fluorescence

Capillary gas chromatography/pulsed supersonic jet/fluorescence excitation spectroscopy for the identification of methylanthracenes in a complex envir...
1 downloads 0 Views 769KB Size
Anal. Chem. 1986, 58, 2825-2830

2825

Capillary Gas Chromatography/Pulsed Supersonic Jet/Fluorescence Excitation Spectroscopy for the Identification of Methylanthracenes in a Complex Environmental Sample Barry V. Pepich, James B. Callis,* David H. Burns, and Martin Gouterman Center for Process Analytical Chemistry, Department of Chemistry, BG-IO, University of Washington, Seattle, Washington 98195

David A. Kalman Trace Organics Analysis Center, Department of Environmental Health, SC-34, University of Washington, Seattle, Washington 98195

A new hyphenated Instrument for the detection of fluorescent polynuclear aromatic hydrocarbonsIn complex environmental samples Is presented. The Instrument comblnes caplllary gas chromatography with laser excited fluorescence detectlon using a pulsed supersonic jet Interface. As a test of the system, a complex environmental sample was analyzed for the geometric lsomers of mmthylanthracene. Prior to the lnvestlgatlon of the environmental sample, low-temperature fluorescence excitation spectra of each Isomer are presented and a selective basts for laser excltatlon Is developed. Detectlon llmits for the monomethylanthracenes are reported to be 2-6 ng and the selectlvlty of thls Instrument Is compared to that of capillary gas chromatography/mass spectrometry. Potentlal applications of the Instrument are revlewed.

It is well established that particular families of polynuclear aromatic hydrocarbons (PAH) exhibit strong structurerelated dependencies in their relative levels of carcinogenicity ( I ) . For example, the 6-, 7-, and 8-methyl derivatives of benzo[a]pyrene (BAP) vary from highly carcinogenic to noncarcinogenic in the order: 6mBAP >> 7mBAP >> 8mBAP = 0 (2). Thus, proper assessment of biological dangers associated with any particular sample suspected of containing PAH requires (i) identification of the sample constituents and (ii) the determination of absolute abundances for each particular isomer. Despite recent advances in analytical chemistry, the separation and unambiguous determination of PAH isomers a t trace levels in complex sample matrices remains a challenging problem. The difficulty in throughly analyzing naturally occurring samples known to contain PAH by capillary gas chromatography has recently been assessed by Herman et al. (3). These workers concluded that several hundred million theoretical plates would be required to resolve a preextracted sample of crude oil with any degree of confidence. This enormous gap between current and required chromatographic reseolution renders even very powerful “hyphenated techniques” such as capillary gas chromatography/mass spectrometry (CGC/MS) inadequate in the analysis of complex mixtures of geometric isomers without the aid of mathematical techniques like numerical deconvolution (4). In theory, fluorescence spectroscopy provides the basis for separation of complex mixtures of PAH (5). Unfortunately, spectral features at room temperature are very broad due to vibrational sequence congestion and strong solvent interactions which impose severe limitations on this method. However, a number of low-temperature fluorescence techniques have recently been successfully applied to the analysis of PAH (6-24). To facilitate discussion of these methods, we have

separated them into two general categories based on differences in their low-temperature environment. The first category of low temperature fluorescence spectroscopy contains all methods in which the analyte of interest is embedded in a frozen matrix. Included in this category are Shpol’skii spectrometry, matrix isolation spectrometry, and fluorescence line narrowing spectrometry. Shpol’skii spectrometry was the first method in this category to emerge. Here the low-temperature environment is generally achieved by freezing the sample into a n-alkane polycrystalline matrix at temperatures below 77 K (6-11). Through appropriate matching of geometric and dimensional characteristics of the solute-solvent combination, it is possible to reduce the distribution of local microenvironments often referred to as “sites” (11). As a result, one realizes a reduction in the extent of inhomogeneous line broadening in the Shpol’skii medium. When optimized, “quasi-linear” fluorescence spectra can be obtained for PAH that exhibit line widths of less than 10 cm-’ (fwhm) (9). A drawback of this low-temperature technique is its limited dynamic range, e.g., usually less than 3 decades, which is believed to arise from the combination of solute aggregation at high concentrations and fluorescence of the matrix at low concentrations (14). This limitation has led investigators to search for a more suitable low-temperature environment. In matrix isolation (MI) spectrometry the analyte of interest is mixed in a large excess of diluent gas, generally nitrogen or argon, and vapor deposited at cryogenic temperatures onto suitable optical substrates (12-15). The large excess of diluent gas (generally in molar ratios of greater than lo4)ensures that all solute molecules have matrix molecules as their nearest neighbors and minimizes formation of aggregates. As a direct consequence, MI techniques offer superior linear dynamic ranges when compared to Shpol’skii techniques. For example, Stroup et al. (14)have reported a linear range of over 5 decades for various four-ring PAH in a synthetic mixture using MI. However, the extreme mismatch in size between the ideal gas host and PAH guest impose a limitation on this technique through inhomogeneous line broadening (16). A deeper understanding of the origin of inhomogeneous line broadening together with the advent of narrow band laser light sources has led to the development of fluorescence line narrowing spectrometry (FLNS). In this technique a monochromatic laser source is used to selectively excite a portion of the inhomogeneously broadened absorption envelope near the u“ = 0 v’ = 0 electronic transition for the molecule of interest. The fluorescence spectra elicited in this manner are substantially narrower than fluorescence obtained with a much broader, incoherent source. FLNS has been successfully applied to both inert gas matrices (13)and organic glasses (17-19) where despite the broad absorption spectra, acceptable line

-

0003-2700/86/0358-2825$01.50/00 1986 American Chemical Society

2826

ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986

narrowing has been observed. Although these solid-phase low-temperature spectroscopic methods properly address the problem of spectral specificity and signal to noise ratio (S/N), they do suffer two major drawbacks: (i) they require a lengthy cool-down period and complex cryodeposition equipment, which excludes the possibility of real time analysis of chromatographic effluent; and (ii) interpretation of results can be complicated due to residual inhomogeneous line broadening. Supersonic jet spectroscopy is a convenient way in which to surpass these limitations. This technique comprises the second category of low-temperature fluorescence methods and is the subject of this paper. In a free jet expansion, cooling is accomplished in the gas phase. Here, the analyte is seeded into a monatomic ideal gas and expanded from high pressure through a small orifice into a vacuum chamber that is maintained a t low pressure. Cooling is accomplished through energy transfer from the solute to the ideal gas by molecular collisions as the molecules travel away from the restrictive aperture. As the expansion proceeds toward the region of collisionless flow, the molecular degrees of freedom become frozen in the order T Y i b > Trot> Ttram.

Under carefully controlled conditions, the analyte molecules are present as isolated ultracold gas-phase species. This particular technique has been of great interest to spectroscopists and physical chemists and has inspired many elegant basic studies over the last decade. Recently, analytical applications of supersonic jet spectroscopy have begun to appear (20-29). Due to essentially instantaneous formation of the cooled species together with their rapid flushing away, one might expect free jet expansions to provide a useful tool for a number of laser-based high-resolution hyphenated methods. Hayes and Small were first to investigate the separation power that could be realized through the direct union of gas chromatography and supersonic jet spectroscopy (22). Their pioneering system employed a short (about 1 ft) packed GC column, directly interfaced to a continuous free jet expansion nozzle. With this instrument, they were able to separate isomers of naphthalene, a-methylnaphthalene, and &methylnaphthalene that were chromatographically unresolved. Realizing that significant chromatographic improvements could be made, we have constructed a “second generation” capillary gas chromatograph/pulsed supersonic jet/fluorescence excitation spectrometer (CGC/PSJ/FES) (25). In our previous communication we set forth a theoretical basis for a novel pulsed free jet system which had sufficiently low dead volume and sufficiently high operating temperature to operate with a capillary GC without loss of chromatographic resolution. Experimental study of the prototype system revealed: (i) its ability to cool GC effluent to rotational temperatures below 10 K; (ii) its compatibility with capillary gas chromatography; and (iii) its ability to yield a 7-fold concentration of GC effluent during the pulse cycle. In this paper we demonstrate the analytical utility of our instrument by quantitation of the geometric isomers of monomethylanthracene (MMA) in a very complex environmental sample.

EXPERIMENTAL SECTION Capillary Gas Chromatograph/Pulsed Supersonic J e t Spectrometer. This portion of the instrument has been previously described (25);it consists of a capillary GC, a pulsed antechamber, and a vacuum chamber. The capillary GC feeds the antechamber through a heated transfer line. The antechamber is maintained at 220 “C to avoid condensation of the GC effluent and is backed with an NRC pulsed molecular beam valve. The pulsed valve operates into the antechamber through a large aperture (760 pm) at 5 Hz and acts to concentrate GC effluent and attain optimal cooling as both gas streams mix turbulently and exit the antechamber through its restrictive orifice (416 pm) and proceed into the vacuum chamber. Thorough mixing of the

backing gas and GC effluent before expansion is ensured by inserting a small plug of quartz wool in the antechamber. The vacuum chamber is kept at low pressures with a conventional diffusion pump (10 cm in diameter). The free jet expansion is crossed 9-14 mm downstream from the antechamber exit with a laser excitation source that enters and exits the chamber through baffled ports. Fluorescence is collected at right angles to the laser excitation with a f / 1 . 5 Pyrex lens and spatially imaged onto a photomultiplier tube (PMT) to facilitate the rejection of warm fluorescence. A glass cutoff filter placed in front of the detector facilitates the rejection of any scattered laser light. Signal from the PMT is gated and digitally normalized for laser intensity by comparison to a monitor diode signal. The entire experiment, e.g., firing of the pulsed valve, laser and signal gating, is synchronized in the time domain with a minicomputer (Terak, Model 8510). Laser Excitation. The Chromatix CMX-4 flashlamp pumped dye laser described in our previous work (25) was replaced by a nitrogen-pumped dye laser (National Research Group, Inc., Model NRG-0.7-5-200) excitation source. The dye laser was designed in our laboratory (30) and uses a Littrow mounted diffraction grating (2400 lines/mm) as the dispersive element. Scanning of the diffractiongrating is achieved with a galvanometer (American Time Products). The galvanometer and data collection are controlled by a second computer (DEC, PDP-11/04) and software developed in OUI laboratory for this application. In order to reduce laser line width, we have employed a very simple double prism beam expander (31,32)which reduces the output line width to approximately 1A (fwhm). The laser dye employed in this study (BPBD-365, Exciton) was continuously tunable from 361 to 391 nm. Spectroscopic Studies. The various parameters affecting the overall temporal behavior of the antechamber and efficiency of cooling were previously studied for iodine expansions (25). These parameters, which included delay time, downstream distance, backing pressure, and choice of backing gas, were reoptimized for the family of monomethylanthracenes and were found to be very similar to the conditions used in our previous study. Here, 9-methylanthracene (Alfa Products, 95%) was seeded into a He carrier gas at 150 OC and allowed to enter the antechamber through the GC transfer line at 2 mL/min. By collecting fluorescence excitation spectra as a function of time delay relative to the firing of the pulsed valve, we were able to determine optimum delay times for laser excitation. Fluorescence excitation spectra were then collected for 1-methylanthracene(Alfa Products, 99%) and 2-methylanthracene (Aldrich,97%) in a similar manner. Comparison of the excitation spectra allowed us to choose an appropriate selective wavelength for each isomer that was later used in the chromatographic studies. Chromatographic Studies. The environmental sample that was the object of this study was obtained from Everett, WA, on the 29th of September, 1984 during the “Everett tire fire”. The sample was collected over a 24-h period during the fire in the form of airborne particulates from a volume of air equal to 516 m3. The collected particulates (118.5 mg) were continuously extracted with benzene and dichloromethane, and the extract was separated into three fractions by size exclusion chromatography. Fraction 2 containing the highest content of low-molecular weight PAH, was selected for analysis by conventional capillary GC/MS and CGC/PSJ/FES. Stock solutions were prepared from the monomethylanthracene standards by dissolving an appropriate amount of each standard in toluene (Aldrich, Gold Label). These stock solutions were used to optimize the gas chromatographic conditions and evaluate the analyte concentrations in the environmental sample. All chromatography was performed on a new 30-m 0.25 mm i.d. capillary column (DB-5, J&W Scientific). For the sake of comparison, identical chromatographic experiments were performed both on our CGC/PSJ/FES system and on a Hewlett-Packard CGC/MS system (Model 5985) employing the same capillary column. RESULTS AND DISCUSSION A. Spectroscopic Studies. In our previous work (25) we set forth a gas dynamic model that explained the pressure behavior of the pulsed antechamber; Figure 1 depicts this behavior. Basically, the antechamber starts out at an initial

ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986

2827

1 0

I

Flgure 1. Tlme dependence of the antechamber molecular number denslty. Trace (A) shows the opening and closing of the pulsed valve and (6) gives the expected number denslty in the antechamber. I5

1 I

I

^

,

I

I

I

I

I

I

I

1

I

I

I

lA

362

364

366 360 370 372 WAVELENGTH (nm)

374

Flgure 3. Comparison of the fluorescence excitation spectra for 9-methylanthracene (trace A), 1-methylanthracene (trace B), and 2methylanthracene (trace C) at a 2500-ps delay. All experimental conditions were Identical with those reported In Figure 2.

0

362

1

I

l L

51

CH3

364

366

368

370

372

374

W A V E L E N G T H (nrn)

Flgure 2. A portlon of the fluorescence excitation spectra for 9methylanthracene (9MA) as a function of delay tlme relatlve to the firing of the pulsed valve wlth (trace A) 0 ps delay, (trace 6 ) 1500 ps delay, and (trace C) 2500 ps delay. I n each case 9MA was seeded into He at a reservoir temperature of 150 OC and allowed to enter the antechamber at 2 mL/mln through a heated capillary transfer line that was malntained at 200 OC. The antechamber was held at 210 OC and had a 416-wm exit aperture. The antechamber was backed at 5 Hz with Ar supplied from the NRC valve which had an open time of 600 ps and reservoir pressue of 1.65 atm. Fluorescence was excited 11 mm downstream from the antechamber exit.

or “steady state” pressure that is determined by the GC flow rate alone. Some time after the firing of the pulsed valve the antechamber exponentially reaches a maximum value in pressure. Once the pulsed valve is closed, the antechamber exponentially returns to its initial steady-state level. This exponential pressure behavior was established to be dependent on the antechamber volume, exit aperture area, and the choice of backing gas, as predicted by the gas dynamic model. Since the efficiency of vibrational and rotational cooling in the expansion process is related to the backing pressure, assuming adequate pumping speed, one would predict most efficient cooling to be achieved coincident with the pressure maxima of the antechamber; this was also demonstrated (25). In Figure 2 we substantiate the remarkable cooling of a PAH molecule as the antechamber reaches its optimum pressure. In each case 9-methylanthracene (9MA) was seeded

into He at elevated temperature and allowed to enter the antechamber at a constant flow rate and Ar was employed as the backing gas. Fluorescence excitation spedra were recorded as a function of laser delay relative to the firing of the pulsed valve with delay times of 0 ps, 1500 ps, and 2500 ps, respectively. At 0 ps (Figure 2A) the spectrum for 9MA is comparatively broad and displays a very poor S / N ratio. It is evident that very little cooling has been accomplished. This is expected because the flow into the antechamber at 0 ps originates from the GC only. Previously, the GC flow alone was shown to be a very inefficient high-pressure source for free jet expansions by nature of its intrinsically low flow rate (25). However, as the delay time is increased, efficient cooling is accomplished as backing gas is added from the NRC valve and a significant gain in instrumental S / N is noted. At 2500 ps (Figure 2C) optimal cooling has been achieved for the Ar expansions and no further increase in spectral resolution or S / N is achieved by proceeding to longer time delays. These findings for the time dependence of cooling for 9MA are in agreement with our previous results for iodine (25). The He-backed expansions proved to be nearly identical in nature with those reported in Figure 2, however, the most efficient cooling is achieved earlier in the expansion process, e.g., 1700 ps. This is also in accord with our previous work. The conditions used in Figure 2 were found to be optimum for the entire family of monomethylanthracenes (MMA) and were employed in all subsequent experiments described in this paper.

2828

ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986

6

8

IO 12 TIME (MINUTES)

14

Flgwe 4. Total ion chromatogram collected for the "Everett tire fire" sample on our HP Model 5885 GC/MS system. The chromatographic conditions were as follows: He carrier gas velocity, 32 cm/min; a temperature program of 80 'C for 1 min and then to 240 'C at 16 OC/min. The capillary column employed in these studies was a 30-m DB-5, J&W Scientific.

Figure 3 show the fluorescence excitation spectra of 1methylanthracene (1MA) and 2-methylanthracene (2MA), together with 9MA shown previously. Each trace was recorded under identical experimental conditions. Comparison of these traces allows a full appreciation of the selectivity attained through fluorescence excitation of these compounds. From these data we determined optimum excitation wavelengths for each isomer (ASMA, XIMA, A P M ) that were later used in the CGC/PSJ/FES studies. B. GC/MS Studies. In order to demonstrate the separation power of the CGC/PSJ/FES and to aid in the initial characterization of the environmental sample, we first conducted a comparative study employing CGC/MS. Figure 4 is a plot of the total ion current for all masses as a function of chromatographic elution time. This figure displays the very complex nature of the sample that was collected during the Everett tire fire. When one considers that this sample has already been subjected to a standard cleanup procedure, a better appreciation is developed for the prerequisite of several hundred million theoretical plates to achieve total separation as proposed by Herman et al. (3). The mass spectroscopist, however, has additional separation power available through computer-based generation of selected ion chromatograms. Since PAH generally tend to exhibit very strong molecular ions, a search of the recorded scans for the molecular ion of interest is often very informative. Figure 5B displays the selected ion trace at m / e 192 as compared to the total ion chromatogram in Figure 5A. While both chromatograms exhibit partially resolved peaks in the expected retention window (10-10.5 min) whose mass spectra generally yielded strong intensities a t m / e 192, they also contained many fragments not characteristic of the family of MMA including strong intensities at higher mass numbers. These were assumed to arise from high molecular weight aliphatics coeluting in this time window. In addition, we could not distinguish between MMA and methylphenanthrenes (MMP), which are known to elute in this region and have virtually identical mass spectra. In order to further elucidate the nature of the components eluting in this retention time window, we injected a standard mixture of the three monomethylanthracenes under the same chromatographic conditions as shown in Figure 5C. The elution order was established to be 2MA, lMA, and 9MA, respectively by injecting standard solutions for each component of the three-component mixture. From the total ion chromatogram of the standard mixture in Figure 5C, we note that base line resolution has been achieved for 9MA, while 2MA and 1MA are only partially resolved. In addition, we note that none of the retention times of the peaks exhibited by the sample in the time window of 10-10.5 min match the

TIME (MINUTES)

Figure 5. GUMS comparison of the "Everett tire fire" sample and a

stock solution containing the three monomethylanthracenes. I n traces A and B the total ion and selected ion at m / e 192 for the unknown are shown, respectively. Trace C displays the total ion chromatogram for the standard stock solution.

standards. The lack of agreement in retention times between the standard and unknown solutions in conjunction with the mass spectral data discussed above illustrates the diffulty in quantitation of the MMA's without either higher GC resolution or further cleanup procedures. Thus, the complexity of this sample renders it a suitable candidate to test the selectivity of our CGC/PSJ/FES system. I t should be noted that the conditions employed in the GC/MS studies were chosen to allow a reasonable analysis time and are certainly not optimal for this instrument. In fact if one chooses a much slower temperature program, e.g., 3 'C/min, it is possible to achieve base line separation for a mixture of all three MMA isomers if one is careful not to overload the column. However, further complications in a real-world sample due to the possible coelution of five monomethylphenanthrene isomers all with very similar mass spectra render a successful separation and verification exceedingly difficult. This is true even under ideal conditions, where the additional burden of coeluting aliphatics has been reduced by further cleanup steps. CGC/PS J/FES Studies. Earlier, we established a basis for selective excitation and detection of each MMA isomer. The complex nature of this sample should therefore pose less of a problem when the PSJ/FES is used as a selective chromatographic detector. First, the family of MMP, which severely complicated the selected ion trace in Figure 5B, should be easily separated spectroscopically from the MMA since their electronic origins are shifted nearly 30 nm to the blue. Second, any aliphatic species would not be expected to fluoresce. The selectivity of our instrument was initially investigated for a standard mixture containing all three isomers of interest at nearly equal concentrations. The selective excitation chromatograms are shown in Figure 6. In each case the laser was tuned to the excitation wavelength that was previously determined to be most selective for the particular isomer, and a 1.O-fiLaliquot of the stock solution was injected containing a mixture of 144 ng of 2MA, 137 ng of lMA, and 128 ng of 9MA. From the traces in Figure 6 we note that sensitivity for the monomethylanthracenes decreases in the order of 9MA > 1MA > 2MA. This is in agreement with the relative intensities as seen in Figure 3. It should be noted that not all of the traces in Figure 6 are completely free of interferences. Of the isomers investigated in this study only 9MA exhibits no interference from the other

ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986

2829

W

m z 0

a

v)

:o 500 600 700 TIME (SECONDS)

400

Il 11

t 0

400

500

600

Figure 7. Selected fluorescence excitation chromatograms for the unknown solution (trace A) optimized for 2MA, (trace B) optimized for lMA, and (trace C) optimized for 9MA. All Instrumental corditions were identical with those reported prevlously.

Table 11. Experimental Results for “Everett Tire Fire Sample”

700

TIME (SECONDS)

Figure 6. Selected fluorescence excitation chromatograms of the standard mixture (trace A) optimized for 2MA, (trace B) optimized for lMA, and (trace C) optimiued for 9MA. In each case a mixture of 144

ng of 2MA, 137 ng of lMA, and 128 ng of 9MA was injected. The chromatographicconditions were identical wlth those reported in Figure 4. All condtons related to the antechamber were identical with those reported in Figure 2 except the antechamber and transfer line temperatures which were elevated to 220 and 240 “C, respectively. Table I. Detection Limits for MMA

compounda 2MA 1MA

9MA

excitation wavelengths, A,, nm 365.05 363.95 371.19

detection limits,bng 6.2 2.7 2.2

a 2MA = 2-methylanthracene, 1MA = 1-methylanthracene,and 9MA = 9-methylanthracene. bDetectionlimits were defined as the quantity of each isomer which yielded a signal 3 times larger than the standard deviation of the noise judged by peak height for standard iniections (33).

MMA. This is attributed to its red shift in electronic origin relative to the other isomers. The selective wavelength of exhibits a small response for the excitation for 2MA (A& 9MA isomer (20.2 f 1.5%) relative to the response of 2MA a t this wavelength. However, since these two isomers are chromatographically resolved, this poses no problem. The possibility of an overlapping interference from 1MA at X z m was investigated by injecting 300 ng of 1MA. No response for 1MA was observed and therefore we concluded there was no interference from 1MA a t this wavelength. We evaluated the instrumental selectivity for 1MA in a similar manner and found that 2MA displayed a response of 8.91 f 0.06% at A 1 w Since 1MA and 2MA were not completely separated chromatographically, this necessitated a small correction in the evaluation of the relative levels in the unknown sample. Prior to the analysis of the unknown sample the instrumental response was determined to be linearly dependent on concentration over a range of 400-25 ng in the case of 9MA. Standard injections of each MMA isomer were then made to determine their detection limits, which are listed in Table I with the selected excitation wavelength for each molecule. Separate experiments indicated that the detection limits were determined by the number of fluorescent photons arriving a t the detector.

excitation extract concn,

compounda ng/1.6 2MA 1MA 9MA

pL

78.5* 1.6 82.5 f 1.4 16.3 0.4

*

wave-

air concn, particulate ng/mg loading, ppm 95.1 100

19.8

414 435 86.1

length, nm

365.05 363.95

371.19

2MA = 2-methylanthracene, 1MA = 1-methylanthracene,and 9MA = 9-methylanthracene

Figure 7 displays the selected wavelength chromatograms for the environmental sample (“Everett tire fire”) collected as described above. As expected the chromatograms display no response to the coeluting aliphatic molecules or to the monomethylphenanthrenes. A reduction in chromatographic resolution is noted in the case of the unknown. This is attributed to an increase in sample volume and column loading. Because of differences in resolution between the standard and unknown, chromatographic peak areas were used to calculate the levels of MMA. Here, replicate injections of the unknown were encompassed by replicate injections of the standard solution to exclude the possibility of laser drift. In Table I1 we have listed the experimental results obtained for the “Everett tire fire” sample. From these data we conclude that the sample does indeed contain all three isomers of MMA. At present the factors governing the differing abundances of the three isomers are not understood; however, it is clear that the CGC/PSJ/FES has allowed unambiguous determination of relative and absolute abundances of each isomer. The present work illustrates the promise of the CGC/ PSJ/FES as an analytical tool. The union of a high-resolution separation technique and a high-resolution laser-based spectroscopic technique has resulted in an instrument capable of separating and identifying structural isomers of PAH in complex environmental samples. Our instrument exhibits significantly smaller spectral interferences for PAH compared to the CGC/MS and is therefore of value for these types of applications. We are of course fully aware that with more extensive sample cleanup and much longer chromatographic runs, one can resolve many components in complex samples. However, with increased spectral resolution less GC resolution is required and analysis time can be reduced. Furthermore, this method should prove applicable to any family of geometric isomers that can be excited with a conventional laser source and possess reasonable fluorescent quantum yields.

2830

ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986

When compared to the initial GC continuous jet expansion system demonstrated by Hayes and Small (22), we see that our second generation instrument attains detection limits a full order of magnitude better, together with a chromatographic resolution 2 orders of magnitude greater. At the same time, further improvements in sensitivity and selectivity are clearly possible. One major limitation is the relatively weak, broad band dye laser system used. Amirav et al. (23) have previously published a similar spectrum for 9MA that displays line widths 5 times narrower than we obtained. Thus a great deal of our broad, already weak excitation energy is being wasted. Furthermore, a state of the art Nd:YAG pumped dye laser with an amplifier cell and proper pumping geometry can attain substantial improvements in pulse energies. Finally, the excitation beam can be multipassed, and considerable improvement in collection efficiency can be achieved with an intrachamber spherical reflector (34). With these improvements, the sensitivity and selectivity of the CGC/PSJ/FES will approach and perhaps begin to surpass that of selective ion monitoring in GC/MS. Further increases in selectivity will be obtained by monitoring a specific wavelength of fluorescence emission. A new type of emission analyzer with high throughput is under development for this purpose (35). However, an increase in selectivity of this nature will be attained only a t some cost in sensitivity. All of these advantages will still not remove certain disadvantages of fluorescence. First, not all molecules fluoresce; for these species, one could use the CGC/PSJ device in conjunction with laser photoionization spectroscopy. Lubman and co-workers have demonstrated the analytical power of this detection scheme (21,27,29).Alternatively a CGC/PSJ would comprise an ideal match to a Fourier transform/mass spectrometer (FT/MS). Current interfaces between CGC and F T / M S suffer large duty factor losses (36). The second drawback to fluorescence is the difficulty in deriving an unambiguous structure from a spectrum, especially compared to mass spectrometry or infrared spectroscopy. However, fluorescence spectra are totally unique and library searches would be feasible once a suitable data base was established. Registry No. lMA, 610-48-0;9MA, 779-02-2;2MA, 613-12-7.

LITERATURE CITED (1) Hoffman, Dietrich; Lavoie, Edmond J.; Hecht, Stephen S. I n pdynuclear Aromatlc Hydrocarbons : Physlcal and Blological Chemistry ; Cooke, Marcus, Dennis, Anthony J., Fisher, Gerald L., Eds.; SpringerVerlag: New York, 1982; p 1.

Lee, Milton L.; Novotny, Mllos V.; Bartle, Keith D. I n Analflical ChemIstv of Po&cycllc Aromatic Hydrocarbons; Academic Press: New York, 1981; p 440. Herman, David P.; Gonnord, Maire-Grance; Guiochon, Georges Anal. Chem. 1984, 56,995. Cailis, James B. I n Ultra Hlgh Resolution Chromatography: Ahuja, S., Ed.; American Chemical Society: Washington, DC, 1984; ACS Symp. Ser. No. 250, pp 171-198. Ho, Chu-Ngi; Christian, Gary D.; Davidson, Ernest R. Anal. Chem. 1981, 50, 1108. Kirkbright, Gordon F.; De Lima, C. G. Anawst (London) 1974, 9 9 , 338. D'Silva, Arthur P.; Ostreich, G. J.; Fassel, Velmer A. Anal. Chem. 1978, 46, 915. Colmsjo, Anders; Stenberg, Ulf Anal. Chem. 1979, 5 1 , 145. D'Silva, Arthur P.; Fassel, Velmer A. Anal. Chem. 1984, 56, 985A. Yang, Yen; D'Silva, Arthur P.; Fassel. Veimer A.; Iles, Malvern Anal. Chem. 1980, 5 2 , 1350. Lamotte, M.; Joussot-Dubien, J. J. Chem. Phys. 1974, 6 1 , 1892. Wehry, Earl L.; Mamantov, Gelb Anal. Chem. 1979, 57.643A. Maple, Jon R.; Wehry, Earl L. Anal. Chem. 1981, 5 3 , 266. Stroupe, Robert C.; Tokousbalides, P.; Dickerson, Rochard B., Jr.; Wehry, Earl L.; Mamantov, Gelb Anal. Chem. 1977. 4 9 , 701. Dlckenson, Richard 6.; Wehry, Earl L. Anal. Chem. 1979, 51, 778. Maple, Jon R.; Wehry, Earl L.; Mamantov, Gelb Anal. Chem. 1980, 5 2 , 920. Chiang, Iris; Hayes, John M.; Small, Gerald J. Anal. Chem. 1982, 5 4 , 315. Brown, Johnie C.; Edelson, M. C.; Small, Gerald J. Anal. Chem, 1978, 5 0 , 1394. Brown, Johnie C.; Ducanson, John A., Jr.; Small, Gerald J. Anal. Chem. 1980, 5 2 , 1711. Warren, Jonathan A.; Haves. John M.; Small, Gerald J. Anal. Chem. 1982, 54, 138. Lubman, David M.; Kronik, Me1 N. Anal. Chem. 1982, 5 4 , 660. Hayes, John M.; Small, Gerald J. Anal. Chem. 1982, 54, 1204. Amirav. Ariv; Even, Uzl; Jortner, Joshua Anal. Chem. 1982, 5 4 , 1866. Yamada, S.;Smith, B. W.; Voigtman, E.; Winefordner, J. D. Appl. SDeCtrOSC. 1985. 3 9 . 513. Pepich, Barry V.; Caliis, James 6.; Danlelson, J. D. Sheldon; Gouterman, Martin Rev. Scl. Instrum. 1988, 5 7 , 878. Imaska, Totaro; Hirata, Kaoru; Ishibashl, Nobuhiko Anal. Chem. 1985, 5 7 , 89. Sin, Chung Hang; Tembreull, Roger; Lubman, David M. Anal. Chem. 1984, 5 6 , 2776. Hayes, John M.: Small, Gerald J. Anal. Chem. 1983, 55, 565A. Tembreull, Roger; Lubman, David M. Anal. Chem. 1984, 56, 1962. Skoropinski, D. Bruce; Callis, James 6.; Danielson, J. D. Sheldon; Christian, Gary D. Anal. Chem. 1986, 5 6 , 2831. Racz, B.; Bor, 2s.;Szatmari, S.: Srabo, G. Optics Commun. 1981, 3 6 , 399. Brosnan, Stephen J.; Byer, Robert L. I€€€ J. Quantum Nectron. 1979, 15, 415. ACS Committee on Environmental Improvement and Subcommittee on Environmental Analytical Chemistry Anal. Chem. 1980, 5 2 , 2242. Pan, C. L.; Prodan, J. V.; Fairbank, W. M., Jr.; She, C. Y. Opt. Lett. 1980, 5, 459. Caliis, James B., work in progress, 1986. Sack, T. M.; Gross, M. L. Anal. Chem. 1983, 5 5 , 2419.

RECEIVED for review March 11,1986. Accepted July 9, 1986. This research was supported in part by NIH Grant GM 22311 and NSF Grant IS1 8415075.