5995
J. Phys. Chem. 1993,97, 5995-6001
Photophysics of Polycyclic Aromatic Hydrocarbons Adsorbed on Silica Gel Surfaces. 3. Fluorescence Quantum Yields and Radiative Decay Rate Constants Derived from Lifetime Distributionsf Yuan S. Liu, Paul de Mayo, and William R. Ware’ Photochemistry Unit, Department of Chemistry, The University of Western Ontario, London, Ontario, Canada N6A 5B7 Received: December 28, 1992
The fluorescence quantum yields of six polycyclic aromatic hydrocarbons (PAHs) adsorbed on both hydroxylated and highly dehydroxylated silica gel surfaces have been measured, using a diffuse reflectance method, an integrating sphere method, and a secondary method based on diffuse reflectance. Consistent results were obtained from these different procedures. Radiative and nonradiative decay rate constants of these systems were derived from these quantum yield data together with the lifetime distribution data obtained previously. These decay rate data, when compared with those obtained for the same PAHs in fluid media, appear reasonable.
Introduction In a homogeneous system, the decay mechanism of excited molecules is routinely studied by measuring fluorescence lifetimes and quantum yields. Then, both the radiative and nonradiative decay rate constants can be calculated, using where 4 is the fluorescence quantum yield, k is the total decay rate constant (the reciprocal of the fluorescence lifetime), and ~ F isM the radiative decay rate constant. When the molecules under investigation are adsorbed on a silica gel surface, however, this type of fundamental study becomes much more complicated. In addition to the problems present in the lifetime analysis discussed elsewhere,’ difficulties also arise in fluorescence quantum yield measurements for these systems and in the derivation of radiative and nonradiative decay rate constants, which must be based on lifetime distributions. In addition, for adsorbed molecules, an accurate measurement of their extinction coefficient as a function of wavelength is extremely difficult. In the last decade, although numerous photophysical studies have been conducted on fluorescent molecules adsorbed on surfaces, only a small number of these studies has been devoted to the detailed examination of the decay mechanism. For those compounds whose transitions to the lowest excited singlet states are weak, a fluorescence quantum yield measurement, in conjunction with a lifetime measurement, is the only simple way to obtain the radiative and nonradiative decay rate constants. Therefore, the determination of fluorescencequantum yields for these systems is of great interest. Oelkrug and co-workers24 have studied this problem since the late 1970s. They have measured the luminescencequantum yields of a number of aromatic molecules adsorbed on preheated alumina surfaces and derived the corresponding radiative and nonradiative decay rate constants. Originally, in these studies, a primary (absolute) method2Jwas used; in a later work? however, quantum yields were determined relative to the value for phenanthrene physisorbed on an alumina surface; the latter was assumed to be equal to the value of that in solution. The fluorescence lifetime was assumed to be a single exponential to permit the calculation of the radiative decay rate constant, although such an assumption should be considered questionable. In the case where the medium can be assumed to possess an infinite scattering power, the absolute fluorescencequantum yield Publication No. 489 from the Photochemistry Unit, University of Western Ontario. +
can be measured by monitoring the diffusely reflected excitation light and the diffusely emitted fluorescence light. Wrighton et al.5 have determined the absolute emission quantum yields of some powdered samples, using a conventional scanning emission spectrophotometer. They assumed that, for their powdered sample in a cuvette, the angular distributions of diffusely reflected light and of emitted light were the same and that Lambert’s cosine law was obeyed. Hence, the same fraction of emitted light and of reflected light was supposed to be collected by the detection system. The fluorescence quantum yield could thus be determined by measuring the so-called apparent quantum yield, defined by
if io-i
9,, = where jf is the fluorescence intensity, j o is the back-scattered intensity of excitation light from a blank, and j is that from a loaded sample. A more difficultcase is that in which a semitransparent medium of finite scattering power, such as silica gel powder, is involved. In such a medium, the incident light and the emitted fluorescence light are “leaking” in all possible directions, and the distribution of the leaking light along the surface of the bulk sample is a complicated function of the optical properties of the sample. Special measures are therefore required for such cases. A remarkable effort to estimate fluorescence quantum yields in such a turbid system has been made by Seely.6-8 He described a sophisticated approach that involved measuring an “apparent quantum yield”, similar to what discussed by Wrighton and coworkers, and then calculating a correction factor based on the Kubelka and Munk (K & M) theory9-ll of light transport in scattering media, in order to deduce the true quantum yield. In this approach, it is assumed that the medium is an infinitely broad plate with a finite thickness and is evenly illuminated from one side with ideally diffused light. These assumed conditions, however, cannot in practice be obtained when real fluorescence instrumentationand real samples are employed. Seely’s treatment corrects, in fact, the error of the theoretical apparent quantum yield against an idealcase, the K & M model, but the discrepancy between the experimentally observed “apparent quantum yield” and the theoretical apparent quantum yield still remains unknown.12
A Primary Method Based on Diffuse Reflectance: Experimental Technique and Results We have used three methods to determine the fluorescence quantum yield of silica gel samples. The first method was,
0022-3654/93/2097-5995~Q4.QQ/Q0 1993 American Chemical Society
Liu et al.
5996 The Journal of Physical Chemistry, Vol. 97, No. 22, 1993
mirror scattered light from the blank corrected by I/b
C I
I
emission spectrum of pyrene XlO
300 Figure 1. Schematic diagram of a sample cuvette with a backup‘mirror, showing the diffuse reflectance of incident light from the sample.
light s o u r c e
400 450 WAVELENGTH/n m
350
Figure 3. Spectra illustrating an example of fluorescence quantum yield measurements with the primary method. The fluorescence quantum yield was determined by the ratio of the area under the emission spectrum to the difference of the integrated areas of scattered light from the blank and the sample.
i r------
TABLE I: Cuvette Orientation Dependence of the Measured Apparent Quantum Yield of Pyrene on Kieselgel-60 (35-70 mesh), 0.16 mg/g
-------,
.-____--_---__-4
diffuser
screen
500
Y
Figure 2. Schematic diagram showing the experimental setup for fluorescencequantum yield measurements with the primary method based on diffuse reflectance, using a mirror-backed sample cuvette.
stimulated by the methodologies of Seely and Wrighton, to compensate experimentally for the discrepancy between a real silica gel sample and an ideal scattering material. As shown in Figure 1, a U-shaped mirror was used to back up the sample cuvette. If the front surface of the cuvette is evenly illuminated by excitation light and the mirror is assumed to reflect all light back into the sample, in such a case the front surface of the cuvette is artificially made the only opening for light input and output. On the other hand, silica gel is known not to absorb light of wavelength longer than 260 nm; thus, the majority of the incident photons above this wavelength should travel into the medium for a distance, be scattered, be reflected, and be refracted many times, until they eventually escape from the front surface. In this process, the directions of movements of these photons are completely randomized. This means that the bulk silica gel powder, under the above-described conditions, can be considered as an ideal diffuser which transforms the collimated excitation light into diffuse light, except for that portion absorbed by the fluorescent molecules in the medium. A part of the absorbed light will be emitted as fluorescence, which, ideally, is also diffused. Therefore, the measured apparent fluorescence quantum yield should approximate the true value as it does in Wrighton’s experiments. The experimental setup is shown in Figure 2. A Perkin-Elmer 650-40 fluorescence spectrophotometer was employed. The sample compartment was equipped with a device that held a samplecuvette in a desired orientation. An (1 X 10)-mm cuvette was backed with the U-shaped mirror made of polished aluminum. In front of the detection monochromator, there was an attenuator to adjust the light intensity to be measured so that both the scattered and emitted light intensities would be in the proper working range of the detector. For most of the measurements,
angle, deg
blank calibrated (method A)
selfcalibrated (method B)
diffuser used (method C)
20 30 60 70 av
0.36 0.34 0.25 0.23 0.30 f 0.07
0.30 0.335 0.29 0.26 0.30 f 0.03
0.465 0.45 0.39 0.40 0.43 f 0.04
the excitation light beam was aimed a t the sample cuvette at an incident angle of 30’ from the normal. The directly illuminated area was about 16 mm2. Two matched cuvettes, one filled with an unloaded silica gel sample and the other with a polycyclic aromatic hydrocarbon (PAH)-loaded silica gel sample, were prepared. Details of the preparation of the samples have been described elsewhere.’ In this paper, only two types of silica gel pretreatments are involved. We refer to “wet” or “hydroxylated” silica gel as that pretreated at 25 OC in vacuum, while “dry” or “highly dehydroxylated” silica gel is that preheated at 800 OC in vacuum for 8 h. The spectrophotometerwas run in the “corrected spectra”mode. Five measurements were needed for the calculations, as described below. (1) Integrated scattered light from the blank. The excitation monochromator was fixed at the selected excitation wavelength X, while the detection monochromator was scanned over a range of A. JO represents the area under this curve. (2) Scattered light intensity, lo, obtained from the blank at a long wavelength, usually 580 nm, where the adsorbed molecules did not absorb. (3) Integrated scattered light from the sample, J, obtained at the excitation wavelength in the same way as Jo. (4) Scattered light intensity, I , from the sample at the same wavelength as l o . (5) Integrated area under the emission spectrum from the sample, Jf. The apparent quantum yield was then calculated, based on eq 2, according to Jf
(3) = Jo(Z/Zo)- J here the factor I l l o was introduced to correct for the error due to the discrepancy of sizes or positions of the twocuvettes. Figure 3 shows the procedures for measuring the absolute quantum yield of pyrene as a representative example. Table I presents the results of repeated measurements of a pyrene sample. The silica gel was not heated (so called “wet”)
Photophysics of Polycyclic Hydrocarbons
The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 5991
TABLE 11: Fluorescence Quantum Yield Measured by the Diffuse Reflectance Method compd
quantum yield
chrysene perylene 1,12-benzoperylene coronene
0.1 1
light source
1
J
0.75 0.28 0.13
and was loaded with pyrene at 0.16 mg/g or 0.1 1% by area. The measurements were made with the cuvette in four different orientations. The angles represent the incident light with respect to the normal of the cuvette surface. The three columns contain three groups of data measured under various conditions. The data in the first column were obtained by the normal method (method A) described above; that data in the second column were obtained using a single sample cuvette (method B). The integrated scattered light, Jo,was measured in the same way as in step 1, but with the sample itself rather than a blank, and a t a wavelength where the adsorbed molecules did not absorb light, normally 580 nm. Since there was only one cuvette and it was not touched throughout the measurement, the factor Z/Zo in eq 3 was equal to unity. In this way, the possible errors due to cuvette position and size were eliminated, but precalibration then became critical. The data in the third column wereobtained with a quartzdiffuser placed in front of the sample, as shown in Figure 2, so that the incident light was partially diffused and the sample cuvette was illuminated more evenly (method C). Intuitively, the results obtained with a diffuser (method C) should be given more weight, because in this case the front surface of the backed cuvette is more evenly illuminated, which is a condition required by the original assumption. If, however, for the moment, we give all these numbers an equal weight, then the average fluorescence quantum yield of pyrene adsorbed on the wet surface is 0.34 f 0.08. The same level of error was reported by Wrighton and co-workers.5 Considering that even in the simplest case, the determination of the quantum yields of fluorophors made in solution, the experimental uncertainties from laboratory to laboratory run 10-20%, the data precision in the present work is satisfactory. Unfortunately, there is no way to estimate the accuracy of these data. Table I1 presents some data obtained for other PAHs adsorbed on the wet surfaces using method A.
A Primary Method Based on the Integrating Sphere: Experimental Technique and Results The main advantage of using an integrating sphere is that the irregularities in the angular distribution of the intensity of the scattered and of the fluorescence light can presumably be averaged out. Hence, the error from this source may be suppressed. In practice, however, the preparation and the use of an integrating sphere is a delicate and difficult technique, and to get reliable results with such a sphere is as difficult as by other methods. An integrating sphere technique was employed in this work to provide a comparison. These measurements were again conducted on the Perkin-Elmer fluorescence spectrophotometer but equipped with an integrating sphere. Figure 4 shows the experimental setup for this method. The integrating sphere was made from a flask approximately 80 mm in diameter. The fluorescence viewing window was about 6 mm in diameter and was off-center so that it viewed a portion of the internal wall of the sphere. The sphere was coated with MgO by burning magnesium ribbon in an O2stream a t the openings. The sample was sealed in a quartz or Pyrex glass tube and was installed on the short arm attached to the stopper. The excitation light came into the sphere through anothe window, also 6 mm in diameter, directly illuminating the sample tube. The orientation of the sample tube was such that light directly reflected from the sample tube to theviewing window was minimized.
"
integrating
sphere
Figure 4. Schematic diagram showing the experimental setup for fluorescence quantum yield measurements with the primary method based on an integrating sphere.
TABLE III: Fluorescence Quantum Yield Measured with the Integrating Sphere Method compd perylene pyrene pyrene pyrene pyrene pyrene
silica gel treatment wet wet 200 'C 400 "C 600 "C 800 "C
auantum vield
heated heated heated heated
0.74 0.37 0.25 0.17 0.096 0.08
The detection system of the fluorescence spectrophotometer was calibrated with a blank silica gel sample installed in the integrating sphere. The instrument was run in the "corrected spectra" mode. Five measurements, as described in the previous section, were needed for the calculations (eq 3). The results are listed in Table 111. The results for pyrene and perylene are reasonably close to thoseobtained from the diffuse reflectancemethod. Thissuggests that the error from an irregular spatial distribution of scattered light in the diffuse reflectance method may not be serious.
A Secondary Method Based on Diffuse Reflectance: Experimental Technique and Results Since confidence had been established in the results (especially for pyrene) obtained by the primary methods used, a secondary (relative) method was then used to measure the relative quantum yields of the six PAHs adsorbed on silica gel. The quantum yield of pyrene on the wet surface was taken to be 0.37, based on the value obtained from the integrating sphere method (Table 111). 1,12-Benzoperylene was measured relatively to pyrene. Chrysene was measured relatively to 1,12-benzoperylene. Coronene and perylene were measured relatively to 1,12-benzoperylene. Phenanthrene and naphthalene were measured relatively to chrysene. These arrangements were based on considerations of a proper overlapping of the absorption and emission spectra involved. The experimental setup is illustrated in Figure 5 . The calculation was based on eq 4, derived from eq 3, where Js and J, were the areas under the corrected fluorescence spectra JXZS
4, = &-
JSIX
(4)
of the standard and unknown samples, respectively. Z, and I, were the fractions of absorbed intensities at the common excitation wavelength. 4sand 4, were the quantum yields. Figure 6 shows the procedures for measuring the relative quantum yield of chrysene (unknown) against 1,12-benzoperylene (standard), as a representative example. The absorption spectra of the two samples were repeatedly measured, and then an appropriate excitation wavelength, 292 nm in this case, was
5998
Liu et al.
The Journal of Physical Chemistry, Vol. 97, No. 22, 1993
TABLE I V Fluorescence Quantum Yields Determined by the Secondary Method
light source
sample
detector
compd
on wet silica gel
on dry silica gel
pyrene 1,12-benzoperylene perylene coronene chrysene phenanthrene naphthalene
(0.37) 0.26 0.66 0.13 0.095 0.098 0.090
0.11 0.24 0.72 0.047 0.061 0.048 0.036
TABLE V Fluorescence Quantum Yields of PAHs in Ethanol Reported in the Literature fluorescence yield
compd
Figure 5. Schematic diagram showing the sample cuvette setup for the flat cuvette. absorption spectrum measurement, using a (1 X IO)-"
I
Q
h
pyrene 1,12-benzoperylene perylene coronene chrysene phenanthrene naphthalene
0.6527 0.8932 0.2330 0. 1729-31 0.1033 0.2129-31.34
0.7228-30 (not available) 0.943i
0.5331
0.1329-31 0.1233
0.1329 0.1935
We start from eq 5 , the definition of decay rate distribution,Ia Z ( t ) = Jabi(k)e-kr dk
where a 1 0 and b I=. Z ( t ) is the total fluorescence intensity as a function of the time elapsed after the excitation impulse. This definition implies that the fluorescence intensity of the above system may be written as a function of k and t , that is,
>. e UI
z
kz
260
292
360
460 b
i'(k,t) = i(k) e-kr (6) Integration of eq 6 over t yields the total fluorescence photon flux in the steady state as a function of k, designated as P(k), that is, the photon population emitted at a decay rate kin steady state:
P ( k ) = Jomi(k)e-ktd t = i ( k ) / k
(7)
On the other hand, we have
P ( k ) = k F M ( k ) m*(W
(8) where k ~ ~ (iskthe ) radiative decay rate constant as a function of k and m*(k) is the molecular population as a function of k. Therefore, the steady-state molecular population distribution may be expressed by
WAVELENGTHInm Figure 6. Spectra illustrating an example of fluorescence quantum yield measurements with the secondary method: (a) the measured absorption spectra of a chrysene sample and a benzoperylene sample, from which l J I x at 292 nm was determined; (b) the corrected emission spectra from these twosamples, at 292 nm excitation, from whichJ,/J, was determined.
selected, as shown in Figure 6a. Z, and Z, were the averages of the absorptions of the unknown and the standard, respectively, at the selected excitation wavelength. At this excitation wavelength, the corrected emission spectra were measured, as shown in Figure 6b. J , and J, were the average areas under these emission spectra. The relative quantum yield was then calculated from eq 4. The results from these measurements are presented in Table IV. These results may be compared with Table V whereliterature data for these PAHs in ethanol are listed.
(9) The overall quantum yield, which is normally measured under steady-state conditions,may be represented in terms of continuous functions by
4=
JabkFM(k) m*(k) dk
f k m * ( k ) dk
(10)
In the case where the molecular state is unique, this equation reduces to eq 1. Substituting the expression for m*(k),given in eq 9, into (lo), we obtain eq 1 1 , which relates the unknown function k ~ ~ (tok ) the measurable variables or distributions:
Derivation of the Radiative and Nonradiative Decay Rate Constants of Six PAHs Adsorbed on Silica Gel Since, for the systems studied in the present work, the lifetimes are represented by distributions, eq 1 is no longer valid for the derivation of a radiative decay rate constant. A basic equation which relates the lifetime distribution, the fluorescence quantum yield, and the radiative decay rate constant can be derived as follows.
where ~ ( k =) l/k, TO(^> = l / k ~ ~ ( kand ) , the bar means the average over i ( k ) . This can be considered as the basic equation for dispersed photophysics corresponding to eq 1 for nondispersed
Photophysics of Polycyclic Hydrocarbons
The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 5999
systems. The physical interpretation of this equation is that, when the “initial intensity distribution in k domain” (decay rate distribution) is available, the quantum yield is expressed as the ratio of the average 7(k) and average 7o(k), both weighted by the k distribution. Equation 11 may be rearranged to
L
e In the simple case when kFM(k) can be assumed to be a constant, we obtain
0
k/ns-’
0.5
Figure 7. Decay rate (k)distribution of pyrene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
This equation implies that, in the case when the dispersion is determined by the nonradiativedecay rate only, then the average lifetime, 7 (‘1 /k), weighted by the initial intensity distribution in k domain (decay rate distribution), should replace the 7 (or l/k)ineq 1. IfkFM(k)isnotaconstant,itispossible,butdifficult, to obtain an approximate numerical solution for the unknown function, kFM(k), in (12), by converting the integral equation into a system of linear equations.13 In previous publications,’ we have shown that the lifetimes of the PAHs studied in this work may be described by bimodal distributions. As a first approximation, we assume the kFM(k)’s within each peak are constant, designated as kFM( 1) and k ~ ~ ( 2 ) , respectively. Under these conditions, eq 12 may be rewritten as Integral(1) kFM(1)
+ Integral(2) -- Integral(3) 9 kFM(2)
(14)
L
e
Figure 8. Decay rate distribution of phenanthrene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
where Integral( 1) = Sci(k) dk Integral(2) = l b i ( k ) dk Integral(3) = Ja
bi(k)
7dk
c is the point on the k axis at which the two models or peaks are separated or the k value where the distribution function is the minimum between the two peaks. An example of the selection of a, b, and c is shown in Figure 9b. To test this approach, the lifetime distribution’ and the fluorescence quantum yield data obtained for the wet and dry surfaces were examined. The initial intensity distributions in k domain (decayratedistributions), required by eq 14,wereobtained by converting the corresponding lifetime distributions into the k domain. In Figures 7-12 are shown the decay rate distributions derived from dehydroxylation experiments where the silica gel was preheated in vacuum at 25 and 800 O C . ’ The fluorescence quantum yields were taken from the results obtained by the secondary method, except for pyrene. The calculated integrals and the corresponding quantum yields were substituted into eq 14. For example, the following two equations are obtained for pyrene on the wet and dry surfaces: at 25 OC 0.00745667’( 1)
+ 0.000241 37’(2)
= 1.458 32/0.37
at 800 OC 0.00061337°(1)
+ 0.0089830~~(2)= 0.293 16/0.084
Figure 9. Decay rate distribution of chrysene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
where @(1) = l/kFM(l) and +(2) = 1 / k ~ ~ ( are 2 ) the natural lifetimes of pyrene on the wet (long lifetime peak) and the dry (short lifetime peak) silica gel. The results of solving this system of equations are fl(1) = 517 ns and ~ ~ ( =2 353 ) ns. On the basis of this method of calculation, we have derived the radiative decay rate constants, the total decay rate constants, and the nonradiative decay rate constants for the six PAHs adsorbed on the wet and dry silica gel surfaces. The results are listed in Tables VI and VII. Also included is a comparison with data from the literature for these PAHs in various s01vents.I~ The results obtainedmay contain large errors, because of errors in the lifetime distribution analysis and the quantum yield measurement. Some common trends, however, are noticeable, and they may be interpreted in terms of our existing knowledge. These trends are the following. (1) These results show that the dramatic change of the total decay rate constants is mostly to be attributed to the nonradiative
6000 The Journal of Physical Chemistry, Vol. 97, No. 22, 1993
0:3 k/ns-1 Figure 10. Decay rate distribution of 1,12-benzoperylene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
0
k/ns-1
0.3
Figure 11. Decay rate distribution of coronene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
e
1 .o
0 k/ns-l
Figure 12. Decay rate distribution of perylene adsorbed on (a) the wet and (b) the dry silica gel surfaces.
TABLE VI: Comparison of the Radiative and Nonradiative Decav Rate Constants of Pvrene decav rate constants (lo6 s-l) environment cyclohexane 2-propanol ethanol acetonitrile dimethyl formamide silica gel (wet) silica gel (dry)
k
kFM
ki M
ref
2.20 2.33 2.59 2.06 2.68 2.94 3.57 5.1 46.1
1.45 1.51 1.68 1.36 1.93 2.06 2.50 1.99 2.80
0.75 0.82 0.91 0.7 0.75 0.88 1.07 3.11 43.3
14 15 15 14 15 15 15
decay pathway. This is consistent with the original assumption, madeineq 14,where kFM(k)isapproximatedbyaconstant kFM(n), n = 1 or 2, within each peak. (2) The effect of the wet and the dry surface on the radiative
Liu et al. and nonradiative decay rate constants of the PAHs may again be grouped in terms of the lowest excited singlet state of the PAH. This is similar to what has been noted in lifetime distribution analysis,' but here we are able to discuss radiative and nonradiative decay rate constants separately. (3) The radiative decay rate constants of ILb-type PAHs adsorbed on the wet surface are similar to those for the same compound in strongly polar solvents. On the dry surface, the radiativedecay rateconstants of thesePAHs becomemuch larger but are still of the same order of magnitude as those in polar solvents. Ware and Hara15have shown that the radiative decay rate constant of pyrene increases with the increase of solvent polarity. Our results are, therefore, quite reasonable, provided that we assume the silica gel surface is equivalent to a strongly polar "local environment". (4) The nonradiative decay rate constants of the ILb-type adsorbed PAHs are more sensitive to the nature of the surface, as compared with the correspondingradiativedecay rate constants. It can be seen that, on the wet silica gel surface, the nonradiative decay rate constants of these PAHs are generally higher than those in polar solvents,and on the dry surface, they further increase by about 1 order of magnitude. By comparison, in solvents, the nonradiative decay rate constant does not change very much from nonpolar to polar solvents. These decay rate data, therefore, suggest again that there should be a second reason, presumably a specific interaction between some surface groups and the PAH molecules, which is responsible for the dramatic increase of the nonradiative decay rate constant of these PAHs on the surface, especiallyon thedry surface. We suggest that this effect is caused by the formation of a surface complex between PAH molecules with isolated hydroxyl (silanol) groups that are dominant on the dry surface. Evidence is available in the literature that an isolated hydroxyl group can, indeed, complex with an aromatic molecule through its .rr system. Nakajima,16J7when observing the Ham effect of pyrene as a function of solvent polarity, noted that the intensity of the forbidden 0-0 band showed a relatively rapid increase on addition of a small amount of a polar solvent such as an alcohol to a nonpolar solution, suggestingthe importance of some specific interaction. Lianos and Georghiou18J9 observed, using IR spectroscopy, 1:1 complex formation between alcohols and pyrene and 1-methylpyrene, in a 0.02 M solution of alcohol in C C 4 . They found a much shortened lifetime component which was then assigned to the complex. In the last decade, the formation of complexes of aromatic molecules with molecules that have hydrogen-bondingcapabilities has been the subject of numerous experimental and theoretical studies. The existence of such complexes has been confirmed by X-ray diffraction,20microwave spectroscopy,21J2 matrix-isolationIR spectroscopy,23and ab initio quantum-chemical calculations.2"26 ( 5 ) The only exception to this behavior is that of perylene. We have discussed elsewhere' that this may be related to the nature of its lowest excited singlet state (as compared with those of the other five PAHs), the IL, state. Its radiative decay rate constant seems to decrease on silica gel surfaces, especially on dry surfaces. Its nonradiative decay rate constant on silica gel surfaces is higher than that in fluid media. It will be interesting to examine the behavior of more PAHs of this type to confirm this interpretation.
Discussion At present, it is very difficult to discuss the accuracy of the quantum yield data obtained in this work. Published quantum yield data for silica gel or similar systems are so rare that there is no commonly accepted "standard" for comparison. At present, however, a reasonable level of confidence has been established because these results, derived from different techniques, seem to be self-consistent. In conjunction with the results of lifetime distribution analysis, a plausible picture about the photophysical properties of PAHs adsorbed on silica gel surfaces has emerged.
Photophysics of Polycyclic Hydrocarbons
The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 6001
TABLE MI: Comparison of the Radiative and Nonradiative Decav Rate Constants of PAHs decay rate constants PAH phenanthrene
chrysene 1,12-benzoperylene coronene perylene
environment
k
kFM
kiM
cyclohexane heptane isobutanol cyclohexanol silica gel (wet) silica gel (dry) cyclohexane silica gel (wet) silica gel (dry) poly(methy1 methacrylate) silica gel (wet) silica gel (dry) poly(methy1 methacrylate) silica gel (wet) silica gel (dry) ethanol benzene silica gel (wet) silica gel (dry)
16.8 15.8 16.5 24.9 145.1 22.4 34.4 147.5 4.08 9.57 49.0 3.12 5.03 26.6 167 202 126 101
3.9 2.7 3.3 4.0 2.63 6.53 2.61 3.47 8.77 1.53 2.62 12.7 0.84 0.78 1.09 145 180 89 76.9
14.1 12.5 12.5 22.3 138.6 19.8 30.9 138.7 2.55 6.95 36.3 2.28 4.25 25.5 21.7 22.4 37 24.5
One might get a feeling that the quantum yields obtained in this work are systematically too low. For example, since the fluorescence lifetime of perylene is not sensitive to the surface, why is its quantum yield about 20%lower than that in solution? This doubt can be removed, at least partly, by looking at the relative fluorescence quantum yields at low temperatures published in another paper of this series.lb In that work, we observed an increase of the fluorescence quantum yield from 295 to 1 1 K by a factor of 1.2 for perylene on the dry surface, indicating that the "true" quantum yield of this PAH at room temperature cannot be higher than0.83. Theresultsfromdirect measurements(shown in the tables as 0.72) are, therefore, quite reasonable. If the same "upper-limit test" is applied to pyrene on the wet and dry surfaces, these low-temperature data suggest that the quantum yields of pyrene should not exceed 0.23 on the wet surface and not exceed 0.24 on the dry surface. Compared with these upper limits, the measured value of 0.1 1 for the dry surface seems to be reasonable,but the measured value of 0.37 for the wet surface is too high rather than too low! For the wet surface, however, such a comparisonmay not bevalid, because the optical properties of the wet silica gel may considerably change at low temperature. Obviously, further more detailed investigations and more delicate experimental designs are demanded in the future. Improved results may be obtained by a number of methods, such as use of improved mirrors or coatings, use of finer silica gel powders, use of more even illumination on the front surface of the sample cuvette and more refined correctionsfor reabsorption, etc.
Conclusion (1) Three techniques for fluorescence quantum yield measurements of molecules adsorbed on silica gel surfaces have been investigated. The results from all these approaches appear to be mutually consistent when applied to PAHson silica gel. Although we have been unable to estimate directly the accuracy of these results, they seem to be reasonable, compared with the corresponding literature values for the PAHs in solution. The deduced radiative and nonradiative decay rate constants are also quite reasonable,further enhancing our confidence in these results and techniques, as well as in the lifetime distribution analysis method employed. I (2) On a silica gel surface, the adsorbed PAHs appear to experience a strongly polar environment. This polar environment effect, however, is not sufficient to account for all the decay rate data, especially for the variation of the nonradiative decay rate constants. We have to conclude, therefore, that there is a specific interaction, presumably the formation of a surface complex between the PAH molecules and isolated surface hydroxyl groups,
ref 14 14 14 14 14 36 36 14 14
which causes a drastic increase of the nonradiative decay rate constant on the dry surface in the 'Lb-type PAHs.
Acknowledgment. The authors wish to acknowledge the financial support of the National Science and Engineering Research Council. References and Notes (1) (a) Liu, Y. S.; Ware, W. R. J. Phys. Chem., first of three papers in this issue. (b) Liu, Y. S.; de Mayo, P.; Ware, W. R. J . Phys. Chem., second of three papers in this issue. (2) Oelkrug, D.; Plauschinat, M.; Kessler, R. W. J . Luminesc. 1979, 18119.434. (3) Honnen, W.; Krablchler, G.; Uhl, S.; Oelkrug, D. J. Phys. Chem. 1983.87. 4872. (4) Plauschinat, M.; Honnen, W.; Krabichler, G.; Uhl, S.; Oelkrug, D. J . Mol. S t r u t . 1984, 115, 351. (5) Wrighton, M. S.;Ginley, D. S.;Morse, D. L. J . Phys. Chem. 1974, 78, 2229. (6) Seely, G. R.; Haggy, G. A. J. Phys. Chem. 1987, 91, 440. (7) Seely, G. R. Biophys. J . 1987, 52, 311. (8) Seely, G. R. J. Photochem. Phorobiol. 1988, AIS, 325. (9) Kubelka, P. J . Opt. SOC.Am. 1948, 38, 448. (10) Goldman, J.; Goodall, R. R. J . Chromatogr. 1968, 32, 4. (1 1) Goldman, J. J. Chromatogr. 1973, 78, 7. (12) The K & M model, on which Seely's approach is based, is called a two-flux model. The use of a moresophisticated six-flux model has also been investigated and is considered more precise in describing powdered samples. See: Gade, R.; Kaden, U. J . Chem. SOC.,Faraday Trans. 1 1990,86, 3707. (13) Liu, Y. S.Ph.D Thesis, The University of Western Ontario, 1991. (14) Birks, J. B. Photophysics of Aromatic Molecules; Wiley-Interscience: London, 1970. (15) Hara, K.; Ware, W. R. Chem. Phys. 1980,51, 61. (16) Nakajima, A. Specrrochim. Acra 1982, 38A, 693. (17) Nakajima, A. Specrrochim. Acra 1974, 30A, 860. (18) Lianos, P.; Georghiou, S. Phoiochem. Photobiol. 1979, 30, 355. (19) Lianos, P.; Georghiou, S. Phorochem. Phorobiol. 1979, 29, 843. (20) Voort, P. V. D.; D'Hamers, I. G.; Vansant, E. F. J . Chem. Soc., Famdav Trans. 1 1990.86. 3751. (2lf Read, W. G.; Campbell, E. J.; Henderson, G.; Flygare, W. H. J . Am. Chem. SOC.1981, 103, 7670. (22) Read, W . G.; Campbell, E. J.; Henderson, G. J. Chem. Phys. 1983, 78, 3501. (23) Engdahl, A.; Nelander, B. J. Phys. Chem. 1985,89, 2860. (24) Cheney, B. V.; Schulz, M. W.; Cheney, J.; Richards, W. G. J. Am. Chem. SOC.1988, 110,4195. (25) Bredas, J. L.; Street, G. B. J. Am. Chem. SOC.1988, 110, 7001. (26) Bredas, J. L.; Street, G. B. J . Chem. Phys. 1989, 90, 7291. (27) Parker, C. A.; Hatchard, C. G. Trans. Faraday SOC.1963,59, 284. (28) Stevens, B.; Thomaz, M. Chem. Phys. Lert. 1968, I , 535. (29) Parker, C. A.; Joyce, T. Trans. Faraday SOC.1966, 62, 2785. (30) Horrocks, A. R.; Wilkinson, F. Proc. R. SOC.London, Ser. 1968, A306, 257. (31) Dawson, W. R.; Windsor, M. W. J . Phys. Chem. 1968, 72, 3251. (32) Melhuish, W. H. J. Phys. Chem. 1961, 65, 229. (33) Weber, G.; Teale, F. W. J. Trans. Faraday SOC.1957, 53, 646. (34) Stevens, B.; Thomaz, M.Chem. Phys. Lert. 1968, I , 549. (35) Parker, C. A. Anal. Chem. 1962, 34, 50. (36) Derived from the data given by: Kropp, J. L.; Dawson, W. R. In Molecular Luminescence; Lim, E. C., Ed.; W. A. Benjamin: New York, 1969; p 39. ~
