Direct Quantitation of PCB Congeners Using a Helium Discharge Detector and Internal Standard Techniques

P E T E R J . McATEER, T O M B. RYERSON, M A R K D. A R G E N T I N E , M A R G A R E T L. WARE, a n d GARY W. RICE* Department of Chemistry, College of William and Mary, Williamsburg, Virginia 23185

Individual PCB congeners have been quantitated at ppm levels, with an average error of _+3.2%, with the use of a helium discharge detector (HDD) for element-selective detection of CI emission. Chlorinated in- ternal standards of known concentrations were added to each solution determined to establish the relative peak areas per unit concentration of CI present. No detector precalibration or response factor formulations were required, since the detector response is based solely on the moles of CI present. The same methodology was utilized to determine the % CI in Aroclor samples without prior identification of the PCB congeners present. Index Headings: Element selective detectors; GC; Atomic emission; Spectroscopic techniques; Internal standard techniques; PCB determi- nations.

INTRODUCTION

The widespread distribution of polychlorinated bi- phenyls (PCB) in the environment and food chain eco- system has been well documented over the past twenty years. 1,2 The number of possible congeners (209) and the complexity of PCB mixtures commonly observed have resulted in a number of methods for quantifying the total PCB content in a sample. These methods have included derivatizations, 3-5 pattern recognition routines in gas chromatography, 6,7 and mass spectrometry2 Several PCB congeners have been found to possess greater mutagenic properties relative to other congeners. 9,1° Thus, selective congener quantitation may be more pertinent for as- sessing the potential ecological impact of specific PCB contaminal;ions.

Selective quantitation of individual PCB congeners has been studied in detail. The most widely used and sensitive techniques involve separation of the congeners by gas chromatography (GC), using an electron capture detector (ECD). 11-1~ The primary disadvantage for quan- titative purposes is that signals generated by each PCB congener axe not directly proportional to the mole amount of eluent present. Substitution patterns on the biphenyl rings have a marked effect on the observed signals. 14 The response fi'om a known amount of each congener must then be related to one or more internal standards to develop response factors for quantitating solutions of unknown composition. In such cases, one also must rely on the long-term detector stability to maintain these values, as well as on whether other ECDs of different construction can validly use such response factors for accurate quantitative measurements.

We recently reported on the development of a helium discharge detector (HDD) as an element-selective, mul-

Received 21 September 1987; revision received 23 November 1987. * Author to whom correspondence should be sent.

tielement detector for GC. 15 The detector functions by means of producing atomic emission from the respective elements dissociated from compounds through collision- al energy transfer processes with He metastable species produced from the discharge.

We later observed that peak areas observed in chro- matograms for a number of chlorinated or brominated aliphatic hydrocarbons were directly proportional to the relative moles of halogen present. 18 Thus, one could di- rectly quantitate a number of compounds using an in- ternal standard containing the same element at a known concentration, without relying on relative response fac- tors.

We now report on the applicability of the HDD as a device which possesses a reliable and attractive alter- native for direct quantitation of individual PCB conge- ners and mixtures.

EXPERIMENTAL

PCB Reference Solutions. Seventeen PCB congeners of known concentrations in hexane were provided by the Virginia Institute of Marine Science, College of William and Mary. An additional twelve congeners (Analabs) were weighed and diluted in hexane to yield concentrations on the order of 1 to 8 #g/mL. Internal references were prepared from 1,2,4,5-tetrachlorobenzene (TCB) and hexachlorobenzene (HCB) (Aldrich). Both standards were purified through two successive sublimations. Standard solutions containing 1000 #g/mL of each internal stan- dard were prepared in hexane (HCB required initial dis- solution in dichloroethane prior to dilution with hexane,

10 % V/V). The TCB and HCB were added to the PCB solutions at concentrations of 4 #g/mL each.

Stock solutions of Aroclor samples (Alltech) were pre- pared in hexane at a concentration of 1000 ~g/mL. Di- luted solutions containing the Aroclors and internal stan- dards were prepared at concentrations of 50 #g/mL and 2 #g/mL, respectively.

A p p a r a t u s . The design and operation of the HDD uti- lized in these investigations have been previously de- scribed. 15 All major instrument components were changed in the present study; however, the overall configuration was similar in design to that of the previous system. Details of the operational characteristics of the HDD, spectrometric system, and GC parameters used for the present investigation are summarized in Table I.

RESULTS AND DISCUSSION

The helium discharge detector provides a simple and reliable means of element-selective detection for GC by

586 Volume 42, Number 4, 1988 0003-7028/88/4204-058652.00/0 APPLIED SPECTROSCOPY © 1988 Society for Applied Spectroscopy

TABLE I. Operating conditions for the GC-HDD system.

He flow rate: Voltage: Frequency: Load power:

Monochromator: Bandpass (FWHM): Analytical wavelenth: Photomultiplier tube: PMT voltage: Amplifier: Integrator:

GC: Injection mode: Carrier gas: Column: Temp. program: Detector interface temp:

(A) Helium discharge detector 60 mL/min 6500 V (RMS)] 176 kHz ~ Model HPG-2 65 W ) (ENI Power Systems)

(B) Spectrometric system 0.5 m (Minuteman, Model 305M) 0.22 nm (100 ~m slits) 837.6 nm R758 (Hamamatsu) 1060 V (Keithley Model 247) Keithley Model 485 picoammeter Shimadzu Model CR3A

(C) GC system Carlo Erba Model 4180 On-column (1.0 ~L injections) Helium DB-5; 30 m × 0.25 i.d. (J&W) 65°C-6°C/min-280°C (15 rain) 300°C

generating atomic emission through the dissociation of analyte molecules present in the discharge. The system is relatively simple to construct and maintain, He gas consumption is low (60 mL/min), and day-to-day repro- ducibility is excellent. A significant advantage of the HDD is that GC effluents are introduced above the region where the primary He discharge is produced; thus the discharge region where solvents and analyte molecules are intro- duced is continually replenished by long-lived energetic He species from the primary discharge. 15

Absolute limits of detection observed for individual PCB congeners ranged from 10 pg for decachlorobi- phenyl to 35 pg for the monochlorobiphenyls, based on a signal-to-background scatter ratio of three. The signal response to C1 concentration was linear over approxi- mately four orders of magnitude. Selectivity over signals generated by nonanalyte species at the C1 analytical wavelength was approximately 350:1.

The ability to quantitate individual PCB congeners, using TCB and HCB as the internal standards, is given in Table II. The number of chlorinated sites and sub- stitution sites was varied as much as possible over the entire range of PCB congeners to validate the general applicability of the method. An average relative area per unit C1 concentration was used from the TCB and HCB peak areas to minimize possible integration errors. The average error observed for experimental concentrations was ___3.2% from the known values at congener concen- trations ranging from 1 to 7.5 ppm (1-7.5 #g/mL).

The unit area determined from the known C1 concen- tration of the internal standards could be directly cor- related to each congener concentration, since the emis- sion signal observed for each compound was solely dependent on the number of C1 atoms present. This as- sumption is well founded with respect to the low errors generally observed with the use of the chlorinated in- ternal standards. Most of the resultant error observed may be associated with how accurately the standard so- lutions could be prepared (considering that only 5-10 mg of each congener was available), as well as the integ- rity of the parameters selected for the integrator in es- tablishing the peak areas.

TABLE II. PCB congener concentrations determined by the HDD with the use of internal standard techniques.

Concentration

Chlorine substitution Known Found BZ# a pat tern (#g/mL) (#g/mL) % Error

1 2 7.32 7.14 -2 .44 3 4 5.96 5.59 -6.17

15 4, 4' 4.55 4.56 +0.22 18 2, 2', 5 1.74 1.77 +1.54 21 2, 3, 4 6.02 5.97 -0 .84 24 2, 3, 6 6.82 6.88 +0.92 31 2, 4', 5 1.23 1.32 +7.45 33 2', 3, 4 4.44 4.49 + 1.20 40 2, 2', 3, 3' 5.25 4.98 -5.20 49 2, 2', 4, 5' 0.98 1.00 +2.00 50 2, 2', 4, 6 3.62 3.56 -1.64 52 2, 2', 5, 5' 1.12 1.11 -1.16 54 2, 2', 6, 6' 6.43 6.67 +3.69 60 2, 3, 4, 4' 3.80 3.75 -1.35 61 2, 3, 4, 5 6.36 6.11 -3 .88 65 2, 3, 5, 6 5.38 5.09 -5.31 66 2, 3', 4, 4' 2.96 2.97 +0.65 70 2, 3', 4', 5 2.13 2.04 -4.38 77 3, 3', 4, 4' 0.912 0.933 +2.33 88 2, 2', 3, 4, 6 4.03 3.72 -7.73 97 2, 2', 3', 4, 5 4.80 4.79 -0.28

101 2, 2, 4, 5, 5' 5.91 5.90 -0.14 105 2, 3, 3', 4, 4' 1.98 1.90 -3.87 116 2, 3, 4, 5, 6 5.58 5.35 -4.16 136 2, 2', 3, 3', 6, 6' 5.00 4.72 -5.64 153 2, 2', 4, 4', 5, 5' 3.62 3.82 +5.51 171 2, 2', 3, 3', 4, 4', 6 4.34 4.58 +5.68 183 2, 2', 3, 4, 4', 5', 6 3.98 3.68 -7.55 209 2, 2', 3, 3', 4, 4', 5, 5', 6, 6' 1.00 1.02 +2.00

a Numbering nomenclature developed by Ballschmitter and Zell (Ref. 17).

One disadvantage of the technique is that, in order to perform a conversion from the C1 weight to PCB con- gener weights, one must know the number of C1 atoms present on the biphenyl ring; however, adequate reso- lution and accurate retention time identification of each congener could minimize such problems considerably.

We further confirmed the applicability of the tech- nique by determining the C1 percentage in standard Aro- clor samples; the results are presented in Table III. The % C1 determined with the use of the HDD was also compared with results obtained from a potentiometric titration method developed in our laboratory. 18 The total peak area obtained from the PCB congeners in the Aro- clor samples was compared with the average relative peak area per unit C1 concentration obtained from the TCB and HCB internal standards to assess the C1/Aroclor

TABLE III. Percent chlorine determined in Aroclor samples.

% CI found by"

Potentiometric Aroclor % C1 ti tration b GC-HDD

1221 21 21.5 (0.4) 1232 32 31.9 (0.6) 1242 42 42.6 (0.2) 1248 48 47.6 (0.4) 1254 54 53.8 (0.4)

21.7 (0.4) 32.0 (0.3) 42.3 (0.5) 48.1 (0.2) 53.4 (0.4)

"Mean of triplicate determinations (_+average deviation). b Ref. 18.

APPLIED SPECTROSCOPY 587

weight percentage. The % C1 de te rmined in all the Aro- clot samples was in close agreement with the nominal values expected, as well as the values ob ta ined f rom the po ten t iomet r ic method. An interest ing aspect of these results is t ha t the % C1 could be de te rmined wi thout prior identification of the individual PCB congeners, again indicating t ha t the C1 emission observed was indepen- den t of the s t ruc ture for each respect ive PCB congener p resen t in the Aroclor samples. A p r imary concern was t h a t several PCB congener peaks would overlap with the in ternal s tandards; however, no significant peaks were observed a t the re tent ion t imes of the in ternal s tandards (~15.6 min for T C B and ~24.4 min for H C B under the presen t exper imenta l conditions).

Compar i sons to ch romatograms obta ined with the use of a flame ionization detec tor indicated t h a t there was no discernible loss of resolut ion or peak qual i ty th rough the hea ted interface between the column and HDD. One advan tage no ted was the absence of detectable column bleed a t the high GC column t empera tu re s (280-310°C) necessary to elute the highly chlor inated P C B congeners. Even a t the highest permissable gain set t ing (4 nA full scale) a t t empe ra tu r e s of 300°C over a 30 min period, no greater t han a 5 % full-scale increase in the baseline was observed. This observat ion is in direct cont ras t to results f rom convent ional nonselect ive GC detectors, in which the background signal can significantly increase over the course of a ch romatogram, especially at low a t tenua t ions and high column tempera tu res . Such prob lems can resul t in in tegrat ion errors for peak area de te rmina t ions or in the need for greater signal a t t enua t ion to compensa te for excessive background shifts.

ACKNOWLEDGMENTS This work was supported by an ACS-PRF Grant (#17417-GB3) and

a Thomas F. and Kate Miller Jeffress Memorial Research Grant (J- 99). This work was presented in part at the 1987 Federation of Ana- lytical Chemistry and Spectroscopy Societies Conference, Detroit, MI.

1. J. H. Koeman, M. C. Debrouw, and R. H. DeVos, Nature (London) 221, 1126 (1969).

2. L. Fishbein, J. Chromatogr. 68, 345 (1972). 3. M. Cooke, G. Nickless, A. M. Prescott, and D. J. Roberts, J. Chro-

matogr. 156, 293 (1978). 4. A. L. Robbins and C. R. Wilhite, Bull. Environ. Contam. Toxicol.

21, 428 (1979). 5. J. M. Lin and S. S. Que Hee, Anal. Chem. 57, 2130 (1985). 6. R. E. Lea, R. Branston-Cook, and J. Tschida, Anal. Chem. 55, 626

(1983). 7. W. J. Dunn, D. L. Stalling, T. R. Schwartz, J. W. Hogan, J. D.

Petty, E. Johansson, and S. Wold, Anal. Chem. 56, 1308 (1984). 8. A. L. Alford-Stevens, T. A. Bellar, J. W. Eichelberger, and W. L.

Budde, Anal. Chem. 58, 2014 (1986). 9. A. Poland and E. Glover, Mol. Pharmacol. 13, 924 (1977).

10. J. Goldstein, P. Hickman, H. Bergman, J. McKinney, and M. Walk- er, Chem. Biol. Interact. 17, 69 (1977).

11. J. W. Rote and P. G. Murphy, Bull. Environ. Contam. Toxicol. 6, 377 (1971).

12. M. Mullin, C. M. Pochini, S. McCrindle, M. Ronkes, S. Safe, and L. Safe, Environ. Sci. Technol. 18, 648 (1984).

13. S. D. Cooper, M. A. Moseley, and E. D. Pellizzari, Anal. Chem. 57, 2469 (1985).

14. T. Cairns and E. S. Siegmond, Anal. Chem. 53, 1183A (1981). 15. G. W. Rice, A. P. D'Silva, and V. A. Fassel, Spectrochim. Acta 40B,

1573 (1985). 16. G. W. Rice, Quarterly Report, Ames Laboratory, Ames, IA, 4/84-

6/84. 17. K. Ballschmitter and M. Zell, Fresenius Z. Anal. Chem. 302, 20

(1980). 18. M. L. Ware, M. D. Argentine, and G. W. Rice, Anal. Chem., 60,

383 (1988).

Spectroscopic Quantitative Analysis of Strongly Interacting Systems: Human Plasma Protein Mixtures

M A R C R. N Y D E N , * G L E N N P. F O R N E Y , and K R I S H N A N C H I T T U R Center for Fire Research, National Bureau of Standards, Gaithersburg, Maryland 20899 (M.R.N., G.P.F.); and National Center for Biomedical Infrared Spectroscopy, Battelle Laboratories, Columbus, Ohio (K.C.)

Blood plasm~L protein infrared spectra, while qualitatively very similar, display subtle differences in the frequencies and intensities of absorption bands. These small differences are sufficient to permit an accurate quan- titative analysis of mixtures of these proteins. In this paper we examine the performance of some alternative methods of spectroscopic quanti- tative analysis in determining the concentrations of proteins in aqueous solutions. T~e widely-used K matrix method, using sloping baselines and intercept functions, was found to be inadequate for these determi- nations. In contrast, a method based on the little-known Q matrix ap- proach, augmented by a robust equation solver, yielded results with a sufficient degree of accuracy to make it a viable tool for use in the study of proteins at solid interfaces and for more general applications in the field of protein chemistry. Index Headings: Infrared; Spectroscopic techniques.

Received 7 December 1987. * Author to whom correspondence should be sent.

I N T R O D U C T I O N

Techniques based on Fourier t r ans fo rm infrared with a t t enua ted to ta l in ternal reflection ( F T - I R / A T R ) have been used to probe the dynamics of the interact ions be- tween synthet ic po lymer surfaces and blood proteins. 1~ In te res t in this area has been mot iva t ed by the real izat ion t ha t the fo rmat ion and behavior of ha rmfu l pla te le t and white cell t h romb i on the surfaces of synthe t ic polymer ic surfaces are de te rmined to a large ex ten t by the na ture of the adsorbed pro te in layer t h a t forms on such surfaces. Recen t papers have highl ighted the kinds of in format ion t ha t can be ob ta ined by s tudying cer tain prote ins such as a lbumin, immunoglobul in G, and fibrinogen on a num- ber of polymeric surfaces of clinical interest . Pa s t efforts were focused on the descr ipt ion of the kinetics and c o n -

588 Volume 42, Number 4, 1988 ooo3-7o28/ss/42o4-o~ss$zoo/o APPLIED SPECTROSCOPY © 1988 Society for Applied Spectroscopy