6361
J. Phys. Chem. 1992, 96,6361-6367
Spectra and Angular Distributions of Fluorescence Emitted from Anthracene on a Silver Surface Shigeru Ohhima* and Yoshizumi Ishibashit Department of Biomolecular Science, Faculty of Science, Toho University, Funabashi, Chiba 274, Japan (Received: August 26, 1991; Zn Final Form: March 27, 1992)
The fluorescence spectra of anthracene molecules adsorbed on a silver substrate were measured as a function of substrate temperature, T. At lower T, the spectra consist of a broad structureless excimer fluoresscenceband; at higher T, a structured band due to monomer fluorescenceappears and grows with T to dominate the spectra. The angular distributions of the excimer fluorescence, arising from interference at the surface, were also obtained. By analyzing the data on the basis of a classical model, the orientations of the e x h e r s at several temperatures were determined. Moreover, the adsorption state of the molecules as a whole and its change with T were deduced from both the data of the spectra and the angular distributions: with increasing temperature, the molecules are rearranged from an amorphous state to a crystalline state, in which the molecules are oriented with their short axis parallel to the surface and their long axis almost perpendicular to it.
1. Introduction
In recent years, there has been a growing interest in the optical properties of excited molecules near surfaces.' This is because the measurements of luminescence from the molecules provide information about the photochemical processes in surface adlayers of a moleculemetal system. The behavior of molecules on a metal surface is particularly of interest, since energy and electrons are transferred from the excited molecules to the metal; they cause luminescence quenching, which becomes markedly strong with decreasing moleculemetal distance.'S2 On the other hand, the photochemical processes also depend on the adsorption state of the molecules on the metal surface, because they are dominated by both the moleculemolecule and moleculemetal interactions. Tro et al. investigated the disord e d r transition and energy transfer in phenanthrene adlayers on AI,O,( 1120) by using laser-induced fluorescence and electronic absorption spectroscopy and found that electronic energy transfer in the adlayer is extremely sensitive to the degree of order of the adlayer.' Such an effect of order on photochemical processes can be more enhanced on the metal surface due to a stronger molecule-metal interaction. So far, however, little is known about the relation of the adsorption state and the photochemical processes. As an initial step toward understanding the relation, we have investigated the adsorption state of the molecules on the metal surface by use of luminescence data. In the present study, an anthracene thin film was prepared on a silver substrate and the spectra and angular intensity distributions of the fluorescence from anthracene were measured as a function of substrate temperature. It has been demonstrated that the angular distribution of the fluorescence can be utilized to reveal the molecular orientation on metal s ~ r f a c e s . ~Anthracene .~ was chosen as a fluorescent molecule since it is of particular interest as one of the prototype aromatic molecules. In addition, the fluorescence spectra of anthracenein solid phases and their temperaturedependences have been reported by several authors.- Thus we have carried out our study by referring to their results, though there are some differences in the experimental conditions. To clarify the effect of the silver surface on the luminescence, a hexatriacontane film was also used as a spacer between the anthracene molecules and the silver surface. 2. Angular Patterns of the Fluorescence from a Molecule on a Metal
When a fluorescing molecule is placed on a metal, interference occurs between the reflected and unreflected parts of the emitted light wave (Figure la). The amplitude of the resultant wave, A(B), can be obtained by superposition of the direct and reflected 'Resent address: Janome Sewing, Ltd., Kyobashi, Chuo-ku, Tokyo 104, Japan.
0022-3654/92/2096-6361$03.00/0
waves from the excited molecule. Here the following assumptions are made: (1) the excited molecule is regarded as an electricdipole oscillator, (2) the metal surface is infinitely flat, and (3) the distance between the oscillating dipole and the metal surface is very short compared to the wavelength of the fluorescence. From assumption 3, the geometrical path difference between the two waves can be neglected. Therefore, the relative phase of the two waves depends only on the reflection phase shift. By taking these into account, A(0) can be calculated, and its square gives the intensity of the radiation in the 0 direction, Z(0). The intensity also depends on the orientation of the dipole axis related to the surface. Three cases can be distinguished, as shown in Figure lb.*O Case 1. The dipole oscillates perpendicular to the surface:
z(e)
= sin2 e (1
+ rIl2+ 2rllcos
(1)
Case 2. The dipole oscillates parallel to the surface and parallel to the plane of incidence: z(e) = cos2 e (1 rIl2+ 2rllCOS 611) (2)
+
Case 3. The dipole oscillates perpendicular to the plane of incidence: z(e) = 1 r L 2- 2r, cos 6, (3)
+
Here rlland r , are the reflectivities and bll and 6, the reflection phase shifts, for parallel and perpendicular polarizations to the plane of incidence, respectively, and they are functions of 0. The radiation distribution for any other case can be obtained by resolving the dipole into the components represented by these three cases. Typical angular distributions for the three cases are shown in Figure 2a for a dipole emitting 470-nm light on a silver surface. The case 1 intensity shows a maximum at about 60°, while both the case 2 and the case 3 intensities decrease monotonically with an increase in the detection angle. Case 3 can be experimentally distinguished from cases 1 and 2; only the case 3 pattern is observed with a polarizer whose polarization direction is set along the case 3 dipole direction (referred to as the "vertical" situation). On the other hand, cases 1 and 2 cannot be observed separately; with a polarizer rotated by 90° from the vertical situation (referred to as the "horizontal" one), both case 1 and 2 patterns contribute to the angular distribution. Therefore, superimposed patterns of the two are provided as shown in Figure 2b. The three curves are obtained by varying the ratio of case 1 to case 2, R; R = 0.05, 0.1, and 0.2 for solid, dashed, and dotted curves, respectively. If the dipole is oblique to the surface, the tilt angles are estimated to be 3O, 6O, and 14' for R = 0.05, 0.1, and 0.2, respectively. Assumption 2 might be very severe since such a flat surface is experimentally unobtainable. In fact, according to Sano et al., 0 1992 American Chemical Society
-
6362 The Journal of Physical Chemistry, Vol. 96, No. 15, 1992
Ohshima and Ishibashi
/m
excitation light
1
A
(a)
metal
-1- -----A emission
--
heater
(b)
ea)
case1 case2
H
case3
polarizer
v
F i p e 1. (a) An oscillating dipole near a metal surface. Direct and reflected waves interfere, causing an angular distribution of radiation. P denotes the plane of observation (see the text). (b) Three orientations of the radiating dipole. Case 1: dipole axis perpendicular to the surface. Case 2: dipole axis parallel to the surface and the P plane. Case 3: dipole axis parallel to the surface and perpendicular to the P plane. The setup of the polarizer (see Figure 3b) is also shown: the horizontal and vertical situations are denoted by H and V, respectively.
I
(0)
(b) . .... ......,
,,
0
30 60 Detection Angle/deg.
90
FIgm 2. Calculated angular distribution of radiation from an d l l a t i n g dipole that emits 470-nm light. Results are (a) for the three orientations defined in Figure l b and (b) for oblique orientations tilted 3 O (solid curve), 6 O (broken curve), and 14' (dotted curve) from the surface.
the emission pattern due to a slightly tilted dipole, as shown in Figure 2b, can also be provided by a dipole lying flat on a surface with small random roughne~s.~ Thus the precise orientation of the dipole cannot be determined uniquely from the radiation pattern. However, it is possible to determine if the dipole is oriented almost parallel or perpendicular to the surface.
3. Experimental Section Figure 3a shows a schematic diagram of the apparatus, which consists of a high vacuum chamber, a variabletemperature sample holder, and an evaporation unit. All experiments, except the preparation of substrates, were performed by use of this apparatus. As substrates, Ag film and hexatriacontane (HTC) (Tokyo Kasei Kogyo Co.)films were used; they were prepared on a copper block by a vacuum evaporation method. In the measurements of the angular distribution, Ag and HTC were evaporated onto a thin glass slide, and the substrate was fixed on another type copper block by an electrically conductive adhesive (see Figure 3b). A substrate thus prepared was mounted on a sample holder in the high-vacuum chamber. After evacuation by an ion pump (60L/s) up to 1 X IO-"Pa, the substrate was oooled down to about 190 K. Anthracene (Tokyo Kasei Kogyo Co., zone refined) was evaporated onto the substrate and a sample such as shown in
Figure 3. (a) A schematic diagram of the apparatus. This setup is for the spectral measurements. MCS and TM mean a multichannel slit and a thickness monitor, respectively. (b) A setup modified for the measurements of the angular distribution of fluorescence. (c) A model of samples prepared by vacuum evaporation. HTC means hexatriacontane, whose thickness is varied from 0 to 400 A.
Figure 3c was prepared. The amount of deposition was monitored by a quartz-oscillator thickness meter, which was also cooled to almost the same temperature as the substrate. The thickness of an anthracene film was about 20 A, which is equivalent to about six monolayers. The thickness of HTC films, d, was varied from 0 to 400 A; a HTC film with a thickness of d is represented by HTC(d). The fluorescence spectra were measured as follows (see Figure 3a): 360-nm light, obtained by filtering the output of a 500-W Xe lamp (Ushio) with a monochromator (Spex Minimate), was irradiated to the sample through a quartz lens and a quartz window. The resulting fluorescence was filtered by a double monochromator (two Spex Minimates) through a tungsten-glass window and detected by a photomultiplier (Hamamatsu R585), whose output was processed by a photon counter (Hamamatsu C-767) and a personal computer (NEC PC-9801VM). The fluorescence spectra were taken at several temperatures between 190 and 300 K. In the measurements of the angular distribution of fluomccnce, the sample holder was modified as shown in Figure 3b. The sample was set in such a way that both incident and reflected light were in the same line. Excitation light was focused into a rectangle 1.5 mm wide and 5 mm long on the sample surface by a cylindrical lens. The fluorescencewas detected by the photomultiplier through a polarizer (Polaroid HN32), a lightgide, and an appropriate interference band-pass filter (Toshiba). The polarizer was mounted on an aluminum head, which was rotatable around the optical axis and controlled electrically. It was set parallel or perpendicular to the plane including the surface normal and the observation direction, referred to as vertical or horizontal, as described in section 2 (Figure lb). The detection angle, 0, defined as the angle between the surface normal and the detection direction, was varied over the range 25O Id I90°. 4. Results
4.1. Temperature and Substrate Dopendencies of the Fluorescence Spectra. Figure 4a shows the fluorescence spectrum of anthracene on a Ag substrate at 197 K, just after evaporation; all spectral data for Figures 4a and 5-10 were taken in the Figure 3a setup. There is no vibrational structure but only a broad band with a peak around 470 nm. This structureless band is due to the fluorescence of anthracene excimers; produced by the interaction of excited and unexcited molecules. The spectrum is distinctly different from that of monomer fluorescence in solution, which exhibits a structured band system with a maximum peak at about 400 nm (Figure 4b). Figure 5 shows several fluorescence spectra of the anthracene/Ag system, when substrate temperature, T,was increased
Fluorescence Emitted from Anthracene on Silver
The Journal of Physical Chemistry, Vol. 96, NO. 15, 1992 6363 *g (Anthracene IOOOA)
Wavelength / nm
Figure 4. (a) Fluorescence spectrum of an anthracene thin film on a silver surface at 197 K. (b) The spectrum of anthracene in a cyclohexane solution ( mol) for comparison.
400
500
600
Waveleng th/nm
Figure 7. Temperaturedependence of the fluorescence spectra of a thick anthracene film (1000 A) on a silver surface.
HTC(5O AVAg
A
.-
ul
c
5
e
4
Y
.-cu>rl C al c C
-
197 K
500
400
600
Wavelength/nm
Figure 5. Temperature dependence of the fluorescence spectra for the
anthracene/Ag system.
Figure 6. Temperature depcndencc of the total fluorescence intensity for the anthracene/Ag system.
from 197 K. Until to about 270 K,the spectrum shape was almost identical, though the fluorescence intensity gradually increased. At T = -270 K, small peaks began to appear in the shortwavelength region; they correspond to the vibrational bands of
500 600 W avel eng t h/nm Figure 8. Temperature dependence of the fluorescence spectra for the anthracene/HTC(SO A)/Ag system.
400
monomer fluorescence. Then both the bands of monomer and excimer fluorescence continued to growh until T = -280 K. As T was further increased, the structureless band due to excimer emission reduced remarkably; contrary to this, the vibrational bands grew rapidly to dominate the spectrum. The total fluorescence intensities integrated in the 390-600-m region are plotted against T (Figure 6). The intensity increases only a small amount in the range 190-220 K and, after a slight decrease around 240 K, gradually increases to attain a maximum at 280 K. The sharp decrease above 280 K is due to desorption of anthracene molecules from the substrate. The temperature dependence of the fluorescence in the anthracene/Ag system is similar to that in the anthracenelquartz system, reported by Maruyama and Ichikawan6However, there are a few differences: (1) immediately after evaporation, a strong band due to excimer fluorescence already appears in their spectrum, but the fluorescence intensity is very weak in ours, and (2) the total intensity of the fluorescenceis considerably reduced in the temperature range 200-270 K,in their results, whereas it stays almost constant in our results. These are mainly due to the
Ohshima and Ishibashi
6364 The Journal of Physical Chemistry, Vol. 96, No. 15, 1992 H T ~ O AVACJ O
---_.
(b) glass
- - - p8. +be
k$\o
T-271 K
-\,
400
500
600
Wavelength/nm
Figure 9. Temperaturedependence of the fluorescence spectra for the anthracene/HTC(lOO A)/Ag system. Detection Angle/deg.
Figure 11. Angular fluorescence distributions for (a) the anthracene/Ag and (b) the anthracene/glass systems: filled and open circles denote
experimental data for the horizontal and the vertical situations, respec-
tively. The best-fitting results are also shown with solid and broken
curves. The error bars indicate the statistical uncertainty.
400
600
500 Waveleng t h/nm
Figure 10. Temperature dependence of the fluorescence spectra for the
anthracene/HTC(400 &/AB system. difference in the anthracene film thickness in both systems. For an anthracene/Ag system, where the anthracene film thickness was increased to IO00 A, both strong excimer emission just after evaporation and reduction of its intensity in 200 5 T 5 270 K were observed (Figure 7). This leads to a closer resemblance to the results of Maruyama and Ichikawa and indicates that the spectra also depend on the anthracene film thickness. However, this lies outside theacope of the present study, which is concerned with the behavior of excited molecules near a metal surface. When the substrate was changed from A to HTC(5O A)/Ag, having a HTC film thickness (d)of 50 , the growth of the excimer band was suppressed, whereas the monomer bands were enhanced as shown in Figure 8. This feature is clearly seen from a comparison of the spectra at T = 283 K in Figures 5 and 8. As d of the HTC film was increased to 100 A, this feature was more emphasized as seen in the more distinct vibrational structure (Figure 9). A further increase in d from 100 A, however, caused only a small change in the spectral shape; the spectra for a HTC(400 A)/Ag substrate are shown in Figure 10. Generally, the total fluorescence intensity was also increased with increasing
w
d. This is accounted for by the reduction of excimer fluorescence quenching due to energy transfer.'.* For all samples with different d, however, the temperature dependence of the total fluorescence intensity was almost the same as that shown in Figure 6. 4.2. Angular Diotributiolrsof the Fluorescence Intemity. Figure 1 l a shows the angular intensity distributions of the excimer fluorescence from anthracene on a Ag substrate. The excimer fluorescence was detected through an interference band-pass filter with maximum transmittance at 470 nm. The filled and empty circles represent the experimental values for the horizontal and vertical situations, respectively. At T = 193 K, the fluorescence intensity is decreased monotonically with increasing detection angle, 0, for the vertical situation; the distribution is the typical case 3 pattern. As T is increased, however, the distributions gradually deviate from the case 3 pattern. On the other hand, the distribution for the horizontal situation at T = 193 K is slightly different from the case 2 pattern; a small and broad hump can be seen around 0 = -40'. With increasing T, the hump disappears so that the distribution becomes the typical case 2 pattern. At T = 268 K, the distributions for both the situations are almost the same. This phenomenon can be attributed to reduction of interference a t the Ag surface (see section 5.2). In fact, an antracene/glass system, where the interference of the fluorescence is very weak, exhibits similar angular distributions for both the horizontal and vertical situations; moreover, the patterns are independent of the substrate temperature (Figure 1lb). More detailed measurements of the angular distributions were performed for the anthracene/HTC(lOO &/Ag system (Figure 12). The temperature dependence of the angular distributions described above can be clearly seen. The improvement of the signal-to-noise ratio is due to the reduction of fluorescence quenching by energy transfer because of an increase in the molecule-metal distance. A further increase in the distance provided stronger signal intensity, but the interference effect was much weaker. The same measurements were carried out for the monomer fluorescence by changing the observation wavelength from 470 to 410 nm. At low temperatures, the signal intensity was so weak that no reliable data could be obtained. The intensity rapidly increased at T > 260 K, but then only "noninterference" patterns
Fluorescence Emitted from Anthracene on Silver
0
30
60
90
Detection Angle/deg. Figure 12. Temperature dependence of the angular fluorescence distributions for the anthracene/HTC( 100 A)/Ag system.
like as shown in Figure 1l b were obtained.
5. Discussion 5.1. Temperature and substrate Dependences of the Adsorption State of Anthracene. The temperature dependence of the spectra is caused by changes in the adsorption state of the anthracene molecules, as pointed out by Tro et al.3 and Maruyama and Ichikawam6Immediately after evaporation, Le., T = 200 K, the molecules are randomly adsorbed and their motions are tightly constrained. In such a state, excited molecules produced by irradiation strongly interact with adjacent unexcited ones, which causes significant internal quenching of the excited states and chemical reactions that yield photodimers. Energy transfer also occurs from monomer sites to excimer sites lying at lower energies, leading to excimer emission. Therefore, monomer fluorescence is barely observed just after the sample preparation. It is also noted that the intensity of the excimer emission is very weak. Two explanations are possible. One is the freezing of the molecular motion which hinders to attain an appropriate arrangement for excimer formation. The other is the presence of nonfluorescent dimers, and its details are described in the next paragraph, compared with the results of Seki and I t ~ h With . ~ increasing T u p to about 260 K, such freezing is loosened by degrees and the molecules become mobile. This facilitates excimer formation and also promotes Crystallization of the molecules together. As a result, excimer emission increases and monomer emission begins to contribute to the spectrum. At higher temperature, the monomer bands go on growing due to the progress of the crystallization, whereas the excimer band becomes smaller. At 280-290 K, most molecules are crystallized and the monomer bands dominate the spectrum. More detail of the change in the adsorption state is discussed in section 5.4. Seki and Itoh investigated dimer formation and crystallization of anthracene ultrathin films (0.5-10 monolayers) deposited onto fused quartz and sapphire at liquid He temperatures under a pressure of lo4 Torr using transmission spectroscopy? Though temperature and ambient pressure during evaporation are different between their and our experiments, film thicknesses are almost the same. Some of their findings are as follows: the films as deposited at T 10 K are disordered but still partially oriented, and some "stable ground-state dimers" begin to form. When the film is heated to about 190 K, the development of the stable dimers reaches a maximum and there is considerable development of
-
The Journal of Physical Chemistry, Vol. 96, No. 15, 1992 6365 crystallites. Above T > 190 K, the crystallization proceeds and the stable dimers begin to decompose. Such stable dimers can also be formed on Ag and HTC/Ag substrates used in our study. In fact, the weakness of fluorescence at low temperatures (190 K 5 T 5 270 K) is explained by considering the formation of the stable dimer, because dimers generally emit no fluorescence" and the amount of the stable dimers reaches a maximum in the temperature range. It follows that with increasing temperature the stable dimers should decompose into fluorescent molecules, resulting in stronger fluorescence. This agrees with the present results. Tro et al. reported the disorder-order transition for phenanthrene adlayers on Al,O,( 1lzO).3 In the phenanthrene/Al,O, system, the adsorption spectrum changes abruptly between T = 195 and 200 K and resembles the crystalline absorption spectra. On the other hand, the fluorescence spectrum shows a large amount of excimer emission still above 200 K the excimer emission is removed completely at about 240 K and a small minority of excimer sites are very effective traps for excited electronic energy. Their discussion holds in the present system as well. Most molecules crystallize at about 260 K, because the emission spectrum has a vibrational structure, showing that the emission is originated from the molecular crystal. A few excimer sites still survive, however, and serve as energy traps. Thus excimer emission dominates the fluorescence spectrum. This interpretation is consistent with the fact that the angular distributions of the molecular fluorescence deviate from the model calculation due to scattering effects at crystal parts. The temperature at which molecular fluorescence begins to be observed is higher by 65 K in the present system than in the phenanthrene/A1203 system. This indicates that anthracene forms excimers more easily than phenanthrene, due to a stronger moleculemolecule interaction. The fluorescence spectra depend not only on temperature but also on substrates, especially, at higher temperatures. When the spectra for the Ag substrate (Figure 5 ) are compared with those for the HTC(100 A) substrate (Figure 9), excimer emission is more enhanced in the former at T > 270 K. This feature can be explained as the difference in the moleculesurface interaction between the Ag and HTC substrates. The anthracene molecules interact more strongly with the Ag surface than with the HTC surface. Therefore, the crystallization is hindered on the Ag surface, leading to the survival of excimer sites. This explanation is also applicable to the finding that the HTC film thickness, d, influences the spectra. The HTC( 100 A) substrate have a coverage of 1; this is estimated from the thickness on the assumption that the HTC molecules are oriented with their long chains perpendicular to the substrate.I2 Thus the HTC molecules are considered to cover the entire Ag surface. The HTC(5O A) substrate has a coverage of 0.5 and there still remains a Ag area uncovered with HTC, which facilitates excimer formation. On the other hand, the HTC(400 A) substrate with a coverage of 4 provides almost the same results as the HTC(100 A) substrate. 5.2. Orientation of the Transition Moment of the Anthracene Excimer. As described in section 2, the transition moment of molecular emission near a metal surface can be determined from the angular intensity distributions of the emission. Immediately after evaporation, a slight rise appears around 0 = 45O in the horizontal distributions for both Ag and HTC substrates (Figures l l a and 12). This corresponds to the pattern given by the case 2 orientation slightly mixed with the case 1, shown in Figure 2b. Therefore, the transition moments are considered to be tilted to the surface. To confrm this, a least-square fitting was performed for the angular distributions on the basis of Greenler's model calculations; the optical constants (ii = n - ik is the complex refractive index) of silver were chosen as parameters. The results are shown in figures by solid and broken lines for the horizontal and the vertical situations, respectively. They reproduce well the experimental data. The n and k values that gave the best fit were n = 0.054 and k = 2.67 for the horizontal situation and n = 0.050 and k = 2.75 for the vertical situation. These values are in good agreement with the literature values, n = 0.045 and k = 2.76 for 470-nm light.I3 The fraction of case 1 was 0.02, from which the
-
6366 The Journal of Physical Chemistry, Vol. 96, No. 15, 1992 oblique angle of the transition moment to the substrate is estimated to be 2 O . As Tis increased, a slight rise in the horizontal distribution disappears and the distribution curve becomes similar to the case 2 pattern. Therefore, the transition moment is parallel to the surface. Using almost the same values of the complex refractive index as the literature values, the distributionscan be fitted. Above T = -260 K, however, those values are no longer applicable to reproduce the experimental results. This can be explained as follows: as the crystallization proceeds, the molecules homogeneously cover the metal surface, reducing the reflection of the molecular emission. Instead of the interference, the molecular emission suffers from scattering by the other molecules. In such a case, excited molecules can be regarded as isolated radiating dipoles, which emit light htropically. Then the resultant intensity distribution is expected to obey the cosine law that gives a cos 6 curve. The experimental data at higher T, indeed, approach such a pattern. 5.3. Structure of Excimen. Hofmann et al. identified three types of the green excimer band' in solid noncrystalline anthracene; the maxima of bands I, 11, and I11 are 457 f 3,465 f 3, and 510 f 4 nm, and their full widths at half-maximum are 2600, 3500, and 4400 cm-l, respectively! They also speculated on the structure of the excimer corresponding to each band; the molecular planes of the two molecules are essentially parallel in the three excimers, and in excimers I and I11 one molecule displaces along both long and short axes largely (I) and slightly (111), respectively, whereas in excimer I1 the short axes form an angle of about 60°. The excimer band in our results have a maximum at about 460 nm and a full width at half-maximum of about 3100 cm-I. From this band shape., excimer I11 is discarded, but excimers I1 and 111 cannot be distinguished. The data of the angular distributions for excimer fluorescence indicate that its transition moment is practically oriented parallel to the substrate. This orientation prefers the structure of excimer I1 rather than that of 111; in the former both the short axes of the anthracenemolecules are parallel, whereas in the latter they are tilted and hence the transition moment should be tilted to the surface. Thus,the excimer emission in our results is considered to originate from excimer 11. 5.4. AdsMptioa State of Anthracene Exclmers. To deduce the adsorption state of anthracene excimers on the Ag and HTC surfaces from the fluorescence angular distributions, it is necessary to determine the orientation of the transition moment of the excimers. The absorption of an anthracene monomer at 300-380 nm is mainly due to the transition to the lowest singlet state ILa; the transition moment is directed along the short axis in the molecular plane.I4 Hence the monomer fluorescence is polarized along this direction. In an anthracene excimer, the two molecules are oriented with the molecular planes and axes parallel to each other and perpendicular to the line joining their centers, as discussed in section 5.3. The stabilization of such an excimer is primarily due to exciton resonance, causing a strongly polarized dipole-dipole interaction.I5 Therefore, the transition moment of the excimer fluorescence is also directed along the short axes of the two molecules. For such an orientation of the transition moment, the excimer can take two possible adsorption states (Figure 13a,b); the plane containing the transition moment vector is parallel or perpendicular to the surface. The latter, referred to as the perpendicular orientation, is favored to the former (the parallel orientation) from the following reason. It has been clarified by the Penning ionization study applied to the solid that films of organic compounds such as naphthacene and pentacme on metal or graphite substrates crystallize in the arrangement with the long axis of the molecules almost perpendicular to the surface,16 which is similar to the perpendicular orientation. At higher T, all molecules take this orientation. Thus the molecules closely adjacent to each other, which are responsible for excimer formation, are also rearranged toward the crystalline orientation. In the case of the perpendicular orientation, the rearrangement is easily attained because of the similarity of both orientations. On the other hand, in the case of the parallel orientation, the molecules are required to rotate
Ohshima and Ishibashi
Figure 13. Orientations of an anthracene excimer: (a) parallel and (b) perpendicular orientation. More geometrical diagrams are also shown on the right of each figure; a molecule is represented as a straight line, which corresponds to the long axis of the molecule. This notation is used below. (c) Temperature dependence of the adsorption states of the molecules. The orientation and intensity of the transition dipole for each adsorption state are also shown by a dotted arrow on the right.
about 90°. Such a large change in the molecular orientation should induce a corresponding large change in the direction of the transition moment of the excimer emission, which would be reflected in the angular distribution patterns. This is not the case; the distributions are monotonicallychanged at higher T. However, a small change is recognized in the horizontal distribution at lower T; the hump appeared just after evaporation stopped at T = -250 K. Therefore, as described above, the molecules are adsorbed in the amorphous state at T = 193 K, there are molecules in the parallel orientation in part. However, in the temperature range 193-250 K the molecular orientation changes to the perpendicular one, which is energetically stable. Taking the discussion described above into account, the change in the adsorption state with increasing Tis illustrated in a stepwise fashion (Figure 13c). Immediately after the evaporation, the anthracene molecules are adsorbed randomly on the substrate; some of them contact other molecules face to face, favorable for excimer formation. A small portion of the molecules in the excimer sites are tilted to the surface. With further increase in T, the molecules gradually move on the surface and begin to rearrange. At this stage, however, the excimer sites near the surface keep their molecular planes parallel to the surface. When T i s raised up to -270 K, the molecules in such excimer sites also rearrange to the crystalline orientation. The final process is faster on the HTC substrate than on the Ag substrate. 6. Conclusions
The fluorescence spectra of anthracene molecules were measured on silver and HTC substrates as a function of substrate temperature, T. At lower T , the spectra consist of a broad structureless band due to excimer fluorescence. As Tis increased, a structured band due to monomer fluorescence appears at higher energies than the excimer fluorescence band; the former grows to dominate the spectra at T S 280 K. The variation of the spectral shape originates from the change of the adsorption state of the molecules. The spectra depend on the substrate as well as the temperature; excimer fluorescence contributes more largely to the spectra for the silver substrate than for the HTC one. The angular distributions of excimer fluorescence and their temperature dependence were also measured. By analyzing the data, the orientation of the excimers was determined. Moreover, the change in the adsorption states of the anthracene molecules as a whole was deduced from data of both the spectra and the angular distributions: Just after evaporation (T = -200 K), the
J. Phys. Chem. 1992,96, 6367-6371
6367
anthracene molecules are adsorbed randomly, and with increasing T, they are rearranged to a crystalline state, where the molecules are oriented, their short axis parallel to the substrate and their long axis almost perpendicular to it. It should be noted that the angular distribution measurements of luminescence are useful for the study of the orientation of photophysical products such as excimers or exciplexes on a metal substrate. Registry No. Ag, 7440-22-4; anthracene, 120-12-7.
(4) Ishibashi, Y.; Ohshima, S.;Kajiwara, T. Surf.Sci. 1988, 201, 311. (5) Sano, H.; Mizutani, G.; Ushioda, S.Surf.Sci. 1989, 223, 621. (6) Maruyama, Y.; Takamiya-Ichikawa,K. Int. J. Quantum Chem. 1980, 18, 587. (7) Ferguson, J.; Mau, A. W.-H. Mod. Phys. 1974, 27, 377. (8) Hofmann, J.; Seefeld, K. P.; Hofberger, W.; Bgssler, H. Mol. Phys. 1979. 37. 973. (9) Seki, H.; Itoh, U. J . Chem. Phys. 1980, 72, 2166. (10) Greenler, R. G. Surf Sci. 1977, 69, 647. (11) Chambers, R. W.; Kajiwara, T.; Kearns, D. R. J. Phys. Chem. 1974,
References and Notes
(12) Seki, K.; Inokuchi, H. Chem. Phys. Lerr. 1982, 89, 268. (13) Johnson, P.; Christy, R. Phys. Reu. 1972, 86, 4370. (14) Michl, J.; Thulstrup, E. J. Spectroscopy with Polarized Light; VCH: New York, 1986; p 405. (1 5) Birks, J. B. Photophysics of Aromatic Molecules; Wiley-Interscience: London, 1970; p 327. (16) Harada, Y.; Ozaki, Y. Jpn. J. Appl. Phys. 1987,26, 1201.
(1) Avouris, P.; Persson, B. N. J. J . Phys. Chem. 1984, 88, 837. (2) Waldeck, D. H.; Alivistos, A. P.; Harris, C. B. Surf Sci. 1985, 158,
103.
( 3 ) Tro, N. J.; Nishimura, A. M.; George, S.M. J . Phys. Chem. 1989, 93,
3276.
Photochemical Reaction of H,FeOs,( CO)
Adsorbed on the Surface of Silica
Sadaaki Yamamoto,* Central Research Institute, Mitsui Toatsu Chemicals, Znc., 1 190, Kasama, Sakae, Yokohama, Japan
Yasushi Miyamoto, Robert M. Lewis, Mitsuo Koizumi, Kuroda Solid Surface Project, Research Development Corporation of Japan, Tsukuba Research Consortium, 5-9-4 Tokodai, Tsukuba, Zbaraki 300-26, Japan
Yoshiyuki Morioka, Department of Chemistry, Faculty of Science, Tohoku University, Aoba- ku, Aramaki, Sendai, Japan
Kiyotaka Asakura, and Haruo Kurodat Department of Chemistry, Faculty of Science, The University of Tokyo, Hongo, Tokyo, Japan (Received: September 3, 1991)
The photochemical reaction of silica-supported H , F ~ O S ~ ( C Oinduced ),~ by irradiation with visible light was investigated by using FT-IRand UV-visible reflectance spectroscopies. The structure of the photoproduct was studied by laser Raman and EXAFS spectroscopies. H , F ~ O S ~ ( C O molecules ) ~ ~ that are adsorbed in a molecularly dispersed state on the surface of silica were found to lose one CO ligand when irradiated with light in the region of the first electronic transition of the carbonyl cluster. The reaction yielded a coordinatively unsaturated species, H,F~OS~(CO),~. It was concluded that this unstable photoproduct is stabilized on the surface of silica and does not undergo secondary reactions because of an interaction with surface hydroxyl groups.
Introduction Metal carbonyl compounds are often used to prepare highlydispersed supported metal catalysts.' Usually, a thermal treatment followed by reduction at an elevated temperature in a hydrogen atmosphere is used to remove the CO ligands or to produce small metal clusters on the support. Decarbonylation of metal carbonyl compounds can also be done by using light irradiation. Because photochemical processes are often more selective than thermal processes, the use of a photochemical process to make unstable surface bound chemical species that cannot be obtained by conventional thermal p r o c a w appears reasonable. Relatively little is known, however, about the photochemical behavior of metal carbonyl compounds on solid s u r f a ~ e s , ~ although -'~ the photochemistry of metal carbonyl compounds in solution has been extensively studied.'* We previously found that the surface bound hydride anion cluster HFe3(CO),,- can be formed when Fe3(C0)', adsorbed 'Also at Kurcda Solid Surface Project.
on the surface of silica is irradiated with visible light.' This species cannot be obtained by thermal treatment of Fe3(CO),2supported on silica. With the success of the photochemical approach in this system, it was important to explore the limits of such a photochemical approach. Another challenging carbonyl system is that of bimetallic carbonyl clusters. These carbonyl clusters often decompose when thermally treated. In the present paper, we report the photochemical reaction of the bimetallic carbonyl cluster compound H2FeO~3(C0)13 supported on silica. Experimental Section H2FeOs3(C0)13 was prepared according to the method reported in the 1iterat~re.l~ n-Hexane used in the sample preparations was dried over CaC12 and then distilled from Na or LiA1H4 prior to use. We used two types of silica as the support, Aerosil 380 (Nippon Aerosil), which has surface O H groups, and RY200 (Nippon Aerosil), in which all surface silanol groups have been replaced by methyl groups. We refer to the first type of silica as 'hydroxy silica" and the second type as "methyl silica". Before
0022-365419212096-6367$03.00/00 1992 American Chemical Society