Annual cycle of polychlorinated biphenyls and organohalogen

---

Environ. Sei. Technol. 1992,26,266-275

Annual Cycle of Polychlorinated Biphenyls and Organohalogen Pesticides in Air in Southern Ontario. 1. Air Concentration Data Raymond M. Hoff,*,+Derek C. G. Muir,t and Norbert P. Grift$

Atmospheric Environment Service, Rural Route 1, Egbert, Ontario, Canada LOL 1L0, and Department of Fisheries and Oceans, Freshwater Institute, 501 University Crescent, Winnipeg, Manitoba, Canada R3T 2N6 ~~~

From July 1988 to September 1989, 143 air samples, obtained at Egbert, ON, Canada, were analyzed for vapor-phase polychlorinated biphenyls (PCBs) and organohalogen pesticides. This data set is believed to be the first high temporal resolution PCB data set obtained over an annual cycle in North America and has obvious use for determining processes of deposition, transport, and atmospheric transformation of these important chemicals. Concentrations of the sum of 91 PCB congeners (CPCB) of IUPAC No. 16 or higher ranged from a subpicogram per cubic meter detection level to over 2 ng m-3. Monthly averages of CPCB ranged from 55 to 823 pg m-3. Organochlorine pesticide maximum (mean) concentrations were as follows: for &HCH, 1ng m-3 (20 pg m 3 ; for CCHLOR, 430 (81) pg m- , for PCCs, 580 (26) pg m"; for CDDT, 560 (90) pg m-3; for dieldrin, 210 (46) pg m-3. Higher concentrations of the locally and regionally used pesticides trifluralin [3.4 ng m-3 (270 pg m-3)] and endosulfan [3.7 ng m-3 (346 pg m-3)] were found. Some of these values were actually higher than reported in other studies since the 1970s. The ratio a-HCH/y-HCH has a well-defined annual cycle peaking in winter at -7 and minimizing in summer at -1. The concentrations of these species suggest a Lorentzian form for the annual cycle of these chemicals. CPCBs have a minimum in this function of 55 pg m-3 and a maximum amplitude 14 times this minimum, occurring in late July and with a half-width of 0.67 month. CDDT has a minimum of 11 pg m-3 with an amplitude of 20 times the minimum value, again peaking in late July but with a half-width of 1.7 month. This function is readily programmable into models which make use of the air concentration data to determine deposition (both wet and dry). ~~

Introduction

Because of their moderate vapor pressure, low solubility, and low reactivity, a number of semivolatile organic compounds (SOCs) have been found to be widely distributed in the atmosphere. Many of these same compounds, containing chlorine, were designed to be toxic pesticide agents and have the capability of bioaccumulating through the food chain. The presence of these chemicals in regions where they were not used is of concern, both for ecological health reasons and because of our inability to control the movement of these chemicals in the environment without having a better knowledge of the processes regarding their movement and fate. The spatial distribution of organochlorine compounds (OCs) and polychlorinated biphenyls (PCBs) in the atmosphere has been studied in remote locations as diverse as the Canadian and Scandanavian high Arctic (I-IO), the Antarctic (11,121,the Indian Ocean (11-13), and the Pacific Ocean (14,15). A significant number of studies have been carried out to determine the air concentrations in industrialized areas (16-43). In addition to these studies, ~

+Atmospheric Environment Service. Freshwater Institute.

*

266

Environ. Sci. Technol., Vol. 26, No. 2, 1992

several important reviews have examined the spatial distribution of these chemicals (44-46). Despite this large volume and breadth of data, as far as we are aware, a single high temporal resolution data set (for example, subweekly samples) of PCB and OC air concentrations for a 1-year period has not been published. The need for such data is obvious. The processes of volatilization, deposition, transport, and transformations of these chemicals in the atmosphere respond to variations in temperature, insolation, and weather and may be reflected first in the ambient concentration of these chemicals, even at a single sampling site. Ideally, annual concentration patterns at a number of sites would provide a more realistic (i.e., less locally biased) view of the overall behavior of the chemicals in air, and several efforts are currently being mounted to give such information. We believe that the work reported here is the first high temporal resolution data set and will be demonstrably useful for determining transport-related phenomena. Several research programs have generated data over either short periods with short sampling times (cf. refs 16-22) or longer study periods with more sporadic sampling (i.e., several days per month, for example) ( 4 9 ,30, 3 2 , s ) . A recent paper (36) presenta PCB data taken from a large number of samples at several sites over a 1-year period. The time series of the data was, unfortunately, not shown in the paper. As will be shown in the following paper (47), there is a strong reason to take data in a contiguous fashion with short time resolution. This is because the transport time scales which carry the ubiquitous pollutants studied here are of synoptic size. These synoptic features are usually of 2- or 3-day duration in the midlatitudes, and thus it is important to generate temporal data of this scale or shorter. Ideally, daily sampling (or shorter time scales) would provide the most useful data for transport analysis [for example, as is generated by inorganic sampling in many long-range-transport monitoring programs (&)I. Analysis of up to 400 organic samples per year (or twice that number if filter/absorbant combinations are used) is a formidable task. Due to the limitations of cost and manpower in this study, we chose the compromise of 2-day sampling over the period of 1year and focused on organics trapped on polyurethane foam (PUF) absorbants. It is believed that the data will show that synoptic scale information and transport pathways can be derived from such a sampling cycle. Because of the volume of data obtained in this study, this work is broken up into two reports. The first part, which is presented in this paper, details the concentration time series of 35 organohalogen insecticides and herbicides and over 90 specific PCB congeners. These annual cycles are compared to previous studies. A recommendation of functional forms of the annual cycles for the species measured is given in this paper as a first step in providing modelers with an analytical parameterization of the annual vapor-phase behavior. The companion paper (47) analyses the time series of the air concentration data in conjunction with air parcel

0013-936X/92/0926-0266$03.00/0

0 1992 American Chemical Society

to this vapor pressure limit. While it is preferable to maintain the samples at temperatures colder than their sampling temperature, it is believed that maintaining the samples in sealed containers at room temperature for this Experimental Section limited period did not jeopardize the quality of the data. Air samples of mid molecular weight chlorinated organic Since the primary purpose of the experiment was to compounds were taken using a Sierra Andersen PS-1 PUF obtain an annual cycle of the PCB concentrations in air, sampler. The sample head contains a 10.2 cm diameter it was believed from existing data that the predominance Whatman GF/A glass fiber filter followed by a 7.2 cm of the PCB sampled mass would reside on the foam plugs diameter by 7.5 cm long PUF plug of density 0.022 g ~ m - ~ . and not on the filters. Studies by other researchers have The sampling efficiency of foams of this type, size, and borne this assumption out (34, 37, 49). Foreman and density has been extensively documented by Bidleman and Bidleman (69) showed that, for Denver at 5 'C and his co-workers (5, 17, 18, 20, 21, 49, 51). Sampling was chemical species of vapor pressure >lo4 Torr, less than carried out at Egbert, ON, Canada (44'14' N, 79'47' W, 40% of the total concentration is on the particulate phase. 251 masl), in a rural area which is some 100 km north of This would apply to the chlordanes, DDT, and PCBs of the city of Toronto. The research site was chosen to be IUPAC number less than 108. For a temperature change representative of the mostly rural nature of southern Onof +10 "C, po for these chlorinated species increases by tario and has no local industrial sources within 15 km. The a factor of 2.7-3.6 (69)and the fraction on particles would site is subject to impact from current agricultural chemicals decrease by a factor of about this same order (50). While from a range of nearby farming activities, which include this supports the intent of measuring most of the mass of hay cultivation, potato, wheat, corn, and barley farming, the chemicals in air, there remains doubt about the amount market gardening, and tree cultivation. of material retained on the filters (since they were not Sampling was started on July 10,1988. After that date, analyzed), and thus, the results presented here are a samples were predominantly taken on a 2-day observaminimum for the total air concentration of these species. tional cycle until July 1, 1989. Subsequent to that date, The total air concentration should not increase on an ansampling was changed to a 1day in 6 cycle. Three record nual average basis by more than 20% above the values gaps (the longest being over 3 weeks) were encountered reported here. during the sampling; other than this, the sampling was Foams were Soxhlet extracted with hexane for 4 h. continuous. Internal standards of aldrin and octachloronaphthalene Before sampling, the foam plugs were precleaned by were added to each sample prior to extraction. The extract Concord Scientific Corp. of Downsview, ON, using a was evaporated to -1 mL and then quantitatively large-volume Soxhlet extraction with distilled-in-glass transferred to a Florisil column (80-100 mesh, 1.2% water grade dichloromethane (DCM, Caledon Laboratories) for by weight; 10-cm length). The column was eluted with at least 12 h. Each foam was dried in air and placed in hexane (Fl), hexane-DCM (85:15) (F2), and hexane-DCM a 250-mL glass sample jar with precleaned PFTE lid liners. (1:l) (F3). The chromatography on Florisil separated The glass fiber filters (GFFs) were not pretreated before PCBs, chlorobenzenes, 4,4'-DDE, and mirex in F1 from use. There is evidence that this type of GFF has a residual most polychlorinated camphenes (PCC),chlordane-related organic material loading on the order of 150 pg (taken from compounds, and 4,4'-DDT in F2. F3 contained heptachlor the ratio of the 10.2-cm circular filters used here to 8 in. epoxide and dieldrin (53). X 10 in. high-volume fiters in use in a current, unpublished Florisil eluates were analyzed by capillary gas chromastudy at AES involving combustible organic carbon). This tography on a Varian 6000 equipped with a 63Ni elecresidual organic material may serve to change the operatron-capture detector (GC-ECD) using a 60 m X 0.25 mm tional definition of the gas and particle fractions of the i.d. DB-5 column with H2 carrier gas (53). PCBs were sample, since gas-phase material may be absorbed onto this quantified using a series of congener mixtures obtained organic material (50). The additional effect of desorption from the National Research Council of Canada (Halifax, of organic material from the untreated fiiters onto the PUF NS), or where congeners were not available, on the basis plug is believed to be negligible for the chlorinated maof patterns and estimated response factors (RFs) for peaks terials studied in this work. in an Aroclor 1254:1260 mixture (53). Total PCB (CPCB) All samples were taken so that the total flow over the was the sum of all measurable congeners of IUPAC no. 16 sample period was less than 800 m3, in order to minimize and above (see discussion below). Chlordane-related sample breakthrough. Each sample had the starting and compounds were quantified with individual standards ending pressure drop logged and the flow was determined except for nonachlor I11 and photoheptachlor where RFs from the median pressure drop across the PS-1 limiting for trans-nonachlor and heptachlor, respectively, were orifice over the sampling period, corrected for ambient used. Minor chlordane components, "C", C1, C2, and C3, temperature. were quantified using the RF for heptachlor while C4 was After sampling, the cartridge was unloaded immediately quantified using the RF of trans-nonachlor. Total and the foam plugs were replaced in the original sample chlordane (CCHLOR) was the sum of all chlordane-rejars and refrigerated at either 4 'C or frozen at MDC (of 143) Min

46 32 640 670

MDC MDC 0.04 4.0

132 133 143 143

>5.3 >8.0 >54 >70

540 28 820 1000

10 MDC 4.0 20

143 96 143 143

145 1.8 60 210

53 30 26 17 9.1 2.7 27 23 20 20 75 76 60 4.1 72 430

0.04 MDC MDC MDC MDC MDC MDC MDC MDC MDC MDC 0.67 MDC MDC MDC 4.1

143 140 127 122 108 56 116 141 41 76 142 143 141 45 142 143

5.4 5.8 3.3 1.2 1.2 0.21 2.0 4.8 0.55 0.51 14 14 11 0.32 16 82

320 6.0 120 130 560

1.0 MDC MDC MDC 1.0

143 86 137 134 143

53 0.89 16 20 90

31

MDC MDC MDC MDC MDC MDC MDC MDC MDC

48 114 5 140 41 75 135 117 142

0.71 26 0.35 28 1.7 270 350 3.7 46

580 22 130 27 3400 3700 33 210

annual mean

Chlorobenzenes reported are minima since incomplete quantitation occurs in summer months. 2500 n

m

'

2000

M

a

w1500

d 0

5 (d

k

1000

4

8

:

u

500 0

July 88

Jan 89

July 89

Flgure 1. CPCB concentration vs time. After July 1, 1989, the sample frequency resolution is lowered to 1 day in 6.

pressure, the total CPCB concentrations may be quite different from those shown here, especially where other sampling media have been used. The authors have chosen to exclude congeners of no. 15 or less since it is expected that the sampling of these congeners by the foam plugs will be nonquantitative in summer months when the total concentration was highest. This arises since these species will break through the foam even at the modest sample volumes collected here (49). No attempt was made to use multiple plugs to monitor breakthrough in this study. A simplified view of the annual cycle of the congeners is seen in Figure 2a-d. In this figure, all data from a given month (with July-September data from 1988 and 1989 combined by month) are plotted. The annual cycle of the PCB congeners in air is strongly peaked in summer with the amplitude, AM, of the peak to minimum ratio of monthly averages being largest for the heaviest PCBs. Clearly, annual estimates of concentration made from other work where, for largely logistical reasons, summer sampling only was conducted will be biased high. The positive view Environ. Sci. Technol., Vol. 26, No. 2, 1992

269

+

PcB33

200 150

100 h

c!

50

10

Do

,a -

0 L .O 50 -3

g g

45

0 I

4 (c)

401 35

301

PCB 101

25

-

-PCB

99

PCBIIO

-

15 -

20

10 -

15 10 5

0

Jan

Mar.

May

July

Sept.

Nov.

Jan

hlar.

May

July

Sept.

Nov.

Flgure 2. PCB congener monthly average concentrationsvs time: (a) PCB homologues; (b) PCBs 52, 33, 47/48, 31, and 16/32; (c) PCBs 64, 70/76, 101, and 99: (d) PCBs 110, 153, 138, and 187.

about earlier summer measurements is that they will be representative of the period when intermedia exchanges will be maximum. Figure 2a shows the annual cycle for the PCB homologues. As the chlorination increases, the “peakedness” of the distribution becomes more apparent. This can be due to two processes: (1)preferential partitioning into precipitation as chlorine number increases, due to increasing Henry’s law coefficients and particle scavenging, and (2) preferential desorption of lighter PCBs from atmospheric particles and the earth’s surface at a given temperature. The work of Hawker (56) shows that the decrease in Henry’s law coefficient with chlorination (actually Hawker used the planar total surface area of the molecule, TSA) is -4 orders of magnitude smaller than the decrease in vapor pressure with the same index, in agreement with ref 70. This suggests strongly that the decrease in atmospheric residence time reflected here in the peakedness of the annual cycle is due to volatilatization effects, not scavenging. The lighter trichlorinated PCBs show the least peaked annual cycle. If the mono/di PCBs could have been measured quantitatively, it is believed that they would be even more uniformly seen throughout the year, and this strengthens the need to develop sampling media which can obtain valid data on even the lightest PCB congeners. Organochlorine (OC) Insecticides. A number of cyclodiene compounds including dieldrin, endrin, mirex, polychlorinated camphenes (PCC), chlordane, nonachlor, heptachlor, and the epoxides have been measured in air during this study (Table 11). These chemicals were widely used before the 1980s as soil fumigants and agricultural and household insecticides. Their usage has been limited in the 1970s and early 1980s because of their persistence in the environment. Volatilization appears to be a major loss mechanism of these chemicals from soil (68) and it is not surprising that their concentrations in air might remain detectable even many years after application. 270

Environ. Sci. Technol.. Vol. 26, No. 2, 1992

Chlordane has been widely detected in the environment, even in the high Arctic (2, 5). Chlordane was used as a residential fumigant as late as the mid-1980s in Canada, and in 1985 all registrations of new chlordane products were suspended by Agriculture Canada. It still can be used by licensed pest control officers. Technical chlordane is a mixture of up to 29 different compounds (57). The annual cycle of CCHLOR (16 peaks in all) is shown in Figure 3b. The cycle of these compounds is not as strongly peaked as the PCBs and shows input of these chemicals to southern Ontario throughout the year. A useful transport diagnostic for the chlordane isomers is the ratio of truns-chlordane (TC) and trans-nonachlor (TN) to cis-chlordane (CC) ( 2 , 3 , 5 ) .The technical mixture of chlordane used in North America would indicate that the ratios should be TC/CC = 1.26 and TN/CC = 0.37 (57). In the southern Ontario winter, the ratio TC/CC is very similar to the technical mixture, while the TN/CC is somewhat elevated from the mixture (Figure 4a,b). In summer months, the ratios converge. As here, TN/CC ratios above unity were occasionally seen in southern Sweden (61). A vapor pressure order of T N > TC > CC with a stronger temperature dependence for TN than TC would explain the annual behavior, although this is not the expected order of vapor pressures of TC > CC > TN > CN (5, 52). In summer, when the temperatures were warm, T N would be preferentially volatilized over TC. This difference in the vapor pressure/temperature slope is supported by Hinckley et al. (52). Arctic air samples have shown ratios strongly differing from this sequence ( 2 , 5 )with a reversal of the abundance of the trans-nonachlor and the trans-chlordane, and both TN/CC and TC/CC well below unity. This air ratio has been reflected in the isomeric ratios in aquatic fauna (4). PCCs are another group of cyclic aliphatic pesticides which were widely used in North America into the 1970s. The measurements of the PCCs (Figure 3b) show atmospheric concentrations less than 100 pg m-3 for all but the

500

I

I

-

250 1 200

ZCHLOR 150 100

50

.s

B

8

8

0 100 I-

90

80

Dieldrin Methoxychlor

+-

Endosulfan

II

70 1200

60

Trifluralin

--C

50

40

800

30

600

20

400

10

0

Jan

Mar.

May

Sept.

E

200

Jan

Mar.

May

’

Sept.

Nov.

Figure 3. Organohalogen concentrations vs time: (a) CCBz and CHCH; (b) CCHLOR, CDDT, and toxaphene; (c) pentachloroanisole, dieldrin, and methoxychlor; (d) endosulfan and trifluralin. 2.0

2

0.0

,

1

I

1 I

~

8.0

4.0

0.0 July 88

Jan 80

July 80 July 88

Jan 80

July 89

Figure 4. Species ratios vs time: (a) trans-chlordanelcis -chlordane;

(b) trans -nonchlorlcis-chlordane; (c) 4,4’-DDE/xDDT; (d) a-HCHlyHCH.

month of July (and this was dominated by an episode of high pollution in southern Ontario). While these concentrations are encouragingly low in relation to earlier measurements in more southern latitudes (29),toxaphene has still not disappeared from the atmosphere despite its restriction. Dieldrin and endrin have been detected in our samples to a maximum concentration of -200 pg m-3 with annual averages of 46 and 4 pg m-3, respectively. Dieldrin is ubiquitous, being detected in all but one sample, and its monthly time series is shown in Figure 3c. Endosulfan, which has been sold by the commerical name Thiodan, is a current-use pesticide which is used as a foliar insecticide on commercial crops and ie tree sprays. Technical endosulfan has a and P isomers. In the environment, the P isomer (measured here) is the most persistent. It has in fact been used as a method of removing unwanted fish from lakes before restocking.

Endosulfan is the most concentrated OC observed in this study, reaching a maximum concentration of almost 4 ng m-3. Despite this high concentration, to our knowledge, it has been reported in few other studies (5, 10,23)and a concentration of 78 pg m-3 was seen in the latter study. Its annual cycle is extremely peaked (see Figure 3d) and indicates that current usage of the chemical probably controls its summertime air concentration. Fourteen tonnes of endosulfan were estimated to have been used in southern Ontario in 1988, mostly on vegetable crops (58). Given its summertime concentration, it appears that the monitoring of endosulfan and its sulfate in precipitation would be important in determining its atmospheric pathway. Four DDT species have been measured in our air samples, 2,4’-DDT, 4,4’-DDT, 4,4’-DDE (which is a dehydrochlorination product of DDT), and 4,4’-DDD. All of these species have been found in commercial DDT mixtures (59). Despite its ban in the United States and Canada, DDT continues to be detected in ambient air in Canada, possibly because of its continued use in Latin America. The sum of the DDT compounds reached a maximum concentration of over 0.5 ng m-3, a level which has only been reported in the early 1980s in Texas (14). The average annual concentration of 90 pg m-3 is comparable to results obtained in the lower United States in the early 1980s and indicates that there is no apparent decrease in DDT concentration in air. This would seem to indicate that either DDT has become ubiquitous in the environment or else t4ere continue to be fugitive sources of the chemical. The annual cycle of CDDT (Figure 3b) is similar to the PCBs and PCCs and indicates that its air concentration may be driven by volatilization processes. It clearly is less markedly peaked than the current-use pesticides (endosulfan, trifluralin, methoxychlor). A M for CDDT is -20, while for CCHLOR it is 8.6 and for EHCH it is -4. The current-use pesticides have A M over 100 and for trifluralin exceeding 1000. Environ. Sci. Technol., Vol. 26, No. 2, 1992 271

Figure 4c shows the ratio of 4,4’-DDE to CDDT. Since DDE is expected to be a degradation product of DDT, an increase in this ratio may indicate the age of the material. There is a peak in the ratio of -3 in winter vs 1.5 in summer, indicating that some conversion of DDT to DDE may be occurring. Confounding this interpretation is that DDE and DDT have different vapor pressures and Henry’s law coefficients (60) and other processes may affect the ratio. Pentachloranisole (PCA) is a microbial metabolite of pentachlorophenol (PCP), a widely used fungicide in Canada and Europe. A survey of PCP (measured as PCA) levels in European air, using pine needles as a monitoring tool, indicated that this compound is a widespread air pollutant (61). Swedish pine needle samples had twice the levels of those in the rest of Europe. This may reflect past use of PCP in Sweden and continued use in neighboring countries such as Finland. PCP use was banned in Canada in 1989 but older stocks are probably still in use. PCA is a minor species in air, peaking at 130 pg m-3, which shows a relatively uniform annual cycle having AMof less than 4 and no clear annual cycle. In 1986, in Ontario, 4053 kg of methoxychlor was sold, and in Canada as a whole, 13775 kg was sold (62), mainly for biting fly control. Its abundance in southern Ontario is low, less than 5 pg m-3, as a monthly average (Figure 3c). The peak concentrations are seen in the months of June, August, and September, corresponding to insect control periods. The detection of methoxychlor thus seems to be an indication of local or regional application. Methoxychlor was not detected from November through February. From this observation (or lack of it), it appears that the residence time of this chemical in air must be short (on the order of 1 month or less). The annual cycle of the sum of the a-, p-, and y-HCH isomers is shown in Figure 3a. AM for CHCH (-4) is limited by the ubiquitousness of the a-HCH, which varies only from a minimum of 77 pg m-3 to a maximum of 260 pg m-3 throughout the year. Lindane shows a more peaked annual cycle. It is known that local usage of lindane occurs near the Egbert research station during limited periods in summer months, even though a survey of southern Ontario agricultural use of lindane reported zero usage in 1988 (58). Use of lindane on coniferous trees (in Christmas tree farming, for example) is still allowed in Canada. The ratio of a-HCH/y-HCH is shown in Figure 4d. A clear winter maximum of 7 is seen, with summer values well below that of the technical mixture. The y isomer, lindane, is the active ingredient in the technical mixture, which contains 5540% a and 8-15% y (63). In the technical mixture the a / y ratio should be about 3.6-10. This indicates that recent usage of HCH is dominated by the active ingredient, y-HCH, with decreased or no usage of the inactive a-HCH isomer. For this reason, the a / y ratio is a useful diagnostic of recent vs old usage of HCH. The values of the ratio are never as large as are seen in the Canadian Arctic (5)and show an opposite annual cycle. The minimum of the a / y ratio is 1 2 in the Arctic in midwinter. The cycles of temperature in the Arctic and in midlatitudes show the same functional cycle (albeit with different ranges), so a volatility-driven explanation to this discrepancy seems unlikely. Another possible explanation is that the ratio is controlled by precipitation-related processes. This would have the correct annual behavior in that rain occurs predominantly in the winter in midlatitudes (when the a / y ratio is highest) and in the summer in the Arctic (also when the a / y ratio is the highest). The argument against this mechanism is that it is believed that

-

272

Environ. Sci. Technol., Vol. 26, No. 2, 1992

not enough rain events exist to be able to greatly affect the a / y ratio by washout (64). Photochemical transformation of y-HCH to a-HCH as a controlling factor in the ratio would not seem likely given the reverse Arctic annual behavior. A more probable reason for the differing behaviors in the a / y ratios is due to transport. During winter in southern Canada, the HCH ratios may represent air that has spent a significant time north of the polar front. This aged air has probably experienced the same processes as has the air in the Arctic (mostly cold and dry) and thus a higher a / y ratio might be expected. In summer, air in the Arctic is cut off from midlatitudinal exchange (since the polar front has moved north of most of the sources), undergoes persistent drizzle in Arctic stratus clouds, and thus experiences even less input of the more soluble yHCH. In southern Canada, in contrast, y-HCH is used regionally or imported readily in summer synoptic systems from other locations and thus the a / y ratio decreases. Trifluralin. Trifluralin is a current-use preemergence herbicide used in wheat and corn planting in order to attack grasses and competing vegetation. It has a relatively high vapor pressure of 6 X Pa at 20 OC but also has a high Henry’s constant of 4.02 Pa m3 mol-l (60). The annual cycle of trifluralin is compared in Figure 3d to another current-use pesticide, endosulfan. The maximum concentration of trifluralin in June exceeded 3.4 ng m-3. Trifluralin is the second most abundant organohalogen measured in this study. Trifluralin is not prevalent in air long into the fall, however, and this must be a due primarily to use pattern. The high Henry’s law coefficient would make washout in rain unlikely and should lengthen the atmospheric residence time. Since over 2 X lo6 kg of trifluralin is used annually in the prairie provinces of Canada to the west, long-range transport of trifluralin is possible and 139 tonnes of trifluralin have been used in southern Ontario (58). A limitation to the atmospheric lifetime would be that trifluralin has been reported to have a number of photolytic breakdown products (65) and thus may not be stable in air for long periods. Chlorobenzenes (CBzs). Four chlorobenzene species were measured in this study, 1,2,3,4-tetrachloro-, 2,3,4,5tetrachloro-, pentachloro-, and hexachlorobenzene. These chemicals have high volatility in relation to other species measured here and do break through the foam plugs in summer months (49). For this reason, the data shown here should be considered to be a minimum of the air concentration and used with some caution. The annual cycle of the chlorobenzenes is shown Figure 3a. There is a minimum in the amount of CCBz detected in summer, which substantiates the speculation that the material rapidly passes through the foam plugs at air volumes of >250 m-3 and at summer temperatures (49). We know of no other physical process which would allow for a minimization of the concentration of these chemicals in warm summer months. Mirex. The final organohalogen routinely searched for was mirex. Mirex was used extensively in the southern United States as a control against fire ants. It has been detected in the Great Lakes, especially in Lake Ontario (66). Mirex was banned from importation into Canada in 1978. The low volatility and high solubility of mirex makes atmospheric transport unlikely, and previous reviews of the atmospheric pathways of OCs have decided that the evidence suggests ignoring this chemical in the atmosphere (45). We have detected mirex in air in only the five most polluted air samples. The highest concentration was 22 pg m-3 for the July 10, 1988, persistent elevated pollution

episode, which had PCB air concentrations of greater than 2 ng m-3. It appears that the detection of mirex in air is rare, and an aquatic source for the chemical in Lake Ontario is probable. Comparison with Previous Observations The PCB concentrations seen in this study are considerably lower than those seen in studies from the early 1980s in the Great Lakes. The summer average concentrations taken from the July average data are comparible to concentrations seen on Lake Superior in 1979 and 1980 (25). The Ontario average yearly average data are similar to that seen by Singer et al. (24)in 1980, using their two-column peak-matching technique. This latter technique is conservative in reporting PCB congener peaks and the results for 1980 were considerably lower than the 1979 Ontario Ministry of Environment results (24). Manchester-Neesvig and Andren have published an annual cycle of PCB concentrations for northern Wisconsin for 1985 (34). The summer, winter, and annual average CPCB concentrations from that study are a factor of -2 higher than those found here. On a congener-by-congener basis, though, summer and winter concentrations of PCB 52 and PCB 118 are actually higher in this study. The concentration of PCBs 28+31 measured in the Wisconsin study are a factor of -2 higher than the sum of PCB 28 and PCB 31 here. This study did not determine filtertrapped PCB congeners and, therefore, cannot confirm their observation of the existence of a February submaximum in the PCB concentration due to particulate-bound PCB (34). In a study on Siskiwit Lake (Isle Royale, Lake Superior) in 1983-1984, Swackhamer et al. (32) found air concentrations which were largely comparable to those found here. The exception to this is in the lightweight PCB congeners 18 and 33, which were 10 times and 2 times higher, respectively, than those seen in this study. There is the possible interpretation that the Minnesota researchers had higher capture efficiency for the lower weight congeners than was observed here. Bruckmann et al. (36)and Duinker and Bouchertall(37) have reported PCB congener concentrations in two urban studies in Germany. The former study, which took place over a l-year sampling time, has a nearly uniform congener spectrum with the heaviest PCBs almost as concentrated as the light PCBs. The study by the group in Kiel (37) gives congener concentrations (from the average of four samples) almost identical to the results here except for the lightest PCB reported, PCB 18, where again our concentration is very low. From the comparison to the PCB congener patterns seen in other studies, it appears that PCB 18 has been undersampled in this study. The study of the frontal movement of PCB 31 (2,4',5-trichlorobiphenyl) through 7.5 cm of foam at 20 "C (49) would indicate less than 10% breakthrough of this species. PCB 18 has -3 times the PCB 31 vapor pressure (70) and would be more susceptible to breakthrough. In July-August, 1988, PCB 18 was not reported from our laboratory, and since this is a hot period, one might expect that the PCB 18 concentrations annually have been biased to the low side. With the exception of the Hamburg results (36),which indicate some local sources, both the results here and in the literature show that a-HCH appears to have reached a widespread concentration averaging about 200-250 pg m-3 in summer months and 50-150 pg m-3 in winter months. Lindane (yHCH) is more variable but appears to be about 10-20% of a-HCH in undisturbed air. The summertime concentration average at Egbert is known to

be disturbed by local sources of lindane and certainly cannot be expected to be representative of the continent as a whole. Nevertheless, the undisturbed northern Wisconsin site (34) has lindane average concentrations which are 60% of the Egbert abundances, showing that the impact of the local sources at Egbert may not be too dominant in the long term. CDDT concentrations appear to be as high as they were since the mid-1970s. If any trend is apparent, it is that summertime inputs are actually higher than they have been in the mid-1980s. This is a disturbing result, if true, and indicates that there must be current inputs of DDT, DDE, and DDD to the North American atmosphere whether by direct or fugative sources. PCC concentrations in the summertime appear to be about the same as were seen by Rice et al. (29)in their 1981 study in Michigan. The quantitation method used here is conservative in comparison to that used by Rice et al. Our quantitation method using GC-ECD compares favorably with a negative chemical ionization method in fish and whale extracts (55). For the air concentration to remain essentially unchanged over the 19808, it is apparent that current sources of this chemical must exist. Dieldrin concentrations are similar to earlier studies, except for the mid-Atlantic coastal missions of Knap and Binkley (39). Chlordanes appear to be significantly lower in abundance than early-1970 reported values, although some of those studies were near known local sources of chlordane (14,18,26). The concentration of chlordanes would indicate that, on the whole, Ontario is a receptor for long-range transport of chlordane from a larger upwind source. Functional Representations of the Annual Cycles of the PCBs and the Organohalogens The strong variation in the annual cycle of these chemicals, with AM varying from a minimum of 2-3 to a maximum of over 1O00, is an important feature to parameterize for use in models which characterize atmospheric deposition. These concentrations will be directly reflected in dry deposition estimates through a flux derived from a deposition velocity times the mean concentration or in wet deposition estimates through fluxes derived from washout or snowout ratios times the air concentrations. In either case, the functional behavior of the annual cycle can be used to scale the deposition by time of year so that the true annual deposition might better be derived, Eisenreich (67) has presented this cycle as a conceptual sketch, which looked like a sinusoid with maximum in summer and minimum in winter. The data here show that for some species, a-HCH for example, that type of estimator might be appropriate. For others, with large AM, clearly a more peaked distribution might be more useful. We have examined three distributions, which can be easily parameterized in models: sinusoidal, Gaussian, and Lorentzian. The best fit for the majority of the species measured was the Lorentzian:

where I? is the half-width in months of the distribution, 7 is the time in months into the year (Le., January 15 would be 0.5), and 7,,, is the time in months of the maximum concentration. Table III shows the results of fitting the monthly average data to the above distributions. On average, for the species shown in the table, the variance of the data from the distribution is 80% higher for the Gaussian and 38 times Environ. Sci. Technol., Vol. 26, No. 2, 1992

273

Table 111. Best Functional Fit for Monthly Data

r,

chemical

Xmim

7mam

pg m-3

month

month

CY-HCH Y-HCH CCHLOR PCC

77 12 20 0.08 11 3.9 0.33 7.1 55 25 19 3.6 3.0

8.2 6.3 6.6 6.8 6.8 6.7 5.6 6.4 6.7 6.8 6.7 6.6 6.6

2.2 0.9 1.7 1.0 1.7 0.8 0.76 2.2 0.67 1.1 0.65 0.51 0.18

CDDT endosulfan trifluralin dieldrin CPCBs C1,-PCB Cld-PCB C1,-PCB C1,-PCB

3.3 17 8.6 1300 20 470 3900 13 14 8.4 14

40 40

1500 OO303 Z-PCB x x x x x Trifluralin

n c1

1

1000

E

2 0

0

0 1 2 3 4 5 6 7 8 9101112

r (months) Figure 5. Comparison of the experimental concentrations with the Lorentziin fii from Table 111: (a) CPCBs and trifluralin; (b) a-HCH and

CDDT.

higher for the sinusoid than for the Lorentzian. Examples of the fit for CPCBs, trifluralin, a-HCH, and CDDT are shown in Figure 5. Since this procedure has been applied to only one site and 1year, this parameterization should not be considered conclusive at this point. However, considering the simplicity of the Lorentzian and its ease of computation, it would be a preferred choice for parameterizing the annual cycle of these species. The position of the annual maxima in months is generally mid- to late-July except for trifluralin and a-HCH. r is generally greater than 1month for the pesticides and less than 1month for the PCBs. The width of the hexachlorobiphenyl should be viewed skeptically since the July monthly average dominates the annual cycle. For short atmospheric cycles such as this, the high-resolution data should be used to determine a half-width of the cycle. Given, however, that for the heavier PCBs one would expect some particulate-bound material to be comparible to the vapor-phase concentrations in winter, it was felt unjustified to extend this technique here. Conclusions

This study has obtained a unique high temporal resolution data set of PCBs and OCs in the Great Lakes region. The results show a large variation in the concentration of the chemicals as a function of time of year, but this dynamic behavior can be reduced to a relatively simple functional format for parameterization in models. In comparison to past data, it appears that PCBs, chlordanes, and a-HCH concentrations are decreasing in 274

Acknowledgments

We thank F. Froude, B. Martin, F. Maclean, and P. Heck for the acquisition of the samples and C. Smith for the analysis of the samples during this project. The preparation of samples by Concord Scientific Corp. (M. Underwood and P. Fellin) is acknowledged. Supplementary Material Available

500

W

F:

air. Other species which have been controlled, such as DDT, PCCs, and dieldrin, are not decreasing at a noticeable rate. The overall concentration trends in air will be difficult to detect, however, given the large amplitudes of the annual behavior. Given past experience in the detection of trends from highly variable signals (COz and tropospheric aerosol loadings, for example), long time series are needed to detect such trends and even then can be inconclusive. Integrative techniques encompassing many years of integration [e.g., peat cores (28)] and modeling may be the most useful diagnostics of the efficacy of control strategies for trace organics.

Environ. Sci. Technol., Vol. 26, No. 2, 1992

Table IV containing the concentrations of each of the analyzed PCB congeners by monthly averages and Table V giving the monthly average concentrations of each of the remaining organohalogen species measured (3 pages) will appear following these pages in the microfilm edition of this volume of the journal. Photocopies of the supplementary material from this paper or microfiche (105 X 148 mm, 24X reduction, negatives) may be obtained from Microforms Office, American Chemical Society, 1155 16th St., N.W., Washington, DC 20036. Full bibliographic citation (journal, title of article, authors’ names, inclusive pagination, volume number, and issue number) and prepayment, check or money order for $10.00 for photocopy ($12.00 foreign) or $10.00 for microfiche ($11.00 foreign), are required.

Literature Cited (1) McNeely, R.; Gummer, W. D. Arctic 1984, 37, 210. (2) Hoff, R. M.; Chan, K.-W. Chemosphere 1986, 15, 449. (3) Bidleman, T. F.; Patton, G. W.; Walla, M. D.; Hargrave, B. T.; V m , W. P.; Erickson, P.; Fowler, B.; Scott, V.; Gregor, D. J. Arctic 1988, 42, 307. (4) Hargrave, B. T.; Vass, W. P.; Erickson, P.; Fowler, B. R. Tellus 1988, 40B, 480. (5) Patton, G. W.; Hinckley, D. A.; Walla, M. D.; Bidleman, T. F. Tellus 1989, 41B, 243. (6) Bidleman, T. F.; Patton, G. W.; Hinckley, D. A.; Walla, M. D.; Cotham, W. E.; Hargrave, B. T. Chlorinated pesticides and polychlorinated biphenyls in the atmosphere of the Canadian Arctic. In Long Range Transport of Pesticides; Kurtz, D. Ed.; Lewis Publishers: Chelsea, MI, 1990. (7) Oehme, M.; Mano, S. Fresenius 2. Anal. Chem. 1984,319, 141. (8) Oehme, M.; Ottar, B. Geophys. Res. Lett. 1984, 11, 1133. (9) Semb, A.; Pacyna, J. M. Toxic Trace Elements and

Chlorinated Hydrocarbons: Sources, Atmospheric Transport and Deposition; Report NORD 198810; Nordic Council of Ministers: Copenhagen, Denmark, 1988. (10) Gregor, D. J.; Gummer, W. Environ. Sci. Technol. 1989, 23, 561. (11) Tanabe, S.; Tatsukawa, R. J.Oceanogr. SOC.J p n . 1980,36, 217. (12) Tanabe, S.; Tatsukawa, R.; Kawano, M.; Hidaka, H. J . J p n . 1982, 38, 137. Oceanogr. SOC. (13) Bidleman, T. F.; Leonard, R. Atmos. Environ. 1982, 16, 1099. (14) Atlas, E.; Giam, C. S. Science 1984, 211, 163. (15) Atlas, E. L.; Bidleman, T. F.; Giam, C. S. Atmospheric

Transport of PCBs to the Oceans. In PCBs in the Environment; Ward, J. S., Ed.; CRC Press: Boca Raton, FL, 1986. (16) Harvey, G. R.; Steinhauer, W. G. Atmos. Environ. 1974, 8 , 777.

(17) Bidleman, T. F.; Olney, C. E. Bull. Enuiron. Contam. Toxicol. 1974, 11, 442. (18) Bidleman, T. F.; Olney, C. E. Nature 1975,257, 475. (19) McClure, V. E. Environ. Sci. Technol. 1976, 10, 1223. (20) Bidleman, T. F.; Christensen, E. J. J. Geophys.Res. 1979, 84, 7857. (21) Bidleman, T. F.; Christensen, E. J.; Billings, W. N.; Leonard, R. J. Mar. Res. 1981, 39, 443. (22) Doskey, P. V.; Andren, A. W. J. Great Lakes Res. 1981, 7, 15. (23) Bidleman, T. F. Atmos. Enuiron. 1981, 15, 619. (24) Singer, E.; Jarv, T.; Sage, M. Survey of PCB in Ambient Air Across the Province of Ontario. In Physical Behavior of PCB in the Great Lakes; Mackay, D., Paterson, S., Eisenreich, S. J., Simmons, M. Eds.; Ann Arbor Science: Ann Arbor, MI, 1983. (25) Eisenreich, S. J.; Looney, B. B.; Hollod, G. J. In Physical Behavior of PCBs in the Great Lakes; Mackay, D., Paterson, s.,Eisenreich, S. J., Simmons, M., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983; pp 115-25. (26) Billings, W. N.; Bidleman, T. F. Atmos. Enuiron. 1983,17, 383. (27) Rapaport, R. A,; Urban, N. R.; Capel, P. D.; Baker, J. E.; Looney, B. B.; Eisenreich, S. J.; Gorham, E. Chemosphere 1985, 14, 1167. (28) Rapaport, R. A.; Eisenreich, S.J. Atmos. Environ. 1986, 20, 2367. (29) Rice, C. P.; Samson, P. J.; Noguchi, G. E. Environ. Sci. Technol. 1986,20, 1109. (30) Bidleman, T. F.; Wideqvist, U.; Jansson, B.; Soderlund, R. Atmos. Environ. 1987, 21, 641. (31) Rapaport, R. A.; Eisenreich, S. J. Environ. Sci. Technol. 1988, 22, 931. (32) Swackhamer, D. L.; McVeety, B. D.; Hites, R. A. Enuiron. Sci. Technol. 1988, 22, 664. (33) Atlas, E.; Madero, M.; Giam, C. S. Water, Air, Soil Pollut. 1988, 38, 19. (34) Manchester-Neesvig, J. B.; Andren, A. W. Environ. Sci. Technol. 1989,23, 1138. (35) Baker, J. E.; Eisenreich, S.J. Environ. Sci. Technol. 1990, 24, 342. (36) Bruckmann, P.; Kersten, W.; Funcke, W.; Balfanz, E.; Konig, J.; Theisen, J.; Ball, M.; Papke, 0. Chemosphere 1988,17, 2363. (37) Duinker, J. C.; Bouchertall, F. Environ. Sci. Technol. 1989, 23, 57. (38) Knap, A. H.; Binkley, K. S. Atmos. Environ., in press. (39) Knap, A. H.; Binkley, K. S. Atmos. Environ., in press. (40) Whelpdale, D. M.; Knap, A. H.; Toom, D. L. Global Biogeochem. Cycles, submitted. (41) Church, T. M.; Tramantano, J.; Whelpdale, D. M.; Andreae, M. 0.;Galloway, J. N.; Keene, W. C.; Tokos, J. J. Geophys. Res., in press. (42) Lane, D. A.; Schroeder, W. H.; Johnson, N. D. Atmos Environ., in press. (43) Hermanson, M., Hites, R. E. Enuiron. Sci. Technol. 1989, 23, 1253. (44) Atlas, E. Long Range Transport of Organic Compounds. NATO Research Workshop on Long Range Transport of Natural and Contaminant Substances from Continent t o Ocean and Continent to Continent; NATO: Brussels, 1988. (45) Strachan, W. M. J.; Eisenreich, S.J. Mass Balancing of Toxic Chemicals in the Great Lakes: The Role of At-

(46)

(47) (48) (49) (50) (51) (52) (53) (54) (55)

mospheric Deposition; International Joint Commission: Windsor, ON, Canada, 1988. Barrie, L. A.; Gregor, D. J.; Hargrave, B. T.; Lake, R.; MU^, D. C. G.; Shearer, R.; Tracey, B.; Bidleman, T. F. Arctic Contaminants: sources, occurrence and pathways. Workshop on Scientific Evaluation of Chemical Contaminants in the North, Ottawa, ON, Canada, February 1989. Hoff, R. M.; Muir, D. C. G.; Grift, N. B. Enuiron. Sci. Technol., following article in this issue. Eliassen, A.; Saltbones, J. Atmos. Environ. 1983,17,1457. Burdick, N. F.; Bidleman, T. F. Anal. Chem. 1981,53,1926. Ligocki, M. P.; Pankow, J. F. Environ. Sci. Technol. 1989, 23, 75. Foreman, W. T.; Bidleman, T. F. Environ. Sci. Technol. 1987, 21, 869. Hinckley, D. A.; Bidleman, T. F.; Foreman, W. T.; Tuschall, J. R. J. Chem. Eng. Data 1990,35, 232. Muir, D. C. G.; Norstrom, R. J.; Simon, M. Environ. Sci. Technol. 1988,22, 1071. Chuang, J. C.; Hannan, S. W.; Wilson, N. K. Enuiron. Sci. Technol. 1987, 21, 798. Muir, D. C. G.; Ford, C. A,; Grift, N. P.; Stewart, R. E. A.; Bidleman, T. F. Environ. Pollut., submitted. Hawker, D. W. Environ. Sci. Technol. 1989,23, 1250.

(56) (57) Chlordane. Its Effects on Canadian Ecosystems and its Chemistry;No. 14094; National Research Council of Canada: Ottawa, ON, Canada, 1974. (58) Moxley, J. Survey of Pesticide Use in Ontario, 1988; Ontario Ministry of Agriculture and Food: Toronto, ON, Canada, 1989. (59) McEwen, F. L.; Stephenson, G. R. The Use and Significance of Pesticides in the Environment; John Wiley and Sons: New York, 1979. (60) Suntio, L. R.; Shiu, W. Y.; Mackay, D.; Seiber, J. N.; Glotfelty, D. Rev. Environ. Contam. Toxicol. 1988,103, 1. (61) Erickson, G.; Jensen, S.; Kylan, H.; Strachan, W. Nature 1989, 341, 42. (62) Commercial Chemicals Branch, Environment Canada, Ottawa, ON, Canada, personal communication, 1989. (63) Metcalf, R. L. Organic Insecticides, Their Chemistry and Mode of Action; Interscience Publishers: New York, 1955. (64) Bidleman, T. F., personal communication, 1990. (65) Leitis, E.; Crosby, D. G. J. Agric. Food Chem. 1974,22,842. (66) Oliver, B. G.; Charlton, M. N.; Durham, R. W. Environ. Sci. Technol. 1989, 23, 200. (67) Eisenreich, S. J. Proceedings, International Association of Great Lakes Research Annual Meeting; Gobas, F., Ed.; University of Windsor: Windsor, ON, Canada, 1990. (68) Glotfelty, D. E. J. Air Pollut. Control Assoc. 1978,28,917. (69) Foreman, W. T.; Bidleman, T. F. Atmos. Environ. 1990, 24A, 2405. (70) Murphy, T. J.; Mullin, M. D.; Meyer, J. A. Environ. Sci. Technol. 1987, 21, 155.

Received for review February 22, 1991. Revised manuscript received July 19, 1991. Accepted August 21, 1991. R.M.H. acknowledges the support of the Wave Propagation Laboratory of NOAAIERL, NASA Jet Propulsion Laboratory, and NASA Langley Research Center during his Professional Development Leave and in the preparation of this article. The financial support of the Great Lakes Program Office made this work possible. K . Brice, D. A. Lane, and two helpful reviewers are thanked for a critical reading of the manuscript.

Environ. Sci. Technol., Vol. 26, No. 2, 1992

275