had fragments at 313 (M' - COOCHJ and 166 (M' COOCHB- COC2FS)amu but had no detectable molecular ion which was calculated to occur at 372 amu (Table I).
DISCUSSION GC-MS methods are capable of providing sensitive and highly specific determinations of aromatic monoamines and their metabolites, provided that suitable volatile derivatives of these compounds can be formed. The present data demonstrate potential problems of reduced sensitivity that may arise in derivatization of monoamine metabolites using pentafluoropropionic anhydride and halogenated alcohols. Such reduced sensitivity is attributed to multiple derivative formation. Although mass fragmentographic methods are known to be specific at high sensitivities, there is always a possibility of detecting ghost peaks from unknown impurities or from multiple products such as those demonstrated. The smaller and simpler the fragment monitored, the more likely is the chance of occurrence of a ghost peak, since such a fragment will be present in many other molecules in a complex biological sample. In the case of 5-HIAA, the completely derivatized product (Ia) showed a fragment at 291 amu while the partially derivatized product (Ib) showed a fragment at 292 amu. If 5-HIAA-d2is used as an internal standard, the corresponding fragments are found at 293 and 294 amu. The product Ib migrates very slowly with a relative retention time of 6.6 and therefore its presence will produce a peak at 292 amu and to a lesser extent at 293 amu due to the isotopic contribution of the elements in that fragment. Hence, for each injection of the 5-HIAA derivatives (Ia and Ib) a ghost peak will be produced at 293 amu which may interfere with the quantitation procedure. Double ion monitoring for both the metabolite and the internal standard will detect such an effect. It is obvious from the structures of these products that they will also be detected by GC using a electron capture detector. In the work of Watson et al. (9) a number of GC-electron capture detection peaks are shown for 5-HIAA. The main product was shown to have a retention time of 10 min with a carrier gas flow rate of 45 mL/min. Comparing with our data, it seems that this product should have the structure Ib instead of Ia, as assigned by these authors. One of the earlier and minor peaks in the chromatogram will be Ia. With respect to 5-HIAA, the formation of the partially derivatized mon-
opentafluoropropionyl product Ib was substantially reduced by increasing the heating interval of the reaction mixture to 3 h. In the case of VMA, extending the duration of the first reaction up to 3 h or running the reaction between room temperature and 75 "C did not reduce the by-product formation in the presence of pentafluoro-n-propanol or trifluoroethanol. Our results are in agreement with Gelpi et al. (6)who found that longer heating times are required for complete derivatization of the indole compounds. They also found that when 3,4-dihydroxymandelicacid was esterified with methanol and HC1, a methylated by-product was formed which matched the Kovats Index of IIe but its mass spectrum indicated methylation of the P-hydroxyl group. In the present work we have confirmed unequivocally the structures of such byproducts of VMA. Pentafluoro-n-propanol and trifluoroethanol used for esterification of p-hydroxy substituted mandelic acids such as VMA or dihydroxymandelic acid partially alkylate the side chain hydroxyl group and, therefore, are not preferable for this purpose. The more suitable esterification reagent is methanol and HC1 which is simple to prepare and inexpensive.
ACKNOWLEDGMENT We thank Mrs. D. Ratansi for the preparation of this manuscript. LITERATURE CITED (1) F. Cattabeni, S.H. Koslow, and E. Costa, Science, 178, 890 (1973). (2) S.H. Koslow, F. Cattabeni, and E. Costa, Science, 176, 177 (1972). (3) E. K. Gordon, J. Oliver, K. Black, and I. J. Kopin, Biochem. Med., 11, 32 (1974). (4) E. Anggard, and G. Sedvall, Anal. Chem., 41, 1250 (1969). (5) B. Sjoqulst and E. Anggard, Anal. Chem., 44, 2297 (1972). (6) E. Gelpi, E. Peralta. and J. Segura, J. Chromatogr. Scl., 12,701 (1974). (7) E. K. Gordon, J. Oliver, and I. J. Kopin, Life Sci., 16, 1527 (1975). (8) S. W. Dziedzic, L. M. Bertani, D. D. Clarke, and S . E. Gitlow, Anal. Biochem., 47, 592 (1972). (9) E. Watson, S. Wilk, and J. Roboz, Anal. Biochem., 59, 441 (1974). (IO) J. D. Pearson and D. F. Sharman, Br. J . Pharmacol. Chemother., 5 3 , 143 (1975). (11) J. Warsh, Ph.D. thesis. University of Toronto, 1974. (12) F. Karoum, J. Gillen, and R. J. Wyatt, J. Neurochem., 25, 653 (1975).
RECEIVED for review November 8,1976. Accepted March 21, 1977. This work was supported by an equipment grant from Ontario Mental Health Foundation of which J. J. Warsh is a research scholar.
Determination of 2,3,7,8-Tetrachlorodibenzo-p-dioxin in Environmental Samples by High-Resolution Gas Chromatography and Low Resolution Mass Spectrometry Hans-Rudolf Buser Swiss Federal Research Station, CH-8820 Wadenswil, Switzerland
High-resolutlon gas chromatography In combination wlth mass spectrometrlc detection (mass fragmentography) was demonstrated to be a powerful means for detecting trace amounts of 2,3,7,8-TCDD in envlronmental samples. The separation of different TCDD Isomers was studied on OV-101 and OV-17 glass caplllary columns and some dlfferences in the mass spectra of these isomers were pointed out. TCDD In envlronmental samples from Seveso, Italy, had the same retentlon times on these columns and the same mass spectrum as standard 2,3,7,8-TCDD. 918
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
The accidental release of a chemical cloud containing 2,3,7,8-tetrachlorodibenzo-p-dioxin(2,3,7,8-TCDD) and 2,4,5-trichloropheno1(2,4,5-TCP) into the atmosphere and over the surrounding area at Seveso on July 10,1976, caused a most serious environmental contamination. 2,3,7,8-TCDD (for structure, see Figure 1)has been recognized as an extremely toxic ( I ) , teratogenic (Z), mutagenic (3), and possibly carcinogenic compound stable in biological systems. It therefore presents a threat to life and environment. The high toxicity of 2,3,7,8-TCDD requires very sensitive
Figure 1. Structure of 2,3,7,8-tetrachlorodibenzo-pdioxin (2,3,7,8numbering according to Chemical Abstracts system
TCDD);
and specific analytical techniques. Detection levels have to be orders of magnitude below the usual limits of pesticide residue analysis. In recent years, several analytical methods (4-9) have been developed for the determination of TCDDs and related compounds in environmental and industrial samples, the most specific methods using mass spectrometric detection. Ultra trace analyses of 2,3,7,8-TCDD with detection limits below the 1-ppt-level (1 part in 1OI2 parts) will ultimately be required in order to prove the absence of untolerable quantities of 2,3,7,8-TCDD in the environment. Any analysis of environmental samples at these low levels is complicated by the presence of a multitude of interfering compounds ranging from natural products to industrial pollutants. The best available separation techniques followed by highly specific detectors have to be used to accurately determine this dangerous compound. Theoretically, a total of 75 different polychlorinated dibenzo-p-dioxins (PCDDs) exist, including 22 TCDD isomers. Different isomers of the same dioxin may vary significantly in their toxicological properties (1)and their identification therefore becomes important. Although no other isomers are expected to be formed in the Seveso accident, an accurate determination of 2,3,7,8-TCDDallowing a differentiation from other TCDD isomers is desirable. The separation of these isomers may be difficult and requires highest separation efficiencies, such as those from glass capillary columns. In previous papers (10, 11),we have reported the application of these columns for the separation of hexachlorodibenzo-pdioxin (HCDD) isomers and of higher chlorinated PCDDs and dibenzofurans (PCDFs) present as contaminants in industrial samples. In this paper, we report results on the application of high-resolution gas chromatography to the analysis of 2,3,7,8-TCDD in environmental samples. Glass capillary columns in combination with mass specific detection (mass fragmentography) were used to achieve high sensitivity and specificity in the detection of this dangerous compound. EXPERIMENTAL Gas Chromatography-Mass Spectrometry. OV-101 and OV-17 glass capillary columns of 20-30 m length and 0.35-0.37 mm i.d. were prepared in this laboratory but may be substituted by columns from commercial sources. Efficiencies of 50 000-70OOO theoretical plates were obtained for 2,3,7,8-TCDDunder operating conditions. The columns were operated at 207-220 "C with helium carrier gas pressures of 0.60 atm resulting in average linear velocities of 35 cm/s. They were mounted in a Finnigan gas chromatograph equipped with inlet splitter, septum flush, and low volume injector. A vaporizer temperature of 275 "C was used. Sample introduction was effected by an isothermal splitless injection technique. Two-pL aliquots of samples in n-tetradecane were injected with the inlet splitter closed. Fifteen s after injection, the valve for inlet splitter and septum flush was opened (60-80 mL/min) and the analysis continued. The glass capillary columns were coupled via a fused 0.15-mm i.d. platinum capillary leading directly into the ion-source of a Finnigan 1015 D quadrupole mass spectrometer. The platinum capillary interface was held at 250 "C. The electron-impact (EI) ion source was operated at an electron energy of 70 eV. The filament of the ion source was switched off during elution of the solvent. For mass specific detection (mass fragmentography) of TCDDs, a programmable multiple ion monitoring attachment (PROMIM) was used and the ions at m / e 320, 322, and 324 simultaneously recorded.
Reference Compounds. 2,3,7,8-TCDDwas obtained from Stickstoffwerke Linz (Linz, Austria); 1,2,3,4-and 1,2,3,8-TCDD were the gift of C. Rappe (Universityof Umea, Sweden). A series of other TCDD isomers was prepared by microscale pyrolysis of potassium trichlorophenates (2,3,4-,2,3,5-, and 2,3,6-TCP)under conditions previously described (IO). They were not isolated in pure form. Preparation of Environmental Samples. Twenty to fifty g of plant or soil were extracted with 100 mL of n-hexane-acetone (1:l)by shaking for 60 min. The extract was partitioned with two portions of 100 and 50 mL of water, respectively; 1 mL of 5 N lithium hydroxide was added to the first portion. In case of incomplete or difficult separations, sodium sulfate was added to break the emulsion. An aliquot of the organic phase (25 mL correspondingto 10-25 g sample) was passed through a prewashed column (20 X 1cm) containing sodium sulfate (3 cm, top layer), Celite-sulfuric acid (5 cm) and silica gel (3 cm, 70-230 mesh, Merck, Darmstadt, G.F.R.). Celite-sulfuric acid was prepared by thorough mixing of 3 mL of concentrated sulfuric acid and 20 g of Celite. Some pressure was applied to the column for fast elution. The column was washed with 20 mL of n-hexane. The combined eluates were concentrated to 1-2 mL and chromatographed on an alumina micro-column as previously described (12), except that TCDDs are eluted with 20% methylene chloride in n-hexane (13). This eluate was carefully brought to dryness in a stream of nitrogen and redissolved in 10-50 pL of n-tetradecane, slightly heating if required. A 2-pL aliquot was used for analysis. Samplesheavily contaminated with TCDDs were injected directly as dilute solutions in n-tetradecane after extraction, partition, and alumina chromatography.
RESULTS AND DISCUSSION Gas Chromatography of TCDDs Using Glass Capillary Columns. Glass capillaries were coated with OV-101 (nonpolar methyl silicone) and OV-17 (semi-polar methyl phenyl silicone) as stationary phases. The columns were of the same type as used in a previous study on PCDDs and PCDFs (11). Thin films of stationary phases allowed lower operating temperatures for the rather high-boiling TCDDs, than when using conventional packed columns. Column temperatures of 207-220 "C were used in order to elute TCDDs within a 4-10 min range. The usual stream splitting used in capillary gas chromatography (90-99% of a sample vented) is not applicable to trace analyses. For such analyses, splitless injections are required. In the present study, we have used a splitless injection technique described by Grob and Grob (14) that allows an isothermal operation of capillary columns by using a high-boiling solvent. n-Tetradecane was found to be an excellent solvent in the analysis of TCDDs under the operating conditions described. At other column temperatures, a different solvent may have to be selected. Separation efficiencies and peak shapes were at least as good as when using a lower boiling solvent with injection a t low temperature followed by subsequent temperature programming of the columns. No adverse effects were observed, although n-tetradecane has a boiling point (253 OC) above the column temperatures used and was directly introduced into the mass spectrometer. Injection periods of 15 s (time between injection and reopening of splitting system) allowed about 95% of a sample to be transferred onto the column. Longer injection periods resulted in some tailing of the peaks observed. Sample aliquo& of 2 pL were injected for best peak shapes; some tailing was observed when smaller sample volumes were injected; larger volumes caused disturbance of the vacuum system of the mass spectrometer. Glass capillary columns were coupled to the quadrupole mass spectrometer via a fused platinum capillary allowing the complete sample to enter the ion source. The columns showed the same efficiencies as with direct coupling to a flame ionization detector. Figure 2 shows a chromatogram of 2,3,7,8-TCDD on a 25-m OV-17 glass capillary column a t 220 ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
919
I
-c
2.558
I74
> I
/2'3'7'8TCDD tr. 285 6
I C-Hf
ni70000
Figure 4. Partial mass spectra ( W e 70-80) of 2,3,7,8-, 1,2,3,8-,and
1,2,3,4-TCDD '5II ,
601
,
"
=lo
20 PB
ov-101
OV-17
5 20
7
B h
30
10 pg
50
1 5
min
,
'j 0
lnlected
Figure 5. Calibration curve (peak heights vs. amounts injected) and mass fragmentograms ( M e 320) of 2,3,7,8-TCDD in low picogram range; isothermal splitless injections at 220 OC on 25-m OV-17 glass capillary column (m.d.q. = minimal detectable quantity) 9
rnin
7
5
9
rnin
7
5
Flgure 3. Partial mass fragmentograms ( W e 320) showing separation of nine TCDD isomers on glass capillary columns, isothermal splitless injections at 207 'C: (a) OV-17, 25 m, 0.37 mm i.d., (b) OV-101, 27 m, 0.35 mm i.d. Peak assignments: 1 = 2,3,7,8-;2 = 1,2,3,8-;3 = 1,2,3,4-TCDD; 4 and 5 = 1,3,6,8- and 1,3,7,9-TCDD or vice versa, from 2,3,5-TCP; 6 and 7 = 1,4,6,9- and 1,2,6,9-TCDD or vice versa, from 2,3,6-TCP; 8 and 9 = 1,2,6,7- and 1,2,8,9-TCDD or vice versa, from 2,3,4-TCP
"C using the isothermal splitless injection technique. A column efficiency of 70 000 theoretical plates with an elution band width at half peak height of 2.5 s was obtained. The almost symmetrical peak indicates negligible adsorption of 2,3,7,8-TCDD on the column or the interface. Figure 3 shows chromatograms of a test mixture containing different TCDD isomers analyzed on a 25-m OV-17 and on a 27-m OV-101 glass capillary column. Some of the isomers of the test mixture were obtained as reference compounds; others were from the pyrolysis of potassium trichlorophenates. Isomers in pyrolysis samples could often not be conclusively assigned, because multiple peaks were usually obtained (Smiles rearrangement, 15). Nine of the 22 theoretically possible TCDD isomers are observed in these chromatograms. Neither column is able to separate all isomers, but a combination of both allows a differentiation between these components. Mass Spectra of TCDDs. The electron-impact mass spectra of PCDDs show strong molecular ions (M'). Fragmentation occurs through the loss of CO and C1 radicals (16). Major ions are at M+ - 63 (M' - COC1) and M+ - 126 (M' - ZCOCl), both characteristic and diagnostically important of PCDDs. Doubly charged molecular ions (M") of some intensity are also observed. Minor fragmentation ions are at M' - 35, M+ - 70 and M' - 98 (loss of C1, Clz, and COCl + C1, respectively). The usual characteristic ion clusterings due to the chlorine isotopes are observed. Based on molecular ions and fragmentation mode, PCDDs are easily distinguishedfrom other chlorinated pollutants, such as polychlorinated biphenyls (PCBs), polychlorinated naphthalines (PCNs),polychlorinated diphenyl ethers (PCDPEs), PCDFs, and chlorine containing pesticides. The mass spectrum of 2,3,7,8-TCDD (Figure 9), shows the molecular ion and its isotope peaks at mle 320,322, and 324. Major ions are at mle 257, 259 (M' - COCl), 194, 196 (Mb' 920
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
- 2COC1); minor ions at mle 285,287 (M' - Cl), 250,252 (M' - Clz), 222 and 224 (M' - COCl - C1) and doubly charged
molecular ions at mle 160, 161, and 162. Only minor differences in the intensity of ions in the higher mass range were observed for different TCDD isomers. However, significant differences in the mass spectra of some isomers were found in the low mass range, e.g., at mle 7475, and 76. Partial mass spectra of 2,3,7,8-, 1,2,3,8-, and 1,2,3,4-TCDD in this mass range are shown in Figure 4. 2,3,7,8-TCDD shows a significant peak at mle 74 (C6H2'); the spectrum of 1,2,3,8-TCDDshows peaks at mle 73,74, and 75 (C6H3+)and that of 1,2,3,4-TCDD at mle 74, 75, and 76 (c6H4'). All isomers prepared by pyrolysis of trichlorophenates, expected to form TCDDs containing 2 chlorine substituents on each ring, gave major ions at mle 74 only. These ions may allow a determination of the type of chlorine substitution of an unknown TCDD isomer, i.e., the number of chlorine substituents in each ring of the dioxin system. Mass Fragmentography of TCDDs. Single or multiple ion detection (mass fragmentography) is carried out by adjusting the mass spectrometer to monitor one or several ions, previously selected and representative of the compound of interest. In this mode, far higher sensitivities are obtained than when scanning complete mass spectra. High specificity is obtained in that only compounds producing those ions are recorded. Interferences from co-eluting components can usually be minimized by monitoring ions at higher mle values. The ions used for the detection of TCDDs were the molecular ions and its isotope peaks at mle 320,322, and 324, the most intense ions in the high mass range. Interferences from co-eluting components are usually smallest at mle 320; PCBs, if present, may interfere at mle 322 and 324 (heptachlorobiphenyl, M' - Clz = 322; pentachlorobiphenyl, M' = 324). Sensitivity and linearity of the mass fragmentographic response in combination with glass capillary columns were examined by injecting a series of 2,3,7,8-TCDD dilutions in n-tetradecane. Splitless injections were made on a 25-m OV-17 column, isothermally operated a t 220 "C. Mass fragmentograms (mle 320) obtained at highest sensitivity are shown in Figure 5. Linearity was established from the low picogram to the nanogram range. A typical calibration curve (peak heights vs. amounts injected) of 2,3,7,8-TCDD in the low picogram (5-50 pg) range is included in Figure 5. The
1
Soil, Zone A, Seveso
Grass,
7 min
5
3
7 rnin
5
Figure 6. Partial mass fragmentograms at m/e 320 (0.2-V f.s.) and 322 (10-V f.s.)of (a)Aroclor 1254 (8 hg) and (b) Aroclor 1254 (8 hg) with 2,3,7,8-TCDD (200 pg) added chromatograms indicate a minimum detectable quantity of 2-3 pg TCDD under the operating conditions used. This value should be considered as a very conservative estimate of the maximum sensitivity obtainable by mass fragmentography. By using more modern equipment, this value may easily be exceeded. Final utilization of the maximum sensitivity will depend on the levels of interfering co-extractants in individual samples. The high specificity of mass fragmentography was established by analyzing 2,3,7,8-TCDD in the presence of large amounts of PCBs. Aroclor 1254, a commercial PCB, was analyzed at mle 320 and 322 before and after addition of 25 ppm of 2,3,7,8-TCDD. Hepta- and to some extent hexachlorobiphenyls, present in Aroclor 1254, yield signals at mle 322 due to M+ - Clz and M+ - HC1 ions, respectively. These compounds were used to examine possible interference at mle 320. In Figure 6, mass fragmentograms obtained at mle 320 and 322 are shown. At m l e 322 and low sensitivity (10-V f.s.) significant peaks are seen due to the presence of hexa- and heptachlorobiphenyls preventing recognition of a peak for 2,3,7,8-TCDD. At m / e 320, these interfering peaks are up to ZOOOX smaller, allowing a much higher sensitivity (0.2-V f.s.) to be used. At this recording, a peak corresponding to 200 pg (25 ppm relative to Aroclor) of 2,3,7,8-TCDD is easily recognizable. The intensity of this peak was not affected by the presence of the high quantity of co-eluting PCBs. Analysis of Environmental Samples. The sample preparation procedure used for environmental samples such as plants and soil included an extraction with n-hexaneacetone, aqueous partition of the extract under basic and neutral conditions, initial clean-up on Celite-sulfuric acidsilica gel, and final chromatography on a alumina microcolumn. Clean extracts were obtained that allowed high final sample concentrations. Recoveries of 2,3,7,8-TCDD added to samples before extraction were excellent (>so%)down to the 500-pg level (10 ppt), not requiring the addition of a carrier. Detector response remained sufficiently stable over a suitable period of time to allow the use of external standards for quantitation. Known quantities of 2,3,7,8-TCDD added to samples gave the expected increases in detector response. Two types of samples were examined during the course of this investigation. A first series of samples, heavily contaminated with TCDD, was from Zone A at Seveso; a second series, not expected to be contaminated with any TCDD, was from two regions in northern and southern Switzerland. Mass fragmentograms ( m l e 320, 322, and 324) of a soil and grass sample from Seveso are shown in Figure 7. Equivalents of 1.6 mg of sample were injected and analyzed on a OV-17 glass capillary column at 210 "C. A major peak is observed on all recordings at the retention time of 2,3,7,8-TCDD. The expected response ratio for a TCDD is obtained and the
4
8 rnl"
3
Figure 7. Mass fragmentograms(m/e320, 322, and 324) of grass and soil samples, Zone A, Seveso; equivalents of 1.6 mg of sample injected; 25-m OV-17 glass capillary column at 210 O C ; sensitivity, 0.5-V f.s.
Figure 8. Chromatogram (total ion monitor, m/e 35-500) of soil sample, Zone A, Seveso; equivalent of 160 mg of sample injected; 25-m OV-17 glass capillary column at 210 O C . Peak assignment: 1 = tri-CDD, 2 and 3 = TCDDs. Sensitivity, 0.5-V f.s. Seveso
I
,I
Soil, Peak 3
M'
Standard
,
1.
70
100
150
=0\
2,3,7,8-TCDD
200
m/e
Mt-CI
I 1
300
Figure 9. Mass spectra of soil sample (peak 3), Zone A, Seveso and standard 2,3,7,8-TCDD component does co-chromatograph with standard 2,3,7,8TCDD on OV-101 and OV-17 glass capillary columns. The contents of TCDD found in these particular grass and soil samples were corresponding to 250 and 500 ppb, respectively. Mass spectrometric analyses were carried out on larger sample equivalents (160 mg). A chromatogram (total ion monitor) of a Seveso soil sample is shown in Figure 8. Peak 1 of the chromatogram was identified as a trichlorodibenzo-p-dioxin (tri-CDD), co-eluting with another, unidentified compound. The mass spectrometric identification of tri-CDD was based on the presence of its molecular ion and on the fragmentation pattern (M+ = 286, Cla; M+ - C1= 251; M+ - COCl = 223, Clz). At present, it is not clear whether tri-CDD was formed in the Seveso accident itself or by environmental degradation of TCDD. Peaks 2 and 3 of the chromatogram were both identified as TCDDs, the latter co-chromatographedwith standard 2,3,7,8TCDD. A complete mass spectrum of peak 3 was identical to that of standard 2,3,7,8-TCDD (Figure 9). The mass spectrum of peak 2 was ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
921
2,SXB-TCDD
chromatography with packed columns would not separate some of these components. The chromatograms demonstrate the usefulness of high-resolution gas chromatography with glass capillary columns to the trace analysis of 2,3,7,8-TCDD in environmental samples.
4 mln
The author thanks H.-P. Bosshardt for helpful discussions and for comments on the manuscript. Two of the reference compounds were obtained from C. Rappe, University of Umea, Sweden. Environmental samples were received from Italian authorities and Givaudan AG, Dubendorf, Switzerland. Part of the clean-up procedure was suggested by A. Cavallaro and co-workers, Institute of Public Health, Milan.
soil, Switzerland /20ppt
ACKNOWLEDGMENT
6 min
4
2
Figure 10. Mass fragmentograms ( m / e 320) of grass and soil samples,
Switzerland, in absence and presence of added 2,3,7,8-TCDD; equivalents of 4-5 g of samples injected; 25-m OV-17 glass capillary column at 220 O C ; sensitivity, 0 . 2 4 f.s. weak and had interferences in the low mass range that did not allow a determination of the type of chlorine substitution. This isomer did not co-chromatograph with any of the pyrolysis or reference TCDDs available (see Experimental and Figure 3). Mass fragmentograms ( m / e 320) of noncontaminated samples are shown in Figure 10. Equivalents of 4-5 g of sample were injected and analyzed on the OV-17 glass capillary column at highest sensitivity. Partial chromatograms of fortified samples (2,3,7,8-TCDD added at the 20-ppt level before extraction) are included for comparison; they show the expected increase in intensity of the 2,3,7,8-TCDDpeak. The maximum level of 2,3,7,8-TCDD possibly present in these noncontaminated samples is estimated to be less than 3-5 ppt. Interfering peaks were observed, some eluting within seconds of the expected elution of 2,3,7,8-TCDD. Conventional gas
LITERATURE CITED (1) B. A. Schwetz, J. M. Norris, G. L. Sparschu, V. K. Rowe, P. J. Gehring, J. L. Emerson, and C. G. Gerbig, Environ. Heailh Perspect., 5, 87 (1973). (2) G. L. Sparschu, F. L. Qunn, and V. K. Rowe, Food Cosmet. Toxicol., 9, 405 (1971). (3) J. P. Seiler, Experientia, 29, 622 (1973). (4) Q . Firestone, J. Ress, N. L. Brown, R. P. Barron, and J. N. Damico, J . Assoc. Off. Anal. Chem., 55, 85 (1972). (5) E. A. Woolson, R. F. Thomas, and P. D.J. Ensor, J. Agrlc. Fow'Chem., 20, 351 (1972). (6)J. W. Edmonds, D.F. Lee, and C. M. L. Nickel, Pestic. Sci., 4, 101 (1973). (7) W. B Crummett and H. R. Stehl, Environ. Health Perspect., 5 , 15 (1973). (8) R. Baughman and M. Meselson, Environ. HeaMPerspect., 5, 27 (1973). (9) H. R. Buser and H.-P. Bosshardt, J. Chromatogr., 90,71 (1974). (10) H. R. Buser, J. Chromatogr., 114, 95 (1975). (11) H. R. Buser, Anal. Chem., 48, 1553 (1976). (12) H. R. Buser, J. Chromatogr., 107, 295 (1975). (13) M. L. Porter and J. A. Burke, J . Assoc. Off. Anal. Chem., 54, 1426 (1971). (14) K. Grob and K. Grob, Jr., J. Chromatogr., 94, 53 (1974). (15) A. P. Gray, S. P. Cepa, and J. S.Cantrell, Tetrahedron Lett., 33, 2873 (1975). (16) N. P. Ruu-Hoi and G. Saint-Ruf, J. Heterocycl. Chem., Q, 691 (1971).
RECEIVED for review November 1,1976. Accepted March 14, 1977. Presented in part at the Workshop on TCDD, Milan, Italy, October 23-24, 1976.
Field Desorption and Electron Impact Mass Spectra of Ionic Dyes C. N. McEwen," S. F. Layton,' and S. K. Taylor2 Central Research and Development Department, E. 19898
I. du Pont de
Nemours and Company, Experimental Station, Wilmington, Delaware
Field desorption mass spectrometry (FDMS) Is shown to be a uniquely useful anaiytlcal method for Mentlfylng submllllgram quantities of dyes extracted from acomplex matrlx. The mabr peaks observed for dyes with slngly charged anlons, An-, and cations, C+, have the formula [Cx+,An,]+, x I0, where Ion Intensity decreases as the cluster-Ion Increases in mass. Electron Impact mass spectrometry (EIMS) sometimes provides complementary fragment Ion Information but Is generally not useful for molecular welght determlnatlon. We flnd, however, that several triarylmethane dyes [Le., crystal violet ( I ) ] and azine type dyes [Le., methylene blue (4)] glve abundant C+ and/or CHf* Ions by EIMS. 'Present address, T e x t i l e Fibers Department, E. I. du P o n t d e N e m o u r s & Co., Old H i c k o r y , T e n n . 37138. Present address, Photo Products Department, E.I. du P o n t d e N e m o u r s & Co., E x p e r i m e n t a l Station, Wilmington, Del. 19898.
922
*
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
Identification of ionic dyes is hindered by the difficulty involved in obtaining useful mass spectral information, The molecular weights of the dye salts cannot be determined by ionization methods which require vaporization of the dyes, except for some cationic dyes (so-called because the chromophore is in the cation portion of the molecule) in which the products of thermal dealkylation or Hofmann degradation can be identified. Identification of the dye pyrolysis products by electron impact or other ionization methods can in some cases be useful for structure elucidation and would be especially valuable if a method were available to identify unambiguously the formula weight of the ion containing the chromophore. The volatility of salts of acid dyes can be enhanced by preparing the free acid, thus increasing the possibility of observing a molecular ion. The free acids of sulfonate-substituted dyes, however, usually remain too nonvolatile to produce molecular ions by conventional ion-
