Temperature dependence of fluorescence decays ... - ACS Publications


Temperature dependence of fluorescence decays...

1 downloads 92 Views 962KB Size

J. Phys. Chem. 1989, 93, 6704-6710

6704

are I+, C2Hx+,CIHx+,and H+. This type of spectrum can be rationalized if it is assumed that with off-resonant wavelengths the intensity of the laser is sufficiently high that the parent ion formed absorbs a large number of photons to produce the energy-expensive small ionic fragments. In this case, iodobenzene is ionized by two-photon absorption. The third or fourth photon absorbed could access a dissociative or predissociative electronic state, producing I+ as described above. It is possible that a ladder-switching occurs a t this step, i.e., further absorption by the C6H5radicals formed could lead to the small energy-expensive fragments. The strong absorption of the iodobenzene and the phenyl radical could lead to a high probability of forming the observed small fragments by multiphoton absorption in this system.

Conclusions The results show that this technique can be used to identify the dissociation mechanisms resulting from laser multiphoton absorption of organic iodides. By selecting laser wavelengths that are resonant with the two-photon electronic states of the iodine

atoms, a ladder-switching mechanism on the neutral manifold dominates in which dissociation occurs faster than energy randomization and molecular rotation. The I+ produced are expected to have the same anisotropic spatial velocity distribution as the iodine atoms. At laser wavelengths that are off-resonance of the iodine atom two-photon absorption, channels forming I+ from the dissociation of the parent ions via a ladder mechanism are observed. For weakly absorbing molecules, e.g., CH31, the ladder mechanism seems to produce only lower excited states that form I+ statistically. This leads to broadened central mass peaks. For strongly absorbing molecules and their ions, e.g., C6H51,or for very high laser intensities, the ladder mechanism could lead to higher excited states of the ion that are dissociative, giving rise to an anisotropic velocity distribution of the I+. Acknowledgment. The National Science Foundation is acknowledged for the financial support of this work. R.v.d.B. acknowledges The Netherlands Organization for Scientific Research (N.W.O.) for a research fellowship.

Temperature Dependence of Fluorescence Decays of Isolated Rhodamine B Molecules Adsorbed on Semiconductor Single Crystals Klaus Kemnitz,* Nobuaki Nakashima,? Keitaro Yoshihara,* Institute for Molecular Science, Myodaiji, Okazaki 444, Japan

and Hiroyuki Matsunami Department of Electric Engineering, Kyoto University, Sakyo-ku, Kyoto 606, Japan (Received: July 25, 1988; In Final Form: January 24, 1989)

Fast nonexponential fluorescence decays of rhodamine B monomers adsorbed on single crystals of a-Sic, &Sic,and GaP have been observed. The decays are fitted with three exponentials and are interpreted by adsorption sites of differing electron-transfer capability. A thermal equilibrium is postulated to exist among the ground-state populations of these sites. The electron-transfer rate is only weakly dependent on temperature but is considerably slowed down by addition of water and by aging of the semiconductor surface.

(I) Introduction From electrochemical measurements of the sensitization effect of dye molecules adsorbed on semiconductors, it is known that electron or hole injection occurs, depending on the relative position of conduction and valence band edge with respect to the energy levels of the dye. In the case of GaP and a-Sic, a cathodic current is observed that is interpreted as electron transfer from the valence band of the semiconductor to the excited, adsorbed dye.I In the case of ZnOl-3 and Sn02?v5 an anodic current indicates electron transfer from the dye to the conduction band of the semiconductor. Energy transfer from the excited dye to the semiconductor, which is an alternative relaxation mechanism, was shown to play a role In this case, the spectral in the case of eosin adsorbed on overlap is sufficient for energy transfer to compete with electron transfer. In the system of rhodamine B (RhB) adsorbed on Gap, energy transfer apparently occurs from the vibrationally nonrelaxed SI state as soon as the excitation energy surpasses 570 nm.6 Recently, detailed studies were performed on the energy transfer of cresyl violet adsorbed on Ti02,' rhodamine 640 adsorbed on Zn0,8 and pyrazine adsorbed on G a A s 9 In the case of single crystals of GaP'OJI and a-SiC,12electron transfer can occur also from surface states that are located about 1.6 eV above the valence band edge in Gap" and in the middle of the forbidden zone in a-SiC.'2 Electron transfer of RhB adsorbed on polycrystalline Present address: Institute of Laser Engineering, Osaka University, Suita, Osaka 565, Japan.

0022-3654/89/2093-6704%01SO10

SnOz5was also shown to involve surface states. Surface states introduced by the creation of dislocations were shown to enhance electron transfer into the conduction band of ZnO from excited RhBa3 The above results indicate that both mechanisms of electron and energy transfer between adsorbed dye and substrate are operative in semiconductor adsorption systems, depending on the individual semiconductor/dye combination, and it is one of the goals of this paper to try to discriminate between both mechanisms. Few direct determinations of the rate of electron and/or energy transfer exist so far, utilizing the fluorescence quenching of the adsorbed m ~ l e c u l e s . ~ ~In~ particular, ~ ' ~ - ~ ~ only a few temperature ( 1 ) Memming, R. Photochem. Photobiol. 1972, 16, 325. (2) Daltrozzo, E.; Tributsch, H. Photogr. Sci. Eng. 1975, 19, 308. (3) Li, B.; Morrison, S. R. J . Phys. Chem. 1985, 89, 5442. (4) Memming, R. Prog. Surf. Sci. 1984, 17, 7. ( 5 ) Kim, H.; Laitinen, H. A. J. Electrochem. Soc. 1975, 122, 53. (6) Memming, R.; Tributsch, H. J. Phys. Chem. 1971, 75, 562. (7) Crackel, R. L.; Struve, W. S. Chem. Phys. Lett. 1985. 120, 473. (8) Anfinrud, P. A,; Caugrove, T. P.; Struve, W. S. J. Phys. Chem. 1986, 90,5887. (9) Whitmore, P. M.; Alivisatos, A. P.; Hams, C. B. Phys. Reu. Left. 1983, 50, 1092. (10) Memming, R.; Schwandt, G. Electrochim. Acta 1968, 13, 1299. (1 1) Beckmann, K. H.; Memming, R. J. Electrochem. Soc. 1969,116,368. (12) Gleria, M.; Memming, R. J. Electroanal. Chem. Interfacial Electrorhem. 1975, 65, 163. (13) Liang, Y.;Ponte Goncalves, A. M.; Negus, D. K. J . Phys. Chem. 1983, 87, 1

0 1989 American Chemical Society

Rhodamine B Adsorbed on Semiconductor Single Crystals

TABLE I: Fluorescence Decay Parameters

c v) E

a

u

The Journal of Physical Chemistry, Vol. 93, No. 18, 1989 6705

sample AI B-SIC (293 K)bvc 0.80 j3-Sic (160 K)bS 0.77 SnO+ 0.53 0.67 Gap a-SIC (dry)b 0.75 a - S i c (wet)b 0.27 & S i c (aged)c 0.46

41

A2 0.176 0.185 0.39 0.28 0.215 0.40 0.38

A3 7Ia/PS ~ Z 0 / p 6 0.024 73 458 0.046 81 503 53 0.08 235 0.05 179 731 0.035 34 301 109 801 0.33 0.16 242 875

r3a/Ps 2030 2070 2050 1910 1940 2520 2590

-

Ok, = 1 / ~ , l/Tc 7 , = 3.5 ns.’* *Freshly etched with HF. cShown in Figure 2. dShown in Figure 1. ‘See text. 4 r log counts 0

200

400

800

Channel Number Figure 1. Nonexponential fluorescence decays of RhB (coverage 0 = 0.04)adsorbed on S n 0 2 (a) and GaP (b). Analyses are given in Table 1. Conditions: A,, = 550 nm; hb> 570 nm; 100 ch = 1.28 ns.

dependence studies seem to have been performed.l&f Dry, solvent-free systems are used in the present work, which are accessible to the study of the temperature dependence of the rate of energy or electron transfer in a wide temperature range. The analyses of the fluorescence decays were performed in terms of three exponentials that revealed the fast character of the first component; two-exponential fits yielded clearly longer lifetimes of the first component. A coverage of l/looth to 1/25th of a monolayer is applied that allows the measurement of electron transfer of isolated dye monomers,”J8 undisturbed by energy transfer from excited dye monomers to dimers or aggregates. (11) Experimental Section

The weak fluorescence of a l/loothdye monolayer was measured with a time-correlated single-photon counting system of picosecond time resolution, which has been described e 1 s e ~ h e r e . I ~ a - S i c , @-Sic,and GaP were p-type semiconductors. a - S i c was epitaxially grown on a silicon wafer and had a carrier density , of 393 cm2/(V s), and a thickness of 3.78 X 10l6 ~ m - a~mobility of 30 pm. @-Sic,also epitaxially grown, had a carrier density , of 91.5 cm2/(V s), and was 310 of 2.43 X IO1’ ~ m - a~mobility pm thick.20-21 G a P was obtained from E & M Co., and SnO, was obtained as Nesa-glass. a - S i c and B-SiC were treated with 30% H F before adsorption of the dye; the other semiconductors were sonicated in acetone and distilled water. Aged samples of &Sicwere obtained by storing the crystals in the dark but exposing them to air for several months before adsorption of the dye molecules. Details of sample preparation, fluorescence decay measurement, and analyses were published e l ~ e w h e r e . ~ ~Since . ~ ~ the * ~ observed ~ fluorescence decays were polarization-independent, a polarizer, set a t the magic angle, was omitted in order to increase the intensity of the collected fluorescence. (14)Liang, Y.; Ponte Goncalves, A. M. J. Phys. Chem. 1985,89,3290. (15) Kajiwara, T.; Hashimoto, K.;Kawai, T.; Sakata, T. J . Phys. Chem. 1982,86,4516. (16)(a) Hashimoto, K.;Hiramoto, M.; Sakata, T. Chem. Phys. Lerr. 1988, 148,215. (b) Hashimoto, K.;Hiramoto, M.; Lever, A. B. P.; Sakata, T. J . Phys. Chem. 1988, 92, 1016. (c) Hashimoto, K.;Hiramoto, M.; Sakata, T. J . Phys. Chem. 1988, 92,4272.(d) Hashimoto, K.; Hiramoto, M.; Kajiwara, T.; Sakata, T. J. Phys. Chem. 1988,92,4636. (e) Sakata, T.; Hashimoto, K.; Hiramoto, M. Submitted for publication in J. Phys. Chem. (17)Kemnitz, K.; Tamai, N.; Yamazaki, I.; Nakashima, N.; Yoshihara, K. Chem. Phys. Lett. 1983, 101,337. (18)Kemnitz, K.; Tamai, N.; Yamazaki, I.; Nakashima, N.; Yoshihara, K. J. Phys. Chem. 1987,91, 1423. (19)Murao, T.; Yamazaki, 1.; Yoshihara, K. Appl. Opt. 1982,21,2297. (20)Suzuki, A.; Mameno, K.; Furui, N.; Matsunami, H. Appl. .. Phys. Lett. 1981,39,89. (21) Suzuki, A.;Ikeda, M.; Nagao, N.; Matsunami, H.; Tanaka, T. J . Appl. Phys. 1976,47, 4546. (22)Kemnitz, K.; Tamai, N.; Yamazaki, I.; Nakashima, N.; Yoshihara, K. J. Phys. Chem. 1986, 90,5094. (23)Kemnitz, K.; Nakashima, N.; Yoshihara, K. J . Phys. Chem. 1988, 92,3915.

3-

9

Figure 2. Temperature dependence of the fluorescence decay of RhB adsorbed on &Sic (0 = 0.01) for 293 and 160 K. Conditions: X, = 550 nm; Aok > 570 nm; 100 ch = 1.28 ns. Analyses are given in Table I.

All measurements were performed under an argon atmosphere, with use of an Oxford cryostat (CF 204) as a housing. The time resolution of all the decays shown is 12.4 ps/ch (ch = channel number) at a fwhm of the system response function of 70 ps; our shortest observed lifetimes of about 30 ps might be, therefore, limited by the system time resolution. (111) Results (a) Fluorescence Decay Analyses. We concentrated this study on the systems of RhB adsorbed on a-Sic, @-Sicand Gap, but a few measurements were also done for RhB adsorbed on polycrystalline Sn02. Fluorescence decays of all these systems were strictly nonexponential. Figure 1 shows the nonexponential fluorescence decays of RbB adsorbed on S n 0 2 and GaP a t a coverage of 1/25th. The convolution with the excitation pulse yields good fits with exponential decays of three components, I(t) = C;,.,pliexp(-t/ri), which in dry systems are dominated by a fast component of 30-200 ps (Table I). The lifetimes of the longest lived component were close to 2 ns and are considerably shorter than the corresponding lifetimes observed on the inert substrates phenanthrene and glass, which were about 3.5 ns.17J8 ( b ) Temperature Dependence of the Fluorescence Decays of RhB Adsorbed on a-Sic, fi-Sic, and Gap. The effect of temperature on the fluorescence dynamics is illustrated in Figure 2 for RhB (l/looth monolayer) adsorbed on @-Sicat 293 and 160

K. The analysis of both fluorescence decays in Table I shows that the lifetimes are independent of temperature within the experimental error and that it is a change of the preexponential factors that is responsible for the visual temperature dependence of the fluorescence dynamics; i.e., the slower fluorescence decay at 160 K is not due to longer lifetimes but mainly due to a 100% increase of the contribution of the slow 2 4 s component. The analysis of fluorescence lifetimes and preexponential factors in the tempera t u r e range 60-300 K is shown in Figures 3 and 4 together with the corresponding analyses of GaP and a-Sic. Figure 4a shows a decrease of the contribution of the fastest component with decreasing temperature, accompanied by a corresponding increase of the slower second and third components (Figure 4b). Figure 4 also shows indications of a leveling off at temperatures below about 100 K. This behavior is qualitatively the same for all three semiconductors studied.

6706 The Journal of Physical Chemistry, Vol. 93, No. 18, 1989 a

a

23.5 23.4-

T

3

6

I

J

4

23.2 23.lZ

1/71

1

i

I

T

t

1/72

?

480psi-5.8%

'

i 1 . I 1/73 __-

78psi9.9%

Y

\

r

I

T

23.3;-

-r -

Kemnitz et al.

216C 21521 4 -

--

21 3 -

2020ps +2%

*

21 :1 200L

I

4

'

5

6

7

8

&

-i'i--it-

l/Temperature (10-3K) I

22.7

0

22.6 22 5 22.4 -

168ps k 10.8%

+

T

678psF9%

1

Temperature/K

f

b

0.42-

2 0.38T

4

1/72

6 0.34-

i

20 0 L

5

+ 1830ps_+6.7%

4

i

2011

100

~

+

213

202t

200

T

*

:2 1 2 -

8

300

T

0.30;

m

-.-m 0.20 LL

1/Z3

A

I L -

4

5

6

I _

7

8

9'&

c

5

0.16-

; C

1/Temperature ( W 3 K )

0.12-

n

0.080.04 I

300 23 2

-

-

100

Temperature/K

Figure 4. Analysis of the temperature dependence of the preexponential factors of the fluorescence decays of RhB adsorbed on Gap, a-Sic, and @-Sic: (a) A, = squares; (b) A2 = triangles; A, = circles. A fit of the data points to a straight line from 293 to 100 K yields linear regression correlation factors of 0.976 ( A l ) ,-0.985 (A2),and -0.905 ( A 3 )for G a P and 0.938 ( A ' ) ,-0.710 (A2),and -0.901 ( A 3 )for &Sic. The error bars of shown correspond to the assumed maximal error bar of an individual measurement. The actual error within a series of measurements at identical optical alignment (e.g., temperature dependence) seems to be much smaller, especially for the dominant first componedt, as can be seen from the small scatter of the date points and the high correlation factors.

23 1~ 21 4 2

1/Temperature (1O-3K) Figure 3. Analysis of the temperature dependence of the fluorescence lifetimes with three exponentials I ( t ) = &,,,'A, exp(-t/q), of RhB (0 = 0.01) adsorbed on P-SiC (a), GaP (b), and a-Sic (c). Conditions: plot of In 1 / ~ ,vs 1 / T ; A,, = 550 nm;

200

> 570 nm.

~~b

The lifetimes of all three components are independent of temperature within the experimental error24 for GaP and @-Sic (Figure 3a,b). a-Sic seems to show a slightly negative activation for the dominant first component (Figure 3c). (c) Effect of Water. A thin film of water on top of the semiconductor/dye system induces a pronounced increase of the --

(24) We derive our experimental and analytical error from measurements of samples with temperature-independent fluorescence lifetimes and/or preexponential factors. The error for lifetimes and preexponential factors is taken from the measurements of RhB adsorbed on phenanthrene,'* anthracene, and pyrenc2, rr 5 10%and A, 5 10% are assumed as upper limit for an individual measurement.

lifetimes of all components of the three-exponential decay, as demonstrated in Table I for RhB adsorbed on a-Sic. (d)Effect of Aging on the Semiconductor Surface. An aging effect was observed in the case of a - S i c and @-Sic. Especially @-Sicshowed a decrease of all lifetimes of the three-exponential decay (Table I), when the semiconductor crystal was exposed to air for a period of two months, before adsorption of the dye. ( e ) Dependence of the Fluorescence Decays on the Excitation Wavelength. The lifetimes of all three components (obtained for an aged sample of @-Sic,freshly etched a-Sic, and acetone-rinsed Gap) were independent (56%)of the exciting wavelength from 540 to 570 nm. (IV) Discussion The coverage of the dye molecules adsorbed was adjusted to 1/25thto l/,wth of a monolayer in order to prevent energy transfer and migration among adsorbed dye species. The absence of surface dimerization of monomerically adsorbed dye molecules was in detail discussed elsewhere.I8 The three-exponential decays observed for RhB adsorbed on a - S i c , @-Sic,Gap, and S n 0 2 are interpreted by electron or hole transfer from the substrate to dye

Rhodamine B Adsorbed on Semiconductor Single Crystals monomers adsorbed on surface sites with differing capabilities of electron/hole t r a n ~ f e r . ~ ~ - ~ ~ The following section (a) discusses potential contributions to the fluorescence decays by energy transfer, section b discusses the temperature dependence of electron/energy transfer induced by the adsorbate-substrate vibration, section (c) tries to justify our preference of the electron-transfer mechanism, and section (d) deals with the temperature-dependent equilibrium among molecules adsorbed at different sites. ( a ) Energy Transfer. The transfer of the excitation energy from adsorbed dye molecules to semiconductor substrates may occur by energy and/or electron transfer, according to the different mutual locations of dye and substrate energy levels. The occurrence of energy transfer can be either due to the spectral overlap of the semiconductor substrate with the adsorbed dye, as in the case of small-band-gap semiconductors, or in the case of wideband-gap semiconductors due to the presence of surface states, which are situated in the forbidden band-gap region. Thus, the situation is different from dye molecules adsorbed on organic single crystals, where electron transfer is the unambiguous mechanism.23.28 We were looking for evidence of energy transfer in our systems by varying the excitation wavelength in the range of the main absorption band of RhB (Le., 540-570 nm). It turned out that the lifetimes of all components of the three-exponential decays were independent (56%) of the excitation wavelength, as determined for a-Sic ( E = 3.00 eV;I2 freshly etched), &Sic ( E = 2.3 eV;29aged), and GaP (acetone-rinsed), From the above excitation wavelength studies, we tentatively conclude that energy transfer is absent in our dry systems or, if present, that it occurs on a time scale that is longer than the vibrational relaxation. This discrepancy of our results with electrochemical experiments mentioned in the Introduction may not be surprising, however, since the location of the energy levels of conduction/valence band edge and of the surface states relative to the dye levels in the present dry systems is presumably different from that in the wet electrochemical systems. Furthermore, the number and distribution of extrinsic surface states,30 which may play a dominant role in energy (and electron) transfer, critically depend on the previous treatment of the semiconductor s u r f a ~ e . ~ ~ A~ lrecent - ~ ~ and comprehensive discussion of surface states can be found in the Iiterat~re.~~,~~ In addition to the above excitation energy studies, further clues as to the presence of energy transfer might in principle also be obtained from the inherent temperature dependence: the fluorescence lifetime T of an adsorbed molecule at a distance d (25) We also considered the possibility that the second or third component of the three-exponential decay could be caused by back electron transfer. This back-reaction, however, should be a strong function of temperature (see discussion). (26) James et al. have described complications in multiexponentialanalyses of fluorescence decays arising from closely spaced lifetime^!^ Since our observed three lifetimes are well separated, with the first component dominating the decay (53-80%), we believe that it is justified to attribute the individual lifetimes of the three-exponential decay to three representative sets of surface sites that in turn, however, might have a certain distribution of lifetimes. These representative sets of surface sites may stand for a multitude of sites, which might be resolved by experiments with higher time resolution. (27) Spitler, M. T. J . Phys. Chem. 1986,90, 2156. (28) Gerischer, H.; Willig, F. Topics of Current Chemistry; Boschke, F. L., Ed.; Springer: Berlin, 1976. (29) Nishino, S.;Hazuki, Y.; Matsunami, H.; Tanaka, T. J . Electrochem. SOC.1980, 127,2674. (30) Ehrenreich, H.; Seitz, F.; Turnbull, D. Solid State Physics; Academic Press: London, 1970; Vol. 25. (31) Kawaji, A.; Gatos, H. C. Surf. Sci. 1964, I, 407. (32) Memming, R.; Schwandt, G. Surf. Sci. 1966.5,97. (33) Li, B.; Morrison, S. R. J . Phys. Chem. 1985,89, 1804. (34) Kobayashi, T.; Yoneyama, H.; Tomura, H. J . Electroanal. Chem. Interfacial Electrochem. 1982,138,105. (35) Gerischer, H.; Hein, F.; Lubke, M.; Meyer, E.; Pettinger, B.; Schoppel, H.-R. Ber. Bunsen-Ges. Phys. Chem. 1973,77,284. (36) Kiselev, V. F.; Krylov, 0.V. Electronic Phenomena in Adsorption and Catalysis; Springer Series in Surface Sciences 7; Springer: Berlin, 1987. (37) Kiselev, V. F.; Krylov, 0. V. Adsorption Processes on Semiconductor and Dielectric Surfaces; Springer Series in Chemical Physics, 32; Springer: Berlin. 1985.

The Journal of Physical Chemistry, Vol. 93, No. 18, 1989 6707

from a dielectric substrate surface at parallel transition moment is given by7938 1/7

=

30x3 -

n2~2d3~01

32r3nl (n12+ nZ2- K

~

+ ~4n22~22 ) ~

(1)

where index 1 refers to the surface and index 2 to the substrate, with T, being the fluorescence lifetime at infinite separation from the surface, X the emission maximum, n the refractive index, K the substrate absorption coefficient, and 0 the quantum yield of the adsorbed dye. The refractive index n, is assumed to be 1.O and independent of temperature. Emission maximum X and quantum yield 0 of the adsorbed dye monomer were shown to be independent of temperature for RhB adsorbed on phenanK-'for the temperature threne.18 With use of dn2/dT = 7 X n2(300 K)39 = 2.66, dependence of the refractive index n2 of d = 2 A, T~ = 4 ns, X = 580 nm, 0 = 1, and a constant K2 = 1.75 X leads to ~ ( 3 0 0K) = 6.87 ps and r(60 K) = 6.75 ps, Le., a small decrease of 2%. The absorption coefficient K ~ however, , depends critically on the wavelength. Due to the general temperature dependence of the semiconductor band gap (for S i c , see ref 411, the adsorption coefficient at wavelengths near the band Especially for GaP gap is a strong function of and P-SiC with band gaps close to 560 nm (2.21 eV), a strong temperature dependence of fluorescence dynamics (in the case of RhB with A, 2 560 nm) can be expected, since, according to eq 1, k a K~ (for K~