Anal. Chem. 1997, 69, 3476-3481
Critical Examination of the Quantification of Aromatic Compounds in Three Standard Reference Materials Bernhard Schmid† and Jan T. Andersson*,‡
Department of Analytical and Environmental Chemistry, University of Ulm, D-89069 Ulm, Germany
Three standard reference materials from the National Institute of Standards and Technology (NIST), namely, SRM 1597 coal tar, 1582 crude oil, and 1580 shale oil, were investigated in detail to determine the concentration of polycyclic aromatic sulfur heterocycles using gas chromatography with the atomic emission detector in the carbon- and sulfur-selective modes. Coelution problems were found to be common, and the use of two capillary gas chromatographic columns with stationary phases of widely differing polarity was necessary for the separation of the important analytes phenanthrene, anthracene, dibenzothiophene, and the naphthothiophenes. On the commonly used nonpolar stationary phases, phenanthrene and dibenzothiophene coelute with the isomeric naphthothiophenes and this leads to too high concentrations being measured for the two major analytes and must be corrected for. For (sulfur and oxygen) heterocycles, individual response factors must be used if the flame ionization detector is employed. The NIST values were obtained without regard to those factors. This is done here for the three SRMs, and it is shown that the adjusted NIST values agree very well with the GC/AED values. It is suggested that the (noncertified) NIST values for several polycyclic aromatic compounds (PACs) should be reexamined. Very probably many other determinations of PACs might suffer from the same shortcomings. Reference materials have received increased importance in the last years with the growing demand for quality control in analytical measurements. By analyzing a well-known material like a commercially available reference material and showing that the results obtained agree with the certified values, a laboratory can demonstrate that its analytical procedure, which may differ considerably from the procedures used in the certification process, leads to accurate data. It is immediately obvious that the concentrations of the analytes in reference materials must be known quite accurately since those materials are used as the basis for quality control of measurements worldwide. Furthermore, reference materials are useful also for the analysis of their noncertified components because the reference material is available for laboratories everywhere and results for the same sample can be compared. † Current address: L. A. B. GmbH & Co., Wegener Str. 13, D-89231 NeuUlm, Germany. ‡ Current address: Department of Analytical Chemistry, University of Mu ¨nster, Wilhelm-Klemm-Strasse 8, D-48149 Mu ¨ nster, Germany. E-mail:
[email protected].
3476 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
Polycyclic aromatic compounds (PACs) are among the most investigated chemical classes in samples of varying types, and it is a logical development that many reference materials have been issued by different agencies with data for PACs. In this work, we will deal with three standard reference materials (SRMs) that have been certified and are marketed by the National Institute of Standards and Technology (NIST) in the United States, namely, SRM 1597 (coal tar), 1582 (crude oil), and 1580 (shale oil). For a review of SRMs, see ref 1. The analysis of PACs has become a routine measurement and is not generally considered particularly difficult, at least not as far as organic trace analysis goes. However, as we will illustrate in this paper, even with well-known analytes for which reference compounds are available, in comparatively favorable concentration ranges, a thorough investigation of the sample and an extensive knowledge of the analytical pitfalls are necessary to obtain accurate results. This is particularly true for samples in which the analytes are part of a very complex mixture of hundreds of other members of the same chemical class, as is the case for PACs. Furthermore, as analytical techniques continue to develop, a reinvestigation using techniques and results that were not available at the time of certification may reveal shortcomings in the certification procedures. For this work, we have drawn heavily on our experience with the gas chromatographic atomic emission detector (AED),2 fluorinated PACs as internal standards,3 and the recently introduced commercial standards for the polycyclic aromatic sulfur heterocycles (PASHs).4 EXPERIMENTAL SECTION The instruments, chemicals, and workup methods have been described previously.5 The quantification of the aromatic compounds in the coal tar was performed using external calibration with the analytes. In the crude and the shale oil, the quantification was done with 2030 mg of the oil, using 5-fluorobenzothiophene and 2-fluorodibenzothiophene as internal standards for the two- and the three-ring PASHs. Two- and three-ring PAHs were quantified through comparison with the internal standards 2-fluoronaphthalene and 3-fluorophenanthrene.5 The standards were added to the sample (1) Wise, S. A. In Environmental Analysis: Techniques, Applications and Quality Assurance; Barcelo´, D., Ed.; Elsevier Science Publishers: Amsterdam, 1993; pp 403-446. (2) Andersson, J. T.; Schmid, B. Fresenius J. Anal. Chem. 1993, 346, 403409. (3) Andersson, J. T.; Weis, U. J. Chromatogr., A 1994, 659, 151-161. (4) Polycyclic Aromatic Sulfur Heterocycles. Reference Solutions. Astec, Nottulner Landweg 90, D-48161 Mu ¨ nster, Germany. (5) Andersson, J. T.; Schmid, B. J. Chromatogr. A 1995, 693, 325-338. S0003-2700(97)00194-7 CCC: $14.00
© 1997 American Chemical Society
Table 1. Concentrations of Four PACs in SRM 1597 (µg/g) our values NIST value GC/AED GC/FIDa benzothiophene dibenzofuran dibenzothiophene phenanthrene
35.8 106 18.2 451
27.5 88.9 23.0 461b
NIST internal standard acenaphthene acenaphthene 1-methylphenanthrene 1-methylphenanthrene
a From ref 6. b Certified value is 462 ( 3 µg/g and is the combined value of LC fluorescence and GC/FID results.
before any workup was undertaken. All quantitative data are the mean of at least three determinations. Gas Chromatographic Conditions. Columns: SB-Biphenyl30, Lee Scientific, 25 m × 0.32 mm, 0.25 µm film; SP 2331 (100% cyanopropyl), Supelco, 60 m × 0.32 mm, 0.2 µm film. Helium flow rates were set to ∼30 cm/s. Samples were injected in the splitless/split mode with a purge delay time of 30 s. Temperature program: for the biphenyl column, 80 °C/3 min, 4 °C/min to 280 °C, isothermal for 15 min; cyanopropyl column, 90 °C/3 min, 10 °C/min to 160 °C, 2 °C/min to 230 °C, 10 °C/min to 250 °C, isothermal for 10 min. The correction for coelution of naphtho[2,1-b]thiophene with phenanthrene in SRM 1597 was carried out by subtracting the calculated response of the carbon atoms in the heterocycle from the measured response of the carbon atoms in the coelution peak. RESULTS AND DISCUSSION SRM 1597 Coal Tar. The coal tar reference material consists of a toluene solution of a purified coal tar. Since the PAC pattern is of only moderate complexity, it is tempting to try to quantify the aromatics with no other sample preparation steps than a dilution and possibly addition of internal standards. Since no losses should occur in those steps, the sample is ideally suited for checking the chromatographic performance and the integration of the peak areas. While on the whole our quantitative data agreed well with those given by NIST, we experienced unsatisfactory agreement between our data3 and the NIST values6 for three compounds: benzothiophene, dibenzofuran, and dibenzothiophene (see Table 1). The NIST data in Table 1 are those which were obtained using GC/FID and, with the exception of phenanthrene, are not certified since only one technique was used to obtain them. Although the AED value for phenanthrene is in good agreement with the certified value, it is included in the table since it will be discussed later. From the detailed description given of the certification procedure,7 it became obvious that one reason for the discrepancy might be that the NIST data for those compounds are not based on individual response factors whereas our results for this sample are based on an external calibration with the compounds in question or using the demonstrated2 molar response of the AED for carbon. In the literature, the concept of response factor has two meanings. It may be the magnitude of the detector signal for a certain mass of analyte divided by the detector signal for the same (6) Wise, S. A.; Hilpert, L. R.; Rebbert, R. E.; Sander, L. C.; Schantz, M. M.; Chesler, S. N.; May, W. E. Fresenius Z. Anal. Chem. 1988, 332, 573-582. (7) Wise, S. A.; Benner, B. A.; Byrd, G. D.; Chesler, S. N.; Rebbert, R. E.; Schantz, M. M. Anal. Chem. 1988, 60, 887-894.
mass of a standard compound. The other use of the concept is the reciprocal of the response factor just mentioned, which turns out to be the factor with which an analyte signal must be multiplied to get a mass value, again relative to a standard.8 In the NIST work, internal standards which eluted fairly close to the analytes were used with an assumed relative response factor (RRF) of 1. While this is generally acceptable for the flame ionization detector and pure hydrocarbonssand the validity of this assumption was demonstrated in the certification procedure for 16 of the compounds in the sample7sheteroatoms, particularly oxygen, can have a large influence on the RRF9 and a response factor of 1 can no longer be assumed to be valid. The three compounds for which we obtained unsatisfactory agreement with the NIST values (Table 1) are heterocycles. A sulfur atom normally has a neglible influence on the response factor,10 which means that the FID response per carbon atom remains unaffected by the presence of the heteroatom. However, since sulfur contributes considerably to the molar mass of a PASH, a response factor of 1 cannot be used per unit weight, as is often done for PAHs and was the case in the NIST certification for this sample, but the response factor has to be adjusted for the percentage of sulfur atoms in the molecule (these do not contribute to the FID response). This is another way of expressing the idea behind the concept “effective carbon number”.10 Based on naphthalene, benzothiophene had a relative response factor of 0.744 and dibenzothiophene 0.837.10 The relative weight of carbon in those compounds, divided by the relative weight of carbon in naphthalene, is 0.744 and 0.826, respectively, demonstrating that the FID response per carbon in those sulfur heterocycles is the same as in naphthalene; i.e., a molar FID response is obtained. The literature value just quoted agrees well with our experimental value of 0.83 for dibenzothiophene.3 This means that the integrated area of the GC peak for those analytes must be divided by a factor of 0.744 and 0.837, respectively, (or multiplied by the inverted values, 1.34 and 1.19, respectively) in order to obtain a number for their masses when naphthalene is used as internal standard. In dibenzofuran, the presence of the oxygen atom reduces the response factor (response per mass unit) significantly, namely, to 0.865.10 The reason for the lower RRF is (at least partly) different for an oxygen heterocycle than for a sulfur heterocycle. In the PASHs, the only effect of the sulfur is a reduction of the relative weight of the signal-forming carbon atoms in the molecule; in an oxygen heterocycle, this effect is combined with a chemical signal-lowering effect because carbon atoms are already combined with oxygen and thus cannot contribute as efficiently as other carbon atoms to the methane formation process which is required for generating the FID signal.11 When these effects are taken together, for dibenzofuran, this corresponds to a “loss” of the response of 0.65 carbon atom, which seems to be typical for many oxygen-containing compounds. The literature factor10 to multiply the GC areas of dibenzofuran is therefore 1.16. In our previous work with another FID, a value of 1.12 was found for dibenzofuran.3 (8) Musumarra, G.; Pisano, D.; Katritzky, A. R.; Lapucha, A. R.; Luxem, F. J.; Murugan, R.; Siskin, M.; Brons, G. Tetrahedron Comput. Methodol. 1989, 2, 1-17 (9) Tong, H. Y.; Karasek, F. W. Anal. Chem. 1984, 56, 2124-2128. (10) Jorgensen, A. D.; Picel, K. C.; Stamoudis, V. C. Anal. Chem. 1990, 62, 683689. (11) Holm, T.; ¨Ogaard Madsen, J. Anal. Chem. 1996, 68, 3607-3611.
Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
3477
Table 2. Literature Values for the Relative Response Factors (RRFs) for Three Heterocyclic Aromatics (Recalculated for Naphthalene as Standard) RRF benzothiophene dibenzothiophene dibenzofuran
1.34,a 1.27,c 1.27e 1.19,a 1.18,b 1.20d 1.16,a 1.21,c 1.12d
a Reference 10. b Reference 9. c Reference 8. d Reference 3. e Reference 19.
Literature values for the response factors of interest here are listed in Table 2. Despite being measured under quite different conditions (but with splitless injection), they are remarkably similar. We chose the multiplication factors from ref 10 and adjusted the peak areas accordingly. The literature factors used here are all based on naphthalene whereas the NIST quantitative data are based on either acenaphthene or 1-methylphenanthrene as standard. However, although alkylated aromatics seem to possess a somewhat lower response factor than the parent compounds, neglect of this dependence introduces an error of no more than perhaps 2%, which is on the same order as the variations in the RRF for different concentrations ranges or other gas chromatographic conditions10 (note the differences in RRF between the various values in Table 2). Adjusted for the appropriate response factors, the data in the fifth column in Table 3 show that the (adjusted) NIST value of 36.8 µg/g for benzothiophene now agrees well with the AED value of 35.8 µg/g. However, the adjusted NIST value for dibenzothiophene, 27.4 µg/ g, deviates even more from the AED value of 18.2 µg/g than the unadjusted concentration of 23.0 µg/g. For dibenzofuran, the adjusted concentration comes to 103.1 µg/g, whereas we found 106 µg/g, again a much better agreement than with the uncorrected NIST value of 88.9 µg/g.7 The adjusted dibenzothiophene concentration is nearly 50% higher than the AED value; however, the latter was determined on a cyanopropyl column which separates dibenzothiophene from naphtho[1,2-b]thiophene.5 On nonpolar stationary phases such as methyl- or methylphenylsiloxanes (e.g., DB-5 used by NIST in the certification procedure, and the biphenyl phase, used in Figure 1), those two compounds coelute. Likewise, phenanthrene and naphtho[2,1-b]thiophene coleute on nonpolar but not on polar phases, which became obvious when commercial reference compounds4 were made available and were used to determine the retention indexes for those and other PASHs.5 Sulfur heterocycles can easily be determined with the AED in the sulfur-selective trace despite coelution with non-sulfur-containing compounds. The coeluting PAHs can be determined either through subtraction of the calculated contribution of the carbon atoms from the (coeluting) PASH or through the use of another stationary phase on which no coelution takes place. On the cyanopropyl phase that we employed, this coelution is not observed so that our data on this column do not need to be adjusted. However, naphtho[2,1b]thiophene elutes together with anthracene on cyanopropyl phases (but not on nonpolar phases), which means another adjustment would be needed or anthracene is first determined on an nonpolar phase. When the contribution of the two isomeric naphthothiophenes to the peak areas for dibenzothiophene and phenanthrene are subtracted from the NIST data, the concentrations reported in 3478 Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
Figure 1. Sulfur-selective AED trace of the three-ring aromatic fraction of SRM 1597 coal tar. Left trace, on the nonpolar HP-2 column; right trace, on the polar 100% cyanopropyl SP 2331 column. The stars indicate peaks from the methyldibenzothiophenes.
the last column in Table 3 are obtained and now agree very well with the AED values, the difference being only 1.4% for phenanthrene and 2.2% for dibenzothiophene. Considering the approximations done regarding response factors, the use of different internal standards, gas chromatographic columns, detectors, and other conditions, this agreement must be regarded as excellent. The discussion above of the relative response factors for the NIST determinations based on the flame ionization detector shows an advantage of the atomic emission detector, for which a molar response is obtained,2 also for compounds containing heteroatoms. Response factors are therefore not needed. In the NIST certification process, two independent methods are typically used and only if the two concentrations thus determined agree is a certified value issued. For phenanthrene, the second independent method was HPLC with fluorescence detection which indicated a concentration of 463 ( 4 µg/g. This is in convincing agreement with the GC result and still only 1.2% above the “adjusted value” in Table 3. At the moment we do not have data on the HPLC determination of naphthothiophenes with fluorescence detection, so we cannot say whether there is an influence of the PASHs on the liquid chromatographic phenanthrene determination on the polymeric C-18 columns (Vydac 201TP) used at NIST. We found that on a monomeric Nucleosil C-18 column all the four three-ring PASHs eluted well ahead of phenanthrene. SRM 1582 Crude Oil. The crude oil matrix is more complex than the coal tar, mainly because of a large number of alkylated aromatics. Furthermore, it is a natural sample, and unlike the coal tar, a cleanup is necessary before the determination. Of the PACs discussed so far, only phenanthrene has a certified value (101 µg/g), determined using GC with mass-selective detection and liquid chromatography with fluorescence detection.6 We were interested in determining the concentration of several PASHs in this crude oil in connection with our work on PASHs in petroleum.12 In particular we wanted to investigate whether it is possible to determine the monomethyldibenzothiophenes without resorting to a mass-selective detector. The relative distribution (12) Andersson, J. T.; Sielex, K. J. High Resolut. Chromatogr. 1996, 19, 49-53.
Table 3. Concentration (µg/g) of Four Aromatic Compounds in SRM 1597 Coal Tar and Their Correction for Response Factors and Coelution as Discussed in the Text compound
GC/AED (this work)
GC/FID (NIST)
resp factor
(NIST value) × (resp factor)
coelution
NIST corr
phenanthrene benzothiophene dibenzothiophene dibenzofuran
451 35.8 18.2 106
462 27.5 23.0 88.9
1.00 1.34 1.19 1.16
462 36.8 27.4 103.1
5.9 0 8.8 0
456.1 36.8 18.6 103.1
Figure 2. Aromatic fraction of SRM 1582 crude oil on the SB-30 biphenyl column. Top trace with sulfur-selective detection at 181 nm; bottom trace with carbon selective detection at 193 nm. Insets: The same samples after removal of the two-ring aromatics and following class separation of PAHs and PASHs on palladium chloride/silica. The bars in the main chromatograms indicate the retention time region of the insets. The fluorinated compounds are internal standards.
of these isomers has been shown to give information on the degree of the maturity of the oil.13 Figure 2 shows the aromatic fraction of the oil with both the carbon- and the sulfur-selective traces from the AED depicted. Obviously neither phenanthrene nor dibenzothiophene and their alkylated derivatives can be quantified directly, mainly because higher alkylated two-ring compounds obscure the GC peaks in question. Liquid chromatographic separation of the two-ring and the three-ring compounds followed by a further group separation of PAHs and PASHs in the three-ring fraction on palladium (13) Radke, M.; Welte, D. H.; Willsch, H. Org. Geochem. 1986, 10, 51-63.
chloride deposited on silica gel5 gave very clean solutions in which the compounds of interest could easily be quantified. The insets in Figure 2 show a segment of the GC of those cleaned up PAH and PASH fractions which correspond to the elution range indicated by bars in the main chromatograms. The much cleaner chromatograms resulting from this workup are readily appreciated. Note in particular the flat baseline in the insets. We have previously shown that the use of fluorinated aromatics offers great advantages for the quantification of polycyclic aromatic compounds.3 The integration was easily performed here for the three-ring fractions with the internal standards 3-fluorophenan Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
3479
Table 4. Concentrations (µg/g) of Selected Polycyclic Aromatic Sulfur Heterocycles in SRM 1597 Coal Tar, SRM 1582 Crude Oil, and SRM 1580 Shale Oila by GC/AED compound
SRM 1597
SRM 1582
SRM 1580
benzothiophene 7-methylbenzothiophene 2-methylbenzothiophene dibenzothiophene naphtho[1,2-b]thiophene naphtho[2,1-b]thiophene naphtho[2,3-b]thiophene 4-methyldibenzothiophene 2-methyldibenzothiophene 3-methyldibenzothiophene 1-methyldibenzothiophene phenanthro[4,5-bcd]thiophene benzo[b]naphtho[2,1-d]thiophene
35.8 1.3 1.2 18.2 8.8 5.9 3.1 1.1 0.7 0.7 nd 8.6 9.7
ndb nd nd 34.3d nd nd nd 72.4 17.0 19.8 37.2 nd nd
101.4c 76.7c 113.9c 29.2 13.5 12.3 3.4 22.5c 4.6c 11.8c 26.7c nd nd
a Mean of three determinations. b nd, not determined. c Mean of two determinations. d NIST certified value is 33 ( 2 µg/g.6
threne for the PAHs and 2-fluorodibenzothiophene for the PASHs.5 The standards were added at the beginning of the workup so that possible analyte losses are compensated for. A chromatogram on a cyanopropyl phase showed the absence of naphthothiophenes in this sample so that coleution problems for phenanthrene and dibenzothiophene do not complicate the determination as was the case for the coal tar. Our value for phenanthrene was 101.9 µg/ g. The NIST value is 101 µg/g.6 The excellent agreement confirms the validity of the determination. The only PASH that seems to have been determined previously in this sample is dibenzothiophene. A GC determination in the NIST laboratory with the flame photometric detector led to a value of 33.9 µg/g and with GC/MS to 32.9 µg/g,14 in excellent agreement with our value of 34.3 µg/g. The certified value1 is 33 ( 2 µg/g. [Note that in ref 6, Table 8, this value is erroneously reported for SRM 1580 instead of for SRM 1582 (S. A. Wise, personal communication, 5 February 1997).] We determined several other PASHs, and their concentrations are given in Table 4. Benzothiophene and its monomethyl derivatives were not present in detectable quantities. SRM 1580 Shale Oil. The shale oil is another very complex sample, but in contrast to the crude oil, it shows a predominance of substituted two-ring rather than three-ring aromatics. The same cleanup procedure was used as for SRM 1582. In this sample, there were abundant alkylated benzothiophenes so that we could quantify the methylbenzothiophenes only after oxidation to the dioxides and separation first of the oxidized PASHs as a class from the PAHs and then the monomethylbenzothiophene sulfones from higher alkylated benzothiophene sulfones5 using a liquid chromatographic step on a diphenyl phase. Even nonpolar GC columns allow the complete separation of all six methylbenzothiophene sulfones, whereas the six unoxidized methylbenzothiophenes always elute in four peaks.15 Higher alkylated benzothiophenes can in principle be quantified in exactly the same way,12 but a lack of reference compounds prohibits their identification. Substituted naphthalenes cannot be determined after the oxidation because many of them are prone to oxidation and severe losses can occur.5 (14) Rebbert, R. E.; Chesler, S. N.; Guenther, F. R.; Parris, R. M. J. Chromatogr. 1984, 284, 211-217. (15) Andersson, J. T. J. Chromatogr. 1986, 354, 83-98.
3480
Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
Figure 3. Three-ring fractions of SRM 1580 shale oil. Top trace: the PAH fraction (after separation from the PASHs on palladium chloride/silica) on the biphenyl phase; carbon-selective detection. Middle and bottom traces: the PASH fraction (after separation from the PAHs on palladium chloride/silica); sulfur-selective detection. On the biphenyl (middle trace) and the cyanopropyl (bottom trace) column. + indicates methyldibenzothiophenes, * naphthothiophenes. IS ) 3-fluorophenanthrene (top trace) and 2-fluorodibenzothiophene (middle and bottom traces) used as internal standards.
The three-ring fraction was subjected to the class separation on palladium chloride-impregnated silica gel. The resulting PAH and PASH fractions are shown in Figure 3. The PAH fraction (upper panel) is strikingly simple with prominent peaks for phenanthrene and anthracene and their monomethyl derivatives only. In contrast, the PASH fraction (middle panel) is quite complex. The sulfur-selective chromatogram on the biphenyl phase shows the presence of naphtho[2,1-b]- and naphtho[2,3-b]thiophenes. The coelution of dibenzothiophene with naphtho[1,2b]thiophene is revealed when the sample is chromatographed on a cyanopropyl column (bottom panel). The ratio of dibenzothiophene to naphtho[1,2-b]thiophene is ∼2:1; i.e., a large error is made if the peak (on a nonpolar column) is considered to consist of dibenzothiophene only. The quantification of methyldiben-
zothiophenes was difficult because of the presence of many methylnaphthothiophenes, and the 1-methyl isomer could therefore not be reliably determined. The use of a mass-selective detector would not solve the problem since the methylnaphthothiophenes are isomeric with the methyldibenzothiophenes and give practically indistinguishable mass spectra. Phenanthrene cannot be quantified on a nonpolar column with an AED or FID since the coelution with naphtho[2,1-b]thiophene would give too high a result, in this case by ∼5%. Noncertified NIST values6 obtained with GC/MS are available for anthracene (88.4 µg/g) and phenanthrene (232 µg/g). No data are available for the PASHs. The determination of the hydrocarbons should not present much of a problem because the mass-selective detection is not affected by the coelutions with the sulfur three rings. Determination of dibenzothiophene on nonpolar columns would lead to false values, becausesas in the work with the coal tar SRMsdibenzothiophene and naphtho[1,2-b]thiophene coelute. A mass-selective detector would be of no help since the compounds give identical mass spectra. Our determination of the PASHs was carried out on the cyanopropyl column where no coelution is observed and resulted in concentrations of 29.2 and 13.5 µg/g for dibenzothiophene and naphtho[1,2-b]thiophene, respectively. Our phenanthrene value of 254 µg/g is in excellent agreement with the 255 µg/g found by NIST using LC;6 the NIST GC value is 232 µg/g, which is too low despite the coleution with 12 µg/g naphthothiophene on the GC column used. The large difference between the LC and GC values means that there is no certified value for phenanthrene. The values we obtained for several PASHs in SRM 1580 are given in Table 4. CONCLUSIONS The problems of quantifying the common environmental pollutant PAHs without regard to the possible presence of sulfur heterocycles are discussed in this work. When universal detectors are used, the coelution possibility of PAHs with PASHs must always be taken into account. If a problem is anticipated, an estimation can easily be reached if a sulfur-selective detector is used in parallel since it will indicate the presence of PASHs, also on nonpolar columns. A polar column is needed for the separation ofsat least the three-ringsPASHs and PAHs so that all the parent compounds can be determined. The use of a mass-selective detector avoids the problems discussed for the quantification of phenanthrene in the presence of naphtho[2,1-b]thiophene but cannot be used for the individual quantification of dibenzothiophene and naphtho[1,2-b]thiophene. It must be expected that a similar situation pertains to larger ring systems since the number of isomeric PASHs of a given number of aromatic rings increases more rapidly than that of the corresponding PAHs16 and the chance of coelution consequently must increase. (16) Andersson, J. T. Int. J. Environ. Anal. Chem. 1992, 48, 1-15. (17) Imanaka, M.; Kadota, M.; Ogawa, N.; Kumashira, K.; Mori, T. Nippon Eiyo, Shokuryo Gakkaishi 1992, 45, 61-70; Chem. Abstr. 1992, 117, 24858. (18) Jacob, J. Sulfur analogues of polycyclic aromatic hydrocarbons (thiaarenes); Cambridge University Press: Cambridge, UK, 1990. (19) U. Weis, Fluorierte Analoge als interne Standards fu ¨ r die gaschromatographische Bestimmung von polycyclischen aromatischen Verbindungen. Ph.D. Thesis, University of Ulm, Ulm, 1994, p 75.
It should be stressed that this situation is not unique to SRMs but is quite general and therefore also applies to the routine determination of PACs by GC. According to our experience, it is rare to find a laboratory that checks for coelution problems before integrating the chromatographic peaks and calculating a concentration. Although the coelution problem was investigated here only for phenanthrene and dibenzothiophene, it is easily conceivable that it occurs for other PACs also. Obviously the magnitude can vary within wide bounds; for instance, the crude oil SRM did not exhibit any signs of coelution of the sort discussed here. However, coelution with alkylated two-ring compounds is severe (Figure 2), and a determination of phenanthrene with the FID without further cleanup would not be advisable. SRMs were used in this work to investigate the problems of quantifying PAHs in the presence of PASHs because they have been studied extensively and the analytical procedure is stated in great detail. Furthermore, they are available to scientists throughout the world who can check their own procedures with the same material. The results obtained in this work may indicate that a detailed investigation of compounds other than those discussed here would show a similar need for an individual response factor determination and a check for coelution problems before quantification. The data given here should not be understood to be replacements of certified values since the latter are always determined using two independent analytical procedures and therefore have a higher reliability than data obtained from a single procedure. Finally it should be stressed that PASHs are not exotic minor components of environmental and fossil fuel samples. One example is the crude oil from the Kirkuk field12 in which the alkylated three-ring PASHs strongly dominate over correspondingly alkylated PAHs. This seems to be the case for many other crude oils. In an Austrian shale oil that we investigated, the phenanthrene concentration would be overestimated by ∼15% if no correction were made for the coelution with naphtho[2,1-b]thiophene.5 This situation is encountered in environmental samples, too. For instance, in a determination of the ratio of total PASHs to total PAHs in tea leaves, it was found that the ratio varied between 0.08 and 1.08.17 Furthermore, if, for example, toxicological conclusions are drawn from analytical data, it might be wrong to look only for the PAHs. Although much less work has been performed on the sulfur heterocycles than on PAHs, sufficient data are available to indicate that they can have the same mutagenic and carcinogenic potential as PAHs.18 ACKNOWLEDGMENT Dr. Stephen A. Wise graciously supplied the SRMs used here. The Fonds der chemischen Industrie is acknowledged for partial financial support. Received for review February 18, 1997. Accepted June 19, 1997.X AC970194+ X
Abstract published in Advance ACS Abstracts, August 1, 1997.
Analytical Chemistry, Vol. 69, No. 17, September 1, 1997
3481
