by B.K. Haines, T.L. King, R. Tokarczyk, J.F. Uthe, and C.L. Chou
Department of Fisheries and Oceans, Marine Chemistry Section, Halifax, Nova Scotia (Canada)

In the Marine Environmental Sciences Division, we have analyzed a variety of marine life for organic and inorganic trace contaminants -- the water, suspended particulate, marine sediment, American lobster (Homarus americanus), snow crab (Chionoecetes opilio), Atlantic cod (Gadus morhua), and marine growth like anemones and corals. The analytical procedure we developed for this wide range of samples can be applied to the extracts of other biological or soil samples with little or no modification.

We used the Hewlett-Packard HP 7686A PrepStation (SPE Module) to develop automated microscale sample preparation methods for the isolation of planar chlorobiphenyls. Carbon chromatography permits the separation of non-ortho substituted chlorobiphenyls (non-o-CBs, sometimes referred to as planar or coplanar chlorobiphenyls) from complex PCB mixtures. Non-o-CBs are similar to the highly toxic polychlorinated dibenzo-p-dioxins. Their concentrations are often expressed in terms of toxic equivalence to 2,3,7,8-tetrachlorodibenzo-p-dioxin.¹

Improved Detection Limits

Earlier carbon chromatographic isolations of non-o-CBs and dioxins required large quantities of benzene and hours of elution. An improvement was the development of a rapid semimicro method using a mixture of activated charcoal and Florisil® with vacuum-aided elution by 500 mL of toluene.² The technique we used improves GC/MS detection limits by reducing the number of ions in the selected ion monitoring method. It also permits the analysis of samples containing other, more abundant, chlorobiphenyls without overloading the GC column or MSD (mass selective detector). Isolation of non-o-CBs from CB mixtures also allows for their measurement by electron capture detection.

The PrepStation system prepares samples at the gas chromatograph (GC) autosampler tray, immediately ready for GC analysis. The Bench Supervisor software controls the preparation and transfer of samples from the PrepStation to the GC. We used an SPE cartridge custom-packed with 300 mg of Florisil to enrich chlorinated fatty acid methyl esters from lipid extracts, which reduced the interference by non-chlorinated fatty acid methyl esters in the GC/MS analysis. Chlorinated fatty acids were recently found in lobster digestive gland.³

Assessing the Recoveries

The table shows the microcolumn recoveries of the non-ortho CBs grouped by class and listed in order of GC elution (Figure 2). Within each class, these recoveries increase with GC elution order and compare well with published values. The CBs of lower recovery (International Union of Pure and Applied Chemistry [IUPAC] Nos. 36, 80, and 127) are the first to elute from the microcolumn, which is consistent with fractionation trials conducted with only non-o-CBs applied to the cartridge.

Using 800 µg of activated charcoal and small solvent volumes provided a clean split between non-o-CBs and tri- or tetra-ortho chlorinated CBs and a close split between non-o-CBs and mono- or di-ortho chlorinated CBs. Most of the non-o-CBs are split 95-5 with respect to the mono-ortho CBs, with the exception of IUPAC Nos. 36, 80, and 127. To increase their recoveries, less than 10 mL of isooctane would have to be used to elute the ortho chlorinated CBs, but this approach would cause ortho chlorinated CBs to be present in the toluene (non-o-CB) fraction. Their presence is undesirable, because CB IUPAC Nos. 80 and 127 coelute with CB IUPAC Nos. 66 and 105, respectively, on the HP-5MS capillary GC column.

Note that IUPAC Nos. 66 and 105 are both mono-ortho substituted. The coeluting pairs 66/80 and 105/127 are indistinguishable using selected ion monitoring on a low-resolution MSD. To verify the microcolumn split of these pairs, the fractions must be checked on another column, such as the HP-35. The retention times of all 159 individual CBs in our mixture have been tabulated only for the HP-5MS column.

IUPAC No.	Mean Recovery (%)	Relative Standard Deviation (%)
36	85.6	2.6
39	91.1	2.8
38	96.6	3.6
35	94.9	3.6
37	93.4	3.9
80	90.3	4.1
79	94.3	2.6
78	93.8	3.2
81	95.5	3.4
77	94.2	3.1
127	88.9	3.4
126	94.2	2.7
169	90.1	2.9
Mean percent recoveries (n=5) of non-ortho substituted chlorobiphenyls from the micro-column. The method recovery of 13C-CB No. 77 added at midscale of the calibration to the marine biota samples before extraction was 93.0 ± 4.5% (n=5).

Decachlorobiphenyl IUPAC No. 209

Fine-Tuning the Separations

Preparing the cartridges one to two days before use did not affect the separation. Separately prepared batches of the support yielded consistent separations when tested with the standard mixture; see Figures 1 (all CBs) and 2 (non-o-CBs). The dominant CBs in the original sample extract and standard mixture are absent from the toluene (non-o-CB) fractions shown in Figure 2. The result is a clean, simple chromatogram showing the CBs of major toxicological interest.

We performed multiple fractionation, flow rate and solvent trials using the eight solvent ports of our system. Acetone, ethyl acetate, methanol, hexane, isooctane, cyclopentane, cyclohexane, dichloro-methane, benzene, toluene and xylenes were all tested. Isooctane and toluene or xylenes offered the best separation; benzene, used in the original methodology, was inferior to both toluene and xylenes.

Sample Preparation and Analysis Marine biota samples were digested with ethanolic potassium hydroxide in an ultrasonic bath and extracted with hexane. Further cleanup by gel permeation chromatography and concentrated sulfuric acid treatment removed unsaponifiable fats and PAHs (polycyclic aromatic hydrocarbons).4 Separation of non-o-CBs from the extracts was the last step before GC/MS analysis (HP 5890 Series II GC and HP 5971A mass selective detector). A mixture containing 159 CBs, each at a nominal concentration of 100 ng/mL (Figure 1), was used to develop the separation. We packed 100±2 mg support (Florisil®:activated charcoal, 128:1, w/w)4 into 100-µL cartridges shortly before use. The PrepStation method in outline: Rinse the entire system with 6.000 mL of isooctane and aspirate 1.400 mL of sample (in isooctane) into the calibrated 2.5-mL sample loop. Load an additional 425 mL of isooctane into the sample loop. Rinse the cartridge with 2.000 mL of isooctane, then apply the sample with 1.750 mL of isooctane at 1.0 mL/min, while directing the eluate to waste. Elute the ortho-substituted chlorobiphenyls from the cartridge to waste with 10.000 mL of isooctane at 1.0 mL/min. Elute the non-o-CBs with 1.700 mL of toluene at 0.5 mL/min. Flush the cartridge with xylenes to ensure complete elution of the non-o-CBs and any dioxins or furans that may be present. Dry the cartridge with nitrogen inside the station before it is returned to the waste port at the tray. Cartridges can then be discarded as regular waste after use. To maximize the amount of sample applied to the cartridge, we aspirated 1.400 mL of the 1.60 mL sample volume from a 2-mL crimp-top vial. A smaller application volume (100-250 µL) is desirable but would require using a smaller-volume sample vial compatible with the PrepStation. We considered microliter inserts inappropriate for use with the PrepStation, but the new High Recovery Vial® should allow a decrease in application volume while increasing the proportion of the sample applied to the microcolumn. We discovered that certain solvents (toluene and xylenes in particular) caused the Teflon-lined rubber or silicon septa on the autosampler vials to swell. The swollen septa blocked gas vent gutters on the evaporation and dispensing needles, which interfered with the concentration of eluted fractions using nitrogen. Septum swelling also caused backpressure in the flow system during elution of cartridges and increased the incidence of system leaks. Therefore, we covered the Target® screw cap vials used for collection of the toluene and xylene fractions with clean aluminum foil instead of septa to prevent backpressure. We replaced the foil with septa immediately after each set was run. No significant solvent loss was measured.

References

1. Battershill, Jon M. Review of the Safety Assessment of Polychlorinated Biphenyls (PCBs) with Particular Reference to Reproductive Toxicity. Human & Experimental Toxicology, 13, 581-597 (1994).

2. King, Thomas L., et al. Rapid Semi-micro Method for Separating Non-ortho Chlorobiphenyls From Other Chlorobiphenyls. Analyst, July 1995, Vol. 120.

3. Curtis, Jonathan M., et al. Dichloromyristic Acid, a Major Component of Organochlorine Load in Lobster Digestive Gland. Environmental Science & Technology, Vol. 31, Number 2, 535-541 (1997).

4. King, Thomas L. and Uthe, John F. Rapid Screening of Fish Tissue for Polychlorinated Dibenzo-p-dioxins and Dibenzofurans. Analyst, Oct. 1993, Vol. 118.