Characterization of polychlorinated biphenyl isomers in Sovol and

Present address: Ecoservice Lab, Ecological Union of Russia,. Krasikov St. 32 .... There was a surprisingly good relationship between the. RRTs of ind...
0 downloads 0 Views 641KB Size
Environ. Sei. Technol. 1992, 26, 2012-2017

Characterization of Polychlorinated Biphenyl Isomers in Sovol and Trichlorodiphenyl Formulations by High-Resolution Gas Chromatography with Electron Capture Detection and High-Resolution Gas Chromatography-Mass Spectrometry Techniques Vassily Ivanov' and Erik Sandeli

Chemical Laboratory, Technical Research Centre of Finland, SF-02150 Espoo, Finland The composition of two technical polychlorinated biphenyl (PCB) formulations, Sovol and Trichlorodiphenyl, has been characterized by high-resolution gas chromatography (HRGC) with electron capture detection (ECD) and HRGC-mass spectrometry (MS) techniques. Identification of the individual PCB components has been made by comparison of relative retention times (RRTs) for seven individual PCB standards, for individual peaks of Sovol and Trichlorodiphenyl, and for different types of Aroclor mixtures. Also, the correlation to literature data was analyzed. The accuracy of PCB determinations with this approach depends on the similarity of the production process and the high reproducibility of retention time data obtained for individual peaks by HRGC-ECD. Average weight percent distribution of the PCB isomer group was calculated for Sovol and Trichlorodiphenyl and compared to Aroclor data. The results indicate that Sovol and Trichlorodiphenyl formulations are fairly close to Aroclors 1254 and 1242, respectively. Introduction Since the discovery of polychlorinated biphenyls (PCBs) in environmental samples in 1966 (I), it has been generally accepted that they are ubiquitous contaminants of every component of the global ecosystem (2-5). The particular attention of public and scientific concern has been directed at commercial PCB production as technical formulations, which are the dominant sources of PCBs in the environment (6-9). PCB technical mixtures with varying degrees of chlorine content (20-80%) have probably been confined to 10 countries in the world, and their total manufacture was estimated at about 1.5 X lo6tonnes (7, 10). Although PCB manufacture and use were severely restricted during the last years, including a total ban, many industrial countries have recognized that they face today a considerable and alarming problem (7, 11). For the past 25 years a lot of reports in the literature have been focused on PCB formulations manufactured in the United States and Western Europe and only contradictory information was known about PCB production in the USSR. In general, it has been reported that technical PCB mixtures bear the name Sovol and mixtures of Sovol and trichlorobenzenes have been produced under the name Sovtol (4). [From an etymological point of view, Sovol (Sovtol) is an abbreviation for Soviet Oil.] The purpose of this article was the identification and quantification of individual congeners in the Soviet commercial PCB mixtures by HRGC-ECD and HRGC-MS as well as a comparison of obtained data with well-characterized Aroclor formulation. We hope that a complete analysis of the individual congeners of PCB technical mixtures will play an essential part in the environmental science research of sources,

* Present address: Ecoservice Lab, Ecological Union of Russia, Krasikov St. 32, 117418 Moscow, Russia. 2012

Envlron. Scl. Technol., Vol. 26, No. 10, 1992

accumulation, degradation pathways, and ecotoxicological implications of materials containing PCBs. A Brief History and Terminology. Sovol, the first Soviet technical PCB mixture, was synthesized in the USSR in 1934 and was prescribed as a nonflammable synthetic dielectric fluid in the manufacture of power capacitors and transformers. Mass production of Sovol started at the end of the 1930s (12, 13). The estimated cumulative production and consumption of Sovol in the USSR during the period from the 19409 to 1990s was about 100O00 tonnes. Now, as a result of scientific concern, Sovol production has been reduced to about 500 tonnes/year. For a long time, Sovol has been used not only in so-called closed systems, such as capacitors, transformers, and hydraulic equipment, but also in so-called open systems, such as sealing paste additives and plasticizers, in paints and plastics and for improvement of wire insulation properties (14).

Three different mixtures of Sovol and trichlorobenzenes (TCBZ) were produced under the name Sovtol from the 1940s to 1980s: Sovtol-1, 75% Sovol and 25% TCBZ; Sovt01-2,64% SoVd and 36% TCBZ; S0vt01-10,90% Sovol and 10% TCBZ (15). In all these formulations, mixtures of three TCBZ isomers were used. Now the production of Sovtol has been terminated. The manufacture of Trichlorodiphenyl (product name; contains substantial quantities of non-trichlorobiphenyl components as reported below) was started in the Soviet Union at the end of the 1960s. Mainly, Trichlorodiphenyl was used in closed systems as a dielectric liquid. Now its production also has ceased for ecological reasons. The total production of Trichlorodiphenyl during the 1960s-1980s may be estimated at about 25 000 tonnes. In addition, Nitrosovol, a mixture of 85% Sovol and a special so-called arctic addition of 15% nitronaphthalene used for the impregnation of some capacitors in small quantities, and Geksol, a mixture of 25% Sovol and 75% hexachlorobutadiene was used as a dielectric fluid for capacitors and transformers (16,17),were manufactured in the 1960s. Experimental Section Materials. Aroclors 1221, 1232, 1016,1242,1248,1254, 1260, and 1262 were obtained from Analabs. Sovol and Trichlorodiphenyl were obtained from the stock of the chemical plant (Minchimprom, former USSR). The seven individual PCB congeners (see Table I)-28,52,101,118, 138,153, and 180-used as calibration congeners for GCECD and GC-MS procedures were obtained from the Community Bureau of Reference [Individual congeners in tables, figures, and text are indicated by their IUPAC numbers according to Ballschmiter and Zell (S)]. PCB congener 11 used for HRGC-MS was from Alltech. Internal standards were 2,4,6-tribromobiphenyl (Phase Separations, Inc.) and octachloronaphthalene (Foxboro). The solvent was n-hexane (Nanograde, Mallinkrodt, Inc.).

0013-936X/92/0926-2012$03.00/0

0 1992 American Chemical Society

IO1 1"t.

st. 5:

1?3

I4

IO1

1000

133

360.0

I10

326.0

1000

76

52

292.0

-9

128

-

I90

A 2a

I I L?

,

25

3a

35

I

'

"1

3:

L

258.0

,

"

10

45

B I1 '8

10

MU /&,

53

20

3'

rwl

I I /I I

4ai

m

95

.u

I

360.0

.,

195

.

.

326.0

292.0 1MO 258.0 15

0

5

25

3e

35

40

TIME

224.0

L5

Flgure 1. Relative abundances of mass peaks in total ion current mass chromatograms of Sovol (A) and Trichlorodiphenyl (B). Injectlon volume corresponds to 1.56 p g of Sovol and 1.12 p g of Trichlorodiphenyl. Coeluting components are given in a vertical line.

Instrumentation and Analytical Conditions. A Hewlett-Packard gas chromatograph Model 5890 equipped with a pulsed 63Nielectron capture detector was used for GC-ECD analysis. All separations were accomplished in a splitless mode with a 30 m X 0.32 mm i.d. fused-silica capillary column with a 0.25-pm film thickness (SPB-5, Supelco, Inc.). The carrier gas was hydrogen with an average linear velocity 40 cm/s. The detector make-up gas was 5% methane in argon at a flow rate of 30 mL/min. An initial temperature of 50 OC was held for 2 min; afterward, it was programmed to 150 "C at 20 deg/min, held for 3 min, followed by a 2 deg/min increase to 250 OC, and immediate heating at a rate of 10 deg/min to 310 "C, where it was held for 5 min. The detectors were maintained at 280 "C, and the injector was held at 250 OC. The relative retention times (RRTs) were calculated by use of the internal standard octachloronaphthalene with an absolute retention time of 59.07 f 0.02 min. Retention times were very reproducible and permitted the use of peak matching in the range of f0.003 unit for all components in reference standard mixtures and in technical formulations. Mass spectral data were obtained with a JEOL SX 102 operated in the electron impact mode (nominal 70 eV) and interfaced with a Hewlett-Packard 5890 A GC equipped with a split-splitless injector port. Separations were accomplished with a 25 m X 0.2 mm i.d. fused-silica capillary column coated with a 0.11-pm film of 5% (phenylmethy1)silicone(HP-5, Hewlett Packard). The carrier gas was helium with a flow rate of 30 cm/s. The source temperature was held a t 240 OC. The GC conditions were the

r

~

' 10

~

l

~

~

'

''

20

'~

~

~

~

~

~ 30

'

~I

I ,i ' . I~

110

~

~

T I M E (PIIN)

Figure 3. Relative abundances of mass peaks in reconstructed single-ion mass chromatograms of Trichlorcdiphenyi.

same as above. An electron multiplier voltage of 1.8 kV and a resolution of 1000 were employed. A HP 9000 Complement data system was used to acquire and manipulate mass spectral data. The MS scan range was from m/z 10 to 500 at a rate of 0.25 s/scan.

Results and Discussion Figure 1 shows the total ion current mass chromatograms for Sovol and Trichlorodiphenyl. The obtained full mass spectral data have been used as the main tool in a proof that all the peaks of interest really have the structure of PCBs. Figure 1shows that the Sovol mixture is characterized by late-eluting peaks that are weak in Trichlorodiphenyl while early peaks are very strong in Trichlorodiphenyl and weak in Sovol. As is usual for PCBs, the mass spectra of Sovol and Trichlorodiphenyl components are characterized by a strong molecular ion cluster (M+)produced by the natural abundances of 36Cland 37Cl, a fragment ion cluster with a loss of two chlorine atoms (M - 70)+,and a weak (M - C1)+ion cluster. Figure 2 for Sovol and Figure 3 for Trichlorodiphenyl represent reconstructed single-ion mass chromatograms (RICs) only for the most important ions of the M+ clusters of these technical formulations. The information derived from RIC gives us further details about the chlorine number of the component(s) in the technical formulations and also data on the weight percentage distribution of each isomer group. ECD chromatograms of Sovol and Trichlorodiphenylare represented in Figures 4A and 5A, respectively. For comparison, ECD chromatograms of similar Aroclor formulations are also shown in Figures 4B and 5B. It is interesting Environ. Sci. Technol., Vol. 26,

No. 10, 1992 2013

~

i

'

I3

Int.

A

Po

St.

14s

35

\

I5

t' 435

I

io0

Iig

I'

tJ I51

B

Int,

st. 05

30

52

i 44

162

:79

58 1128 I

Flgure 4. ECD chromatograms of Sovol (A) and Aroclor 1254 (e). Chromatographic conditions are as in the Experimental Section. Also included: 3CB, brichlorobenzenes; 4CB, tetrachlorobenzenes; HCB, hexachlorobenzene; 7, unknown peak.

to note that the chromatograms of Sovol and Trichlorodiphenyl were virtually identical with those of Aroclors 1254 and 1242, respectively. Ow decisions concerning the determination of individual congeners in Sovol and Trichlorodiphenyl were based on (1) RRT calculations for individual congeners, (2) detailed analysis and comparison with the basic data existing in the literature for individual PCB congeners and for components of different Aroclors and Clophens, obtained under similar chromatographic conditions (8, 18-21), and (3) finally, calculations of correlations between these data and hypothetical RRTs for available and nonavailable com2014

Environ. Sci. Technol., Vol. 26, No. 10, 1992

ponents. Taking into account RIC for each peak of Sovol and Trichlorodiphenyl as well as the Aroclors used in this study, the elution order of PCB congeners was assigned to the ECD chromatographic peaks. The relative contributions of some closely eluting components with different chlorine numbers were evaluated from the RICs. There was a surprisingly good relationship between the RRTs of individual PCBs used in this study (Table I) and the RRT data obtained for these congeners on a 50-m SE-54 fused-silica narrow-bore capillary column by Mullin et al. (B),with R2 = 1.00 and a standard error of prediction of 0.0038. The availability of the RRT values for all 209

A

2 b

IS I8 3 2

I

'2

'I I 1

q 9 66 5

'O

26

60

56 74

\

91

2!

LJ

34

Ib4

1 JOI 0d

Id0

171 I56

,110

92 118 ?,

9 87 I13

I90

I70

OCN

I05

B Inc. St.

5 8

I!

XI

32

53

j6

j3

7 31 2t

44 52

\

21

6

I9

2

1

PCB congeners from ref 18 led us to continue the detailed analysis of the composition of Sovol and Trichlorodiphenyl mixtures. Figure 6 shows fairly unequivocal results of correlative analysis of the RRTs for each component from the Sovol mixture and the corresponding congener from literature data (18). A nondecisive difference at the beginning and at the end of the chromatographic runs can be explained by variances in the temperature-programmed conditions between the two studies. The chlorine numbers of the Components were completely supported by single-ion mass chromatograms (Figures 2 and 3). The present assignments of PCB numbers for Sovol and Trichlorodiphenyl also agree with the elution order of individual congeners in different well-known Aroclor formulations (shown in Figures 4 and 5).

. 1

Flgure 5. ECD chromatograms of Trichlorodiphenyl (A) and Aroclor 1242

(e).

Also included: 4CB, tetrachlorobenzene; 7, unknown peak.

Semiquantitative Analysis. The quantitation of the molar percentages for Sovol and Trichlorodiphenyl components requires much more attention, since their ECD relative response factor (RRF) values are more dependent upon instrumental operating parameters, different carrier/make-up gases, and conditions of column and can be changed dramatically by a slight variation of integration parameters (18,21). Therefore any attempts at creation of RRF correlations from literature data would be erroneous, and as far as it is possible RRFs must be determined for any individual components of PCB mixture. In our study we have used the semiquantitative method established by Duinker and Hillebrand (19). The amount of individual congeners for each formulation was expressed as a percentage in relation to the area of the most abundant peak in a mixture. Figures 7 and 8 summarize the Environ. Sci. Technol., Vol. 26, No. 10, 1992

2015

Table I. PCB Isomer Components of Calibration Mixture, Their Relative Retention Times and Relative Response Factors isomer group int Std

isomer no.

c1-2 C1-3 C1-4 C1-5 C1-5 c1-6

Ild 28 52 101 118

153 138 180

c1-6 (31-7

RT

RRT"

RRTb

RRF'

19.02 24.10 26.94 34.35 39.78 41.86 44.22 50.28

0.322 0.408 0.456 0.581 0.673 0.709 0.749 0.851

0.324 0.403 0.456 0.582 0.669 0.704 0.740 0.836

2.05 1.98 1.65 1.50 1.32 0.98 0.90 0.81

1o

o y - - -

"RRTs calculated by use of the internal standard octachloronaphthalene. *Reference 18. RRFs calculated using GC-MS. Isomer 11 was used only for semiquantitative analysis. + +

P

I

I

-

P

=

I

Aroclor 1254-1

Flgure 8. Semiquantitative comparison of the most abundant PCB congeners in Sovol and Aroclor 1254.

+

Table 11. Average Molecular Composition of Technical Mixtures PCB tech mix

1

Sovol TCDb Aroclor 1254 1242 1016

+++

041

+

0 3-03

04

05

-~ --

06 07 Sovol components RRT

08

-~

09

Flgure 6. XY relationship between predicted ( 78) and found RRTs for Sovol components. Number of components, 62;slope. 0.97: R 2 = 0.99;standard error of prediction, 0.0033.

33 53

w t % of PCB"

I

2

O

2

6

7

23 32

53 4

22 1

1

14

1 49

3

4

5

1 42 55

16 34 23

57 8

25 1

1

15 20

"For n (=1-7) in Cl,Hl+,,Cl,,. *Trichlorodiphenyl.

vironmental and biological samples, we have selected these compounds as well as congener 11 for quantitation of Trichlorodiphenyl technical formulations. These eight individual PCBs were used for calibration of the MS response to compounds of technical mixtures eluted from a GC column. One isomer from each level of chlorination (from 2 to 7) was used to calibrate the MS RRFs to all isomers in that group (22). Data shown in Table I1 provide distinguishing characteristic features for each of these formulations and three Aroclors for comparison. Conclusions

I

Trichlorodiphenyl

Aroclor 1016

0Aroclor 1242

Flgure 7. Semiquantitative comparison of the most abundant PCB congeners in Trichbodiphenyland Aroclors 1242 and 1016. For each formulation, the relative percentage contribution height was approximated to loo%, 75%, 50%, 25%, and 10%.

resulta of semiquantitative estimates of Trichlorodiphenyl, Sovol, and different Aroclors. It is interesting to note not only that the components are identical in these formulations but also that the relative amounts of the components are very similar among the groups of TrichlorodiphenylAroclor 1242 (1016) and Sovol-Aroclor 1254. Since congeners 28,52,101,118,138,153, and 180 are widely used for quantitation of PCB concentration in en2016 Environ. Sci. Technol., Vol. 26, No. 10, 1992

Information about Soviet technical PCB formulations was obtained by application of HRGC-ECD and HRGCMS. The results from correlative analysis of RRTs of the individual components in conjunction with the data of average molecular composition of Trichlorodiphenyl and Sovol provide the distinguishing characteristic details regarding the composition of these formulations. I t was found that the composition of Sovol and Trichlorodiphenyl technical mixtures is similar to Aroclors 1242 and 1254, respectively. Acknowledgments

We gratefully acknowledge Dr. E. Hlisiinen, A. Kiviranta, and co-workers for support and technical assistance and Prof. B. Jansson, Dr. P. de Voogt, and U. Eriksson for helpful discussions and reviews of the manuscript. We also thank Dr. V. Tulchinsky for his comments. Registry NO.3CB, 12002-48-1; 4CB, 12408-10-5; 6CB, 118-74-1; sovol, 37353-77-8; trichlorodiphenyl, 25323-68-6.

Literature Cited (1) Jensen, S. New Sci. 1966, 32, 612.

Envlron. Sci. Technol. 1002, 26, 2017-2022

Hutzinger, 0.;Safe, S.; Zitko, V. The Chemistry of PCBs; CRC Press: Cleveland, OH, 1974; Chapters 1, 2. Erickson, M. D. In Analytical Chemistry of PCBs; Erickson, M. D., Ed.: Ann Arbor Science: Ann Arbor, MI, 1986; Chapter 1. Paasivirta, J.; Mhtykoski, K.; Paukku, R.; Piilola, T.; Vihonen, H.; Siirkka, J.; Granberg, K. Aqua Fenn. 1986, 16, 17-23. Waid, J. S., Ed. PCBs and the Environment; CRC Press: Boca Raton, FL, 1986. Safe, S.; Safe, L.; Mullin, M. In PCBs, Environmental Occurrance and Analysis; Safe, S., Ed.; Springer: Berlin, Germany, 1987; pp 1-13. Voogt, P. de; Brinkman, U. A. Th. In Halogenated Biphenyls, Terphenyls, Naphthalenes, Dibenzodioxins and Related Products; Kimbrough, R. D., Jensen, A. A., Eds.; Elsevier: Amsterdam, The Netherlands, 1989; pp 1-45. Ballschmiter, K.; Zell, M. Fresenius 2. Anal. Chem. 1980, 302,20-31. Alford-Stevens, A. L.; Bellar, T. A.; Eichelberger, J. W.; Budde, W. L. Anal. Chem. 1986,58, 2014-2022. Weaver, G. Environ. Sci. Technol. 1984, 18, 22A-27A. Burruss, R. P. Assessment of the Environmental and Economic Impacts of the Ban on Imports of PCBs; EPA560/6-77-007; U.S. Government Printing Office: Washington, DC, 1977. Andrianov, K. A., Ed.Sovol. New unflammable insulating fluid; ONTI: Moscow, 1938 (in Russian). Andrianov, K. A,, Ed. Sovol and Sovtol; Gosenergoizdat:

Moscow-Leningrad, 1941 (in Russian). Maiophis, I. M., Ed. In Chemistry of Dielectrics; Higher School Publishing House: Moscow, 1970; pp 311-312 (in Russian). Andrianov, K. A.; Skipetrov, V. V., Eds. Synteticheskie 1962; Chapter Zhidkie Dielectric; Gosenergoizdat: MOSCOW, 2 (in Russian). Gulevitch, A.; Kireev, A. I., Eds. Production of Power Capacitors; Higher School Publishing House: Moscow, 1970; pp 65-66 (in English). Shachnovitch, M. I. Elaboration and investigation of new types of synthetical liquids for transformers. Ph.D. Dissertation, Moscow Institute of Oil and Gas Industry, Moscow, 1977; pp 30-31, 34 (in Russian). Mullin, M. D.; Pochini, C. M.; McCrindle, S.; Romkes, M.; Safe, S. H.; Safe, L. M. Environ. Sci. Technol. 1984, 18, 468-476. Duinker, J. C.; Hillebrand, M. T. J. Environ. Sci. Technol. 1983, 17, 449-456. Ballschmiter, K.; Schiifer, W.; Buchert, H. Fresenius 2. Anal. Chem. 1987,326, 253-257. Onuska, F. A.; Terry, K. A. J . High Resolut. Chromatog. Chromatogr. Commun. 1986,9, 671-675. Slivon, L. E.; Gebhart, J. E.; Hayes, T. L.; Alford-Stevens, A. L.; Budde, W. L. Anal. Chem. 1985, 57, 2464-2469.

Received for review January 3,1992. Revised manuscript received May 11, 1992. Accepted May 12, 1992.

Infinite Dilution Activity Coefficients and Henry's Law Coefflclents of Some Priority Water Pollutants Determined by a Relative Gas Chromatographic Method Ginger Tse, Hasan Orbey, and Stanley I . Sandier*

Center for Molecular and Engineering Thermodynamics, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716

A simple, fast relative measurement method based on gas chromatography developed recently has been used to determine the infinite dilution activity coefficients and Henry's law coefficients in water of some priority pollutants. We show that this simple method can be used to obtain accurate data quite rapidly, which is especially valuable for screening studies. Further, the infinite dilution activity coefficient and Henry's law coefficient data reported here can be useful for directly estimating environmentally important properties such as solubilities in water, multimedia partitioning, and octanol-water partition coefficients. Introduction

To predict bioaccumulation, the distribution of pollutants in the environment, and multimedia partitioning ( I ) , and to design equipment to purify polluted aqueous streams, it is necessary to have information on the fugacity of pollutants in water. For a species which is a liquid as a pure component at the conditions of interest, the fugacity of species i in solution, f i , is fi(TP,xi)= xiyi(TP,xi)fi"(TP) (1) where f i o is the pure component fugacity at the temperature T and pressure P of the solution and y i and x i are its activity coefficient and mole fraction. Since pollutants are generally present at low concentrations, our interest is with activity coefficients at very high dilutions, indeed, in the limit of infinite dilution. 00 13-936X/92/0926-2017$03.00/0

There are two general methods of determining infinite dilution activity coefficients. The first is to extrapolate activity coefficient information obtained in the midcomposition range (typically from vapor-liquid equilibrium data) to high dilutions. This is subject to great inaccuracy and cannot be done for strongly hydrophobic chemicals which have only limited solubility in water, such as the volatile organic pollutants of interest to us. The second class of methods is the direct measurement of infinite dilution behavior; this is what we have used here. Before consideration of the specifics of the measurements we have made, it should be pointed out that infinite dilution activity coefficients are useful in a number of ways. First, if water is relatively insoluble in the pollutant, aqueous solubilities of water contaminants can be estimated from infinite dilution activity coefficients. In liquid-liquid equilibrium the fugacity of each component must be equal in each phase, so we can write for the component i

f?(TP,xi')= fi"(TP,xi")

(2) Here the f i terms are the fugacities of component i in water-rich phase I and in chemical-rich phase 11, respectively. These fugacities can be rewritten in terms of activity coefficients as xiIyiIfi,pureO

I1 I I f . 0 xi yi &,pure

(3)

where fi,pureo is the pure component reference fugacity of component i a t the system temperature, the yi terms are

0 1992 American Chemical Society

Envlron. Sci. Technol., Vol. 26,

No. 10,

1992 2017