310 -4600 -4420 -6920 -5570 -3740 -5360 -6700 -4770 -7580 -4890 -4520 -4280 -6320 -3570 -3870 -4570 -7790 -5250 -5720 2180 -4340 -9250 -4160 -1000

290 250 330 540 280 300 500 650 410 480 580 460 910 590 naa 780 1920 naa 1000 820 1330 870 1190 720 870 1170 1500 naa 910 1960 1040 2280

0.48 0.63 0.48 0.43 0.67 0.63 0.36 0.45 0.47 0.51 0.66 0.68 naa 0.55 naa 0.41 naa 0.44 0.38 0.40 0.37 0.31 0.23 0.54 0.70 0.38 0.33 0.05 0.47 0.71 0.41 0.01

Tagish, Yukon Gotska sando¨ n, Sweden

78 29

320 -780

240 naa

Finokalia, Crete Bjuro¨ klubb, Sweden Vasa, Finland Norrbyn, Sweden Docksta, Sweden Holmo¨ gadd, Sweden Kalix, Sweden

36 32 31 28 26 25 25

-1730 1400 2060 -4620 -680 -3270 -2570

1050 naa naa naa naa naa naa

Alert, Nunavut Ny-Ålesund, Norway

107 52

-580 -1380

300 850

a

“na” ) not available.

b

ref

site

Inland Temperate Sites IADN IADN IADN (4) IADN IADN (5) (6) (7) (5) (5) (5) (8 ) (5) (9 ) (2,10) (11) (12) (10) (5) (13) (5) (10) (5) (2,14) (15) (2,16) (10) (17) (10) (5) (18)

Bloomington, Indiana Carter, Anniston, Alabama Mars Hill, Anniston, Alabama Stillpond, Maryland Alloway Creek, New Jersey Salaspils, Latvia Northwest England, U.K. Bloomington, Indiana Moody Brook, Falkland Is. Bloomington, Indiana Indiana Dunes, Indiana Chiwaukee Prairie, Wisconsin South Haven, Michigan Muskegon, Michigan Manitowoc, Wisconsin Lake Michigan, Illinois Northern Wisconsin Gårdsjo¨ n, Sweden Ansung, South Korea Concord, California Esthwaite Water, U.K. Stockholm, Sweden Chicago, Illinois Stockton, New York Rice Creek, New York Sterling, New York Aspvreten, Sweden Green Bay, Wisconsin Potsdam, New York Zagreb, Croatia Augsburg, Germany

N

Coastal Sites (35) Lahemaa, Estonia (20) O ¨ land, Sweden (20) Breana¨ s, Sweden (20) Stockholms ska¨ rgård, Sweden (20) Ventes Ragos, Kaliningrad (20) Vilsandi, Estonia (20) Kap Arkona, Germany

Polar Sites (Latitude > 66o) 0.06 (37) Dunai, Eastern Siberia 0.05 (2,38) Brunt Ice Shelf, Antarctica

SE

r2

ref

23 -7600 1040 0.72 (2,14) 23 -6770 1620 0.45 (19) 23 -4360 1050 0.45 (19) 23 -2120 1100 0.15 (15) 20 -3700 1100 0.38 (5) 20 -2810 naa 0.11 (20) 19 -2850 naa 0.16 (10) 19 -6970 1210 0.66 (2,21) 19 -10550 4300 0.26 (22) 18 -7560 1850 0.51 (2,14) 18 -7610 850 0.80 (23) 18 -6610 770 0.82 (23) 18 -6100 790 0.79 (23) 18 -8830 1100 0.84 (23) 18 -5490 590 0.84 (23) 16 -5270 690 0.81 (24) 14 -4090 510 0.84 (2,25) 14 -6940 760 0.87 (26) 14 -2950 naa 0.49 (27) 14 -1030 870 0.10 (28)b 12 -700 2620 0.01 (29) 11 -6830 1020 0.83 (2,30) 11 -5140 5690 0.08 (31) 10 -1450 1120 0.17 (32) 10 -5460 1020 0.78 (32) 10 -3020 950 0.56 (32) 10 -12520 910 0.96 (2,30) 9 -390 3890 0.00 (31) 9 -4790 1450 0.61 (32) 9 -9350 1260 0.89 (33) 8 -5360 670 0.91 (2,34)

Remote Island and Mountain Sites 0.02 (37) Teide, Canary Islands 20 0.04 (20) King George Island, Antarctica 16 0.07 0.05 0.21 0.33 0.01 0.17 0.17

slope, a1

-1980 naa 0.20 3680 9680 01

(40) (39)b

24 21 21 21 17 9 6

-1430 -3820 2470 -3800 120 -3080 -420

0.05 0.29 0.25 0.41 0.00 0.20 0.30

(20) (20) (20) (20) (20) (20) (36)

34 13

-400 370 0.29 -3150 2540 0.12

(37) (22)

naa naa naa naa naa naa 320

Temperature data from an online archive was used in calculations.

cartridge. The samplers are operated at a flow rate of 34 m3/h. The PCBs are removed from the XAD by 24-h Soxhlet extraction using a 1:1 acetone-hexane mixture (OmniSolv, EM Science). Surrogate recovery standards (PCB congeners 14, 65, 166) are added just before extraction. The extracts are concentrated by rotary evaporation, and the solvent is exchanged to hexane. Silica gel (Aldrich Chemical, Davisil, Grade 634, 100-200 mesh), deactivated to 3.5% with water, is used to fractionate the concentrated hexane extract. This column is eluted with 25 mL of hexane. The cleaned-up samples are further concentrated by rotary evaporation and N2 blow-down before addition of quantitation internal standards and injection into the gas chromatograph. The PCBs are analyzed by gas chromatography on Hewlett-Packard 5890 and 6890 instruments equipped with 63Ni electron capture detectors and DB-5 and DB-1701 (J&W Scientific) 60-m columns (250-µm i.d. and 0.1-µm film thickness). Quantitation is done by the internal standard method. Surrogate standards are used to estimate recoveries of each compound in each sample. Recoveries of each compound per batch are determined by matrix spike

experiments. More details on IADN sampling, analytical methods, and data analysis have been published elsewhere, including information on congeners quantified, internal standards, and recoveries (41, 42). The software package SigmaPlot 2002 was used for modeling the data. The regression slopes of the Clausius-Clapeyron equation were compared by testing for parallelism at the 5% significance level.

Results and Discussion Literature Data. All locations and the slopes calculated at each location are shown in Table 1, arranged by type of location and the size of the data set. The locations in Table 1 are divided into four categories: Inland temperate sites are those that are > 5 km from the oceanic coast in temperate latitudes or those with clear urban influences, such as Sandy Hook, New Jersey, which is adjacent to New York City. Remote island and mountain locations are those at which atmospheric samples clearly represent oceanic concentrations, but not polar; two of these sites (Tagish and King George Island) are close to the Arctic or Antarctic Circles but they have much warmer temperatures than the polar sites (see VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

741

FIGURE 1. Distribution of Clausius-Clapeyron slopes for the sum of all PCB congeners for each year at each U.S. IADN site (48 total regressions). Line indicates a Gaussian fit with a mean of -5030 and a standard deviation of 1070. below). Coastal sites are those located as close as possible to the shore ( 285 K). From this simulation, it is clear that low PCB concentrations can give misleading Clausius-Clapeyron slopes. If even a portion of the data is near the noise, the Clausius-Clapeyron slopes will be skewed toward lower values. This error can be a major problem for slopes calculated for PCB concentrations that are just above field blank levels; these are typically data obtained at very remote or very cold locations. Temperature Dependence of Individual PCB Congener Concentrations. With an awareness of the problems of calculating slopes using too few data points, too low of a temperature range, or too low concentrations, we can now apply this insight to the analysis of congener-specific PCB data. Many studies have calculated the temperature dependence of individual PCB congener atmospheric concentrations and noted that this dependence is sometimes a function of the degree of chlorination (2, 4, 6, 8, 9, 11, 12, 19, 35, 40, 48, 49). We have reinvestigated these observations, but to

FIGURE 4. Comparison of (A) modeled background noise, (B) hypothetical vapor phase concentrations of low-level PCBs, and (C) the sum of the modeled background noise (plot A) plus the vapor phase concentrations (plot B). avoid the problems outlined above we have included only those PCB congeners that meet the following criteria: (a) The average winter concentrations must be greater than twice the calculated limit of detection. (b) The average atmospheric temperature during sampling must be > 273 K, to avoid a bias toward low slopes. (c) The congener must be detected > 80% of the time. Applying these inclusion criteria for the Chicago data resulted in the acceptance of 77 congeners or congener groups. Applying these criteria to the other IADN sites resulted in too few remaining congeners for a thorough analysis (only those with 2, 3, or 4 chlorines were left). We note that a prior paper from our group (44) included all congeners in a similar analysis, the majority of which we now feel should be excluded based on the above criteria. Figure 5 shows the C.C. slopes for the included congeners as a function of number of chlorines. Note that there is a large amount of variation within each homologue group and that the overall regression equation would usually be a poor predictor of the slope for an individual congener. The large amount of variation within homologue groups has also been noted for the vapor pressures of PCBs (50). A simple explanation for this variation is that not all chlorine atoms have an equal effect on vapor pressure; for example, the presence of more than one chlorine atom on the ortho ring positions forces the molecule out of planarity, which results in a lower vapor pressure (50). Previous work has suggested that the strong dependence of the Clausius-Clapeyron slope on the degree of chlorination would only occur in areas remote from PCB sources (2). Figure 5, however, clearly shows a strong dependence of the C.C. slope on chlorination at the Chicago IADN site, an urban location with known nearby sources. Other urban sites show VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

743

FIGURE 5. Clausius-Clapeyron slopes at Chicago (T > 273 K) of individual PCB congeners (b) that meet the inclusion criteria and averages of each homologue group (O). Line is a linear fit to the homologue averages (r2 ) 0.97). a similar pattern (8, 9, 48), and some authors have concluded that the dependence of the C.C. slope on chlorination means that local sources are present (13, 44). However, the previous discussion of C.C. slopes of individual PCB congeners at remote locations (2, 4, 12, 49) convinces us that the same dependence on chlorination is seen at remote sites. The limited number of congeners at the more remote IADN sites (after we apply our exclusion criteria) also indicate a dependence on chlorination (in contrast to prior conclusions (44) based on analyses with all congeners). Therefore, we conclude that such behavior is a function of the intrinsic properties of the PCB congeners and not a function of extrinsic factors such as transport distance. Conversely, transport distance cannot be inferred from the observation of C.C. slopes that depend on degree of chlorination. Models That Show Nonlinear Clausius-Clapeyron Plots. As outlined above, the steepness of C.C. slopes of PCBs has been used to infer whether the PCBs are from local sources or are subject to long-distance transport. These inferences are based on the interpretations of two similar models used to explain the apparent range in C.C. slopes and the observed temperature independence at low temperatures. Hoff et al. (3) described atmospheric vapor phase PCB concentrations as the sum of a constant background concentration from long-range transport and a local, temperature-dependent partitioning with the surface of the Earth. This can be represented by the following equation

ln P ) ln(ea0+a1/T + P0)

(2)

where P0 is the constant atmospheric background partial pressure. Wania et al. (2) derived a similar equation (in form) from a fugacity model that, when corrected for an algebraic error, can be represented as follows (their eq 17, corrected)

ln P ) ln(eR/T+β + TX′)

(3)

where R is the difference in the temperature dependence of the Henry’s law constant and the octanol-water partitioning coefficient; β is the difference of the intercepts of these two parameters plus a factor that accounts for soil PCB concentrations, soil organic content, and meteorological conditions; and X′ is a constant that is a function of the constant background concentration of the analyte, meteorological conditions, and molecular weight. Wania et al. (2) suggest that the apparent temperature independence of PCB concentrations at Egbert, Ontario, at low temperatures could be explained by constant PCB concentrations in the inflowing air. They suggested that the 744

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005

PCB concentrations would be constant because they are determined by air-water exchange with the Great Lakes, which have a relatively constant temperature in the winter compared to the land. Presumably, this hypothesis would also apply to the U.S. IADN sites, which are all located near the shore of one of the Great Lakes. We can test this suggestion using data from the five U.S. IADN sites. Table 2 gives the fitted parameters and correlation coefficients for the PCB data from the five U.S. IADN sites using eqs 1-3. All of the correlation coefficients are significant at p < 0.001, except for a0 of eq 2 and β of eq 3 at Chicago. We note that the five calculated C.C. slopes (a1 or R) are not significantly different from one another for the IADN sites using any of the three models. This observation is contrary to what might be expected from the Hoff and Wania models, both of which suggest that Chicago (a major PCB source region) should have a significantly steeper C.C. slope than the other four sites. The similarity of the all five C.C. slopes is notable despite the vast differences in human influences at each site. Table 2 also shows the calculated values of ln P0 and ln X′ from fits of the IADN PCB data using eqs 2 and 3; these logarithms were used in the models to avoid problems of scale. If the values of P0 and X′ (and the apparent temperature independence at low temperatures) were truly related to the background PCB concentration of the Great Lakes region, then the values would be expected to be the same at all sites. However, we note that eq 2 at Chicago gives a P0 value that is ∼30 times larger than this parameter at the other four sites, while eq 3 at Chicago gives an X′ value that is ∼40 times larger than this parameter at the other four sites. In other words, the results of the model fits are contrary to the assumptions of the models. The explanation for the observed differences is likely one of scale: The atmospheric PCB concentrations in Chicago are ∼10 times higher than those at the other four sites (1, 42-44). Incidentally, eqs 2 and 3, although derived in different fashions, are similar in form. The only difference is in the dependence on temperature; in eq 3, the factor relating to background concentrations (X′) is multiplied by temperature, but in eq 2 it is not. Table 2 shows that the correlation coefficients (r2, indicating goodness of fit) for data fitted with both equations are virtually identical. Plots of these two fitted equations are not distinguishable from one another (see Figure 6), perhaps because of the small range of temperatures (∼50 K range with an average of ∼280 K). Still Another Model. Why does the temperature dependence of atmospheric PCB concentrations tend to disappear at low temperatures? In other words, why does the C.C. slope vary with atmospheric temperature as shown in Figure 3? We propose that the temperature measured is not the relevant temperature that should be considered. PCB atmospheric concentrations are, in part, the result of partitioning processes, typically between air on the one side and water, soil, or vegetation on the other. Because this partitioning process is a function of temperature, it follows that it is the temperature of the microenvironment where the partitioning occurs that is important. The only temperature we have available, however, is the average atmospheric temperature at the sampling site during the day the sample was collected. Generally, we have assumed that this measured temperature roughly tracks the temperature at the microenvironmental site where partitioning occurs. This assumption is often valid; for instance, the surface temperature of the Eagle Harbor area in the summer is usually similar to the surface temperature of nearby Northern Wisconsin. However, when atmospheric temperatures drop below the freezing point of water, our measured temperature and the temperature of the location of partitioning are no longer the same. Partitioning will almost always occur between an aqueous

TABLE 2. Results of Fitting Data for Total PCB Concentrations with Eqs 1-3, with Standard Errors eq 1 eq 2

eq 3

eq 1 modifieda a

a0 a1 r2 a0 a1 ln P0 r2 R β ln X′ r2 a0 a1 r2

Chicago

Brule River

Eagle Harbor

Sleeping Bear Dunes

Sturgeon Point

-12.1 ( 1.0 -5126 ( 295 0.63 -1.9 ( 3.9 -8199 ( 1154 -31.3 ( 0.2 0.64 -9307 ( 1251 1.8 ( 4.2 -36.7 ( 0.2 0.65 -9.9 ( 1.1 -5762 ( 321 0.64

-13.6 ( 1.0 -5370 ( 278 0.67 -9.6 ( 2.4 -6539 ( 700 -35.0 ( 0.5 0.69 -7739 ( 812 -5.5 ( 2.8 -40.1 ( 0.2 0.69 -5.3 ( 1.5 -7762 ( 428 0.64

-15.1 ( 1.0 -4952 ( 291 0.48 -9.8 ( 3.3 -6482 ( 950 -34.6 ( 0.5 0.48 -6549 ( 1010 -9.7 ( 3.5 -40.1 ( 0.5 0.48 -9.7 ( 1.4 -6514 ( 398 0.46

-13.3 ( 1.0 -5422 ( 328 0.48 -9.6 ( 3.4 -6517 ( 1001 -34.7 ( 0.8 0.48 -6213 ( 1023 -10.6 ( 3.5 -40.5 ( 1.0 0.48 -9.5 ( 1.5 -6506 ( 411 0.46

-11.8 ( 0.9 -5644 ( 253 0.63 -9.2 ( 2.6 -6419 ( 766 -34.3 ( 0.8 0.62 -5957 ( 754 -10.8 ( 2.6 -40.6 ( 2.2 0.62 -9.4 ( 1.1 -6348 ( 306 0.59

Using T ) 273 K for all temperatures below 273 K.

FIGURE 6. Data for total PCB at (A) Eagle Harbor and (B) Chicago. Red lines indicate model fits to eqs 2 and 3, which are indistinguishable from one another. Green lines indicate model fit to eq 1, using an effective temperature of 273 K for all measured temperatures below 273 K. phase and the air; for example, between the surface of a lake and the atmosphere or between moist soil and the soil’s interstitial air. When the temperature drops sufficiently, the aqueous phase will freeze, and partitioning into the air will no longer occur to any appreciable extent. The PCB concentrations in the atmosphere will then depend on partitioning processes occurring elsewheresthe nearest location where there is liquid water. Clearly, this location will be warmer than the below-freezing temperatures measured at the sampling site but not much warmer than freezing. For instance, this location could be unfrozen soil beneath a porous layer of snow and frozen earth, or it could be open lake water further away from shore. Under these conditions, the effective temperature will be the temperature of the closest available partitioning matrix, which will be very close to 273 K. Therefore, we propose that whenever the measured

temperature falls below 273 K at a sampling site, an effective temperature of 273 K should be used in the C.C. analysis instead. If 273 K is substituted for all temperatures < 273 K, then the atmospheric PCB concentrations can be analyzed once again using the Clausius-Clapeyron equation. Results are shown in Table 2 and in Figure 6. Based on r2 values, we note that the fit is just as good as that using the other three models discussed above. It seems clear that the C.C. slopes are a function of thermodynamic parameters and not how far the PCBs have been transported. Non-IADN Sites. For sites without local sources, the inflowing PCB concentrations will affect the local atmospheric concentrations, as mentioned above. If the regional climate is more or less uniform, the temperature nearby will be similar to the temperature at the sampling site, and in this case, using the sampling site temperature will be acceptable. However, at coastal sites or near mountain ranges, the temperature of nearby regions will not necessarily be similar to the temperature at the sampling site, and in these cases, the inflowing PCB concentrations would not be related to the sampling site temperature, and this artifact may result in apparently temperature-independent atmospheric PCB concentrations. In cases where the inflowing air is from an ocean, which would have relatively constant year-round temperatures, the inflowing PCB concentrations would also be relatively constant, assuming atmospheric-oceanic equilibrium. Indeed, PCB concentrations over the Atlantic Ocean have been observed to broadly follow atmospheric temperature, although a small temperature-independent diurnal cycle was also noted (51). The models of Wania et al. (2) and Hoff et al. (3) would, therefore, be applicable. The one site in this category with a sufficiently large number of data points is Tagish, in the Yukon Territory of Canada (see Table 1). At this site, the majority of air trajectories are coming from the northern Pacific Ocean (37). From this constant-temperature regime, the air moves over snow- and ice-covered mountains (which likely provide little opportunity for partitioning) before reaching Tagish. The incoming air at the sparsely populated Tagish site, therefore, has relatively constant PCB concentrations, resulting in a calculated slightly positive ClausiusClapeyron slope. Table 1 also lists several other remote sites with clear climatic discontinuities. For example, the coastal sites were located in areas intended to capture a maritime signal. Unfortunately the number of measurements at these sites is low (less than 37 in all cases). From Figure 2, we can see that a great deal of uncertainty is, therefore, associated with each of the calculated slopes. Nevertheless, as a group, the coastal sites tend to have shallow slopes, as seen in Figure 2, which VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

745

suggests that coastal sites are more likely to exhibit temperature independence of PCB atmospheric concentrations when there are no local sources, as we and others have predicted (2, 3). Table 1 also lists four polar sites, characterized by very low temperatures and high latitudes; often, the nearest partitioning environment will be the ocean. It is, therefore, easy to rationalize the observed temperature independence of PCB atmospheric concentrations at these sites, which are near climatic discontinuities and subject to below-freezing temperatures for most of the year. One of these sites, Alert, has 107 data points, giving us confidence in the calculated slope. Final Arguments. The variation in slopes calculated at different sites can be explained by the small number of samples collected in most cases, by a bias toward lower slopes for data collected in cold climates, and by low concentrations at very remote locations. Clearly, the true behavior of atmospheric PCB concentrations is dependent on temperature and geography. Temperature independence will be observed at temperatures below freezing, and this can be accounted for using our “frozen water” model. In general, temperature dependence will be seen in geographically uniform areas, where local and regional air-surface partitioning will be the dominant factor. Temperature independence will be seen in regions of constant or very cold temperatures. The models described previously (2, 3) still may apply but only to sites with constant or temperatureindependent PCB concentrations in the inflowing air (as is clear from the description of these models). Based on the arguments presented here, we conclude that little or no temperature dependence of atmospheric PCB concentrations at a site does not mean long-distance transport is an important factor bringing PCBs to that site, nor does a strong temperature dependence mean local sources predominate.

Acknowledgments We thank the U.S. EPA Great Lakes National Program Officer for funding (Grant GL995656; project director, Melissa Hulting); Ilora Basu and Team IADN for the operation of the network; and Will Hafner and Stephanie Buehler for data compilation and calculations.

Literature Cited (1) Buehler, S. S.; Hites, R. A. The Great Lakes’ Integrated Atmospheric Deposition Network. Environ. Sci. Technol. 2002, 36, 354A-359A. (2) Wania, F.; Haugen, J.-E.; Lei, Y. D.; Mackay, D. Temperature dependence of atmospheric concentrations of semivolatile organic compounds. Environ. Sci. Technol. 1998, 32, 10131021. (3) Hoff, R. M.; Brice, K. A.; Halsall, C. J. Nonlinearity in the slopes of Clausius-Clapeyron plots for SVOCs. Environ. Sci. Technol. 1998, 32, 1793-1798. (4) Haugen, J.-E.; Wania, F.; Lei, Y. D. Polychlorinated biphenyls in the atmosphere of Southern Norway. Environ. Sci. Technol. 1999, 33, 2340-2345. (5) Reinfelder, J. R.; Totten, L. A.; Eisenreich, S. J. The New Jersey Atmospheric Deposition Network. Final report to the New Jersey Department of Environmental Protection, 2004. (6) Lee, R. G. M.; Jones, K. C. The influence of meteorology and air masses on daily atmospheric PCB and PAH concentrations at a UK location. Environ. Sci. Technol. 1999, 33, 705-712. (7) Hoff, R. M.; Muir, D. C. G.; Grift, N. P. Annual cycle of polychlorinated biphenyls and organohalogen pesticides in air in Southern Ontario. 2. Atmospheric transport and sources. Environ. Sci. Technol. 1992, 26, 276-283. (8) Garcı´a-Alonso, S.; Pe´rez-Pastor, R. M. Occurrence of PCBs in ambient air and surface soil in an urban site of Madrid. Water, Air, Soil Pollut. 2003, 146, 283-295. (9) Harrad, S.; Mao, H. Atmospheric PCBs and organochlorine pesticides in Birmingham; UK: concentrations, sources, temporal and seasonal trends. Atmos. Environ. 2004, 38, 14371445. 746

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005

(10) Halsall, C. J.; Gevao, B.; Howsam, M.; Lee, R. G. M.; Ockenden, W. A.; Jones, K. C. Temperature dependence of PCBs in the UK atmosphere. Atmos. Environ. 1999, 33, 541-552. (11) Panshin, S. Y.; Hites, R. A. Atmospheric concentrations of polychlorinated biphenyls at Bloomington, Indiana. Environ. Sci. Technol. 1994, 28, 2008-2013. (12) Hornbuckle, K. C.; Eisenreich, S. J. Dynamics of gaseous semivolatile organic compounds in a terrestrial ecosystem - effects of diurnal and seasonal climate variations. Atmos. Environ. 1996, 30, 3935-3945. (13) Currado, G. M.; Harrad, S. Factors influencing atmospheric concentrations of polychlorinated biphenyls in Birmingham. U.K. Environ. Sci. Technol. 2000, 34, 78-82. (14) Hermanson, M. H.; Hites, R. A. Long-term measurements of atmospheric polychlorinated biphenyls in the vicinity of Superfund dumps. Environ. Sci. Technol. 1989, 23, 1253-1258. (15) Bamford, H. A.; Ko, F. C.; Baker, J. E. Seasonal and annual airwater exchange of polychlorinated biphenyls across Baltimore Harbor and the Northern Chesapeake Bay. Environ. Sci. Technol. 2002, 36, 4245-4252. (16) Brorstro¨m-Lunde´n, E. Measurements of semi-volatile organic compounds in air; deposition. Ph.D. Dissertation, Go¨teborg, 1995; pp. (17) Yeo, H.-G.; Choi, M.; Chun, M.-Y.; Sunwoo, Y. Gas/particle concentrations and partitioning of PCBs in the atmosphere of Korea. Atmos. Environ. 2003, 37, 3561-3570. (18) Park, J.-S.; Wade, T. L.; Sweet, S. Atmospheric deposition of organochlorine contaminants to Galveston Bay, Texas. Atmos. Environ. 2001, 35, 3315-3324. (19) Hermanson, M. H.; Scholten, C. A.; Compher, K. Variable air temperature response of gas-phase atmospheric polychlorinated biphenyls near a former manufacturing facility. Environ. Sci. Technol. 2003, 37, 4038-4042. (20) Agrell, C.; Okla, L.; Larsson, P.; Backe, C.; Wania, F. Evidence of latitudinal fractionation of polychlorinated biphenyl congeners along the Baltic Sea Region. Environ. Sci. Technol. 1999, 33, 1149-1156. (21) Wallace, J. C.; Hites, R. A. Diurnal variations in atmospheric concentrations of polychlorinated biphenyls; endosulfan: Implications for sampling protocols. Environ. Sci. Technol. 1996, 30, 444-446. (22) Ockenden, W. A.; Lohmann, R.; Shears, J. R.; Jones, K. C. The significance of PCBs in the atmosphere of the southern hemisphere. Environ. Sci. Pollut. Res. 2001, 8, 189-194. (23) Miller, S. M.; Green, M. L.; Depinto, J. V.; Hornbuckle, K. C. Results from the Lake Michigan mass balance study: concentrations and fluxes of atmospheric polychlorinated biphenyls and trans-nonachlor. Environ. Sci. Technol. 2001, 35, 278-285. (24) Zhang, H.; Eisenreich, S. J.; Franz, T. R.; Baker, J. E.; Offenberg, J. H. Evidence for increased gaseous PCB fluxes to Lake Michigan from Chicago. Environ. Sci. Technol. 1999, 33, 2129-2135. (25) Manchester-Neesvig, J. B.; Andren, A. W. Seasonal variation in the atmospheric concentration of polychlorinated biphenyl congeners. Environ. Sci. Technol. 1989, 23, 1138-1148. (26) Brorstro¨m-Lunde´n, E.; Lo¨fgren, C. Atmospheric fluxes of persistent semi-volatile organic pollutants to a forest ecological system at the Swedish west coast and accumulation in spruce needles. Environ. Pollut. 1998, 102, 139-149. (27) Yeo, H.-G.; Choi, M.; Chun, M.-Y.; Sunwoo, Y. Concentration distribution of polychlorinated biphenyls and organochlorine pesticides and their relationship with temperature in rural air of Korea. Atmos. Environ. 2003, 37, 3831-3839. (28) Tsai, P.; Hoenicke, R.; Yee, D.; Bamford, H. A.; Baker, J. E. Atmospheric concentrations and fluxes of organic compounds in the Northern San Francisco Estuary. Environ. Sci. Technol. 2002, 36, 4741-4747. (29) Gevao, B.; Hamilton-Taylor, J.; Jones, K. C. Polychlorinated biphenyl and polycyclic aromatic hydrocarbon deposition to and exchange at the air-water interface of Esthwaite Water; a small lake in Cumbria, UK. Environ. Pollut. 1998, 102, 6375. (30) Bidleman, T. F.; Wideqvist, U.; Jansson, B.; Soederlund, R. Organochlorine pesticides and polychlorinated biphenyls in the atmosphere of southern Sweden. Atmos. Environ. 1987, 21, 641654. (31) Cotham, W. E.; Bidleman, T. F. Polycyclic aromatic hydrocarbons and polychlorinated biphenyls in air at an urban and a rural site near Lake Michigan. Environ. Sci. Technol. 1995, 29, 27822789. (32) Chiarenzelli, J.; Pagano, J.; Scrudato, R.; Falanga, L.; Migdal, K.; Hartwell, A.; Milligan, M.; Battalagia, T.; Holsen, T. M.; Hsu, Y.; Hopke, P. Enhanced airborne polychlorinated biphenyl (PCB)

(33) (34)

(35) (36) (37)

(38)

(39) (40)

(41)

(42)

concentrations and chlorination downwind of Lake Ontario. Environ. Sci. Technol. 2001, 35, 3280-3286. Romanic, S. A. H.; Krauthacker, B. Levels and seasonal variations of polychlorinated biphenyls in ambient air, Zagreb, Croatia. Fresenius Environ. Bull. 2003, 12, 143-147. Kaupp, H.; Do¨rr, G.; Hippelein, M.; McLachlan, M. S.; Hutzinger, O. Baseline contamination assessment for a new resource recovery facility in Germany. Part IV: atmospheric concentrations of polychlorinated biphenyls and hexachlorobenzene. Chemosphere 1996, 32, 2029-2042. Mandalakis, M.; Stephanou, E. G. Study of atmospheric PCB concentrations over the eastern Mediterranean Sea. J. Geophys. Res. 2002, 107, 4716-4729. Bruhn, R.; Lakaschus, S.; McLachlan, M. S. Air/sea gas exchange of PCBs in the southern Baltic Sea. Atmos. Environ. 2003, 37, 3445-3454. Stern, G. A.; Halsall, C. J.; Barrie, L. A.; Muir, D. C. G.; Fellin, P.; Rosenberg, B.; Rovinsky, F. Y.; Kononov, E. Y.; Pastuhov, B. Polychlorinated biphenyls in arctic air. 1. Temporal and spatial trends: 1992-1994. Environ. Sci. Technol. 1997, 31, 3619-3628. Oehme, M.; Haugen, J.-E.; Schlabach, M. Seasonal changes; relations between levels of organochlorines in Arctic ambient air: first results of an all-year-round monitoring program at Ny-A° lesund, Svalbard, Norway. Environ. Sci. Technol. 1996, 30, 2294-2304. Montone, R. C.; Taniguchi, S.; Weber, R. R. PCBs in the atmosphere of King George Island, Antarctica. Sci. Total Environ. 2003, 308, 167-173. Van Drooge, B. L.; Grimalt, J. O.; Garcı´a, C. J. T.; Cuevas, E. Semi-volatile organochlorine compounds in the free troposphere of the Northeastern Atlantic. Environ. Sci. Technol. 2002, 36, 1155-1161. Simcik, M. F.; Hoff, R. M.; Strachan, W. M. J.; Sweet, C. W.; Basu, I.; Hites, R. A. Temporal trends of semivolatile organic contaminants in Great Lakes precipitation. Environ. Sci. Technol. 2000, 34, 361-367. Buehler, S. S.; Basu, I.; Hites, R. A. Causes of variability in pesticide and PCB concentrations in air near the Great Lakes. Environ. Sci. Technol. 2004, 38, 414-422.

(43) Hillery, B. R.; Basu, I.; Sweet, C. W.; Hites, R. A. Temporal and spatial trends in a long-term study of gas-phase PCB concentrations near the Great Lakes. Environ. Sci. Technol. 1997, 31, 1811-1816. (44) Simcik, M. F.; Basu, I.; Sweet, C. W.; Hites, R. A. Temperature dependence and temporal trends of polychlorinated biphenyl congeners in the Great Lakes atmosphere. Environ. Sci. Technol. 1999, 33, 1991-1995. (45) Honrath, R. E.; Sweet, C. I.; Plouff, C. J. Surface exchange and transport processes governing atmospheric PCB levels over Lake Superior. Environ. Sci. Technol. 1997, 31, 842-852. (46) Currado, G. M.; Harrad, S. Factors influencing atmospheric concentrations of PCBs. Organohalogen Compd. 1999, 43, 163166. (47) Franz, T. P.; Eisenreich, S. J. Snow scavenging of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in Minnesota. Environ. Sci. Technol. 1998, 32, 1771-1778. (48) Brunciak, P. A.; Dachs, J.; Gigliotti, C. L.; Nelson, E. D.; Eisenreich, S. J. Atmospheric polychlorinated biphenyl concentrations and apparent degradation in coastal New Jersey. Atmos. Environ. 2001, 35, 3325-3339. (49) Hung, H.; Halsall, C. J.; Blanchard, P.; Li, H. H.; Fellin, P.; Stern, G.; Rosenberg, B. Are PCBs in the Canadian Arctic atmosphere declining? Evidence from 5 years of monitoring. Environ. Sci. Technol. 2001, 35, 1303-1311. (50) Falconer, R. L.; Bidleman, T. F. Vapor pressures; predicted particle/gas distributions of polychlorinated biphenyl congeners as functions of temperature and ortho-chlorine substitution. Atmos. Environ. 1994, 28, 547-554. (51) Jaward, F. M.; Barber, J. L.; Booij, K.; Dachs, J.; Lohmann, R.; Jones, K. C. Evidence for dynamic air-water coupling and cycling of persistent organic pollutants over the open Atlantic Ocean. Environ. Sci. Technol. 2004, 38, 2617-2625.

Received for review June 17, 2004. Revised manuscript received October 6, 2004. Accepted October 19, 2004. ES049081F

VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

747