Detection of polycyclic aromatic hydrocarbons by ... - ACS Publications

---

1182

Anal. Chem. 1989, 6 1 , 1182-1185

Detection of Polycyclic Aromatic Hydrocarbons by Low-Temperature Phosphorimetry with a Moving Sample Cooling Device B r a d l e y T. Jones, N a n c y J. Mullins, a n d J a m e s D. Winefordner* Department of Chemistry, University of Florida, Gainesville, Florida 32611 A recently described moving sample cooling device has been employed in the determination of several polynuclear aromatic hydrocarbons (PAHs) by low-temperature molecular luminescence spectrometry (I). Several of these PAHs were later identified in cooked beef samples by the same method ( 2 ) . The device employs a thin metal belt connected a t the ends t o form a loop. At one end, the loop is in contact with the cold stage of a closed-cycle helium refrigerator. Samples are injected through a septum in the vacuum chamber onto the cold portion of the belt. After the analysis, the belt carries the sample away from the cold stage where it warms and sublimes. The belt is therefore self-cleaning. This device has several advantages over previously reported moving sample cooling devices. Sampling may be done continuously without the requirement of the individual sample compartments found with similar systems (3-6). T h e arrangement has only two small moving parts, the metal belt and a rotary motion vacuum feedthrough, so elaborate gearing systems or indexing tables are not required for the movement of the entire cryostat as they are with other devices (7,8). Finally, the self-cleaning property of this device is matched by no other previously described research or commercial system. As a result, the refrigerator need not be warmed t o room temperature and samples can be run continuously, requiring 5 min or less for each. The initially described system did have some disadvantages ( I 1. Obtaining reproducible injected samples was difficult. Liquid samples (50 p L ) were injected with a syringe through a rubber septum in the vacuum chamber. Invariably, the injected droplet was a different size and shape each time, certainly causing precision problems. Also, after many injections, the rubber septum became worn and vacuum leaks occurred. In this case, the refrigerator had to be warmed, the vacuum broken, and the septum replaced. Finally, the vacuum chamber and cold tip were designed in the laboratory and built by the departmental machine shop, so the device was difficult to reproduce, especially for those without the luxury of a nearby machinist. These three problems have been addressed in the latest design of the moving belt sample cooling device. The precision problems are reduced by the selection of a suitable internal standard. The problem with the rubber septum is alleviated by changing to a different sample introduction technique. Finally, the design of the vacuum chamber is simplified so that all of the components may be obtained commercially. EXPERIMENTAL SECTION Apparatus. The sample cooling device and the spectrometric system were constructed from the commercially available components listed in Table I. Drawings of the cooling system are shown in Figure 1. The metal belt was cut from 0.002 in. thick stainless steel stock. Stainless steel was chosen as the belt material rather than brass as in previous publications ( I , 2). Stainless steel had about the same flexibility as the brass, but it was less likely to crimp, tear, or tarnish. The belt was in. wide and 40 in. long. The ends were spot-welded together so that the belt formed a continuous loop. One end of the loop was threaded around a copper disk bolted to the refrigerator cold stage. The disk is 2 in. in diameter with a thickness of 3 / 4 in. The belt was threaded around the perimeter of the disk along a groove in. deep and 8 ( ' in. wide. The other end of the loop was threaded around a similar disk made of aluminum and attached to the rotary motion vacuum feedthrough. The groove along the perimeter of this 0003-2700/89/0361-1182$01.50/0

rotating spool was lined with rubber to provide adequate friction to slide the belt along the stationary copper cold head. A flexible stainless steel coupling was positioned in the middle of the vacuum chamber, between the two metal disks. This coupling could be compressed or expanded a distance up to 1 in. to ensure that the metal loop was tight. This adjustment could be made while the system was under vacuum and at low temperature. The sample introduction system consisted of a normal liquid chromatography (LC) injector, 1/16-in.steel tubing, and a 1/16-in. Cajon Ultratorr vacuum fitting (Macedonia, OH). The fitting was welded into the top center of the front portion of the vacuum chamber, directly over the cold stage. One end of the steel tubing was passed through the fitting so that the tip was positioned about in. above the cold portion of the belt. The other end was connected to a three-way LC switching valve mounted beside the vacuum chamber. With the valve in the closed position, the vacuum chamber was isolated. With the valve in the open position, the chamber was in series with a short piece of Teflon tubing open to the atmosphere. Upon injection of a sample, the open end of the Teflon tubing was immersed in a small amount of the liquid to be tested, and the valve was switched so that the liquid was in series with the chamber. The low pressure in the chamber pulled about 50 pL of sample onto the cold belt in 1s. The injector was then quickly returned to the closed mode. A schematic diagram of the spectrometric system is shown in Figure 2. The entire output of the xenon arc lamp was modulated with the chopper and focused onto the frozen sample through the quartz window in the vacuum chamber. Phosphorescence from the sample was collected by a lens, passed through the opposite side of the chopper wheel, and focused through an adjustable aperture onto the entrance slit of the 0.2-m-grating spectrograph. The phosphorescence spectrum was detected a t the focal plane of the spectrograph by a linear photodiode array. An exposure time of 15 s was used for each spectrum. The length of the array was such that the 1024 photodiodes corresponded to the wavelength region 200-800 nm. The entrance slit of the spectrograph was chosen so that one photodiode corresponded to one resolution element, or 0.8 nm. The chopper wheel had three equally spaced openings that together allowed light to pass 33% of the time. The wheel was turned so that the source and emission were modulated at 300 Hz, with an excitation time of 1.1ms and a delay between excitation and observation of 1.1ms. This arrangement assured that no scattered source radiation reached the detector during the observation time. Reagents. Benzo[e]pyrene, phenazine, and triphenylene were obtained from Aldrich (Milwaukee, WI), coronene was from Fluka (Ronkonkoma, NY), and phenanthrene was from Eastman (Rochester, NY). All were labeled 98% pure or better and were used as received. Spectroscopic grade hexane was obtained from Burdick & Jackson (Muskegon, MI) and was used for all solutions. Stock solutions of 100 mg/L were prepared for each PAH, and dilutions were made to cover a concentration range of 5 orders of magnitude. All solutions contained 10 mg/L phenazine as an internal standard. Procedure. Initial evacuation of the chamber was achieved with the roughing pump. A pressure of 100 mTorr was reached in approximately 15 min. Next, the turbopump was started, and a pressure of Torr was reached in 15 s. The helium compressor was then started, and a temperature of 15 K was reached at the cold head in about 2 h. Samples were injected as described above. Once the sample was frozen, the belt was moved via the feedthrough until the sample was positioned directly in the focused excitation beam. A phosphorescence spectrum was obtained, and the belt was rotated so that the sample was carried away from the cold stage. Another sample could be injected immediately. At least four injections were made of each solution, and the average peak height C 1989 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989

1183

Table I. Commercially Available Components of the Sample Cooling Device and Spectrometric System component

company

address

Vac-U-Flat five-way cross with 64x1. flanges Vac-U-Flat three-way cross with 6-in. flanges Vac-U-Flat flexible coupling 6-in. quartz viewport direct-drive rotary motion feedthrough Turbo-V80A turbomolecular pump SD-90 direct-drive mechanical pump Model 860A cold cathode ionization gauge model 801 thermocouple gauge displex Model CS-202 closed-cycle refrigeration system Model 7000 three-way flow switching valve Model 8500 200-W xenon arc lamp Model UFS-200 flat field spectrograph Model 382A optical chopper Model IRY-1024G optical spectrometric multichannel analyzer

Huntington Mechanical Laboratories

Mountain View, CA

Varian Vacuum Products

Lexington, MA

ApD Cryogenics, Inc. Rheodyne, Inc. Oriel Corporation Instruments SA, Inc. Ithaco, Inc. Princeton Instruments, Inc.

Allentown, PA Cotati, CA Stamford, CT Metuchen, N J Ithaca, NY Princeton, NJ

TOP VIEW n INJECT ION

\CRYOSTAT

\

I BELLOWS

+FEEDTHROUGH

-% .

COhTROL

\ APERTURE

SIDE VIEW INJECTION

I,

STATIONARY TIP ,COLD

MOVING BELT I

QUARTZ

WI mow

I \

VACUUM GAUGES

-0

ROTAT I NG SPOOL

Flgure 1. Top view of the vacuum chamber housing the moving sample cooling belt (upper panel). Simplified cross-sectional side view of the vacuum chamber (lower panel). at the phosphorescence wavelength maximum was used to prepare analytical calibration curves.

RESULTS AND DISCUSSION Phosphorescence Spectra. Examples of the low-temperature phosphorescence spectra observed for the samples studied are shown in Figure 3. The relatively narrow emission bands result from the well-known Shpol’skii effect, which was first reported in 1952 and whose various analytical applications have been the subject of several extensive review articles (S12). The experimental setup used to obtain these spectra has several advantages. The white light excitation ensures that all phosphors in the sample are simultaneously illuminated with the highest intensity possible a t their excitation maxima. The rotating disk phosphoroscope allows the separation of phosphorescence from fluorescence and scattered source radiation. The photodiode array detector allows rapid spectral acquisitions, broad spectral coverage, and convenient data manipulation such as background subtraction and spectral averaging. Taken as a whole, the greatest advantage of the spectrometric system is that no prior knowledge of the compounds

Figure 2. Schematic diagram of the spectrometric system for lowtemperature phosphorescence measurements. present in the analytical sample is required. No excitation or emission wavelengths need to be mechanically selected since all are monitored continuously. This is a time-saving advantage of this system over even the fastest rapid scanning phosphorimeters. A complete phosphorescence spectrum may be acquired in 33 ms regardless of sample constitution. This quality makes this system ideal for low-temperature phosphorescence detection in liquid or gas chromatography. Internal Standard. Since poor reproducibility in the sampling procedure was the major cause for poor precision in previous studies, careful attention was given to the selection of a suitable internal standard. Several workers have suggested particular compounds such as coronene, perylene, and benzo[ghi]perylene as internal standards for Shpol’skii fluorescence measurements (12). For phosphorescence measurements benzo[e]pyrene has been suggested as an internal standard for those compounds with emission wavelengths below 500 nm, and phenanthrene has been suggested for those compounds with phosphorescence spectra beyond 500 nm (13). In each of these cases, some prior knowledge of the analyte is required. After a thorough search of the literature, and some trial and error, phenazine was chosen as the internal standard for this study. Several factors contributed to this decision. Phenazine exhibits phosphorescence in the red region, beyond 640 nm, where it interferes spectrally with very few of the PAHs routinely investigated. The phenazine spectrum is characterized by extremely narrow phosphorescence bands under broad band excitation in Shpol’skii solvents ranging from n-pentane to n-octane (14). Phenazine is rarely present in analytical samples. Finally, as demonstrated in this work, quantitative analysis using phenazine as an internal standard results in excellent analytical figures of merit for the deter-

1184

ANALYTICAL CHEMISTRY, VOL. 61, NO. 10, MAY 15, 1989

Table 111. Analytical Figures of Merit Oberved for Low-Temperature Phosphorimetry with a Moving Sample Cooling Device

phospho-

lin dynam

rescence

range,b detectn,a orders of ng/mL magnitude

wavelength max, nm

compd benzo[e]pyrene

coronene phenanthrene triphenylene

553 578 514 479

lim of

0.5 1 2 0.4

4.7 4.7 4.7 4.7

slope of log-log

plot 0.96 1.00 0.96 0.99

LOD was calculated as that concentration of sample giving rise to a signal equal to times the standard deviation of the blank signal. Linear dynamic range was limited at both ends by the linear ranee of the detector and blank noise.

2 0 0

a

5 0 0

WAVELENGTH

0

nm)

Flgure 3. Low-temperature phosphorescence spectra of benzo[e]pyrene (A), coronene (B), phenanthrene (C), and triphenylene (D). The arrows denote peaks due to the internal standard phenazine. The solvent is n-hexane, and the cold stage temperature is 15 K.

Table 11. Precision Improvements Observed in Low-Temperature Phosphorimetry with Phenazine as an Internal Standard

compd

ng/mL

70 RSD" with no internal std

benzo[e]pyrene

100 500 160 280

46 47 44 67

concn,

coronene phenanthrene triphenylene

7oRSDb with phenazine internal std 14 10 11 10

%RSD reported for compound peak height measurements of %RSD reported for the ratio of compound peak height to phenazine peak height for 10 injections. a

10 injections.

mination of PAHs ranging in size from three to seven aromatic rings. Table I1 shows the precision improvements observed in low-temperature phosphorimetry with phenazine as an internal standard. In each case studied, a 4-fold improvement in the relative standard deviation for 10 measurements was obtained by using the internal standard method. The rather poor precision resulted from the variability in the freezing procedure and sample introduction. The combination of the internal standard and standard addition methods should further improve the precision as suggested previously (12). Analytical Figures of Merit. Analytical calibration curves were plotted for each compound, and the analytical figures of merit calculated from those curves as reported in Table 111. Curves were plotted as the ratio of analyte intensity (I,) at the given wavelength to internal standard intensity (Ist) at 650 nm versus analyte concentration. The limit of detection was taken as the lowest ZX/Zst ratio detectable. This was calculated as that concentration of analyte giving rise to a signal equal to 3 times the standard deviation of the blank signal a t the analyte wavelength, while the intensity a t 650 nm for 10 mg/L phenazine was just below the saturation value

of the detector. This condition was achieved for a given sample by manually opening the aperture a t the entrance of the emission monochromator until Zst was just below saturation (Figure 2). Opening the aperture increases both Z, and Is,by a constant factor, so linearity is preserved for the calibration curves. Similarly, the upper limit to the linear dynamic range was taken as the highest Zx/Istratio detectable. This was calculated as that analyte concentration giving rise to a signal, I,, just below detector saturation, while ZSt for 10 mg/L phenazine was equal to 3 times the standard deviation of the blank signal a t 650 nm. Again, this condition was achieved by closing the aperture a t the emission monochromator until the analyte signal was just below saturation. This procedure resulted in a linear dynamic range of 4.7 orders of magnitude for each compound tested, since the limiting Zx/Zst ratios are the same in all cases. Linearity for each curve was very good, with log-log slopes of 1.00 A 0.005.

CONCLUSIONS The combination of the self-cleaning moving sample cooling device and the unique spectrometric system described here produces a powerful technique for the determination of PAHs by low-temperature phosphorimetry. The procedure is very fast, with rapid sample changes a t 15 K and no need for prior knowledge of the sample constitution. This makes the technique particularly attractive as a possible low-temperature "on-the-fly" detector for chromatography. Also, the identification of a nearly "universal" internal standard for Shpol'skii phosphorescence measurements improves the precision and the analytically useful range of the technique.

LITERATURE CITED ( 1 ) Jones, B. T.; Winefordner. J. D. Anal. Chem. 1988, 6 0 , 412-415. (2) Jones, 8. T.; Giick, M. R.; Mignardi, M. A.; Winefordner, J. D. Appl. Spectrosc, 1988, 42, 850-853. (3) Paturei, L.; Jarosz, J.; Fachinger, C.; Suptil, J. Anal. Chlm. Acta 1983, 747. 293-302. (4) Conrad, V. 6.; Carter, W. J.; Wehry, E. L.; Mamantov, G. Anal. Chern. 1983. 55. 1340-1344. (5) Hauge, R: H.; Fredin, L.; Kafafi, 2. H.; Margrave, J. L. Appl. Spectrosc. 1986, 40, 588-595. (6) Reedy, G. T.; Bourne, S.; Cunningham, P. T. Anal. Chem. 1979, 57, 1535- 1540. (7) Brown, R . S.; Wilkins, C. L. Anal. Chem. 1988, 6 0 , 1483-1488. (81 Reedv. G. T.: Ettinaer, D. G.: Schneider. J. F.; Bourne, S. Anal. Chem. 1085; 57, 1602-7609 (9) Shpol'skii, E. V.: Il'ina, A. A.; Klimova, L. A. Dokl Akad. Nauk SSSR 1052 ..-, 87. . , 935-938 ... ... (10) Nurmukhametov, A . N. Russ. Chern. Rev. (Engl Transl.) 1969, 38, 180- 193. (11) D'Silva, A. P.; Fassel, V A. Anal. Chern. 1984. 56, 985A (12) de Lima, C. G. CRC Grit. Rev. Anal. Chem. 1086, 16. 177-220. (13) Pazhnina. S. V.; Khesina, A. Ya.; Eidel'shtein, 9. A,; Gaevaya, T. Ya. Ind. Lab. (Engl. Transl.) 1974, 40, 1291-1293. (14) Dinse, K. P.; Winscom. C. J. J . Lumin. 1979, 78/19, 500-504. I~~

~~

RECEIVED for review December 2 , 1988. Accepted March 1, 1989. This work was supported by NIH-GM11373-26.