19764
J. Phys. Chem. 1996, 100, 19764-19770
Ultrafast Investigation of Vibrational Relaxation in the S1 Electronic State of HITC Ignacio Martini and Gregory V. Hartland* Department of Chemistry and Biochemistry, UniVersity of Notre Dame, Notre Dame, Indiana 46556 ReceiVed: July 31, 1996X
The condensed phase relaxation dynamics of electronically excited 1,1′,3,3,3′,3′-hexamethylindotricarbocyanine iodide HITC (a cyanine dye) has been examined by transient bleach/stimulated emission experiments. These measurements were performed using tunable pump and probe laser pulses with ∼200 fs time resolution. The dynamics observed was assigned to vibrational relaxation in the S1 state of HITC. Solvation effects make a negligible contribution to these experiments because the dipole moment of HITC only changes by a small amount when the S1 r S0 transition is excited. Experiments performed with variable wavelength pump and probe pulses show that vibrational relaxation is faster at high energies in the S1 state. At low energies in the S1 state the vibrational relaxation times depend on the solvent. The measured relaxation times at low energies are1.7 ps in acetonitrile, 3.2 ps in dimethyl sulfoxide, 2.4 ps in methanol, 3.5 ps in ethanol, 7.1 ps in 1-butanol, and 6.4 ps in ethylene glycol. These results show that the vibrational relaxation rate decreases with solvent viscosity and increases with solvent dipole moment. To explain these observations, we propose that the torsional motion associated with isomerization in the S1 state of HITC is responsible for vibrational deactivation at low energies. Rotation about the central C-C bond in the HITC polyene chain produces an internal charge transfer state, creating a large dipole moment along the long axis of the molecule. Thus, the torsional vibration generates an oscillating dipole which can couple to the dipole moments of the solvent molecules, providing a mechanism for energy exchange between HITC and the solvent. The rate of energy exchange will increase with the solvent dipole moment and decrease with the solvent viscosity because highly viscous solvents hinder the torsional vibration and, so, reduce the magnitude of the induced dipole moment in HITC.
Introduction Two important factors in determining the rates of condensed phase reactions are energy exchange between the reactant and the solvent and how the solvent environment responds to changes in the charge distribution of the reacting species.1,2 One way of probing the time scales for these processes is to examine the relaxation of electronically excited dye molecules in solution.1-4 Ultrafast dynamic Stokes shift measurements can reveal detailed information about solvation dynamics (the reorganization of the solvent in response to a change in the dipole moment of the solute).4-7 In the past decade these measurements have been performed for a wide variety of systems.5-7 On the other hand, vibrational relaxation measurements give information about the rate of energy exchange between the solvent and the solute. For example, the time scale of vibrational relaxation for azulene in the ground electronic state was found to vary from 20 to 180 ps, depending on solvent.8,9 Recent measurements for trans-stilbene (also in the ground electronic state) gave vibrational relaxation times of 1-2 ps, with a subsequent 10 ps time scale cooling process for the solvent/solute system.10 Studies of relaxation in excited electronic states of dye molecules show that intramolecular vibrational energy redistribution (IVR) occurs in 200-500 fs.11-13 However, very few studies of the subsequent vibrational population relaxation process have been reported. A recent exception is the measurement of a 1 ps vibrational relaxation time in the S1 state of an IR dye, IR-125.14 A major difficulty with studying vibrational relaxation in excited electronic states is eliminating the effects of solvation dynamics. A great deal of effort has been expended in the reverse problem, i.e., measuring solvation dynamics without * To whom correspondence should be addressed. E-mail: hartland.1Ind.edu. X Abstract published in AdVance ACS Abstracts, November 15, 1996.
S0022-3654(96)02335-0 CCC: $12.00
complications from vibrational relaxation.5-7 For example, in a time-resolved fluorescence study of coumarin 153 (C153) Maroncelli and co-workers showed that IVR from the FranckCondon excited modes is very fast (200 ps.24 Thus, overall rotation of HITC does not contribute to our 10-30 ps time scale experiments, and parallel pump and probe beam polarizations were used. For acetonitrile and methanol, where rotational effects may occur on time scales of 1 ns HITC lifetime. The two exponentials account for the stimulated emission signal. Our analysis shows that the stimulated emission signal has a fast 200-300 fs rise followed by a slower rise that varies for different solvents. The time scale for the slower process is 1.7 ps for acetonitrile and 6.4 ps for ethylene glycol. (The 6.4 ps rise time was determined from data recorded over 30 ps delay.) The stimulated emission signal makes up ∼30% of the total signal. This data is qualitatively very similar to the data obtained using ethanol as a solvent.20 The major difference is the slow rise in the stimulated emission signal, which is 3.5 ps in ethanol.26 The steady state absorption and emission spectra of HITC in acetonitrile and ethylene glycol are shown in Figure 3. Clearly, the absorption and emission spectra are very similar for these two solvents. The Stokes shift for HITC, which is calculated as the wavenumber difference between the peaks in the absorption and emission spectra, is 670 cm-1 in acetonitrile and 640 cm-1 in ethylene glycol. In comparison, the Stokes shift for C153 (a molecule that has been extensively used to study solvation dynamics) is 6030 cm-1 in acetonitrile and 5920 cm-1 in ethylene glycol.15 Note that the Stokes shift arises from the
Figure 4. SPARTAN AM1/CI structures of HITC in the all-trans geometry (top) and with a torsional angle of 90° around the central C-C bond (bottom).
frequency change due to solvation and from the displacement of the Franck-Condon envelopes of the absorption and emission bands. By analyzing absorption and emission spectra recorded in a nonpolar solvent, Maroncelli and co-workers were able to determine that the frequency shift due to solvation for C153 is 2230 cm-1 in acetonitrile and 1870 cm-1 in ethylene glycol.15 The large solvation effect arises from the large 8-10 D change in the C153 dipole moment upon excitation of the S1 r S0 transition.27 Also note that the frequencies of the C153 absorption and fluorescence maxima scale linearly with the solvent polarity (as measured, for example, by the π* parameter).15 This correlation is expected when solvation effects make a significant contribution to the relative energies of the S1 and S0 states. For HITC there is no correlation between the fluorescence and absorption maxima and solvent polarity. The change in dipole moment of HITC upon electronic excitation was estimated by semiempirical AM1/CI calculations using SPARTAN 4.0.28 The ground state geometry was fully optimized at the AM1 level, and the ground and first excited singlet state wave functions were generated by a single-point CI calculation with the AM1 Hamiltonian (single and double excitations including 10 electrons and 10 orbitals). The calculations were performed for a series of rotational angles (0°-180° in steps of 15°) around the central C-C bond in the polyene chain. The 0° (all-trans) and 90° (perpendicular) structures are shown in Figure 4 for reference. The all-trans structure (the normal form) of HITC is responsible for the steady state absorption and emission spectra.22,23 The calculated energy difference between the two
Vibrational Relaxation of S1 HITC
J. Phys. Chem., Vol. 100, No. 51, 1996 19767
electronic states in the all-trans configuration is 1.68 eV, yielding a S1 T S0 transition wavelength of 738 nm. This is in excellent agreement with the experimental absorption spectra; see Figure 3. The shape of the ground state potential energy surface as a function of angle is also reasonably well reproduced (i.e., there is a maximum at the perpendicular isomer). However, the S1 potential energy surface predicted by our AM1 calculation does not agree with the accepted form for HITC. (The discrepancies may arise, in part, because the effect of the solvent is not taken into account in these calculations.) Thus, we consider that the calculated energies and dipole moments for the 0° structure are reasonably accurate, but the results for other angles should be viewed as only qualitatively correct. The AM1/CI calculations show that the dipole moment for the all-trans structure of HITC changes from 1.3 D in the ground state to 2.8 D in the S1 excited state and that the direction of the dipole moment is along the y axis, as shown in Figure 4, for both electronic states. Similar calculations for C153 give a dipole moment of 7 D for the S0 state and 19 D for the S1 state, in reasonable agreement with the experimental values27 and with previously reported AM1/ CI calculations.15 The small 1.5 D change in the HITC dipole moment upon electronic excitation is consistent with the small Stokes shift observed in the steady state spectra. To estimate the actual size of the frequency shift ∆ν due to solvation, we use the reaction field theory expression29
∆ν )
[
]
2µ b1‚(µ b1 - b µ 2) 0 - 1 n2 - 1 + 0 + 2 n2 + 2 hca3
[ ]
µ12 - µ22 n2 - 1 (1) hca3 n2 + 2
where 0 is the dielectric constant, n is the refractive index, a is the radius of the assumed spherical solvent cavity which µ2 are the dipole moments in contains the solute, and b µ1 and b the ground and excited electronic states of the solute, respectively. Rather than use eq 1 to calculate ∆ν directly, the ratio of ∆ν for HITC versus C153 was determined from the values µ2 and by assuming that HITC is 2 times larger than of b µ1 and b C153.30 We find that ∆ν(C153)/∆ν(HITC) ≈ 50. Using a typical value of ∆ν ) 2000 cm-1 for C153,15 the frequency shift due to solvation for HITC is estimated to be ∆ν ≈ 40 cm-1. Displacing the steady state fluorescence spectra shown in Figure 3 by 40 cm-1 to the blue changes the Franck-Condon factors for emission at 780 nm by 320 cm-1. We feel that this is unrealistically large, given that the frequency difference between the peaks in the steady state absorption and emission spectra is only 600-700 cm-1. Another piece of evidence to show that solvation plays a minor role in our experiments is presented in Figure 5 where transient bleach/stimulated emission data are shown for HITC in ethanol. These experiments were performed with a 780 nm probe laser and a 650 or 780 nm pump laser. Fits to the data using a step function to represent the transient bleach and one or two exponentials to represent the stimulated emission signal are also shown in Figure 5. The 650 nm pump/780 nm probe data contain a stimulated emission signal that has a biexponential rise with time constants of 300 fs and 3.5 ps. This signal accounts for ca. 30% of the total signal. In contrast, the degenerate 780 nm pump/probe data shows an instantaneous
Figure 5. Transient bleaching/stimulated emission data for HITC in ethanol recorded with a probe wavelength of 780 nm and a pump laser wavelength of 780 nm (top) or 650 nm (bottom). The solid lines are the experimental data, and the dashed lines are the fitted signals.
rise due to ground state depletion, a strong coherence spike at t ) 0 that obscures the short time dynamics, and a 1.1 ps singleexponential decay that accounts for 10% of the signal amplitude. The coherent coupling artifact only occurs when the pump and probe laser have the same wavelength.4,31 The fit to the 780 nm pump data was obtained by setting the zero in time to the peak of the coherent coupling artifact. Solvation dynamics should make an identical contribution to the signal for the two experiments shown in Figure 5, i.e., in the Born-Oppenheimer approximation the change in dipole moment upon electronic excitation does not depend on the vibrational energy content of the S1 state. Because the dynamics observed in the 780 nm pump and 650 nm pump experiments are completely different, we conclude that the biexponential rise in the 650 nm pump/780 nm probe signal cannot be assigned to solvation dynamics. Note that the small 1.1 ps decay in the 780 nm pump/780 nm probe data also cannot be due to solvation dynamics. Solvation always causes a red shift in the emission spectrum. Because the 780 nm probe laser wavelength corresponds to the maximum of the steady state (i.e., fully solvated) fluorescence spectrum, the red shift due to solvation would increase (not decrease) the stimulated emission signal. There are two possible explanations for the 1.1 ps decay signal. First, the 780 nm pump laser creates molecules with a small amount of vibrational energy in the S1 state. The decay in the 780 nm pump/probe data is then due to vibrational relaxation of these species. A second possibility is that the 780 nm pump laser excites some hot band transitions, and the decay arises from a ground state thermalization process. Our experiments cannot distinguish between these two possibilities. Vibrational Relaxation as a Function of Pump and Probe Laser Wavelength. In our previous communication data recorded for HITC in ethanol with a 780 nm probe laser and pump laser wavelengths of 600, 650, and 700 nm was presented.20 The fast and slow rise times determined from these fixed probe/variable pump laser experiments were the same (within experimental error) for all three pump laser wavelengths. Transient bleach/stimulated emission data recorded with a fixed pump laser wavelength of 600 nm and probe laser wavelengths of 600, 700, and 780 nm are shown in Figure 6 for HITC in ethanol (600 nm is the lowest pump laser wavelength we can use to excite the HITC S1 r S0 band). Also shown in Figure 6 are fits to the data using a step function plus a sum of exponentials. In addition to the transient bleach signal, the 600 nm pump/600 nm probe data has a coherent coupling signal (the spike centered around t ) 0) and a slow 3 ( 1 ps decay that accounts for ∼5% of the total signal. This small-amplitude
19768 J. Phys. Chem., Vol. 100, No. 51, 1996
Martini and Hartland TABLE 1: Relaxation Times and Steady State Spectral Characteristics for HITC in the Different Solvents Used for This Study solvent
τvib (ps)
acetonitrile dimethyl sulfoxide methanol ethanol 1-butanol ethylene glycol
1.7 ( 0.1 3.2 ( 0.7 2.4 ( 0.2 3.5 ( 0.1 7.1 ( 0.2 6.4 ( 0.3
νabs νem τsolva µb ηb (103 cm-1) (103 cm-1) (ps) (D) (cP) 13.49 13.33 13.51 13.46 13.39 13.33
12.82 12.63 12.81 12.72 12.63 12.69
0.15 0.90 2.3 10.9 47 9.3
3.5 .34 4.1 2.0 1.7 0.54 1.7 1.1 1.8 2.3 2.3 18
a Taken from ref (15). b Dipole moments and viscosities were obtained from: Riddick, J. A.; Bunger, W. B. Organic SolVents, 3rd ed.; Wiley: New York, 1970. The viscosities are for 25 °C.
Figure 6. Transient bleach/stimulated emission data for HITC in ethanol recorded with a pump laser wavelength of 600 nm and probe laser wavelengths of 600, 700, and 780 nm: (s) Experimental data; (‚‚‚) fitted signal.
decay is probably due to a ground state thermalization process. The 700 nm probe data show a stimulated emission signal that has a fast rise and decay followed by a slower