Energy Relaxation Dynamics of Photoexcited C60 Solid - The Journal

May 30, 1996 - The bands were attributed to free exciton states and self-trapped exciton states, which decay through diffusive recombination and activ...
2 downloads 16 Views 580KB Size
J. Phys. Chem. 1996, 100, 9223-9226

9223

Energy Relaxation Dynamics of Photoexcited C60 Solid Sung Ik Yang, Yung Doug Suh, Seung Min Jin, and Seong Keun Kim* Department of Chemistry, Seoul National UniVersity, Seoul 151-742, Korea

Jeunghee Park Department of Chemistry, Korea UniVersity, Jochiwon, Chungnam 339-700, Korea

Eun-joo Shin and Dongho Kim* Spectroscopy Laboratory, Korea Research Institute of Standards and Science, Taedok Science Town, Taejon 305-600, Korea ReceiVed: December 6, 1995; In Final Form: February 29, 1996X

The time-resolved photoluminescence (PL) of C60 solid film was measured at various detection wavelengths, excitation laser fluences, and temperatures. Two emission bands were identified which possess different decay profiles, and these profiles exhibited completely opposite temperature dependence. The bands were attributed to free exciton states and self-trapped exciton states, which decay through diffusive recombination and activated intersystem crossing, respectively. For the latter case, a distinct rise component in the PL time profile was observed at low temperature. This strongly suggests that there exists a nonnegligible barrier between the free exciton states and self-trapped exciton states.

1. Introduction As fully conjugated molecular organic systems, the fullerenes have attracted much attention for their potential applications as optical and nonlinear optical materials. In contrast to other organic conjugated systems, the lowest electronic transition in C60 is dipole-forbidden, because the electronic ground state (Ag) and the first excited state (T1g) have the same parity. Negri et al.1 explained the S1-S0 optical spectra in terms of the Herzberg-Teller scheme, i.e., the lowest singlet excited state acquires ungerade character from energetically higher states by adiabatic vibronic coupling, and thus the optical transition becomes partially dipole allowed. Another feature of the photoluminescence (PL) spectra of solid C60 is that they demonstrate certain variations over samples and method of preparation.2,3 This variation can be ascribed to, at least partially, structural disorder, which is common in molecular crystals because of their relatively weak intermolecular forces. The relaxation dynamics of photoexcited C60 in solid has been much studied under various conditions,4-16 which revealed complex nonexponential relaxation dynamics. This unusual nonexponential behavior has led to a great deal of speculation as to the physical origin of the relaxation process, and a wide range of mechanisms have been proposed to explain the decay dynamics. Cheville and Halas4 monitored the change in timeresolved optical transmission following the 633 nm photoexcitation and found that the relaxation behavior is nonexponential and independent of temperature between 150 and 400 K. But Juhasz et al.5 observed that as the sample was cooled from room temperature to 5 K, the nonexponential decay of time-resolved optical transmission becomes an exponential decay and shows dependence on excitation laser fluence. Both Brorson et al.6 and Thomas et al.7 observed the dependence of time-resolved photoinduced absorption on laser fluence. On the other hand, Ebbeson et al.8 reported no laser fluence dependence of their transient absorption spectra. Dexheimer * Authors to whom correspondence should be addressed. X Abstract published in AdVance ACS Abstracts, April 15, 1996.

S0022-3654(95)03642-2 CCC: $12.00

et al.9,10 observed nearly identical decay profiles of photoinduced transmission at 77 K and room temperature, while the excitation laser fluence was found to affect the decay profile. Timeresolved PL spectrum measured by Byrne et al.11 also showed little dependence on temperature. In the present study, we have carried out time resolved PL measurements on pure C60 solid film following excitation to the first electronically excited state with 2 ps optical pulses at 2.1 eV photon energy. Time-resolved PL of a C60 film was measured as a function of laser fluence between 0.9 and 9.0 µJ/cm2, and over the temperature range of 13-300 K. The emission was detected over the range of 660-780 nm. The motivation for this study is directed toward the elucidation of the relaxation dynamics of photoexcited C60 molecular solid under various experimental conditions of excitation fluence, detection wavelength, and temperature. 2. Experimental Section TCSPC (time-correlated single photon counting) technique has been used in the present work to study the decay dynamics of photoexcited C60 solid. TCSPC technique was described elsewhere,17 and only a brief description is given here. The excitation source is a picosecond dual jet dye laser (Coherent 702) pumped by a mode-locked argon ion laser (Coherent Innova 200). The cavity-dumped beam from the picosecond dye laser with a cavity dumper (Coherent 7220) had a 2 ps pulse width and an average power of 20 mW with Rhodamine 6G as a gain dye and DODCI as a saturable absorber at 3.8 MHz dumping rate. The intensity of laser beam was adjusted by neutral-density filters. The beam was focused to an estimated beam diameter of 100 ( 5 µm. The emission was collected at 45-deg angle with respect to the excitation laser beam by 5 and 25 cm focal length lenses, focused onto a monochromator (Jobin-Yvon HR320), and detected with a microchannel plate photomultiplier tube (Hamamatsu R2809U). The signal was amplified by a wideband amplifier (Philip Scientific), sent to a Quad constant fraction discriminator (Tennelec), a time-to© 1996 American Chemical Society

9224 J. Phys. Chem., Vol. 100, No. 22, 1996

Yang et al.

Figure 2. PL decay profiles at various wavelengths under high laser fluence condition at 100 K. The excitation laser wavelength was 590 nm.

Figure 1. (a) Time-resolved PL spectra measured at 13 K for various delay times. (b) PL spectrum at zero time delay fitted with two Lorentzian peaks at 680 and 725 nm.

amplitude converter (Tennelec), a counter (Ortec), and a multichannel analyzer (Tennelec/Nucleus), and stored in a computer. C60 was extracted from the soot produced by the contact-arc method and then purified by repeated column chromatography to obtain a purity of 99.95%. In order to make a thin film, C60 powder was sublimated in vacuum onto a Si(100) wafer at a rate of 50 Å/min to obtain a typical film thickness of 1 µm. Electron diffraction studies of the film show that the sample has a rather high crystalline character. The sample was cooled down using a Janis C-210 cold head with a Leybold-Heraeus helium compressor. The temperature was controlled by a Lakeshore 330 autotuning temperature controller. The temperature fluctuation during the measurement was less than 1 K and the cryostat was evacuated to 10-4 Torr. In order to avoid possible contamination by oxygen, the sample was kept in vacuum at all times. 3. Results and Discussion Figure 1a shows the time-resolved PL spectra measured at 13 K at various delay times. As the delay time becomes longer, the PL intensity at shorter wavelengths decreases more rapidly, indicating that the temporal behavior of high- and low-energy emission bands is different. The time-resolved PL spectrum of C60 solid at zero time delay was reconstructed as a combination of two Lorentzian peaks at 680 and 725 nm (Figure 1b). Figure 2 shows PL decay profiles at various wavelengths under high laser fluence at 100 K. There is a correlation between decay profile and emission wavelength, again showing that the average PL decay is faster at the high-energy side of the PL band than at the low-energy side. Such dependence of PL decay profile on emission wavelength can be explained in the following section. At constant exciton concentration, the decay mode depends on how rapidly excitons diffuse around in solid. If diffusion is fast, singlet exciton-singlet exciton

interaction becomes an efficient decay mode, so that the PL decay profile shows a pronounced fast decay component. On the contrary, if diffusion is slow enough, the exciton collision frequency is going to be too small for the exciton-exciton interaction to be important, and the fast decay component becomes reduced. This suggests that the high-energy band may be taken as due to free excitons whose diffusion coefficient is relatively larger, and thus decay more rapidly through excitonexciton collisions, than localized excitons which are assumed to give rise to the low-energy band resulting from monoexcitonic processes. Byrne et al.11 observed a single-exponential PL decay for C60 film at 10 K with a lifetime of 1.2 ns, which is similar to our monoexcitonic decay time. The monoexcitonic decay may involve processes such as intersystem crossing. A good candidate for the localized exciton is what is known to be self-trapped exciton whose binding energy in C60 solid is known to be 0.17 eV,18 which is much larger than those of semiconductors. As a result, the self-trapped excitons are known to exist in C60 solid even at room temperature. Self-trapped excitons result from an electronic band-to-band excitation through which the excited electron and the remaining hole create a local deformation of the lattice and thus localize themselves into a state below conduction band. The time-resolved PL of C60 film was measured at various excitation fluences. The laser fluences were varied from 3.3 to 9.0 µJ/cm2 to detect the emission at 680 nm and from 0.9 to 9.0 µJ/cm2 for the 725 nm emission. Figure 3 presents PL decay profiles monitored at (a) 680 nm and (b) 725 nm at room temperature, which reveals dependence on laser fluence. Both PL decay profiles turn out to be double exponential and possess two lifetime components of about 0.25 and 1 ns. Table 1 lists the fitting parameters obtained from PL decay curves of Figure 3. It is to be noted that the relative amplitude for the fast decay component shows a large drop upon decreasing the excitation laser fluence from 9.0 to 3.3 µJ/cm2. The laser fluence dependence of the relaxation process can be explained as follows. The exciton model has often been employed6,7,10,12,16 to explain the laser fluence dependent relaxation behavior of photoexcited C60 solid. In C60 solid, exciton annihilation is a nonradiative process that results in a net depopulation of the excited state. The simple model usually applied to this process is given by the following kinetic equation:19

∂[S]/∂t ) Rs - βs[S] - γs[S]2

(1)

where [S] represents the concentration of singlet excitons and Rs is the singlet exciton generation rate. For a low concentration of singlet excitons, βs[S] . γs[S]2; then eq 1 is reduced to ∂[S]/ ∂t ) Rs - βs[S]. In other words, at low exciton concentrations,

Energy Relaxation Dynamics of Photoexcited C60 Solid

J. Phys. Chem., Vol. 100, No. 22, 1996 9225

Figure 3. Laser fluence dependence of PL decay profiles at (a) 680 nm and (b) 725 nm at 300 K. The laser fluence was varied from 3.3 to 9.0 µJ/cm2.

TABLE 1: Fitting Parameters for PL Decay Profiles at 680 and 725 nm under Various Laser Fluences at Room Temperature 680 nm laser fluence (µJ/cm2) 9.0 5.4 3.3

725 nm

decay time (ns)

relative amplitude (%)

decay time (ns)

relative amplitude (%)

0.25 1.00 0.25 1.01 0.25 1.08

66.8 33.2 41.7 58.3 27.1 72.9

0.25 1.02 0.25 1.03 0.25 1.06 0.25 1.17

56.5 43.5 37.2 62.8 22.3 77.7 16.2 83.8

0.9

the singlet exciton-singlet exciton interaction is going to be negligible, and the major relaxation route is expected to be monoexcitonic decay channel, which in many cases is the intersystem crossing. For a high concentration of singlet excitons, βs[S] , γs[S]2; then eq 1 reduces to ∂[S]/∂t ) Rs - γs[S]2. In this case the typical singlet exciton annihilation process involves the following exciton-exciton collision processes:

S1 + S1 f S1* + S0 f S1 + S0 + phonon

(2)

S1 + S1 f e + h

(3)

(e, electron; h, hole)

where * denotes a state of high vibrational excitation (hot excitons). In the case of (2), one exciton is promoted to high vibrational states at the expense of the other exciton’s annihilation. Typically, fast intramolecular relaxation process is expected, so S1* deactivates nonradiatively to the S1 state rapidly with a very short lifetime (∼10-12-10-13 s). Previous transient photoconductivity measurements20 suggest that the generation of charge carriers (case 3) mainly results from exciton-exciton collisional ionization, especially in the weakly allowed absorption region of C60 solid.

Figure 4. PL decay profiles at (a) 680 nm and (b) 725 nm at various temperatures under the high laser fluence condition. For the 680 nm band PL decay becomes faster at lower temperatures, but on the contrary, it becomes slower for the 725 nm band. The inset shows a rise component.

Under extremely high laser fluence condition, the inelastic collisional processes with multiple particle interactions can occur in such diverse ways as singlet exciton-singlet exciton, singlet exciton-triplet exciton, and singlet exciton-charge carrier collisions. These processes would dominate the overall decay behavior, especially in the excited state dynamics investigated by pump/probe experiments (transient absorption or transmittance),4-10 which gives much faster relaxation than that monitored by PL decay such as in the present study using a much lower laser fluence condition. It should be pointed out that a photon density higher than 1019 cm-3 was employed in most of pump/probe experiments, which corresponds approximately to one photoexcitation per C60 molecule. In the present PL decay profile measurements, however, low-intensity optical pulses were employed to create relatively low density of photoexcited molecules (10-3 photoexcitation per C60 molecule or less), which drastically reduces the number of exciton-exciton collision processes. As a result, the contribution from the fast decay component due to exciton-exciton collision is suppressed, while the monoexcitonic decay channel becomes more important. The not-so-prevalent exciton-exciton collision channel does account for the observed dependence of photoexcited decay behavior on the laser fluence. We also examined the temperature dependence of the relaxation of electronically excited state. Figure 4 shows PL decay profiles at (a) 680 nm and (b) 725 nm at various temperatures under the high laser fluence condition. The PL decay becomes faster at lower temperature for the 680 nm band, but it becomes slower for the 725 nm band. For the latter band, a distinct rise component also appears below 100 K (shown in the inset of Figure 4b). If the dominant mechanism limiting the mean free path in exciton diffusion is scattering by phonons,21 then the diffusion coefficient of free excitons, D(T)free, should be proportional to T-1/2. Rose et al.22 reported that the exciton diffusion coefficient of anthracene increases

9226 J. Phys. Chem., Vol. 100, No. 22, 1996

Yang et al. of self-trapped excitons has been reported to be fast in molecular solid systems such as polymers.25 In summary, a photoluminescence study was carried out to investigate exciton decay dynamics in C60 solid under the conditions of low exciton concentration. Two emission bands were identified at 680 and 725 nm, which were respectively assigned to free exciton states and localized exciton states probably resulting from self-trapped excitons. The free exciton states were found to decay faster at lower temperatures through diffusion recombination. On the contrary, the localized excitons were shown to decay slower at lower temperatures due to the quenching of exciton-exciton processes resulting from slow diffusion, and thus the predominance of thermally activated monoexcitonic intersystem crossing. A rise component in PL time profile for localized exciton emission below 100 K suggests there also exists a nonnegligible barrier between free exciton state and localized exciton state.

Figure 5. Schematic potential energy diagram for energy relaxation in photoexcited C60 solid. The excitation laser wavelength was 590 nm. Free excitons emit at 680 nm (PL1), while self-trapped excitons emit at 725 nm (PL2).

Acknowledgment. The present work was supported by the Non-directed Research Fund from Korea Research Foundation and the KOSEF fund through Center for Molecular Science. References and Notes

from 0.8 to 10 cm2/s as the temperature was decreased from 20 to 1.8 K. Since the high-energy band in the present study corresponds to free excitons, its lifetime depends predominantly on the exciton diffusion coefficient in C60 solid. The faster decay of this band at lower temperatures (Figure 4a) strongly suggests that the major decay channel for free excitons is diffusive recombination controlled by processes such as exciton-phonon scattering. Since monoexcitonic decay through intersystem crossing is a thermally activated process, its relative importance will be negligible in the low temperature limit. On the other hand, when exciton-exciton collision is not a predominantly effective decay channel, due perhaps to a very slow diffusion as in the case of localized excitons, thermally activated monoexcitonic decay into triplet states will prevail, and thus the decay will be slower at lower temperatures. In the present study, the slower decay of the 725 nm band at lower temperatures (Figure 4b) suggests that this band is indeed due to localized exciton states. In addition, a rise component starts to show up below 100 K, presumably due to an activation barrier between free exciton state and self-trapped exciton state. Figure 5 shows a schematic potential energy diagram for energy relaxation in photoexcited C60 solid based on the experimental observations. Photoexcitation produces free excitons which emit at 680 nm (PL1) and decay through either exciton-exciton collisions (0.25 ns) or monoexcitonic channels (1 ns). The barrier between the free exciton state and the localized exciton state is so small that the emission from the latter (725 nm; PL2) commences almost immediately upon photoexcitation and decays again by exciton-exciton processes or monoexcitonic ways. This small barrier hinders the production of localized excitons, however, at sufficiently low temperatures (below 100 K), and a rise component is seen in the emission time profile. The major monoexcitonic decay channel for both exciton states is the thermally activated intersystem crossing to the triplet exciton state, which is characterized by a time scale of ca. 1.0 ns, in general accordance with its value in solution.23,24 The activation potential barrier has often been invoked to explain the temperature dependence for exciton decay,19 mostly in semiconductors, organic molecular crystals, and ionic crystals. Since self-trapped excitons are believed to form by local deformation of the lattice, it is usually assumed that there is only a small activation barrier between free exciton state and self-trapped exciton state. Indeed, the formation time

(1) Negri, F.; Orlandi, G.; Zerbetto, F. J. Chem. Phys. 1992, 97, 6496. (2) Guss, W.; Feldmann, J.; Go¨bel, E. O.; Taliani, C.; Mohn, H.; Mu¨ller, W.; Ha¨usler, P.; ter Meer, H. U. Phys. ReV. Lett. 1994, 72, 2644. (3) Shin, E.-j.; Park, J.; Lee, M.; Kim, D.; Suh, Y. D.; Yang, S. I.; Jin, S. M.; Kim, S. K. Chem. Phys. Lett. 1993, 209, 427. (4) Cheville, R. A.; Halas, N. J. Phys. ReV. B 1992, 45, 4548. (5) Juhasz, T.; Hu, X. H.; Suarez, C.; Bron, W. E.; Maiken, E.; Taborek, P. Phys. ReV. B 1993, 48, 4929. (6) Brorson, S. D.; Kelly, M. K.; Wenschuh, U.; Buhleier, R.; Kuhl, J. Phys. ReV. B 1992, 46, 7329. (7) Thomas, T. N.; Taylor, R. A.; Ryan, J. F.; Mihailovic, D.; Zamboni, R. Europhys. Lett. 1994, 25, 403. (8) Ebbeson, T. W.; Mochizuki, Y.; Tanigaki, K.; Hiura, H. Europhys. Lett. 1994, 25, 503. (9) Dexheimer, S. L.; Vareka, W. A.; Mittleman, D.; Zettl, A.; Shank, C. V. Chem. Phys. Lett. 1992, 196, 427. (10) Dexheimer, S. L.; Vareka, W. A.; Mittleman, D.; Zettl, A.; Shank, C. V. Chem. Phys. Lett. 1995, 235, 552. (11) Byrne, H. J.; Maser, W.; Ruhle, W. W.; Mittelbach, A.; Honle, W.; von Schnering, H. G.; Movaghar, B.; Roth, S. Chem. Phys. Lett. 1993, 204, 461. (12) Flom, S. R.; Pong, R. G. S.; Bartoli, F. J.; Kafafi, Z. H. Phys. ReV. B 1992, 46, 15598. (13) Fleischer, S. B.; Ippen, E. P.; Dresselhaus, G.; Dresselhaus, M. S.; Rao, A. M.; Zhou, P.; Eklund, P. C. Appl. Phys. Lett. 1993, 62, 3241. (14) Farzdinov, V. M.; Lozovik, Yu. E.; Matveets, Yu. A.; Stepanov, A. G.; Letokhov, V. S. J. Phys. Chem. 1994, 98, 3290. (15) Rosker, M. J.; Marcy, H. O.; Chang, T. Y.; Khoury, J. T.; Hansen, K.; Whetten, R. L. Chem. Phys. Lett. 1992, 196, 427. (16) Hess, B. C.; Forgy, E. A.; Frolov, S.; Dick, D. D.; Vardeny, Z. V. Phys. ReV. B 1994, 50, 4871. (17) Lee, M.; Kim, D. J. Opt. Soc. Korea 1990, 1, 52. (18) Matus, M.; Kuzmany, H.; Sohmen, E. Phys. ReV. Lett. 1993, 68, 2822. (19) Kao, K. C.; Hwang, W. Electrical Transport in Solids, Pergamon press, Oxford, 1981. (20) Lee, C. H.; Yu, G.; Moses, D.; Srdanov, V. I.; X. Wei, X.; Vardeny, Z. V. Phys. ReV. B 1993, 48, 8506. (21) Powell, R. C.; Soos, Z. G. Phys. ReV. B 1972, 5, 1547. (22) Rose, T. S.; Righini, R.; Fayer, M. D. Chem. Phys. Lett. 1984, 106, 13. (23) Kim, D.; Lee, M.; Suh, Y. D.; Kim, S. K. J. Am. Chem. Soc. 1992, 114, 4429. (24) Lee, M.; Song, O.-K.; Seo, J.-C.; Kim, D.; Suh, Y. D.; Jin, S. M.; Kim, S. K. Chem. Phys. Lett. 1992, 196, 325. (25) Kobayashi, T. Relaxation in Polymer; World Scientific: Singapore, 1993; pp 1-79.

JP953642V