[:re,~,enius' Journal of Fresenius J Anal Chem (1994) 348:226-239

@ Springer-Verlag1994

Levels of polychlorinated biphenyls in the lower troposphere of the North- and South-Atlantic Ocean Studies of Global Baseline Pollution XVII* J6rn Schreitmiiller and Karlheinz Ballschmiter Department of Analytical and Environmental Chemistry, University of Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Received June 5, 1993

Summary. Fourteen polychlorinated biphenyl (PCB) congeners were quantified in air samples of the tropospheric boundary layer of the Atlantic Ocean. The samples were taken on the German research vessel "Polarstern" during north-south cruises across the Atlantic Ocean (1990, 1991), and on the Capo Verde islands in the North Atlantic Ocean (1992). Values for the sum of PCB were between 48 pg/m 3 (values for the seven indicator congeners were [in pg/m3]: PCB 28: 1.3; PCB 52: 5.0; PCB 101: 3.0; PCB 118: <0.5; PCB 138:1 ; PCB 153: 1; PCB 180: < 0.2) in the Westwind Belt of the eastern North Atlantic and 22 pg/m 3 (values for the seven indicator congeners were [in pg/m3]: PCB 28: 2.3; PCB 52: 3.4; PCB 101: <0.5; PCB 118: <0.2; PCB 138: <0.2; PCB 153: <0.2; PCB 180: <0.2) in the Westwind Belt of the central South Atlantic. Up to 385 pg/m 3 (values for the seven indicator congeners were [in pg/m 3] : PCB 28: 2.6; PCB 52: 11.7; PCB 101: 28.4; PCB 118: 9; PCB 138: 21 ; PCB 153: 18; PCB 180: 5.5) were measured of the coast of South Patagonia. A difference depending on latitude and on terrestrial influenced air masses between the lower and the higher chlorinated congeners was observed. The levels of three- and tetrachlorinated congeners were highest in the Trade Wind regions. The contents of the higher chlorinated congeners had maxima in samples influenced by continental air masses. A correlation of the levels of the lower chlorinated congeners in air over the South Atlantic with the surface water temperature and thus with the temperature dependent gas/water partition coefficient Kg w w a s observed.

1 Introduction The global distribution of a chemical is the result of the sum of all local and regional, horizontal and vertical transportation processes in the atmosphere and hydrosphere, including the exchange between environmental compartments. These translocations are superimposed by abiotic and biotic transformation processes. The general aspects of the transport and the fate of organic molecules in the global environment have recently been reviewed [1 - 3]. * Part X V I : Fischer RC, Kr/imer W, Ballschmiter K (1991) Chemosphere 23:889-900 Correspondence to: K. Ballschmiter

During long-range transport of a compound either in the atmosphere or in the oceanic system, multi-phase distribution has to be taken into account [4]. The question remains open whether the multi-phase distribution has to be discussed in terms of a thermodynamic equilibrium, exemplified by the fugacity based unit world approach of Mack_ay [5], or whether regional concentration dependent global mass flow and molecular diffusion based models are more realistic in describing the heterogeneity of the actual and often changing situations at the various places of the globe [6]. The time scale of a transition from non-equilibrium conditions, which are given by any type of input per se, to the equilibrium status has far-reaching consequences for the long-range transport of pollutants. This problem can be examined by determining the concentration of the compound in two compartments at source-near and source-distant locations and comparing the obtained values with the equilibrium conditions given by the physico-chemical properties of the compound. The time scale of a relaxation to attain the troposphere/ oceanic surface water equilibrium of anthropogenic compounds has to our knowledge not yet been determined under environmental conditions. For this reason we have started a program of a series of measurements of semivolatile organohalogenated compounds (SOC) in the Atlantic Ocean, of which we present here the results for the polychlorinated biphenyls (PCB) in the lower troposphere. The complex spatial and temporal picture of the reality can only be described by problem oriented and accurate analytical measurements, the classical challenge in environmental chemistry. The analytical chemistry has to include representative sampling, highly effective and evaluated methods for the enrichment, separation, detection, identification of the analytes, and validation of the quantitative results. Any possible source of contamination has to be examined and to be excluded. The discussion of the environmental fate of the polychlorinated biphenyls makes a congener specific approach mandatory [7]. The analytical tools to determine PCBs and other SOCs by high resolution capillary gas chromatography have been available for about 15 years [8] and have been developed ever since. However, even recent studies either treat the PCB mixture as one. chemical [9] or take mean values of too many measurements so that details for specific congeners may be obscured. In this work we report on the occurrence of fourteen selected polychlorinated biphenyl congeners (C1xB, x -- 3 -

227 7) in the boundary layer ( 0 - 1 5 0 0 _+ 500 m above sea level) of the Atlantic Ocean from 50 ° North to 50 ° South. We discuss the possible input and output on the basis of the given meteorological conditions in the sampling areas and on the basis of the physico-chemical properties of the individual PCB congeners. The air samples were collected during north/ south cruises on the German research vessel "Polarstern" (Cruises ANT IX/I and ANT IX/4) of the "Alfred-WegenerInstitute for Polar Research, Bremerhaven" across the Atlantic Ocean in fall 1990 and spring 1991. In addition, we take into account two air samples collected in March 1992 on the Capo Verde islands in the Trade Wind region of the northern Atlantic Ocean. From the numerous PCB congeners detectable in the air samples we selected for a quantitation the trichlorobiphenyls 2,4,4'-C13B (PCB 28), 2,4',5-C13B (PCB 31), the tetrachlorobiphenyls 2,2'3,5'-C14B (PCB 44), 2,2',4,5'-C14B (PCB 49), 2,2',5,5'-C14B (PCB 52), the pentachlorobiphenyls 2,2',3,4,5'-C15B (PCB 87), 2,2',4,5,5'-C15B (PCB 101), 2,3,3',4',6-C1sB (PCB 110), 2,3',4,4',5-C1sB (PCB 118), the hexachlorobiphenyls 2,2',3,4,4',5'-C16B (PCB 138), 2,2',3,4',5',6-C16B (PCB 149), 2,2',3,5,5',6-C16B (PCB 151), 2,2',4,4',5,5'-C16B (PCB 153) and the heptachlorobiphenyl 2,2',3,4,4',5,5'-C17B (PCB 180). The PCB numbering system as suggested by Ballschmiter and Zell is used [10]. The fourteen PCB congeners selected in this work cover a wide range of chlorination (trichloro up to heptachloro). They differ in the substitution pattern and have widely divergent physical properties, which means that they can be considered as representative for their respective group of congeners in the various technical PCB mixtures. The physico-chemical properties of the selected PCB congeners are summarized in Table 1. We have tried to make a critical selection of the data reported in the literature, also including our own measurements. For the gas/water partition coefficient Kgw, which describes the gas/water equilibrium distribution, the data in the literature are quite inhomogenous. It is surprising that this fundamental set of data needed for the discussion of the environmental fate of the PCBs is only partly available as validated data.

,xz .¢o

c~

o

o

V

x-~

o

c~

o

b~

o

0 {3

.o

i i

i

i

i

i

i

i

i

i

i

i

l

~ ~.~ I

t-q

i i

i

1

~

1

i i

i

i

i

1

II

~

.

~

& © o

~d2 ,..o

2 Experimental section ,2o

2.1 High-volume sampling of air

¢,, 0

©

. ~ ....o +.~ o

.~

~.2 " ~

.~m

0

o ~ t

? © o

Z :h < L;

°

&d 2z

i

i

i

i

i

1

~

i

i

i

i

i

©

~

~o

~~

..

~ 0

4.a

To exclude local contamination the air sampling procedure on the research vessel "Polarstern" was always accomplished windward on the upper-most deck of the ship about 20 m above sea level. The air samples on Capo Verde were taken at the south eastern tip on the small north eastern island Isla do Sal of the Capo Verde Archipelagos. The sampling place was on top of a two storey building on the windward side of a village. Samples of 5 5 0 - 1 1 5 0 m 3 of air were collected by a Str6hlein high-volume air sampler (Type: EM 101, Str6hlein, Kaarst, Germany) at a flow rate of 3 0 - 3 5 m3/h using an adsorption technique based on extensively precleaned silica gel 60 (Merck, Darmstadt, Germany). The application of this method for determining baseline levels of PCBs and SOCs in air is described in references [11, 12]. Each sampling gave as subsamples: 1. the particle filter (glass fiber filter 3481, Schleicher & Schuell, Dassel, Germany, separation efficiency 99.97% for oil droplets <1 gin, maximum at 0 . 5 - 0 . 3 ~tm), 2. the first adsorption layer

228

U 96

I

amoc 6 ,~n

I

175~c 21,5~

101

110

149

I

z~0~C 46.5n~n

153 /38

180

i

~5~: n.5 rran

Fig. 1. HRGC/ECD (column 3: CP-Sil 8 + 50% C 18) of the blank, aliquot corresponding to 20 m 3 of air (at a hypothetical sample volume of 1000 m 3 air) (100 g silica gel 60, mesh range 35-70), and 3. the second adsorption layer (100 g silica gel 60, mesh range 35 - 70). The second silica gel layer was used to check the breakthrough of the first main layer and thus to verify the collection efficiency. Only three- and tetrachlorinated PCBs occurred in detectable amounts in the second layer in samples 5, 6 and 7. In these samples the air temperature (> 25°C) and the humidity of the air ( > 80 %) were high, resulting in a stronger deactivation of the adsorbent. In this case the PCB amounts of the two silica gel layers were added together. The silica gel was transported prior and after sampling in modified 250 ml Erlenmeyer flasks closed by melting the added glass tube [11]. This technique effectively excludes any contamination during transport and long-time storage. It is simple, convenient and easier to handle than the widely used polyurethane plug technique [13]. The latter is superior in excluding the water vapour in air during sampling. To control the sampling procedure and the following sample preparation steps, a defined amount of 2 - 1 0 ng 1,2,3,4-tetrachloronaphthalene (TCN) was added as internal standard onto the first particle filter before sampling of air. The recovery rate of TCN was always between 80 and 105%. The measured values for the PCB congeners were not corrected for these recovery rates.

2.2 Sample preparation All sample preparation steps were accomplished under stringent control in a laboratory designed for extreme organic trace analysis and using a "metal/glass only" clean-bench with charcoal air filters to minimize any possible contamination of the sample from the laboratory environment. The solvents used (hexane, dichloromethane, isooctane) were of nanograde quality (Promochem, Wesel, Germany). All glass-

ware used has been cleaned by heating to 350°C and by rinsing with hexane and dichloromethane before use. The silica gel used for air sampling was packed in a glass column (length 20 cm, diameter 4 cm) and eluted with 400 ml dichloromethane. This amount of dichloromethane is sufficient to elute all the polychlorinated biphenyls as well as more polar substances like hexachlorocyclohexans quantitatively from the adsorbent. The glass fiber filters were Soxhletextracted for 24 h with dichloromethane and hexane. In the next step, the volume of the eluate as well as of the extract was reduced to about I ml transferred to hexane in a specially cleaned rotary vacuum evaporator (Bfichi, Flawil, Suisse). The analytes were preseparated on 4.5 g silica gel 60, 3 5 - 7 0 mesh, 3% H20 [14]. The first fraction (LC 1), eluted with 30 ml hexane, contains the weakly polar substances (polychlorinated benzenes, polychlorinated biphenyls, pentachloroanisole and part of the polychlorinated naphthalenes including the internal standard 1,2,3,4tetrachloronaphthalene, 4,4'-DDE). The second fraction (LC 2), eluted with 40 ml hexane/dichloromethane (3:i), contains more polar substances like the hexachlorocyclohexanes, endosulfan, dieldrin, toxaphen compounds, part of the polychlorinated naphthalenes, the main part of the chlordane group and also the main part of the chlorinated methyl-phenylethers (anisoles) [11]. The fractions were concentrated to 5 0 - 2 0 0 ~tl after adding isooctane as keeper. The absolute amount handled per PCB congener was in the range of 0.2 to 35 ng. No systematical losses of polychlorinated biphenyls have been observed during these preparation steps when checked with adsorbents spiked with these compounds. A chromatogram of the blank of the whole preparation procedure is given in Fig. 1. The isooctane solutions were kept in brown vials (Zinsser, Frankfurt, Germany), which minimize losses during long-time storage.

229

1'468

$8

PCSz

07 49

138

232

146 141 174 I~ 8 It [ 1~1871831177 1~ z \128 156

liT,

15,5 rrdn

175°C

170

~,

2(]~C

40,5 ~n

65,5 rr~

>

Fig. 2. HRGC/ECD (column 2: CP-Sil 8 + 10% C 18) of air sample 1 (46°N, 12°W), aliquot: 5 m 3

2.3 High-resolution gas chromatography

90°C (3 min) - 2 0 ° / m i n - 1 5 0 ° C ( 0 m i n ) - 1 . 2 ° C / m i n (10 rain); carrier gas: hydrogen (100 kPa). Injection technique: Splitless (250 ° C). The advanced separation of PCB congeners on methyloctadecyl polysiloxane coated capillary columns has been reported recently [15]. 270°C

The gas-chromatographic separation and identification of the substances was accomplished by capillary gas chromatography using the following capillary columns of different polarity:

Column 1. CP-Sil 5 + 10% C 18 (methyl-polysiloxane plus 10%-octadecyl-polysiloxane (Chrompack, Middleburg, Netherlands); length: 110 m; inner diameter: 0.25 mm; film thickness: 0.25 gin; silanized retention gap (Chrompack): 2 m; temperature program: 90°C (3 rain) - 20°/rain 150°C (0 rain) - 2°C/rain - 250°C (10 min); carrier gas: hydrogen (300 kPa). Injection technique: Splitless (250 ° C). Column 2. CP-Sil 8 + 10% C 18 (methyl-5%-phenylpolysiloxane plus 10%-octadecylpolysiloxane) (Chrompack); length: 50 m; inner diameter: 0.25 mm; film thickness: 0.20 gin; silanized retention gap (Chrompack): 2 m; temperature program: 90°C (3 rain) - 2 0 ° / r a i n - 140°C (0 rain) - 1° C/rain - 230 °C (10 rain); carrier gas: hydrogen (J 25 kPa). Injection technique: Splitless (250 ° C). Column 3. CP-Sil 8 + 50% C 18 (methyl-5%-phenyl-polysiloxane plus 50%-octadecylpolysiloxane) (Chrompack); length: 105 m; inner diameter: 0.32 ram; film thickness: 0.1 gm; silanized retention gap (Chrompack): 2 m; temperature program: 90 ° C (3 min) - 20°/min - 160 ° C (0 min) 1° C/min - 245 ° C (5 min); carrier gas: hydrogen (185 kPa). Injection technique: Splitless (250 ° C). Column 4. DB 1701 (7%-cyanopropyl-7%-phenyl-methylpolysiloxane) (J & W, Folsom, CA, USA); length: 30 m; inner diameter: 0.20 mm; film thickness: 0.25 gin; silanized retention gap (Chrompack): 2 m ; temperature program:

2.4 Quantitation of the PCB congeners The substance were detected with an electron capture detector (ECD) of a Varian 3700 GC (Varian, Palo Alto). The temperature of the ECD was always 300 ° C. In addition, a mass-spectrometric detector of the ion trap type (Saturn II, Varian) was further used for the identification. The quantitation by the ECD was carried out using relative response factors of the selected 14 PCB congeners to PCB 53 (2,2',5,6'-C14B) or PCB 103 (2,2',4,5',6-C15B), two congeners which did not occur in the air samples. Before these congeners were added, the sample solutions were injected into the gas-chromatographic system in order to verify if the internal standards were suitable for a quantitation of the individual sample. On none of the C 18 columns used, PCB 103 coetutes with another PCB congener nor with other unidentified compounds. PCB 53 was not suitable for quantitation on column 3 as it coeluted with another unidentified compound. In some of the samples octachlorostyrene was used as internal standard. The standard solutions taken for quantitation were prepared by weighing the individual PCB congeners (Promochem, Wesel, Germany). In addition, the PCB calibration solution SRM 2262 and its supplement (NIST, Washington DC), which together contain 39 PCB congeners of a certified concentration, were used for the quantitation.

230 $8 l.St.

UL,

II~

HI 90 101

110 138

151 N L

141 '

I

I

I

~SIY'C 15,5 rr~ Fig. 3.

HRGC/ECD (column

146

175~ 40,5 ~n

/8~ ~187 I

180 174 156 I

,1,771

17o

I

200¢C 65,5 rr~

2: CP-Sil 8 + 10% C J8) o f air sample 5 (12°N, 29°W), a l i q u o t : 15 m 3

S8

l.St. /

101 90

:15,5rrfin

i7'I 5°C 40,5 rrfin

200~C 65,5 rain

Fig. 4. HRGC/ECD (column 2: CP-Sil 8 + 10% C 18) of air sample 9 (31°S, 44°W), aliquot: 21 m 3

Data system. The chromatograms were stored on a Shimadzu CR4-A data system (Shimadzu, Kyoto, Japan) for further processing. Examples of H R G C (CP-Sil 8 + 1 0 % C18)-ECD chromatograms of air samples collected in the North

Atlantic (sample 1, aliquot of 5 m 3, Fig. 2), before the Intertropical Convergence Zone (ITCZ) (sample 5, aliquot of 15 m 3, Fig. 3), and in the South Atlantic (sample 9, aliquot of 21 m 3, Fig. 4), are presented. Of the many organic trace compounds detected by the ECD only the major PCB con-

231

•

•

@

77~

~Z~

1 I~

Q

~Z

~

;~'o

~

~z

~

I

~ ~ ~ ' ~

~ -.~ Z~Z

77~ om.

~4

t..-,i m

t"xl

I

~ 4 oo

~c'q

~m~A

~

..

~11

dm ~ ~

III ~

~

~

IIm ..

g

c q c~3 c ~

2

77~ ~

~

~

Z

~1

(',1

t

~

ll~

~

©

8

~

~

~

t~

~

I

l

¢

c~ r~

232

Fig. 5. Sampling locations (1 - 14) and main tropospheric air movements. The cruise ANT IX/1 of RV "Polarstern" from Germany to Chile is marked with a dashed line. The actual wind directions during sampling are marked with short arrows

geners and some other major compounds, e.g. pentachlorobenzene (PCBz), hexachlorobenzene (HCB), pentachloroanisole (PCA), sulphur ($8), tetrabromodiphenylether (TBDE), 1,2,3,4-tetrachloronaphthalene (TCN) and the internal standards (I.St.) have been assigned. Results concerning the other non-PCB semi-volatile organochlorinated compounds will be reported in a separate paper. In addition, major peaks of unidentified compounds are marked (U). As can be seen in Figs. 2 - 4 , the PCB 149 is coeluting with PCB 107 on column 2, which otherwise gives an excellent separation of the other trace compounds collected. The separation of these two congeners on column 3 showed that PCB 107 contributes less than 5% to the peak area of PCB 149. To ensure the results of the other congeners, the quantitation was accomplished with at least two of the C18 columns. PCB 90, which is always coeluting with PCB 101 on the C18 columns, contributes less than 3% to the amount of PCB 101 [16]. PCB 138 could not be separated from PCB 163 on the C18 columns. According ref. [16] this congener contributes about 16% to the peak of PCB 138 in the technical mixture Aroclor 1254. The mean deviation of the quantitation procedure is in the range of 5 - 2 0 % , depending on the specific congener and the absolute amount of the congener.

2.5 Global orientation of the sampling locations

Fig. 6. Sampling locations (1-14) and main ocean surface water currents. The cruise ANT IX/1 of RV "Polarstern" from Germany to Chile is marked with a dashed line

The general geophysical parameters of the global mass flow, related to the transport of organic chemicals in general [3, 17], and especially the flux of PCBs in the oceanic system [18, 19] have been previously reviewed. The meteorological and hydrospheric parameters related to the samples are listed in Table 2. The sampling locations are representative for a wide range of aspects of the generalized global mass flow in the lower troposphere (Fig. 5) and in the surface water of the Atlantic Ocean (Fig. 6). Figure 5 also shows the actual wind directions during the air sampling. A differentiation for the basic spatial elements of the general circulation - namely 1) northern Westwind Belt, 2) N o r t h East Trade Winds, 3) South East Trade Winds and 4) southern Westwind Belt - can be made. This distinction is in part a compensation that no trajectories have been calculated. The actual weather situation has always been accounted for. A comparison between the generalized picture of global wind pattern and the actual wind directions during the respective sampling periods will follow in Part 3.1. The northern Westwind Belt functions as a source since most of the use and disposal of PCBs occurs in the industrial regions of the northern hemisphere. On the cruise A N T

Table 3. Composition of Ascarels based on PCB mixtures of the Clophen (Bayer) type and trichloro/tetrachlorobenzene (in %) (e) Years of production

A60 (60% C1)

A50 (54% C1)

A30 (42% C1)

C6H3Cla

C6HzC14

- 1950 (a) 1950-1977 (a) 1963-1977 (b) 1977-1980 (c) 1981 - 1984 (d)

60 50 27.5 -

20 27.5 70 -

80

40 30 34.5 30 20

10.5 -

(a) Clophen T 64; (b) Clophen T 241 ; (c) Clophen T 64 N; (d) Clophen T 82; (e) ref. [28]

233 IX/I of RV "Polarstern" from Bremerhaven (Germany) to Punta Arenas (Chile) (Fig. 5) the northern Westwind Belt reached to about 27°N including samples 1, 2 and 3. The other regions of the globe function more or less as a dilution space for the input from the northern Westwind Belt. A specific role is played by the Trade Wind system generalized by the two Hadley cells. The counterflowing North East Trades and South East Trades divide the troposphere of both hemispheres through a high-reaching air curtain, the Intertropical Convergence Zone (ITCZ). The Trades are a rather constant wind system throughout the year. The air in the Trade Winds is well mixed and influenced by the surface water layer up to the height of the Trade Wind inversion at 1500 + 500 m. Above the inversion the back flowing air from the ITCZ wilt be part of the air column. The North East Trades extended on our cruise from about 27°N to 7°N, including air samples 4 and 5. The air masses of the northern Trades of the Atlantic have basically two possible sources: the Westwind Belt and the descending air of the subtropical high-pressure belt over the North Atlantic and northern Africa. The crossing of the ITCZ was indicated at 7°N by a change in wind direction and by a small drop in the carbon monoxide (CO) mixing ratio, also determined on the cruise A N T IX/I [20]. The air sample 6 (4.3°N 28.5°W - 1.0°S 28.4°W) must therefore already be attributed to the Trades of the southern hemisphere. The South East Trades ranged on our cruise from 7°N to about 38°S (including samples 6 - 10), however sampling in the southern Trades was partly influenced by air masses originating from South America. The samples 11 and 12 can be attributed to the southern Westwind Belt. Sample 12 has its origin in the Westwind Belt of the western part of the South Atlantic, far away from any continental input. It was collected during the cruise A N T IX/4 of the research vessd "Polarstern" from South Africa via Bouvet Island (54°S 3°E) to Germany. The air samples 13 and 14 are from the island Isla do Sal (16.5°N 22.8°W), Capo Verde, in the eastern North Atlantic about 800 km west of Africa. These air samples represent the output by the North Eastern Trades from the western part of North Africa, including possible equilibration processes with the surface water of the Canary Stream (Figs. 5 and 6).

oom

.--~ C'.I

C'qO0

I

VI

.-~m~eq'a'-;~°~"-;°~eqcS~°eq~cq {-q

I

Vl

Vl

Vl

V

2= ~ m c,i ~

I

d d d d d d d d d ~

I

v v l v v v v v v v o

~

+

+

~

~

m

'

~

II

<.--. -._.. e'~

._=

,8

~ 0

~

"

II

O

VI

;>

kc I

,a=

I

VI

8 o.=

ea

q)

o ca e~

<

D--

I

I

VI o o o o o o o o

m

sg

~

I

m

ea >-,

0

..=

e..~

~.._:'z~'zz_

,

"~z'~

•

t~

I

VI

m

ea) © ¢) " ~ m'"

I

I

o

3 Results and discussion

V

VI

V[

V

V[

× ~..-~ ' ~

0,,.>

ddd~dddddddddd

0

2~ +.a

-d

8 0

,4

Z

~.~ .= =

09

g~ ~

d

o

~

~

P

=

l

¢

=== mm~

= m

= m

=m mm

85 m

V

The selected fourteen PCB congeners are representative of the occurrence of PCBs deriving from the major technical PCB mixtures, like the Aroclor, Clophen, Phenodor and Kanechlor brands. The technical PCB mixtures were often blended to the Ascarels, the actual "form" in which the PCBs were used. The composition of the Ascarels changed throughout the years (Table 3), mainly as a consequence of regulations on the use of PCBs [21]. In a 1 : 1 : l-mixture of the three most important technical mixtures e.g. Aroclor 1242, 1254, 1260 or Clophen A30, A50, A60, the selected fourteen congeners contribute about 40% to the sum of all PCB congeners. The concentrations of the PCB congeners in the marine air samples are listed in Table 4. The sum of PCB (ZPCB) approximated as Aroclor 1242/1254/1260 (1:1:1) [11] can be calculated from the sum of the seven indicator congeners (PCB 28, 52, 101, 118, 138, 153, 180) multiplicated by a factor of 4, if the composition of the

234 [ppm]

ITCZ

100

[pg/mEt]

1

I TCZ

100 "

ill

80

:~

~

8

7

6

4

80

3

60

11 10 40

~8

= --

4

60

9

'

O

40 20 2O -60

-50

South

-40

-30

-20

-10 0 10 latitude [°]

20

30

40

50

60 North

Fig. 7. Average course of the tropospheric levels of carbon monoxide, derived from measurements made on board of RV "Polarstern" [30] during the cruise ANT IX/1 from Germany to Chile. The sampling locations are indicated by numbers. - CO mixing ratio

technical mixtures listed in ref. [21] is used. These indicator congeners are often used as indicator compounds in legal threshold value measurements of the occurrence of PCBs. Polychlorinated biphenyls are also found as a complex mixture of their own in emission from waste burning [22 28]. Apparently, this PCB source is only of local relevance. This mixture of PCBs is deriving from chlorobenzenes as precursors and it is substantially different from that of the technical mixtures synthesized by FeCla-chlorination of biphenyl [29]. It is dominated by the higher chlorinated congeners. As the consequence of a reaction mechanism involving radicals, the concentrations of PCB congeners 114 (2,3,4,41,5-C15B), 122 (2,3,3',4,5-C15B), 124 (2',3,4,5,5'C15B), 189 (2,3,3',4,4',5,5'-C17B) and others are strongly increased as compared with the technical mixtures [27, 28]. In all samples discussed here all the PCB congeners were found at > 9 9 % in the gaseous state. Any discussion of the physical residence time [3, 4] has therefore to focus on absorption phenomena by water either as cloud water (fog, sea spray) and rain drops or by absorption at the sea surface, the gas-dry deposition. The path of the molecules to the sea surface is governed by aerodynamic and finally rate determining diffusion processes summarized by the overall deposition velocity va [3, 4]. The basic air/water phase distribution ratio is given by the value of Kgwand its temperature dependence, including the differences effected by the salt content of the sea water [4]. Samples collected under the partial influence of continental air masses are revealed by an increase of marker molecules like gamma-hexachlorocyclohexane (7-HCH), H C B and PCBs 153 and 180, as discussed below. In addition, the tropospheric levels of carbon monoxide (Fig. 7) and ozone have been measured on the cruise A N T IX/1 [20, 30]. These molecules - particularly carbon monoxide - also function as markers for the input of continental air masses into the Atlantic Ocean [30].

3.1 Concentration of PCB-congeners in marine air along a north/south axis between 50°N and 50°S Our measurements gave values for the sum of PCB (ZPCB) between 48 pg/m a in the Westwind Belt of the eastern N o r t h Atlantic and 22 pg/m a in the Westwind Belt of the central South Atlantic up to 385 pg/m 3 before the coast of South

0 -60

12 J

i

i

~

-50

-40

-30

-20

South

i

i

i

-10 0 10 latitude [°]

i

i

~

i

L

20

30

40

50

60 North

Fig. 8. Level of the sum of the seven indicator congeners PCB 28, 52, 101, 118, 138, 153, 180 in the tropospheric boundary layer of the Atlantic Ocean in pg/m a. The corresponding samples are indicated by numbers

[pg/rn3]

ITCZ

60

7

3O 3 8

10 9 ~

20

"~/

10

i

O

-60

,/"

"~,

41, L~"

1

~k

\

'~, :

2,"

t

i

i

i

p

i

i

i

=

i

I

i

-50

-40

-30

-20

-10

0

10

20

30

40

50

60

South

latitude P]

North

Fig. 9. Concentrations of PCB 28 (2,4,4'-C13B) and PCB 52 (2,T,5,51-C14B) in the tropospheric boundary layer of the Atlantic Ocean in pg/m 3. The baseline-levels are marked with strong lines. The corresponding samples are indicated by numbers. ~ • PCB 28 (C13B); + . . . . + PCB 52 (C14B)

Patagonia (Table 4). The course o f the sum of the seven indicator congeners is depicted in Fig. 8. The samples I (46°N 12°W), 8 (27°S 40°W) and 11 (46°S 59°W) are definitely characterized by a partial input of PCBs containing air from the continents, as discussed later. The course of the levels is not equal for both hemispheres. The air over the N o r t h Atlantic has a minimum between 15°N and 20°N. A much more detailed picture evolves, if the North/South courses of the levels of the individual PCB congeners (Fig. 9, 10) in the marine boundary layer are considered. We pointed out that the quantification of the polychlorinated biphenyls as the sum o f PCB congeners or as an approximation to a technical mixture, e.g. Arochlor 1242 or Arochlor 1254, suppresses valuable information about their function as markers for a regional input and about their environmental fate with respect to the individual physical and chemical properties [7, 8, 11, 31]. Our results indicate that there is a general difference depending on latitude and on terrestrial influenced air masses between the lower chlorinated and higher chlorinated congeners considered here. The trichlorinated and tetrachlorinated congeners (vapour pressures > 10-2 Pa, Table 1) - represented by the

235 30

[og/rn31 11

ITCZ

25 i

'~

5"

6

i

eOi,,

~,""

'~

3

1

"I5

i \'~

tO

~'

9/

:;'

2

,/ x

0

i

- 6 0 -50 South

-40

-30

-20

-10 0 10 latitude [°]

20

30

40

50

60 North

Fig. 10. Concentrations of PCB 101 (2,2',4,5,5'-C15B), PCB 153 (2,2',4,4',5,5'-C16B) and PCB t80 (2,2',3,4,4',5,5'-C17B) in the tropospheric boundary layer of the Atlantic Ocean in pg/m 3. The baseline-levels are marked with strong lines. The corresponding samples are indicated by numbers. ~ - - e PCB 10t (C15B); + . . . . + PCB 153 (C16B); × - - - x PCB 180 (C17B)

trichlorobipheny128 and the tetrachlorobipheny152 - show a north/south profile (Fig. 9), which is characterized by an increase of the concentrations towards the ITCZ in both the Trade Wind regions. This increase reflects partly the increase of the surface water temperatures (Table 2) and thus the temperature dependence of the gas/water partition coefficient Kg~v(see part 3.2.). In addition there is a slight increase at 46°N 12°W (sample 1) and a substantial break near the region 15 ° North within the northern Trades. The congeners with vapour pressure _<10-3 Pa (Table 2) - represented by the hexachlorobiphenyl 153 and heptachlorobiphenyl 180 (Fig. 10) - also show an increase of the contents in the Trades. Distinct local maxima are observed however at 46°N 12°W (sample 1), 27°S 40°W (sample 8) and 46°S 59°W (sample 11). In the Westwind Belt of the central South Atlantic (sample 12: 47°S 2°W) the levels of the higher chlorinated congeners in the atmosphere fall below the quantification limit of 0.2 pg/m 3 (Table 4) per congener. The congeners with vapour pressures between 10-2 and 10 .3 Pa - exemplified by the pentachlorobiphenyl 101 (Fig. 10) - lay between these two extremes: the elevated concentrations at 46°N 12°W (sample 1), 27°S 40°W (sample 8) and 46°S 59°W (sample 11) are in the range of the increase of the levels in the Trade Wind regions. The increase of the PCB concentrations at 46°N 12°W (sample 1) in the North Atlantic can clearly be traced back to the specific meteorological situation during sampling: an extended high-pressure system over Europe led to the input of continental air masses as seen by the easterly wind at sampling route I (Table 2, Fig. 5). A high content of the insecticide 7-HCH [32] and a high CO mixing ratio, also determined in the sampling region of sample I (Fig. 7) [20, 30], are further indicators of continental air masses. This supports the assumption of an input of continental air. Another region in the northern hemisphere, where continental air masses were encountered, was around 8°N and 12°N. These air masses presumably coming from NorthWest-Africa (Fig. 5) were indicated by an increase in the CO mixing ratio [30] (Fig. 7) and in addition by elevated levels of the cyclodiene insecticide endosulfan [32], which is applied in regions affected by the tse-tse fly. These continental air masses are however not indicated by an increase of the PCB

contents - there is even a decrease of the levels - nor by a clear change of the PCB pattern, as observed in the South Atlantic [33]. The local maxima of the higher chlorinated PCB congeners in the South Atlantic at sample 8 (25.5°S 38.8°W 28.7°S 41.7°W) also should have their origin in an input of continental air masses. Despite the easterly winds at sampling position 8 (Table 2, Fig. 5), an increase in the CO mixing ratio just before this sampling region (Fig. 7), accompanied by a decrease in the tropospheric levels of ozone between 20°S to 25°S [30], indicates freshly contaminated air masses [30], which only could originate from Brazil. Strong westerly winds in this region during the days before sampling support this conclusions [34]. At sampling position 6, however, the elevated CO mixing ratio (Fig. 7) was not accompanied by an increase of the levels of the higher chlorinated congeners. A decrease of the contents of tropospheric ozone, which would mark freshly contaminated air masses, was also not observed [30]. The wind direction in this region during the days before sampling was south east to east [34], leading to the transport of air masses from the central South Atlantic to sampling region 6. This indicates that the older continental air masses in this region, characterized by an increase of the CO-levels, probably had enough time to equilibrate with the surface water of the ocean with respect to the levels of the PCBs. The sample 11 (46°S 60°W) was influenced by air masses from South Patagonia (Table 2, Fig. 5). The oil hauling plants and other industries at Patagonia could be the reason for the elevated levels of the higher chlorinated congeners. A rise of the contents of hexachlorobenzene in the samples 8 and 11 is also observed [32]. I f one excludes the samples with a proven input of continental air (Fig. 9 and 10: dashed lines) one can derive a corrected air/sea exchange regulated "baseline-level" of the polychlorinated biphenyls in the boundary layer of the troposphere of the Atlantic Ocean (Fig. 9 and 10: strong lines).

3.2 An air~sea exchange of PCBs as a global source? It has been discussed for long that water bodies may act as sources of PCBs to the overlaying air [4]. Recent congener specific measurement showed that Green Bay (Lake Michigan) is a major source of these compounds to the local atmosphere [7]. If we assume a sea/air exchange as the relevant source of PCBs to the atmosphere of the open ocean, a decline of the levels in air to higher latitudes should be given as the result of the temperature dependence of the gas/water partition coefficient Egw. Such an influence of the temperature is observed in the South Atlantic: In Fig. 11 the decrease of the levels of C13B 28, C13B 31 and C14B 52 in air from sample 7 (9°S 31°W) to sample 12 (47°A 2°W) is compared with the calculated decrease based on the Kgw value at the actual surface water temperature. The temperature dependence of Kgw (Table 5) was taken from ref. [35]. For the calculation of the equilibrium levels the contents measured in sample 7 are taken as a fixed point and the assumption is made that the levels in the surface water of the South Atlantic are homogenous. The vertical bars in Fig. 11 give the mean deviation of the quantitation procedure, including the range of the recovery rate of the surrogate standard TCN.

236 Although the measured levels slightly differ from the calculated levels, a covariant trend of both results is obvious. Church et al. [36] found during a cruise from Dakar (West Africa), via the Azores, to N o r t h America in 1984 that the PCB levels (ZPCB) in air were highest in the Trade Wind region around 35°N 25°W (669 pg/m3), whereas in the

•

PCB 28 calculated - - - D r

PCB 28 measured

pg/m •

0 t

-50

-40

-30

-20

-10

latitude [°et]

pg/m 3

•

P C B 31 c a l c u l a t e d ~

PCB 31 measured

O! -50

-40

-30

-20

-10

latitude [°=]

PCB 52 calculated ~

•

northern Westwind Belt the levels fell down to values below 100 pg/m 3 in some samples. The authors of this study explained their results with the enhanced volatilization of PCBs from the terrestrial surface in South Europe and North-West Africa due to the higher temperature and the transport of the PCB enriched air masses to the Atlantic Ocean. They did not consider the influence of the temperature on a possible sea/air exchange. The remarkable decrease of the lower chlorinated PCB congeners within the northern Trades around 15°N cannot yet be definitely explained. One reason could be an extremely low PCB activity ai = activity coefficient fix total concentration ci, with fl ~ 1, in the ocean surface water as the result of a high proportion o f PCB associated with particulate organic matter (POM) and colloidal dissolved organic matter (DOM). The consequence will be a reduced PCB level in air, if this level is defined by the Kgw values. The upwelling water at the West African Shelf with its proven high primary production rate would lead to high values in P O M and DOM. Pankow and McKenzie have discussed the parameters which affect the equilibrium distribution of a chemical between the dissolved, particulate and colloidal phases [37]. The equilibrium partitioning of PCB between water and marine plankton as well as other adsorbing agents was experimentally investigated and compared with field data by Pavlou and Dexter [38] and Hiraizumi et al. [39]. A Freundlich type isotherm equation (1):

PCB 52 measured

Cad s =

pg/m ~

K • ~diss ol/2

(1)

Cads: adsorbed concentration [txg sorbate/g sorbent] Cdiss: dissolved concentration [gg/ml] K : equilibrium constant [ml/g] n: Freundlich constant -50

-40

-30

-20

-10

latitude [ ° a ]

Fig. 11. Comparison of the levels of C13B 28, C13B 31 and C14B 52 in air from sample 7 (9°S 31°W) to sample 12 (47°S 2°W) with calculated values based on Kgw values at the actual surface water temperature. For this calculation the contents measured in sample 7 are taken as a fixed point and the assumption is made that the levels in the surface water of the South Atlantic are homogenous. The temperature dependence of Kgw was taken from ref. [36]. The vertical bars give the mean deviation of the quantitation procedure, including the range of the recovery rate of the surrogate standard TCN. The vertical bars at the calculated values are given by the mean deviation of the quantitation of sample 7

with Freundlich constants n between 0.9 and 1.2 and values for K from 105 to 106 for the distribution of PCBs between water and suspended particulate matter was observed [7, 38, 39]. For offshore regions with a suspended load m < 1 rag/l, values for K for individual PCB congeners o f about 10 6 were measured [40]. Considering the suspended load in mg/1 leads to an equation for the adsorbed portion f in % similar to the Junge equation, which describes the portion in air adsorbed to particles: f[%] = 1 0 0 K m / ( K m + 1)

(2)

m: suspended particulate matter [g/ml]

Table 5. The gas/water distribution ratio Kgw = H/RT (dimensionless) in dependency of the temperature [ x 103]. The values in brackets are corrected for the salt content of the sea water (Kgw(..... tor) = 1.4 _+0.1 Kgw [4])

0° C 5°C 15°C 25°C 30°C

PCB 28

PCB 52

PCB 101

PCB 118

PCB 138

PCB 153

PCB 180

0.8 (1.1) 1.3 (1.8) 3.3 (4.6) 7.9 (11.1) 12.0 (16.8)

0.8 (1.1) 1.3 (1.8) 3.3 (4.6) 8.0 (11.2) 12.2 (17.1)

0.4 (0.5) 0.6 (0.8) 1.5 (2.1) 3.7 (5.2) 5.6 (7.8)

0.2 (0.2) 0.3 (0.4) 0.7 (0.9) 1.6 (2.2) 2.4 (3.4)

0.08 (0.1) 0.1 (0.2) 0.3 (0.5) 0.8 (1.1) 1.2 (1.7)

0.09 (0.1) 0.1 (0.2) 0.4 (0.5) 0.9 (1.3) 1.4 (2.0)

0.04 (0.06) 0.06 (0.08) 0.2 (0.2) 0.4 (0.6) 0.6 (0.8)

The values for 25°C are from ref. [55], the temperature dependance of Kgw is taken from ref. [35]

237 [pglm3l

ITCZ

45

,4-.

40

,,"

35

[~/od

38

i

""+

!

"

37

30

36

25 20

35

15 10

34

5 0 -60

~

t

,

~

-50

-40

-30

-20

South

,

,

-10 0 latitude

'

~

~

,

~

,

10 [o]

20

30

40

50

this question. Surprisingly, the pattern as expressed by the similarity index and evaluated by principal component analysis [33] does not change for the low level samples in the northern Trades (samples 5 and 13) as compared with the foregoing and following samples 4 and 6. In all these samples the marine air type pattern is found. We therefore favour an input of PCB depleted air masses coming in from the African desert as part of the subtropical high-pressure belt to explain the reduced levels in samples 5, 13 and 14.

33 60

North

Fig. 12. Concentrations of PCB 28 (2,4,4'-C13B) and PCB 52 (2,2',5,5'-C1~B in the tropospheric boundary layer of the Atlantic Ocean in pg/m 3 compared with the salinity of the surface water in % o. The baseline-levels are marked with strong lines. = • Salinity; e - - - - • PCB 28; + + PCB 52

Compounds with K = 10 s for a given suspended load of 100 ~tg/1 will primarily reside in the water. Only for K = 10 6 the adsorbed portion exceeds a few percent. The measured values of m on our cruise ranged from 1 to 5 gg/1 in the Central South Atlantic (0°S-20°S), the values in the North Atlantic ( 4 0 ° N - 0 ° N ) were between 0.3 and 50 ~tg/1 [41]. In the eastern North and South Atlantic near the African coast the values are in a range of 1 0 - 1 0 0 gg/1 [41]. This leads to the conclusion that at the maximum a few percent of the PCBs in sea water are adsorbed to particulate organic matter. More recently the association constants Kooc for single PCB congeners and marine humic substances were determined by Lara and Ernst [42]. They found values of 5.9 x 10 ~ for PCB 101 (ClsB) up to 5.3 x 105 for PCB 180 (C17B) for DOC contents of 5 mg C/1. The association constants are highly correlated to various molecular descriptors, in particular to those describing the size and coplanarity of the PCB molecules [43]. The average DOC values for open oceans are in the range of 0 . 5 - 1 rag/1 [44]. Using DOC = 1 mg/l and the above KDoc values the truly dissolved fration in water is about 95% for PCB 101 and only 65% for PCB 180. That means that a significant portion of the PCB loading of the surface water could be associated with dissolved organic carbon, depending on the congener specific value of KDoc [43]. The lower chlorinated congeners however should not be associated with D O M at a significant portion due to their lower lipophility, expressed by the octanol/ water-partition coefficient Kow (Table 2). The correlation between the course of the salinity of the surface water and the content of PCB 28 and PCB 52 (Fig. 12) does not give a reasonable explanation for the northern decrease either. Ocean areas with a high salinity in the surface water are marked by high evaporation of water and low precipitation, which should favour the evaporation of PCB from the surface water too. The decrease in air could also be affected by an input of PCB depleted air masses from the desert in northern West Africa from where the eastern North East Trades originate (Fig. 5). Only some isolated urban areas will be possible sources for PCBs at the West African coast. Measurements of the PCB content in the water and in the air near the coast of West Africa and in the desert mainland could answer

3.3 Formerly reported PCB levels in the air of the Atlantic Ocean A discussion of formerly reported PCB levels in air in relation to our measurements is difficult, as the format of the data, e.g. single congeners versus total PCBs (ZPCB), is different. Possible regional sources are often not clearly indicated. Bidleman et al. determined ZPCB contents in air at Barbados (1978) and Newfoundland (1977) of < 5 - 3 7 0 pg/ m 3 (mean: 57 pg/m 3) and 4 2 - 1 5 0 pg/m 3 (mean: 115 pg/ m3), respectively, (calculated as Aroclor 1254) [45]. Knap and Binkley determined ZPCB levels in air over the Bermudas by aircraft up to a height of 3500 m. They found values between 51 and 1066 pg/m 3 [46]. The values for ZPCB measured by Church et al. during a cruise through the North Atlantic in 1984 were supplemented by air trajectories: the levels in the Westwind Belt ranged from 5 8 - 4 4 9 pg/m 3, the samples with the highest values clearly were influenced by continental air masses from North America [36], levels in the Trade Wind region were even higher as discussed before. The recent congener specific measurements of Bidleman et al. at Sable Island [47] were also supported by air trajectories. In this case the soil surface/atmosphere and sea surface/ atmosphere exchange can be differentiated in the discussion of an input/output regulation. The summer and winter mean values, 42 pg/m 3 and 63 pg/m 3, respectively, for the sum of 38 PCB congeners come close to the lower contents measured by Church et al. [36] and also to our low value in the Westwind Belt of the North Atlantic (48 pg/m3). The level in the Westwind Belt of the South Atlantic, which leads over to the Antarctic Ocean, is still lower (22 pg/m3). Levels in arctic atmosphere (about 1 - 2 0 pg/m 3) are generally lower than values in warmer regions [13, 48, 49]. A critical evaluation of global atmospheric measurements in oceanic regions by the GESAMP group yielded a mean PCB (ZPCB) concentration of 290 pg/m 3 for the North Atlantic [50]. Due to the lack of data for the South Atlantic the authors of this study extrapolated a value of 33 pg/m 3 derived from measurements in the South Pacific to the South Atlantic. For the three oceans the following fluxes for 2PCB out of the atmosphere were calculated in the GESAMP study [51]: North Atlantic: 1.81 mg/m2/year, South Atlantic: 0.27 mg/mZ/year, North Pacific: 0.40 rag/ mZ/year, South Pacific: 0.26mg/m2/year, South Indian Ocean: 0.77 mg/m2/year. These calculations do however not consider the possible source function of the upper water layer of the oceans which would lead to a cycling effect in depositions. Sanders et al. estimated the flux of PCB congeners to the sediments of a rural lake near the west coast of Great Britain in the Lake District [52]. The total flux of PCBs remained constant over the last twenty years, 1968 until 1988, with a mean of 21 4- 1 mg/mZ/year. The dominance of lower chlori-

238 nated PCBs in the sediment samples underlines the i n p u t from the marine air. The constancy of the flux suggests that in a first approximation the levels of PCB in air of the marine b o u n d a r y layer of the Westwind Belt were the same over the last twenty years. I n contrary to these findings, peat core profiles indicate for eastern N o r t h America a decline of PCB depositions near terrestrial sources after 1971 by a factor of 2 - 4 [53].

Conclusions There is evidence that the occurrence of PCBs in clean marine air, which is not overlain with continental air masses, is not a remainder of an incomplete air-to-sea deposition but reflects the PCB levels in the marine surface water and thus indicates an air/sea equilibrium. The correlation of the PCB levels in the air over the South Atlantic with the surface water temperature and therefore with the temperature dependent gas/water partition coefficient Kgw support our conclusion. The distribution of PCBs between surface water and the deeper water layer of the oceans will supplemented the long-term fate of PCBs in the oceanic system.

Acknowledgement. This work has been financially supported by the German Science Foundation (DFG-Ba 371-11-2). We gratefully acknowledge the support by the Government of the Republic of Capo Verde and its Embassy in Germany. We thank the Captain and the Crew of the "FS Polarstern" for their support of our work during the cruises ANT IX/1 and ANT IX/4. We thank the "Alfred Wegener Institut fiir Polarforschung", Bremerhaven, Germany, for the support. Support for our expedition to Capo Verde by the airline "Condor" is also acknowledged.

References 1. Atlas EL, Schauffler S (1990) In: Kurtz DA (ed) Long range transport of pesticides. Lewis, Chelsa, Michigan, pp 161 - 1 8 4 2. Bidleman TF, Atlas EL, Atkinson R, Bonsang B, Burns K, Keene WC, Knap AH, Miller J, Rudolph J, Tanabe S (1990) In: Knap AH (ed) The long-range atmospheric transport of natural and contaminant substances. Kluwer, Dordrecht Boston London, pp 259-301 3. Ballschmiter K (1992) Angew Chem 104:501-528; Angew Chem Int Ed Engl 31:487-515 4. Schwarzenbach RP, Gschwend PN, Imboden DM (1993) Environmental organic chemistry. Wiley, New York 5. Mackay D, Paterson S, Cheung B, Neely WB (1985) Chemosphere 14: 335-374; Mackay D (1991) Multimedia environmental models: the fugacity approach. Lewis, Chelsa, Michigan 6. Ballschmiter K (1979) Nachr Chem Tech Lab 27:542-546; ibid (1985) 33 : 206 - 208 7. Achman DR, Hornbuckle KC, Eisenreich SJ (1993) Environ Sci Technol 2 7 : 7 5 - 87; ibid 27:87-98 8. Zell M, Neu H J, Ballschmiter K (1978) Fresenius Z Anal Chem 292: 97 - 107; ibid 293 : 193 - 200 9. Gregor D J, Gummer WD (1989) Environ Sci Techno123:561 565 10. Ballschmiter K, Zell M (1980) Fresenius Z Anal Chem 302:2031 11. Wittlinger R, Ballschmiter K (1987) Chemosphere 16:24972513 12. Wittlinger R, Ballschmiter K (1990) Fresenius J Anal Chem 336 : 193- 200 13. Oehme M, Stray H (1982) Fresenius Z Anal Chem 311:665673 14. Jansson B, Vaz R, Blomkvist G, Jensen S, Olsson M (1979) Chemosphere 4:181 - J 90

15. Ballschmiter K, Mennel A, Buyten J (1993) Fresenius J Anal Chem 346: 396- 402 16. de Boer J, Dao QT (1991) J High Res Chrom 14:593-596 17. Ballschmiter K (1991) Environ Carcino and Ecotox Revs C9:1-46 18. Atlas E, Bidleman TF, Giam CS (1986) In: Waid JS (ed) PCBs and the Environment, CRC, Boca Raton, pp 79-100 19. Tanabe S, Tatsukawa R (1986) In: Waid JS (ed) PCBs and the Environment, CRC, Boca Raton, pp 143-161 20. Slemr F, Langer E (1992) Nature 355:434--437 21. Schulz DE, Petrick G, Duinker JC (1989) Environ Sci Technol 23 : 852- 859 22. Ballschmiter K (1988) In: Fresenius W, Giinzler H, Huber W, Kelker H, Lfiderwald I, T61g G, Wisser H (eds) Analytiker Taschenbuch. Vol 7, Springer, Berlin Heidelberg New York, pp 393-432 23. Tiernan TO, Taylor ML, Garrett JH, van Ness GF, Solchy G, Dais DA, Wagel DJ (1983) Chemosphere 12:595-606 24. Ballschmiter K, Zoller W, Scholz C, Nottrodt A (1983) Chemosphere 12: 585 - 594 25. Sch/ifer W, Ballschmiter K (1986) Chemosphere 15 : 755 - 763 26. Schreiner M, Spitzhiittl E, Vierle O (1986) Chemosphere 15: 2093 - 2097 271 Ballschmiter K, Niemczyk R, SchS.fer W, Zoller W (1987) Fresenius Z Anal Chem 328 : 583- 587 28. Ballschmiter K, Sch/ifer W, Buchert H (1987) Fresenius Z Anal Chem 326 : 2 5 3 - 257 29. Hutzinger O, Safe S, Zitko V (1974) The Chemistry of PCBs. CRC, Cleveland 30. Slemr F, Langer E (personal communication) 31. Zell M, Neu HJ, Ballschmiter K (1977) Chemosphere 6 : 6 9 76 32. Schreitmfiller J, Ballschmiter K (unpublished results) 33. Schreitmfiller J, Ballschmiter K (1992) Poster presented at 12th International Symposium on Dioxins and Related Compounds, 24. - 28. August, Tampere, Finland. Schreitmiiller J, Ballschmiter K (1993) Poster presented at Symposium on Methods and Improvements on Analytical Chemistry ANAKON '93, 19.21. April, Baden-Baden, Germany 34. Meteorological Office ofRV "Polarstern" (1990) Weather Reports 35. Tateya S, Tanabe S, Tatsukawa R (1988) In: Schmitdke NW (ed) Toxic contamination in large lakes. Lewis, Michigan, pp 2 3 7 28116 36. Church TM, Tramantano JM, Whelpdale DM, Andreae MO, Galloway JN, Keene WC, Knap AH, Tokos J (1991) J Geophys Res A 96:18705-18725 37. Pankow JF, McKenzie SW (1991) Environ Sci Technol 25 : 2046- 2053 38. Pavlou SP, Dexter RN (1979) Environ Sci Technol 13:65-71 39. Hiraizumi Y, Takahashi M, Nishimura H (1979) Environ Sci Technol 13:581 - 584 40. Duinker JC (1986) Netherlands J Sea Res 20:229-238 41. Helmers E, Schulz-Baldes M (1992) Labor 2000. Eine Sonderpublikation der Labor Praxis 91/92. Vogel, Wfirzburg, pp 6 - 1 1 42. Lara R, Ernst W (1989) Chemosphere 19:1655-1664 43. Sabljic A, Lara R, Ernst W (1989) Chemosphere 19:16651676 44. Head PC (1976) In: Burton JD, Liss PS (eds) Estuarine Chemistry, Academic Press, New York, p 58 45. Bidleman TF, Christensen EJ, Billings WN, Leonard R (1981) J Mar Res 39:443-464 46. Knap AH, Binkley KS (1991) Atmos Environ 25A: 1507-1516 47. Bidleman TF, Cotham WE, Addison RF, Zinck ME (1992) Chemosphere 24:1389-1412 48. Hargrave BT, Vass WP, Erickson PE, Fowler BR (1988) Tellus 40B :480--493 49. Patton GW, Hinckley DA, Walla MD, Bidleman TF (1989) Tellus 41B : 243 -- 255

239 50. GESAMP, Joint Group of Experts on Scientific Aspects of Marine Pollution Rep Stud-GESAMP 1989, No 38, 1 -111 51. Sanders G, Jones KC, Hamilton-Taylor J, D6rr H (1992) Environ Sci Technol 26 : 1815 - 1821 52. Rapaport RA, Eisenreich JS (1988) Environ Sci Technol 22:931-941 53. Brodsky J, Ballschmiter K (1988) Fresenius Z Anal Chem 331:295-301

54. Fischer RC, Wittlinger R, Ballschmiter K (1992) Fresenius J Anal Chem 342:421-425 55. Brunner S, Hornung E, Santl H, WolffE, Piringer OG, Altschuh J, Brfiggemann R (1990) Environ Sci Technol 24:1751 -1754 56. Atkinson R (1987) Environ Sci Technol 21:305-307 57. Kirst GO, Wolff H, Thiel C, University of Bremen, Germany (personal communication)