Anal. Chem. 1907, 59, 965-970
965
Extended Luminescence Calibration Curves for Studying Surface Interactions in Solid-Surface Luminescence Analysis Gregory J. Burrell and Robert J. Hurtubise* Chemistry Department, University of Wyoming, Laramie, Wyoming 82071
Callbratlon curves extended well beyond the normal linear range for solld-surface fluorescence and solld-surface phosphorescence showed some unique characterlstlcs for benro[f ]quinoline adsorbed on a slllca chromatoplate under neutral and acldlc condltlons. The solid-surface fluorescence curves leveled off, whereas the solld-surface phosphorescence curves passed through a maxlmum and then decreased. The extended callbratlon curves comblned with fluorescence and phosphorescence spectra and phosphorescence llfetlmes for benr~l]qulnollnerevealed dlfferences In the fluorescence and phosphorescence phenomena. I t was determined that fluorescence could occur from molecules adsorbed on the surface and In multilayers of molecules while phosphorescence only occurred from molecules adsorbed on the surface of the chromatoplate and not In the multllayers.
Room-temperature solid-surface luminescence spectroscopy has found widespread utility as an analytical technique. Recent reviews (1, 2) and monographs (3, 4 ) discuss the theories, techniques, and applications of this approach. The majority of work in this area, thus far, has focused on the optimization of experimental conditions necessary to evoke maximum luminescence intensities and the development of an understanding of the solid-surface interactions leading to the luminescence phenomena (1,5-14). The present study considers the use of benzo[flquinoline (B[flQ) to investigate new interactions of the molecule with silica gel containing a polymeric binder. Ford and Hurtubise (7,8) originally showed that the polymeric binder was necessary for inducing strong room-temperature phosphorescence (RTP) from B[flQ adsorbed on silica gel chromatoplates. Mathematical models for solid matrices have been developed by various researchers to predict the luminescence intensity vs. amount of adsorbed compound. Goldman (15), starting with equations originally presented by Kubelka and Monk (16),derived expressions describing the reflected and transmitted fluorescence from molecules adsorbed on thinlayer chromatoplates. These expressions have been shown to predict the intensity vs. concentration behavior for roomtemperature solid-surface fluorescence systems by Hurtubise (17) and Prosek et al. (18). Zweidinger and Winefordner (19) have demonstrated the applicability of the Kubelka and Monk equations to low-temperature phosphorescence from solutions. The various models were based on such factors as absorption and scattering coefficients; however, they were not predicated on the nature or extent of surface-adsorbate interactions. Part of this work involved obtaining luminescence calibration curves well beyond the linear portions of the curves to obtain insights into adsorbate-surface interactions for the room-temperature fluorescence (RTF) and RTP of BMQ. In addition, several luminescence excitation and emission spectra were obtained for BMQ at moderate to high levels of adsorbed B[flQ. The use of physisorbed molecules, as opposed to chemically bound molecules, to probe the adsorption site characteristics
of the silica gel chromatoplate would give an understanding of interactions important for luminescence on this surface that cannot be obtained by chemically bound species. A relatively large body of information exists on the effects of adsorption of organic molecules on silica surfaces (20-22). Some of the luminescent properties of the probe molecules used in this work, BmQ and BmQH', have been detailed elsewhere (7-14, 23-27).
EXPERIMENTAL SECTION Reagents. 5,6-Benzoquinoline (Gold Label, Aldrich Chemical Co., Milwaukee, WI) and reagent grade hydrochloric acid were used as received. Ethanol was purified by distillation. Silica gel 60 aluminum backed thin-layer chromatography (TLC) chromatoplates (E. Merck) were developed three times with distilled ethanol to elute impurities to one end. Spectrometers. Room-temperature fluorescence and roomtemperature phosphorescence spectra and intensity data were obtained with a Perkin-Elmer LS-5 luminescence spectrometer equipped with a Model 3600 data station (Perkin-Elmer,Norwalk, CT). Excitation radiation was provided by a Xe flash lamp pulsed at the line frequency. Scan rate (240 nm min-'), response factor 2, ordinate scaling (auto range), mini-floppy disk spectral storage, and data manipulation were accomplished through the use of a PECFS applications program. The instrumental parameters of excitation slit width (10 nm), emission slit widths (5 nm for RTP, 3 nm for RTF), phosphorescence delay (3.0 ms), and gate time (7.0 ms) were manually selected. A Perkin-Elmer Model 660 printer was used to obtain hard copies of spectra. A Cary 2300 W/Vis/near-IR spectrophotometer equipped with a C a y diffuse reflectance attachment was used to obtain the diffuse reflectance spectra of B[flQ and BMQH' on the EM chromatoplate. Cell Holder. A cell holder was designed and constructed specifically for the LS-5, which permitted rapid sample position optimization. This device accommodated a 1-cm-square quartz cell into which a 2.0 X 1.4 cm section of TLC chromatoplate could be placed. The position of the quartz cell containing the chromatoplate section (spotted with adsorbate), could be vertically, horizontally, and torsionally adjusted manually to maximize luminescence signals. Procedures. Solutions containing various concentrations of BMQ were prepared in neutral and acidic (0.2 M HCl) ethanol. One-microliter aliquots of these solutions were spotted on small sections of the chromatoplate (resulting in 0.05-50 pg). Multiple 1-yL applications were required for spot concentrations exceeding 50 pg. The spots were located on the chromatoplate such that the cell could be raised or lowered vertically to obtain background spectra and intensities from blank areas of the chromatoplate while retaining the angle of incident radiation used for measurements of the spot. All samples were dried 20 min at 110 "C prior to luminescence experiments. After being dried, the chromatoplate section was placed in the sample holder and RTF and RTP intensities were maximized by positional optimization. Excitation and emission spectra were then obtained from the spot, and background spectrum was subtracted when necessary. Diffuse reflectance spectra were obtained for B[flQ and BMQH' after adsorbing 5-10 yg onto the EM chromatoplate from 10 yL of ethanol or 0.2 M HC1-ethanol mixture. The chromatoplate was positioned in the diffuse reflectance apparatus such that the adsorbed compound spot covered the sample port and then spectra were obtained.
0003-2700/87/0359-096580 1.50/0 0 1987 American Chemical Society
966
ANALYTICAL CHEMISTRY, VOL. 59, NO. 7, APRIL 1, 1987
a
0.00
2
1.00
2.00
3. M
4.00
5. 00
M P O 3 5 o 4 0 0 4 5 0 5 m a O B m
1.00
WAVELENGTH (NM)
LOG C ng B I C I O 1
Figure 1. log (RTF intensity)vs. log ng for neutral solutions of B[f]Q adsorbed on an EM chromatoplate. Each data point represents the average of three to five runs.
b
Lifetime Measurements. A computer program was written for the Perkin-Elmer instrumentation which controlled the collection and display of RTP intensity information as a function of time. Phosphorescencedecay was recorded after manual closure of the incident beam shutter. Phosphorescence lifetime ( T ~ was ) calculated as the inverse of the slope of a In (RTP intensity) vs. time plot.
RESULTS AND DISCUSSION Extended Graphs of Luminescence Intensity vs. Nanograms of Analyte. The linear range for a typical RTP calibration curve for BWQ adsorbed on a silica gel chromatoplate is about 0-125 ng, and a typical RTF calibration curve on the same surface is about 0-100 ng (7). In this work it was of interest to adsorb amounts of B[flQ in the linear range and well beyond the linear range to gain insights into the interactions of B[flQ with the silica gel chromatoplate in these regions. It is important to emphasize that the chromatoplates used in this work contained a salt of polyacrylic acid as a binder (8);thus interactions with the polymer and the silica gel had to be considered. As discussed earlier, Goldman (15) derived a set of differential equations to mathematically predict the reflected fluorescence behavior of compounds adsorbed on thin-layer chromatoplates. Hurtubise (17),using simplified versions of Goldman's equations, experimentally validated their utility in estimating linear calibration curve ranges and approximate points of first slope change. Prosek et al. (18) gave experimental evidence for the complete Goldman equations by measuring the fluorescence of ergot alkaloids out to 13 pg/spot. Similar work has not been carried out for solid-surface RTP calibration curves. However, theoretical calibration curves and equations for solution low-temperature phosphorimetry have been derived (19). The equations were obtained for clear glasses, snows, and densely cracked glasses. In general, there were good correlations between the theoretical and experimental curves. The various theoretical equations are very important in defining the useful linear range of a luminescent component; however, they do not describe the interactions of molecules with the solid phase at the molecular level. Figure 1shows the RTF graph for neutral solutions of BMQ adsorbed onto an EM silica gel chromatoplate. Neutral solutions of B[flQ adsorbed on the chromatoplate gave rather weak RTP signals, which were not useful analytically. Thus, RTP from neutral samples will not be discussed. It was shown previously that a salt of polyacrylic acid is present in the EM chromatoplates as a binder (8),and this was resubstantiated in this work by using the same infrared experiments employed earlier. The presence of this polymeric binder complicates a discussion of the surface interactions of the BMQ with the chromatoplates. For simplicity, it will be assumed that silanol
1 M
I
XT)
35o
400
450
5M
550
Bm
WAVELENGTH (NM)
Figure 2. (a) Solution fluorescence excitation and emission spectra of B [ f ] Q in ethanol (-), A,, = 270 nm and A,, = 385 nm, and B[f]Q in 0.1 M HCI ethanol (---), k,, = 280 nm and A,, = 425 nm. Spectra normalized to the strongest emitting species. (b) RTF spectra of 100 ngof B [ f ] Q (-), A,, = 343 nm and A,, = 410 nm, and 100 ng B[f]QH+ (---), k,, = 365 nm and A,, = 420 nm, on an EM
chromatoplate. Spectra normalized to the strongest emitting species. (Si-OH) and carboxylate groups (COO-) can interact with B[flQ under neutral conditions. Lloyd (27) has shown that B[flQ can undergo excited-state protonation in the singlet state through interaction with surface silanol groups in silica gel. Evidence for this in our work is given in Figure 2. In Figure 2a the corrected solution fluorescence excitation spectra of B[flQ and B[flQH' are compared. Clearly, major spectral differences are observed between the B [flQ and BWQH' species. Also in Figure 2a major differences between the solution fluorescence emissions from BMQ and BMQH+ are observed. The RTF graph in Figure 1 represents emission from BMQH' formed in the excited-state by the protonation of BMQ by acidic silanols. This is supported by the spectra in Figure 2. In Figure 2b, the corrected RTF excitation spectra for BWQ and B[flQH+ adsorbed on an EM chromatoplate are shown. By comparison of the neutral solution excitation spectrum (Figure ea) with the neutral solid-surface excitation spectrum (Figure 2b), it can be seen that the neutral form of BMQ is adsorbed on the solid surface when spotted from a neutral solution. This is further substantiated by the spectral shifts that occur for the excitation spectra of the acidic solution (Figure 2a) and solid-surface excitation spectra when spotted from an acidic solution (Figure 2b). These spectra are shifted to longer wavelengths. In addition, the solid-surface fluorescence emission spectra are practically identical for both neutral and acidic conditions on the EM chromatoplate (Figure 2b). Thus, when the neutral solution of B[flQ is spotted on the chromatoplate, excited-state protonation occurs because the excitation spectrum is that of the neutral form, but the emission spectrum is that of the protonated form. Other evidence for excited state protonation was provided by obtaining the diffuse reflectance spectra for B[fJQ and B-
ANALYTICAL CHEMISTRY, VOL. 59, NO. 7, APRIL 1, 1987
907
Table I. Changes in Luminescence Wavelengths as a Function of the Amount of B[f]Q or B[f]QH+ Adsorbed" 100 ng adsorbed on
1.0 gg adsorbed on
20 pg adsorbed on
chromatoplate
chromatoplate
chromatoplate
Xex,
BWQ, RTF BWQH', RTF BMQH+, RTP
nm
261, 272, 332, 344 279, 365 280, 366
Xem,
nm
414 419 485, 516
A,,
nm
Xem
260, 274, 334, 345 280, 365 285, 368, 376
nm
417 421 486, 517
Xex,
nm
Aem, nm
274, 334, 345, 362 s 277, 346, 363, 385 283, 365, 379
419 435 498, 517
'Reproducibility of wavelengths was fl nm. V]QH+ adsorbed on the chromatoplate. The diffuse reflectance spectra showed the same differences between the adsorbed BmQ and BwQH+ as were observed from the solution phase. The maximum wavelength for the diffuse reflectance spectrum for BMQ was 343 nm, and for BMQH' it was 365 nm. There was no indication from the diffuse reflectance spectra that BWQ was protonated in the ground state under neutral conditions. Similar diffuse reflectance spectral results were reported earlier for BMQ adsorbed on an EM chromatoplate by Ramasamy and Hurtubise (13). The RTF graph in Figure 1 approximates the theoretical graphs given by Goldman (15). He predicted that a graph of the fluorescence from a compound adsorbed on a chromatoplate as a function of the amount of compound would first be linear and then level off and be parallel to the concentration axis. The results in Figure 1 and Figure 2 are important analytically because the species emitting fluorescence is defined and experimental evidence is provided for Goldman's theory. Figure 3 shows graphs of the RTF and RTP relative intensities as a function of the amount of BV]QH+ for acidic solutions of B[LlQH+ adsorbed onto an EM silica gel chromatoplate. Both the RTF and RTP signal are relatively strong and both signals are useful analytically. The RTF and RTP curves become nonlinear at approximately the same point, but the RTF curve leveled off and was parallel to the nanogram axis. The RTP curve exhibited a decrease in RTP intensity with increasing amounts of B[flQH+. The maximum RTF intensity for BMQ adsorbed from an acidic solution onto the EM silica gel chromatoplate (Figure 3) is slightly greater than the maximum RTF intensity of BWQ adsorbed from a neutral solution (Figure 1). Additionally, in the rising portions of the RTF curves in Figures 1 and 3, the species from the acidic solution showed a greater sensitivity. The RTP intensity of BmQ adsorbed from an acidic solution onto the EM chromatoplate results from the protonation of the carboxylate groups of the salt of polyacrylic acid binder in the EM chromatoplates upon acid treatment (8). The carboxyl groups formed can interact strongly with B[LlQH+ and allow the RTP signal to be observed (8). The RTP curve in Figure 3 is particularly important in analytical RTP work because it shows that the overall shape of the RTP curve is very different than the RTF curve in Figure 3. A similar RTP curve was obtained previously for p-aminobenzoic acid adsorbed on sodium acetate (9). These results show that the Goldman equations (15) are not applicable over a wide range of adsorbed phosphor. This aspect is in need of future investigation. The RTF curve in Figure 3 for B[flQ under acidic conditions levels off at a lower amount (1pg) than does the RTF curve in Figure 1for BMQ adsorbed under neutral conditions (50 pg) indicating different interactions for the neutral and acidic conditions. For neutral conditions, the singlet B[flQ molecules interact primarily with silanol groups. However with acidic conditions both carboxyl groups and silanol groups are available for interaction with adsorbate (8). Because carboxyl groups are more acidic than silanol groups (28)they should interact more strongly with BmQH+ in the singlet state than the silanol groups. The flat portions of the RTF curves
/ am I .M
I
2.00
3.00
4.M
5.00
6.00
LOG C ng BCfIQH+ 1 b
-0.9
1.~0
2. m
I
3.00
4.M
5.00
6.00
LOG C ng BCflQH+ 3
Figure 3. (a) log (B[f]QH+RTF intensity) vs. log (ng of B[f]QH+) adsorbed on an EM chromatoplate. Each data point represents the average of three to five runs. (b) log (B[f]QH+RTP intensity) vs. log (ng of B[f]QH+)adsorbed on an EM chromatoplate. Each data point represents the average of three to five runs. in Figures 1and 3 show that the RTF signals in these regions are independent of the amount of adsorbate. However, the RTF intensity in Figure 1 does not become independent of the amount of BMQ until 50 pg as mentioned above. Table I shows changes in the RTF excitation and emission wavelengths for the neutral B[flQ as a function of the amount of adsorbed B[flQ. In general the excitation wavelengths are very similar except the 20-pg sample shows a shoulder at 362 nm. Also, the 20-pg sample shows a 5-nm red shift for RTF compared to the RTF of the 100-ngsample. The general shape of the curve in Figure 1 and the shift of A,, to longer wavelengths at 20 pg indicate that multilayer formation is occurring for neutral BWQ at higher amounts of B[flQ. As mentioned, the RTF intensity of B[flQH+ in Figure 3 becomes independent of the amount of B[flQH+ near 1 pg. Table I shows that the 100 ng and 1pg BmQH+ samples have almost the same excitation and emission wavelengths. However, the 20-pg sample shows new bands in the excitation spectrum and the emission spectrum has been shifted 16 nm compared to that of the 100-ng sample. The wavelength results in Table I for the 20-pg BMQH' sample indicate that multilayer formation is occurring in the plateau region of the RTF plot in Figure 3. However, multilayer formation occurs in a different fashion for B[LlQH+compared to that for BMQ because both carboxyl groups and silanol groups can interact with B[LlQH+. Also, the shapes of the RTF curves in Figures 1 and 3 are different, indicating different interactions.
966
ANALYTICAL CHEMISTRY, VOL. 59, NO. 7, APRIL 1, 1987
Table 11. RTP Lifetimes for Different Amounts of B[f]QH' Adsorbed on a Chromatoplate
a
B[flQHt, pg
phosphorescence lifetime," s
av corr coeffb
0.10 1.0 5.0 20 100
1.4 f 0.04 1.3 f 0.1 1.4 f 0.04 1.1 f 0.1 0.9 f 0.1
0.997 0.993 0.995 0.988 0.970
Reproducibility at the 95% confidence level. Average values obtained from at least five determinations for each amount. Linear correlation coefficients for graphs of In (intensity)vs. time. The RTP intensity for BMQH+ begins to decrease above lo00 ng in Figure 3. This is almost at the same point at which the RTF curve for B[flQH+ levels off. The decrease in the RTP curve at 1000 ng is most likely due to two major effects. Because the RTF wavelength results in Table I indicated multilayer formation and the RTF plateau begins at approximately the point that the RTP curve starts to decrease, then both the inner-filter effect and collisional deactivation can be occurring the triplet-state molecules. For example, it was shown earlier that for strong RTP to be observed from the chromatoplate B[flQH+ has to interact directly with the carboxyl groups of the binder in the chromatoplate (8). B[flQH+ molecules in the multilayer would not give R T P because the interactions would not be strong enough and the B[flQH+ molecules would not interact directly with the carboxyl groups. Thus, the multilayers of molecules act as an inner filter and prevent maximum excitation of the BMQH+ molecules interacting directly with the carboxyl groups in the chromatoplate. In addition, collisional deactivation most likely occurs because several B[flQH+ molecules can compete for an adsorption site a t the higher amounts of B[flQH+. The collisional deactivation would be favored by the fact that the experiments were carried out at room temperature. In order to gain more insight into the interactions responsible for the RTP of B[flQH+ in Figure 3, phosphorescence lifetime (T,) values were obtained at several different amounts of BMQH+ as indicated in Table 11. The 0.1-, 1.0-, and 5.0-pg samples showed good linearity for the In (intensity) vs. time plots. The 20-pg and 100-pg samples showed curvature for the In (intensity) vs. time plots with the 100-pg sample showing the greatest curvature. The curvature in these graphs indicated nonexponential decay of the RTP. It can be seen in Table I1 that the RTP lifetime is approximately constant from 0.10 to 5.0 pg. At 20 pg and 100 pg the T~ values decreased. Because the 7, value at 5 pg is about the same as the T, values at 0.10 and 1.0 pg, but the R T P intensity was decreasing as a function of the amount of BV]QH+ (Figure 3) near 5 pg, then multilayers were forming and the inner-filter effect was most likely operative. This can be seen by considering eq 1where 1 7, = (1)
k , + k , + k,[qI
k , is the rate constant for phosphorescence, k, is the rate constant for a radiationless transition, k, is the rate constant for bimolecular quenching such as with oxygen, and [ q ] is the concentration of the quencher. For T, to remain constant, most likely K,, k,, and k , [ q ] are constant a t 0.10, 1.0, and 5 pg. It has been proposed by Niday and Seybold (29) that k, is a measure of the rigidly held mechanism for R T P and k, is a measure of the effect of oxygen on RTP or how efficiently the matrix protects the phosphor from oxygen. Because the rate constants apparently do not vary in eq 1 at 0.10, 1.0, and 5.0 pg the decrease in RTP intensity must be due to an inner-filter effect, whereby the layer or layers of molecules above the molecules interacting directly with the carboxyl groups are prevented from being excited as efficiently compared to
WAVELENGTH (NM) I
b
WAVELENGTH (NM)
Figure 4. (a) B [ f ] Q RTF spectra for 100 ng adsorbed on the EM chromatoplate; ,A, = 270, 300, 330 and 344 nm; A,, = 390, 410, and 440 nm. Spectra normalized to the strongest emitting species. (b) B [ f ] Q RTF spectra for 20 pg adsorbed on the EM chromatoplate; A,, = 270, 300, 330, and 344 nm; A,, = 390, 410, and 440 nm. Spectra were normalized to the strongest emitting species. The excitation spectra correspond to emission wavelengths of 420,440, and 390 nm, respectively, in decreasing order of relative intensity at 346 nm.
lower amounts of B[LlQH+. As Table I shows, the 20-pg and 100-pug samples show a decrease in lifetime, and Figure 3 shows that a t these amounts the RTP intensity of B[flQH+ is very low. The decrease in the RTP lifetime is most likely a combination of the inner-filter effect and collisional deactivation. Additional experiments would be needed to sort out the extent of the changes in the rate constants in eq 1 under these conditions. The amazing aspect of the T~ values for the 20-pg and 100-pg samples is the magnitude of the values. Even though the RTP intensities are very low a t these amounts, the T~ values are relatively high. The high T~ values indicate that the molecules yielding RTP are interacting strongly with the surface carboxyl groups. RTF Spectra on Silica Gel Chromatoplates from Neutral Solutions. The wavelength dependency of luminescence spectra has been related to the microenvironment of a fluorescent molecule and the heterogeneity of adsorbate site interactions (30). Changing the emission wavelength and recording the excitation spectra yield information on ground-state species while changing the excitation wavelength used to generate emission spectra provides some insights into excited-state species environments and interactions. In Figure 4, parts a and b, RTF excitation and emission spectra recorded with various excitation and emission wavelengths are shown for 100-ng and 20-pg B[flQ, respectively. For the 100-ng sample in Figure 4a little change in either the excitation or emission spectra occurred by changing the emission and excitation wavelengths for the corresponding excitation and emission spectra. This indicated that the ground-state species were experiencing similar microenvironments. In addition, the fluorophores in the excited-state
ANALYTICAL CHEMISTRY, VOL. 59,
NO.7, APRIL 1, 1987 969
I
a
WAVELENGTH (NM)
WAVELENGTH (NM)
m
WAVELENGTH (NM) Figure 5. (a) B[f]QH+ RTF spectra for 100 ng adsorbed on the EM chromatoplate: A,, = 280, 300, 340, 365, and 385 nm; A,, = 405, 420, and 450 nm. Spectra normalized to the strongest emitting species. The excitation spectra correspond to emission wavelengths of 420, 450, and 405 nm, respectively, in decreasing order of relative intens@ at 265 nm. (b) B[f]QH+ RTF spectra for 20 pg adsorbed on the EM chromatoplate: A,, = 280, 330, 365, and 390 nm; A,, = 405, 430, 450, and 480 nm. Spectra normalized to the strongest emitting species. The excitation spectra correspond to emission wavelengths of 480, 450, 430, and 405 nm, respectively, in decreasing order of relative intensity at 379 nm.
were interacting in a similar fashion under different excitation conditions and thus experienced similar microenvironments in the excited state. Bauer et al. (31) carried out similar experiments for naphthalene adsorbed on silica gel that was heated for 4 h at 800 "C. They found that the set of excitation and emission spectra were very different compared to each other and concluded that the surface was very inhomogeneous and the naphthalene molecules found themselves in many different environments. For the 20-pg sample of BMQ, the RTF spectral changes in the excitation spectra indicated somewhat differing ground-state interactions and environments for adsorbed BWQ (Figure 4b), which would be expected for multilayer formation. Neither the 100-ng BMQ RTF emission spectra (Figure 4a), nor the 20-pg B[flQ RTF emission spectra (Figure 4b) were affected greatly by the wavelength of exciting radiation indicating a homogenous excited-state environment for a given amount of B[flQ. It should be recalled that the 20-pg-sample spectra showed a 5-nm shift compared to that of the 100-ng sample in Table I; thus the environments experienced by the 100-ng and 20-pg samples are not the same. RTF Spectra on the Silica Gel Chromatoplate from Acidic Solutions. The RTF excitation spectral wavelengths for the 100-ng BWQH+ sample were found to be independent of the emission wavelength (Figure 5a). However, the band at 365 nm was altered somewhat in magnitude with variation in the emission wavelength suggesting some nonhomogeneity in the ground-state environment. The emission spectra in Figure 5a were independent of the excitation wavelength. These results indicate a relatively homogeneous environment
~
~
4
m
4
5
0
5
w
~
WAVELENGTH (NM) Flgure 6. (a) B[f]QH+ RTP spectra for 100 ng adsorbed on the EM chromatoplate: A,, = 280, 320, 370, and 390 nm; A,, = 516 and 550 nm. Spectra normalized to the strongest emitting species. (b) B[f]QH+ RTP spectra for 20 pg adsorbed on the EM chromatoplate: ,A, = 280, 320, 370, and 390 nm; A,, = 516 and 550 nm. Spectra normalized to the strongest emitting species.
for the excited state of the 100-ng sample. At 20 pg of BLflQH', the excitation spectra showed an emission wavelength dependency which indicated multiple fluorophore ground-state environments and interactions existed (Figure 5b). The RTF emission spectra for the 20-pg sample of B[fJQH+were essentially wavelength independent indicating a homogeneous excited-state environment. However, the 16-nm red shift for the 20-pg sample compared to that of the the 100-ng sample indicated that different environments exist for the two samples (Table I). RTP Spectra on the Silica Gel Chromatoplate from Acidic Solutions. Table I shows the RTP excitation and emission wavelengths for BMQ adsorbed from acidic solutions onto an EM chromatoplate. A t 1.0 and 20 pg the excitation spectra had practically the same excitation wavelengths (Table I). However, the 100-ng sample did not have an excitation band near 376 nm as did the 1.0-pg sample. The RTF and RTP excitation wavelengths for the 100-ng B[flQH+ samples in Table I are the same within experimental error. Comparing the RTF and RTP excitation wavelengths for the 1.0-pg samples in Table I shows that the wavelengths are not the same. This is seen more dramatically for the 20-pg of BLflQH+ RTF and RTP samples in Table I. The difference in the excitation wavelengths for these samples is most likely a result of the singlet and triplet emitting molecules being in different environments. The singlet emitting molecules would fluorescence from the silica gel matrix and in the multilayers, whereas the triplet emitting molecules would phosphoresce only from the silica matrix because direct interaction with carboxyl groups is needed for RTP. Table I shows that the RTP emission wavelengths are the same for the 100-ng, 1.0-118, and 20-pg BWQH+ samples, except the band at 485 nm was shifted to 498 nm for the 20-pg sample. These results show that the B[flQH+ in the triplet state is interacting with the
m
970
Anal. Chem. 1987,59,970-973
carboxyl groups in the silica gel matrix [8]because of the similarity of the R T P emission wavelengths. This result is in contrast to the RTF results for BMQ and BwQH+ which shifted to longer wavelengths as the amount of analyte increased (Table I). The B[flQH+ R T P excitation spectra, while exhibiting changes with the amount of B[LlQH+,were found to be almost independent of emission wavelength monitored for a given amount (Figure 6a,b). The excitation spectra for the 20-pg sample were not as smooth as the 100-ng sample and the relative intensities of the bands at 280 and 366 nm showed different relative magnitudes for the 100-ng and 20-pg samples. Thus, the phosphorescent B[flQH+ molecules at a given amount were experiencing similar ground-state interactions and environments but not necessarily the same environments. The BMQH+ R T P emission spectra were found to be essentially independent of the wavelength of the exciting radiation a t all amounts (Figure 6a,b). The RTP emission spectra in Figure 6b for the 2 0 - ~ gsample did show greater relative intensities changes with excitation wavelength compared to that of the 100-ng sample. This is probably due to the weak RTP signals at 20 pg of B[flQH+. Because of the similarity of the maximum emission wavelengths, one can conclude that the B[flQHf molecules were interacting with carboxyl groups. This is supported by the results in Table I, which show that the RTP emission wavelengths were independent of the amount of B[flQH+, except for one wavelength. Registry No. Benzo[flquinoline, 85-02-9.
LITERATURE CITED Parker, R. T.; Freedlander, R. S.;Dunlap. R. B. Anal. Chim. Acta 1980, 779, 189. Parker, R. T.; Freedlander, R. S.;Dunlap, R. B. Anal. Chim. Acta 1980, 720, 1. Vo-Dlnh, T. Room Temperature Phosphorimetry for Chemical Analy sis; Wiley: New York, 1984.
Hurtubise, R. J. Solid Surface Luminescence Analysis ; Marcel Dekker: New York, 1981. Schulman, E. M.; Parker, R. T. J . Phys. Chem. 1977, 8 7 , 1932. Ford, C . D.; Hurtubise, R. J. Anal. Chem. 1978, 5 0 , 610. Ford, C . D.; Hurtubise, R. J. Anal. Chem. 1979, 51, 659. Ford, C.D.; Hurtubise, R. J. Anal. Chem. 1980, 5 2 , 656. Von Wandruszka, R. M. A.; Hurtubise, R. J. Anal. Chem. 1977, 49, 2764. Hurtubise, R. J. Talanta 1981, 28, 145. Ramasamy, S. M.; Hurtubise, R. J. Anal. Chem. 1982, 5 4 , 1642. Ramasamy, S.M.; Hurtubise, R. J. Anal. Chem. 1982, 54, 2477. Ramasamy, S . M.; Hurtubise, R. J. Anal. Chim. Acta 1983, 752, 83. Senthilnathan, V. P.; Ramasamy, S. M.; Hurtubise, R. J. Anal. Chim. Acta 1984, 757, 203. Goldman, J. J. Chromatogr. 1973, 78, 7. Kortum. G. Reflectance Spectroscopy; Springer-Verlag: New York, 1969. Hurtubise, R. J. Anal. Chem. 1977, 49, 2160. Prosek, M.; Kucan, E.; Katic, M.; Bano, M.; Medja, A. Chromatographia I978, 1 1 , 578. Zweidinger, R.; Winefordner, J. D. Anal. Chem. 1970, 4 2 , 639. Kiseiev, A. V.; Lygin, V. I. Infrared Spectra of Surface Compounds; Wiley: New York, 1975. Robin, M.; Trueblood, K. N. J . Am. Chem. SOC. 1957, 79, 5138. Leermakers, P. A.; Nicholls, C. H. Advances in Photochemistry;WileyInterscience: New York, 1971. Nakamizo, M. Specfrochlm. Acta 1966, 2 2 , 2039. Masetti, F.; Mazzucato, U.; Birks, J. B. Chem. Phys. 1975, 9 , 301. Favaro, G.;Masetti, F.; Mazzucato, U. Spectrochim. Acta, Part A 1971, 27A, 915. Mataga, N.; Kaifu, Y.; Koizumi, M. Bull. Chem. SOC.Jpn. 1956, 2 9 , 33. Lloyd, J. B. F. Anahst (London) 1975, 700, 529. Thin Layer Chromatography; Stahl, E. Ed.; Springer-Verlag: New York, 1969; p 13. Niday, G. L.; Seybold, P. G. Anal. Chem. 1978, 5 0 , 1577. Bauer, R. K.; de Mayo, P.; Natarajan, L. V.; Ware, W. R. Can. J , Chem. 1984, 62, 1279. Bauer, R. K.; de Mayo, P.; Ware, W. R.; Wu, K. C. J . Phys. Chem. 1982, 86, 3781.
RECEIVED for review April 28, 1986. Resubmitted November 17, 1986. Accepted December 1,1986. Financial support for this project was provided by the Department of Energy, Division of Basic Energy Sciences, Contract No. DE-ACO280ER10624 and DE-FG02-86ER13547.
Direct Determination of Polycyclic Aromatic Hydrocarbons in Extracts of Particulate Matter A. E. Elsaid, A. P. D’Silva,* V. A. Fassel, and R. L. M. Dobson Ames Laboratory-USDOE
and Department of Chemistry, Iowa State University, Ames, Iowa 50011
Laser-exclted Shpd’skll spectrometry has been d z e d for the dlrect determlnatlon of polycycllc aromatic hydrocarbons (PAH) in high-temperature extracts of partlculate matter. Ouantltative data on seven PAH present in two National Bureau of Standards standard reference materials (SRM 1649 and 1650) are reported.
The combustion of fossil fuels and biomass, be it in utility power plants, trucks, and automobiles or in home fireplaces, usually introduces finely divided particulate matter into the environment. The large surface area of these particles results in the preferential adsorption and deposition of highly carcinogenic or mutagenic polynuclear organic materials (POM) that are usually present in the particulate laden effluent streams. One class of compounds of particular interest is the polynuclear aromatic hydrocarbons (PAH). Because this class 0003-2700/87/0359-0970$0 1.50/0
of compounds contains many mutagenic and carcinogenic compounds, there is serious concern about the potential environmental impact resulting from the widespread dispersal of this particulate matter. This concern is intensified by the realization that adsorption of the PAH on the surface of the particulate matter may enhance carcinogenic and mutagenic potency (1). As a consequence of these considerations, extensive studies designed to obtain PAH profiles of urban air particulate matter have been performed (2). Generally, the profile data were obtained via the extraction of the PAH into appropriate low boiling solvents (3-5), followed by chromatographic isolation of the compounds of interest. Quantitation has usually been performed via mass spectroscopic or ambient-temperature, luminescence approaches. In a recent communication, we enumerated the deficiencies inherent in these approaches, and recommended an overall procedure that satisfies the key criteria desired in an analytical method for this purpose (6). This procedure was based on 0 1987 American Chemical Society