Anal. Chem. 1983, 55, 2407-2413
(5) Although this method is limited to apply to 1:l complexes, the experimental procedure is very simple in which only the metal concentration must be treated as a known parameter.
LITERATURE CITED (1) Ohyoshl, E.; Ohyoshi, A. J . Inorg. Nucl. Chem. 1071, 33, 4265. (2) Ohyoshl, A.; Ohyoshl, E.; Ono, H.; Yamakawa, S. J . Inorg. Nucl. Chem. 1072, 3 4 , 1955. (3) Ohyoshl, E.; Oda, J.; Ohyoshl, A. Bull. Chem. SOC.Jpn. 1075, 48, 227. (4) Ohyoshi, E.; Ohyoshi, A. Bull. Chem. SOC.Jpn. 1980, 53, 805. (5) Benesi, H. A.; Hildebrand, J. H. J . Am. Chem. SOC.1040, 77, 2703. (6) McConnell, H.; Davidson, N. J . Am. Chem. SOC. 1050, 72, 3164.
2407
(7) Sokol, L. S. W. L.; Ochrymowycz, L. A,; Rorabacher, D. 8 . Inorg. Chem. 1081, 2 0 , 3189. (8) Rlngbom, A. "Complexation in Analytical Chemistry"; Intersclence: New York, 1963. (9) Pollard, F. H.; Nickless, G.; Samuelson, T. J.; Anderson, R. G. J . Chromatogr. 1064, 76, 231. (IO) Kolat, R. S.; Powell, J. E. Inorg. Chem. 1982, 1 , 293. (11) Greary, W. J.; Nickless, G.; Pollard, F. H. Anal. Chlm. Acta 1082, 2 7 , 71. (12) Tanaka, M.; Funabashl, S.; Shirai, K. Znorg. Chem. 1968, 7 , 573. (13) Sommer, L.; Novotna, H. Talanta 1067, 74, 457.
RECEIVED for review December 7,1982. Accepted September 9, 1983.
Time-Resolved Fluorometry in Detection of Ultratrace Polycyclic Aromatic Hydrocarbons in Lake Waters by Liquid Chromatography Naoki Furuta* and Akira Otsuki Division of Chemistry and Physics, National Institute for Environmental Studies, 16-2 Onogawa, Yatabe, Tsukuba, Ibaraki 305, J a p a n
For determination of polycyclic aromatic hydrocarbons In natural lake waters at ultratrace levels, a time-resolved fluorescence measurement system was developed. By replacing the Xe lamp of a conventlonal fluorometer with a N, laser-pumped dye laser as an excitation source, we could Improve the detection capablllty for polycyclic aromatlc hydrocarbons by 1 to 2 orders of magnitude. The measurement system was used as a detector for hlgh-performance liquid chromatography. A detection limit of 180 fg was achieved for bento[a]pyrene. The measurement preclslon was 3.9%. The Instrument was applied to measure polycyclic aromatic hydrocarbons in the water of Lake Mashu. After allowlng for extraction efflclency, we obtained analytical results of 0.009 ng/L, 0.007 ng/L, and 0.014 ng/L for benzo[kyiuoranthene, benzo[a Ipyrene, and benro[ghl]peryiene, respectively.
Some polycyclic aromatic hydrocarbons (PAHs) are known as mutagenic and carcinogenic substances ( I ) . They are biosynthesized by plants and soil bacteria and they are also produced in high-temperature reactions of forest fires and volcanic activities. On the other hand, far greater quantities of PAHs originating from man-induced combustion processes are released to the atmosphere. A significant proportion of PAHs, formed by either natural sources or human activities, is decomposed by photooxidation. However, some PAHs, which are not decomposed, contaminate the upper layers of the earth, where fallout of PAHs occurs. Some of this fallout will be carried to rivers, lakes, and, eventually, oceans (2). Until the beginning of the present century, a natural balance between formation and degradation of PAHs ensured a low and constant natural background concentration. However, with increasing industrial development, the PAHs originating from human activities have disturbed the balance and the PAH concentrations in the environment are gradually increasing (3). The concentrations of PAHs can be used as an indication of whether a district is polluted or not. In such 0003-2700/83/0355-2407$01.50/0
cases, the concentrations of PAHs existing under natural conditions are used as the "normalnlevel of PAHs. According to recent reviews concerning PAHs in the aquatic environment ( 4 , 5 ) ,the concentration of PAHs in unpolluted water can be estimated to be less than 1ng/L (ppt). In order to determine such low levels of PAHs, it is necessary to improve conventional analytical methods (6). A fluorescence measurement system utilizing a pulsed laser has been developed to fulfill the demand for extremely sensitive detection of PAHs. Richardson and Ando (7) obtained detection limits, which they defined as the concentrationwhich produced a signal equivalent to twice the noise (SIN = 2)) of 19 pg/L, 1.3 ng/L, 8.9 ng/L, 1 ng/L, and 0.5 ng/L for bezene, naphthalene, anthracene, fluoranthene, and pyrene, respectively, by using a N2laser-pumped dye laser. Voigtman et al. (8)also used a Nzlaser-pumped dye laser and reported that detection limits for 30 PAHs were in the range of 0.4-60 ng/L. In their case, the detection limits were defined as the concentration which produced a signal three times the standard deviation of blanks ( 3 4 There is no doubt that laser fluorometry has a big advantage with respect to sensitivity in the determination of PAHs a t ultratrace levels. However, even more selectivity is necessary when the laser fluorometry is applied to measure PAHs in environmental samples. Van Gee1 and Winefordner (9) tried to improve selectivity by using the lifetime difference between anthracene and pyrene, but that method is rather difficult to apply to a complex mixture of PAHs. Another promising method may be low temperature fluorometrywith a laser. When the PAHs are cooled down to less than 15 K in an appropriate matrix, excitation and fluorescence spectra are sharpened. Therefore, the selective excitation of a given PAH present in a complex mixture can be obtained by utilizing narrow bandwidth, tunable dye laser excitation. There have been several demonstrations of the enhanced selectivity of PAHs in frozen n-alkane matrices (IO), in frozen inert gas matrices (II), and in frozen glass matrices (12). In addition to low temperature fluorometry, Dickinson and Wehry (13) demonstrated that 0 1983 American Chemlcal Socletv
2408
ANALYTICAL CHEMISTRY, VOL. 55,NO. 14, DECEMBER 1983 N2 Laser
Dye Laser
Fluorometer
HPLC
measurement of a PAH was carried out with a gate width of 10 ns. The HPLC consisted of a Waters Associates Model M-6000 1MI Wcharge Channel -tliquid chromatograph pump and Model U6K injector. The Hlgh Voltage column used was a Waters Associates Radialpak A (8 mm x 10 cm) and pBondapak CI8(4 mm X 30 cm). The sample chamber consisted of a Model FP-1030 (Japan Spectroscopic Co., Japan) flow cell system. The active volume of the flow cell was 15 pL. Comparison of a Xe Lamp and a Laser. Usually, a 150-W Xe lamp is used as an excitation source of a fluorometer. The sensitivityimprovement was investigated by replacing a Xe lamp with a N2 laser-pumped dye laser. When a Xe lamp was used as an excitation source, the output signal of the photomultiplier Flgure 1. Laser-induced fluorescence measurement system In con(PMT) was amplified through a current-to-voltage amplifier junction with HPLC: M,, mirror for N, laser; M, mirror for dye laser; (LF356). In the case of the pulsed laser excitation, the output MSand M, toroidal mirrors; M, plain mirror; Me, spherical mirror; L, and L, cylindrical lenses; G,,ruled grating; G,, concave grating; PM, signal of the PMT was directly measured by placing a 5 0 4 load photomultiplier; PD, photodiode. resistor between the anode and ground. The PMT base wiring was modified to allow a large anode current by placing charging time-resolved fluorometry was useful to alleviate the difficapacitors between the last four dynodes (21). When the detection culties of spectral overlap and base line definition. However, capability was compared between a Xe lamp and a pulsed laser in the case of low-temperature fluorometry, the sample cooling as excitation sources, a standard 1-cmquartz cuvette cell was used. Sampling. Lake Mashu is situated in northern part of Japan; procedure has to be carefully controlled to ensure reproducit has the highest recorded value of transparency depth (41.6 m) ibility. Thus, it is time-consuming to handle a large number in the world. There is no river flowing into it and no road near of samples. In addition, the assignment of low-temperature the lake, and so it is remote from human activities. A sDecial luminescence spectra has not been established yet. device was constructed to collect water samples from Lake Mashu In recent years, together with gas chromatography (GC), directly into a 3-L separating funnel. The water sample was the separation technique of PAHs with high-performance collected at a depth of 5 m in the center of Lake Mashu on liquid chromatography (HPLC) has been improved (14),and September 9, 1982. separation with HPLC has been widely applied to the deExtraction Procedures. The water sample of 2 L was placed termination of PAHs in the aquatic environment (15,16). The in a 3-L separating funnel, and 100 mL of redistilled cyclohexane was added as an extraction solvent, at the shore of Lake Mashu. use of lasers as a detection system for HPLC is generating The separating funnel was fixed in a rugged, lighttight box, and great interest (17,18). Diebold and Zare (19)have succeeded brought back to our laboratory by car. In total, the funnel had in obtaining a detection limit (SIN = 2) of 750 fg for aflatoxin, been shaken during transportation for 6 days until the extraction and Folestad et al. (20)have established a minimum detection procedure could be conducted in our laboratory. After the seplimit (SIN = 10) of 20 fg for fluoranthene by utilizing HPLC arating funnel was shaken for 30 min by a shaker, the emulsion separation followed by CW laser-induced fluorescence dewas allowed to stand for 1 day. The cyclohexane layer was tection. A major shortcoming of these fluorometers is less collected into a 200-mL distilling flask through a prewashed flexibility in wavelength selection due to a CW laser. Tunanhydrous sodium sulfate layer. After the cyclohexanelayer was ability is much larger with a pulsed laser through a wider separated, the separating funnel was rinsed with 20 mL of redistilled cyclohexane. The combined cyclohexane solutions were selection of both dyes and doubling crystals. In this study, evaporated to small volume by distillation in a rotating evaporator. time-resolved fluorometry based on a pulsed laser was used The concentrated solution was transferred to a 10-mL Kuderas a detection system for HPLC. na-Danish tube. After the solution was transferred, the 200-mL EXPERIMENTAL SECTION distilling flask was rinsed with 5 mL of redistilled cyclohexane. The combined cyclohexane solutions were evaporated to dryness Instrumentation. A block diagram of the instrument is shown and 0.5 mL of acetonitrile was added. For the extraction proin Figure 1. The pumping source of the dye laser was a Molectron cedure blank, 100 mL of redistilled cyclohexane was added to a Model UV-14 nitrogen laser. The peak power and the pulse 3-L separa$ing funnel without a water sample, and the same duration of the N2laser were 425 kW and 10 ns, respectively. It extraction procedures as described above were conducted. For was operated at a repetition rate of 16 Hz. The dye laser used the extraction procedure, 125 mL of redistilled cyclohexane was was a Molectron Model DL14. The two dyes used for tuning the laser were PBD [2-(1,l’-biphenyl)-4-yld-phenyl-l,3,4-oxadiazole] used. Therefore, in order to check the solvent blank, 125 mL of redistilled cyclohexane was evaporated to dryness and dissolved and BBQ [4,4”’-bis[(2-butylody~)oxy]quaterphenyl], respectively, in 0.5 mL of acetonitrile. for the 360-386 nm and 373-399 nm wavelength regions. The Recovery. For the recovery experiment of PAHs at ultratrace energy of the dye laser measured with a Molectron 53 pyroelectric levels, organically free water was prepared. Distilled water was joulmeter was 0.4 mJ at 366 nm and 0.96 mJ at 386 nm. A small purified by the Millipore Corp. Milli-Q system equipped with two portion of the N2 laser beam was detected by a photodiode to ion-exchange resin cartridges (Ion-Ex)and one activated carbon provide a trigger pulse for a digital boxcar integrator (Model cartridge (0rganex-Q). The water so collected was passed through BX-531, NF Circuit Design Block Co., Japan). The laser-induced a trace organic removal cartridge (Millipore Norganic) and then fluorescence was collected at right angles to the laser beam and filtered under vacuum through a Nucleopore 1-pm filter. The was dispersed by a monochromator with a Rowland circle of 15 NBS Standard Reference Material (SRM 1647),which contained cm diameter. A side-on photomultiplier (Hamamatsu R446) with 16 PAHs at pg/mL levels in acetonitrile,was used for the recovery base wiring modification (described later) was employed at an experiment. Two liters of the organically free water was taken applied voltage of 1 kV with a Model PH-7A (Japan Electronic in a 3-L separating funnel, and a 1-mL portion of SRM 1647 Measuring Instrument Co., Japan) power supply. The fluoresacetonitrile solution diluted by a factor of 25 OOO was added. The cence signal was delayed for 150 ns by a Tektronix Model 7Mll separating funnel was then shaken continuously for 1day to mix delay line to circumvent the inherent boxcar gate pulse delay (150 thoroughly. In order to reproduce the analytical conditions used ns for the digital boxcar wed), and then fed into the digital boxcar. for thd Lake Mashu water sample, the separating funnel was The digital boxcar enables measurement of the fluorescencesignal allowed to stand for 6 days after the addition of 100 mL of during the gate width with an appropriate delay time after the redistilled cyclohexane. The subsequent extraction procedures trigger pulse is acquired. There are two gate units: a Tektronix were also the same as those used for the Lake Mashu water sample. S-5 sampling head unit which has a fixed gate width of 1ns, and Reagents and Glassware. Ethanol and acetonitrile used in a NF Circuit Design Block variable gate unit whose gate width this experiment were purchased from Wako Pure Chemical Inrange is 10 ns to 5 ms. When the fluorescence decay curve of dustries, Ltd. The former was guaranteed reagent grade and the PAHs was measured, a gate width of 1ns was used. Quantitative N2Gas!-
,
’
.
a[-
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
3-
Xe Lamp
0.3
?! 5
2409
Excitation
-
* 0.10.0
I\
1.0-
Laser Excitation
B
3W 0 I
0
J
J
I
,
J
,
,
,
,
I
I
I
I
I
I
I
I
I
100
Time (ns) Figure 2. Fluorescence decay curves: (A) ethanol solutions with B(a)P (a)and without B(a)P(b); (B) resultant decay curve for 0.83ng/mL B(a)P obtained by subtraction treatment (a) - (b). latter was a guaranteed grade for the analysis of pesticides. They were used without further purification. Cyclohexane was ultraviolet spectrometricgrade obtained from Wako Pure Chemical Industries, Ltd., and was distilled twice at 26 O C under vacuum in a rotating evaporator. Anhydrous sodium sulfate was also obtained from Wako Pure Chemical Industries, Ltd., and was fluorescence spectrometric grade. All PAH compounds were obtained from commercial sources and used without further purification: naphthalene, phenanthrene, acenaphthene, acenaphthylene, anthracene,pyrene, benz[a]anthracene,benzo[a] pyrene, and dibenz[a,h]anthracene (all from Wako Pure Chemical Industries, Ltd.), fluorene, fluoranthene, triphenylene, chrysene, and perylene (all from Nakarai Chemicals, Ltd.), benzo[b]fluoranthene, benzovlfluoranthene, and benzo[k]fluoranthene (all from P. K. Chemical, Ltd.), benzo[e]pyrene and benzo[ghilperylene (all from Aldrich Chemical Co., Ltd.), and indeno[ 1,2,3-cd]pyrene (Tokyo Kasei Kogyo Co., Ltd.). All glassware was cleaned with acetone and detergents and then thoroughly rinsed with distilled water. The cleaned glassware remained in contact with a solution of potassium dichromate for more than 1week, and prior to use was rinsed with deionized and redistilled water.
-
RESULTS AND DISCUSSION Comparison of a Xe Lamp and a Laser. By use of a PBD dye, the pulsed laser at 363 nm irradiated an ethanol solution of B(a)P (0.83 ppb), and the fluorescence decay curve was measured by the digital boxcar at 403 nm. Fluorescence decay curves for ethanol solutions with B(a)P and without B(a)P were stored in two memories of the digital boxcar integrator, respectively (Figure 2A). Subtraction provided the channel-by-channel difference of the two memories, and the fluorescence decay curve due to B(a)P could be obtained (Figure 2B). The signals observed during the pulse on time when the pulsed laser is used are due to stray light and Raman scattering. These signals disappeared immediately when the laser irradiation terminated. However, the fluorescence of B(a)P continued for about 70 ns after the laser irradiation ceased. The boxcar integrator enables measurement of the induced fluorescence during the gate width with an appropriate delay time. With increasing delay time, the stray light and the Raman scattering decreased and only compounds which have long fluorescence lifetimes could be detected. Fluorescencespectra for an ethanol solution of B(a)P (0.83 ppb) were also measured by Xe lamp excitation and laser excitation. Figure 3 illustrates an advantage of a pulsed laser
350
4bO
450
Wavelength (nm)
5bO
550
Figure 3. Fluorescence spectra for ethanol solutions with B(a)P (a) and without B(a)P(b) measured by Xe lamp excitation and laser excitation.
Table I. Comparison of Detection Limits (SIN = 2) of B(a)P between Xe Lamp and Laser Excitation Sources Xe lamp laser nm nm time constants, s detection limits, ndmL
A,,, A,,
363 403 0.25 0.5
386 403 0.25 0.2
363 403 0.25 0.05
386 403 0.25 0.03
and the gated detection system over a Xe lamp and the corresponding detection system. In the case of the Xe lamp excitation, Rayleigh scattering (at 363 nm) of irradiated light and Raman scattering (1415 cm-l (at 383 nm); CH bending vibration: 2832 cm-' (at 404 nm); CH stretching vibration) due to ethanol are clearly observed. On the other hand, in the case of laser excitation, Rayleigh and Raman scattering are decreased drastically and fluorescence spectra with improved signal-to-noise (SIN) ratio are obtained. For comparison, the detection limits obtained by Xe lamp excitation and laser excitation are summarized in Table I. In this case, the detection limits are defined as the concentration which produces a signal equivalent to twice the noise (SIN = 2). In the case of Xe lamp excitation, detection limits could be improved by changing the excitation wavelength from 363 nm to 386 nm, because the fluorescence signal of B(a)P can be detected without overlap of the Raman scattering due to CH stretching vibration. In the case of laser excitation, detection limits could be improved by replacing a PBD dye (363 nm) with a BBQ dye (386 nm), because the pulse energy of a BBQ dye is higher than that of a PBD dye. As shown in Table I, the detection limits become lower by 1 order of magnitude when the laser source is used instead of the Xe lamp. Time-Resolved Fluorometry as a Detection System for HPLC. Time-resolved fluorescence chromatograms obtained at different delay times are shown in Figure 4. The NBS standard (SRM 1647) was diluted with acetonitrile by a factor of 250, and 25 p L of the solution was injected into the HPLC column. The separation of PAHs was performed by a Radialpak A column with the use of isocratic elution (80% acetonitrile in water) a t a flow rate of 3.0 mL/min. By use of a PBD dye, a tuned dye laser at 366 nm irradiated the eluent, and the induced fluorescence was measured at 403 nm. With a delay time of 0 ns to 10 ns after excitation, strong base line drift results from drift of stray light, Raman scattering, and short-lived fluorescence from the HPLC eluent. With
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
2410
Table 11. Comparison of Detection Limits (SIN= 2) of B(a)P Obtained by Using a Cuvette Cell and a Flow Cell (ng/mL) cuvette cell injection volume bandwidth, nm, em (403 nm)
flow cell 25 M L
5
10
5
0.043 0.020 0.009
0.033
0.168 (4.2 pg)" 0.065 (1.6 pg)"
50 p L 10
10
time constants, s 0.25 1 4
16
0.014 0.005
0.008
0.100(2.5 pg)"
0.035 (0.9 pg)"
0.003
-
-b
a Values in parentheses represent the injected amount of B(a)P. constant of 16 s.
0.055 (2.7 pg)" 0.030 (1.5 pg)" 0.015 (0.75 pg)a
0.045 (1.1pg)" 0 . t 3 3 (0.8 pg)a
-b
Fluorescence response was too slow with the time 1
0.0
0.0
0.0 ,
0.0 ,
0
A V
5 10 15 Retention Time (mln.)
0
5
10
15
Retentlon Time (rnin.)
) 21010-3
IO-*
16'
1oo
10'
Flgure 4. Liquid chromatograms for a mixture of 16 PAHs (SRM 1647) with time-resolved fluorometry. Chromatographic condltions were as follows: Radialpak A column; 8020 (v/v) acetonitrile-water; flow rate, 3.0 mL/mln. Fluorescence detection conditions were as follows: bandwidth, 5 nm; time constant, 1 s.
Concentration of B(a)P (ppb) Figure 5. Linearity of HPLC-time-resolved fluorometry. Chromatographic colldltions were as follows: MBondapak Cl8 column; 80:20 (v/v) acetoqfthle-water; flow rate, 0.8 mL/mln. Fluorescence detection conditions were as follows: bandwidth, 10 nm; time constant, 4 s.
a delay time from 20 ns to 30 ns, the base line drift decreased, and the chromatogram with the best SIN ratio was obtained. After a 40 ns delay, the S I N ratio decreased because the fluorescenceintensity decreased. With a delay time adjusted from 0 ns to 20 ns, the chromatogram peak observed at a retention time of 3.8 min is assigned to anthracene. Since anthracene has a relatively short fluorescence lifetime of 4.9 ns (22), the fluorescence is greatly reduced in magnitude after a 30 ns delay. It is clear from Figure 4 that even with the moderate gate width of 10 ns, short-lived PAHs can be distinguished from long-lived PAHs. However, more precise separation of PAHs by using the lifetime difference requires the improvement of time resolution of the total system. Imasaka et al. (23) constructed a transversely excited atmospheric-pressure N2laser and achieved time resolution of 1.4 ns by using a subnanosecond tunable dye laser as an excitation source, a subnanosecond response photomultiplier, and a sampling oscilloscope with a bandwidth of 1 GHz. Optimum operating conditions for the determination of B(a)P were determined by changing such parameters as the delay time, the column, the dye, the time constant of the boxcar integrator, and the bandwidth of the monochromator. Although, compared with a Radialpak A column, it gave poorer separation performance for B(a)P and B(k)F, it was decided to use a NBondapak C18column for the PAH separation because the fluorescence signal was twice as large due to a lower flow rate. As mentioned previously the pulse energy of a BBQ dye is higher than that of a PBD dye; therefore, the detection capability was improved by using a BBQ dye. Detection limits of B(a)P were obtained by optimizing the time constant of the boxcar integrator and the bandwidth of the monochromator. Table I1 summarizes the detection limits
obtained by using a cuvette cell and a flow cell, respectively. The detection limits were improved by increasing the bandwidth from 5 nm to 10 nm, and were almost inversely proportional to the square root of the time constants. When a cuvette cell is used, fast response is not required because the sample concentration does not change rapidly with time. Therefore,lower detection limits can be obtained by increasing the time constants. Detection of 0.003 ng/mL of B(a)P could be achibyed with a SIN ratio of better than 2. On the other hand, when a flow cell is used, reasonably fast response is required to follow the sample change. When isocratic elution whs used at a flow rate of 0.8 mL/min, fluorescence response was too slow with the time constant of 16 s. Therefore, the time coptant of 4 s was used. Although lower detection limits can be achieved by increasing the injection volume, there is a limitation because excessive injection deteriorates the column efficiency. Comparison of the detection limits obtained by the cuvette cell with those obtained by the flow cell indicates that the sample dilution factor due to the mobile phase is 4 to 5 with an injection volume of 25 p L and is 2 to 3 with an injection volume of 50 pL. Under optimum operating conditions for B(a)P, the linearity of the instrument was checked. As shown in Figure 5, a linear dependence between fluorescence intensity and concentration extends over 2 orders of magnitude. The chromatograms measured at low concentrations are shown in Figures 6 and 7. If the detection limit is defined as the amount which produces a signal equivalent to twice the noise ( S I N = 2), a detection limit is 750 fg for B(a)P (Figure 6), while a detection limit of 180 fg can be achieved if the detection limit is defined as the amount which produces a signal three times the standard deviation of the background noise
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
2411
Table 111. Retention Time and Relative Fluorescence Intensity Obtained under Optimum Conditions for each PAH re1 retention fluorescence time, intens min compound he,, nm he,, nm no.
10 11 12 13 14 15 16 17 18 19 20 a
0.033 0.63 0.12 0.17 0.0064 0.75 0.092 0.27 0.036
5.68 6.70 6.72 6.80 6.88 7.40 8.44 8.80 9.68 9.99 10.15 12.34 12.42 12.48 12.66 12.84 13.41 15.03 16.68 17.28
0.32 0.55 D.16 0.085 2.8 0.10 1.85 1.00 0.43 0.049 0.17
Represents 16 PAHs contained in the NBS standard (RSM 1647).
h
Oso71 U 0.06
h
E
Y
c
0.07 0.06-
U E 0.05
0.05
Y
c
c
5
0
:
334.7 303.0 304.0 323.0 336.4 400.5 467.1 373.0 353.1 381.8 387.0 428.6 397.2 437.0 437.9 410.0 404.6 395.0 503.0 406.4
276.4 266.0 268.6 289.6 288.7 251.2 284.0 333.0 285.5 26 7.1 286.0 298.5 329.5 407.0 408.2 306.5 382.5 395.5 379.4 381.0
naphthalenea fluorenea phenanthrenea acenaphthenea acenaphthylenea anthracenea fluoranthenea pyrenea triphenylene chrysenea benz [e ]anthracenea benzo [ b 3 fluoranthenea benzo[ e Ipyrene pery 1ene benzo[j] fluoranthene benzo[ h J f l ~ o r a n t h e n e ~ benzo [e ]pyrenea dibenzEa,h]anthracenea indeno[ 1,2,3-cd]pyrenea benzo[ghi]perylenea
1 2 3 4 5 6 7 8 9
c
3 P c
C
2
0.04
5
0
0.030.02 0.01 0.00
-
0.04
c
I
:
a
:
.c
:
a 0.0 0.00 1 I
Figure 0. Timaresolved fluorescence chromatogram for the injection of 50 pL of 0.0176 ng/mL B(a)P (880 fg). Chromatographic and fluorescence detection conditions were the same as in Flgure 5.
5
The measurement of 2.65 pg of B(a)P by HPLC was repeated seven times. The measurement precisions expressed by relative standard deviations (RSD)were 0.37% and 3.9% for the retention time and the fluorescence intensity, respectively. Determination of Benzo[a]pyrene in the NBS Standard. The 16 PAHs in the SRM 1647 and the other 4 PAHs commonly encountered in the aquatic environment (25) were chosen and standard solutions of the 20 PAHs were prepared by dissolving the PAHs in actonitrile. Excitation and fluorescence spectra for the 20 PAHs were measured by using a Xe lamp, and optimum excitation and fluorescence wavelengths were determined for each PAH. Table I11 shows retention times and relative fluorescence intensities for the 20 PAHs together with the abbreviations used in this paper and the optimum wavelengths. The relative fluorescence intensities listed in the last column were calculated by reference to B(a)P and obtained without correction for the
20
15
0 Retention Time (rnin.) Flgure 7. Timaresolved fluorescence chromatogram for the injection of 50 pL of 0.0088 ng/mL B(a)P (440 fg). Chromatographic and fluorescence detection condltions were the same as in Figure 5.
Table IV. B(a)P Concentration of the NBS Standard
( 3 4 (Figure 7). Richardson et al. (24)obtained detection limits
(SIN = 2) of 10 pg and 4 pg for fluoranthene and pyrene, respectively, with a gate width of 1ns. The detection limits are almost the same as those obtained in this work when molar absorptivities and quantum yields of the PAHs are considered.
10
SRM 1647 1/12500 1/25000 1/50000
concentration, ng/mL determined certified 0.43 0.21 0.10
0.424 0.212 0.106
wavelength dependency of the Xe lamp intensity and the PMT sensitivity. The relative fluorescence data imply that Per and B(k)F will give lower detection limits than B(a)P due to high molar absorptivities and high quantum yields. The excitation and fluorescence spectra and retention times listed in Table I11 provided useful information for the assignment of chromatogram peaks. In order to evaluate the accuracy of the instrument, SRM 1647 was diluted with acetonitrile by a factor of 12500,25000, and 50000, and B(a)P in those samples was quantitatively analyzed as unknown samples. Figure 8 shows the time-resolved fluorescence chromatograms obtained by injecting 50 p L of the diluted NBS standard (SRM 1647) into the HPLC column. The concentration of B(a)P in the diluted NBS standard was determined by using the analytical curve shown
2412
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
Table V. Analytical Results of PAHs in Water of Lake Mashu (without Correction for Extraction Efficiency)
standards no. 1 no. 2 no. 3 no. 4
retention time, min 12.61 ?r 0.07 12.64 12.66 12.70 12.56
av RSD
B(k)F, ng/L 0.0078 0.0088 0.012 0.0055 0.0085 32%
retention time, min 13.36 + 0.20 13.35 13.36 13.48 13.34
B(a)P, ng/L
retention time, min 17.05 * 0.06 17.02 17.08 17.17 17.10
0.0064 0.0049 0.0074 0.0056
B(ghi)P, ng/L 0.020 0.013 0.015
0.0061 18%
0.009 0.014 32%
0.21 ng of B(a)P spiked recovered, ng yield, %
0.16 ng of B(ghi)P spiked recovered, ng yield, %
Table VI. Extraction Efficiency of PAHs Spiked in 2 L of Water 0.20 ng of B(k)F spiked recovered, ng yield, % no. 1 no. 2 no. 3 no. 4 av RSD blank
0.18 0.18 0.21 0.21 0.19
90 87 103 104 96 9.2
0.028
in Figure 5. The data summarized in Table IV attest to the accuracy of the developed instrument. There is excellent agreement of the determined values with the certified NBS values. Determination of PAHs in Water of Lake Mashu. Time-resolved fluorescence chromatograms for the sample extracted from the water of Lake Mashu were measured by injecting 50 KLof the extract into the HPLC column. The measurement of the chromatograms was performed under optimum operating conditions for B(a)P. Four extraction experiments were carried out independently, and the chromatograms obtained are shown in Figure 9 as well as those obtained by blank experiments. The chromatogram peak observed at a retention time of 5.42 min is due to impurity found in anhydrous sodium sulfate. Assignment of the other peaks was performed by means of the excitation and fluorescence spectra and retention times for the 20 PAHs listed in Table 111. The concentration of B(a)P was determined by using the analytical curve shown in Figure 5, and the concentrations of B(k)F and B(ghi)P were derived from chromatograms for the diluted NBS standard shown in Figure 8. Table V summarizesthe analytical results for 3 PAHs present in the water of Lake Mashu, as well as retention times for each chromatographic peak. These times are less than those obtained earlier (Table 111)due to slight deterioration of column performance with time. The relative standard deviations of the four extraction experiments were 32%, 18%,and 32% for B(k)F, B(a)P, and B(ghi)P, respectively. The blank of extraction procedures is equivalent to the concentration of 0.004 ng/L, 0.003 ng/L, and 0.010 ng/L for B(k)F, B(a)P, and B(ghi)P, respectively. Therefore, the concentration of PAHs determined in the water of Lake Mashu corresponds to about twice the blank. The reliability of analytical results at ultratrace levels depends on the extraction efficiency. The extraction efficiency has been determined at about 100 times higher concentration than the analytical value because it is very difficult to remove trace organics from water (26). Extraction efficiencies obtained by using the organically free water are summarized in Table VI. The recovery experiment was conducted at about 10 times higher concentrationthan the analytical value of Lake Mashu. Four recovery experiments were carried out inde-
0.18 0.17 0.20 0.21 0.19
84 80 96 96 89 9.7
63
0.10 0.13 0.21 0.20 0.16
81
128 126 100
33
0.020
0.13
2
!I
I
0.4.
1
0.3-
g
g
0.2-
0 0.1
\n
-
0.0
2
I
0.3-
I
2 25000
h
1 50000
R
0.0
2 0.2E
I
5
0.0
U
5
I
0
lb
io
115
Retention Time (Inin.) Flgure 8. Time-resolved fluorescence chromatograms for the diluted NBS standard (SRM 1647). Chrornatographlc and fluorescence detection conditions were the same as in Figure 5. B 0.0 0.1-
B(a)PN~. 1 Lake Mashu ( k ) e No.2 Lake Mashu
2 0.0
N0.3 Lake Mashu
I
e E
E 0.0
3 O.lf 0.0 ,p 0.1-
a
0.0 0.1-
0
N0.4 Lake Mashu
A
-
-
Blank of Extraction Procedure
,.
Blank of Cyclohexane
5 10 15 Retention lime (mln.)
20
Flgure 9. Time-resolved fluorescence chromatograms for the samples extracted from Lake Mashu and the blanks. Chromatographic and fluorescence detection conditions were the same as in Figure 5.
ANALYTICAL CHEMISTRY, VOL.
Table VII. Concentration Ranges of B(a)P in Various Lake Waters sampB(a)P, ling ng/L vol, L date Lake Erie,a at Buffalo, USA Lake Boden,&GFR (Lake Constance) Zhizhitskoe LakesC at Pskov, USSR Lake Mashu,d Japan Cited from ref 27. ref 29. This work. a
0.3
30
1.3
500
5/1964 1971
0.01-0.1
0.007
12/1976
2
Cited from ref 28.
9/1982 Cited from
pendently, and the relative standard deviations of 9.2%, 9.7%, and 33% were obtained for B(k)F, B(a)P, and B(ghi)P, respectively. The poorer relative standard deviation for B(ghi)P results from the blank contained in the organically free water. Table VI1 summarizes literature values for B(a)P measured in various lake waters. B(a)P present in lake water has seldom been determined and, thus, few data have been published so far. Lake Erie is one of the Great Lakes in the U.S.A. and is contaminated with industrial discharge. A concentration of 0.3 ng/L of B(a)P has been found in water of Lake Erie (27). Basu and Saxena (27) considered that the low level of B(a)P observed occurred because the sample was collected during a severe snowstorm period. Lake Constance, which is situated between West Germany and Switzerland,is famous for sightseeing. The lake water has been found to contain 1.3 ng/L of B(a)P (28). Pskov, which is located 160 miles south from Leningrad, is the least polluted region in the USSR. Less than 0.01-0.1 ng/L of B(a)P has been reported by Russian workers (29) as the “normal”level of B(a)P. The data of Lake Mashu (0.007ng/L) are analytical resulta obtained from this study after correction for extraction efficiency. It should be noted that the sampling volume has a big influence. Analytical procedures for the data of Pskov were not described in detail, but for the others the data were obtained with conventional fluorometry except for Lake Mashu. When conventional fluorometry was used, a large sample volume of 30 or 500 L was necessary to extract a detectable amount of B(a)P. In this work, the detection capability is improved by 1 to 2 orders of magnitude by replacing the Xe lamp with the pulsed laser, so that the required sample volume is reduced to 2 L. A small sample volume makes the analytical pocedures simple and enables PAHs at ultratrace levels to be determined more precisely.
CONCLUSIONS It has been verified that time-resolved fluorometry based on a pulsed laser in conjunction with HPLC provides not only high sensitivity but also high selectivity for the determination of PAHs. By utilizing HPLC-time-resolved fluorometry, a detection limit of 180 fg could be achieved for B(a)P. With even more powerful dye laser systems (systems which provide more than 2 orders of magnitude greater pulse energy than the laser used in this work are commercially available) and more efficient flow cells with a smaller active volume (20,30), one should be able to improve the detection limit. Ideally PAH in natural water should be determined directly without preconcentration. However, since the sensitivity is insufficient at the present stage, an extraction procedure is necessary. By replacing a Xe lamp of the conventional fluorometer with a
55,NO. 14, DECEMBER 1983 2413
N2 laser-pumped dye laser as an excitation source, the detection capability for PAHs was improved by l to 2 orders of magnitude. Therefore, the required sample volume was reduced to 2 L and the blank due to the extraction procedure was decreased. In this study, Lake Mashu was chosen as a lake free from pollution by human activities in Japan. By extracting PAH from 2 L of the water sample, and taking into account the extraction efficiency,we obtained analytical results of 0,009ng/L, 0.007 ng/L, and 0.014 ng/L for B(k)F, B(a)P, and B(ghi)P,respectively. These values establish the “normal” level of PAHs existing in natural lake water. The degree of pollution of a given lake water can be estimated by reference to these values. Registry No. Naphthalene,91-20-3; fluorene, 86-73-7; phenacenaphthylene, 208anthrene, 85-01-8;acenaphthene,83-32-9; fluoranthene, 206-44-0;pyrene, 12996-8;anthracene, 120-12-7; 00-0; triphenylene, 217-59-4;chrysene, 218-01-9;benz[a]benzo[ blfluoranthene, 205-99-2; benzo[e]anthracene, 56-55-3; pyrene, 192-97-2; perylene, 198-55-0; benzoLjIfluoranthene, 20582-3;benzo[k]fluoranthene,207-08-9; benzo[a]pyrene,50-32-8; dibenz[a,h]anthracene,53-70-3; indeno[1,2,3-~d]pyrene, 193-39-5; benzo[ghi]perylene, 191-24-2; water, 7732-18-5. LITERATURE CITED Pelkonen, 0.; Nebert, D. W. Pharmacol. Rev. 1982, 3 4 , 189-222. Hase, A.; Hites, R. A. I n “Identification and Analysis of Organic Pollutants in Water”; Keith, L. H., Ed.; Ann Arbor Science: Ann Arbor, MI, 1976; Chapter 13, pp 205-214. Suess, M. J. Sci. Total Environ. 1978, 6 , 239-250. Borneff, J. Adv. Environ. Sci. Technol. 1977, 8 (Part 2), 393-408. Neff, J. M. “Polycycllc Aromatic Hydrocarbons in the Aquatic Environment”; Applled Science: London, 1979. Lee, M. L.; Novotny, M. V.; Bartle, K. D. “Analytlcal Chemistry of Poiycyciio Aromatic Compounds”; Academic Press: New York, 1981. Richardson, J. H.; Ando, M. E. Anal. Chem. 1977, 49, 955-959. (London) Voigtman, E.; Jurgensen, A.; Winefordner, J. D. Anakst . 198% 107, 408-4i3. Van @el, T. F.; Winefordner, J. D. Anal. Chem. 1978, 48, 335-338. Yana, Y.; D’Sliva, A. P.; Fassel, V. A. Anal. Chem. 1981, 5 3 , 894499. Wehry, E. L.; Mamantov, G. Anal. Chem. 1979, 51, 643A-656A. Brown, J. C.; Duncanson, J. A., Jr.; Small, G. J. Anal. Chem. 1980, 52, 1711-1715. Dlckinson, R. B., Jr.; Wehry, E. L. Anal. Chem. 1979, 51, 776-780. Futoma, D. J.; Smith, S. R.; Tanaka, J.; Smith, T. E. CRC Crk. Rev. Anal. Chem. 1981, 72, 69-153. Sorrel, R. K.; Reding, R. J. Chromatogr. 1979, 785, 655-670. Ogan, K.; Katz, E.; Siavln, W. Anal. Chem. 1978, 57, 1315-1320. Yeung, E. S. I n “Lasers in Chemical Analysls”; Hieftje, 0. M., Travis, J. C., Lytle, F. E., Eds.; Humana Press: Clifton, NJ. 1981; Chapter 14, pp 273-290. Green, R. B. Anal. Chem. 1983, 55, 20A-32A. Diebold, 0. J.; Zare, R. N. Sclence 1977, 796, 1439-1441. Folestad, S.; Johnson, L.; Josefsson, 8.; Galle, B. Anal. Chem. 1982, 5 4 , 925-929. Lytle, F. E. Anal. Chem. 1974, 46, 545A-557A. Berlman, I. E. “Handbook of Fluorescence Spectra of Aromatic Molecules”; Academic Press: New York, 1971; p 356. Imasaka, T.; Ishibashi, K.; Ishibashi, N. Anal. Chlm. Acta 1982, 742, 1-12. Richardson, J. H.; Larson, K. M.; Haugen, G. R.; Johnson, D. C.; Clarkson, J. E. Anal. Chim. Acta 1980. 116, 407-411. Harrison, R. M.; Perry, R.; Wellings, R. A. Water Res. 1975, 9 , 331-346. Monarca, S.; Causey, B. S.; Kirkbright, 0. F. Water Res. 1979, 13, 503-508. Basu, D. K.; Saxena, J. Envlron. Sci. Technol. 1978, 72, 795-798. Borneff, J.; Kunte, H. Arch. Mg.1964, 148, 585-597. Il’ntskli, A. P.; Rozhnova, L. 0.; Drozdova, T. V. Hyg. Senif. 1971, 36, 316-3 17. Hershberger, L. W.; Callis. J. 8.; Christlan, G. D. Anal. Chem. 1979, 51, 1444-1446.
RECEIVED for review June 3,1983. Accepted September 14, 1983. This work was supported in part by the Grant-in-Aid for Scientific Research (Grant No. 58740266)from the Ministry of Education of Japan and was presented at the 23rd Colloquium Spectroscopicum Internationale in Amsterdam, The Netherlands, June 28, 1983.
