19 October 2001

Chemical Physics Letters 347 (2001) 108±114

www.elsevier.com/locate/cplett

Excitation-wavelength dependence of the femtosecond

¯uorescence dynamics of 7-azaindole dimer: further evidence
for the concerted double proton transfer in solution
Satoshi Takeuchi a, Tahei Tahara a,b,*
a
Institute for Molecular Science, Myodaiji, Okazaki 444-8585, Japan
b
The Institute of Physical and Chemical Research (RIKEN), 2-1 Hirosawa, Wako, 351-0198, Japan
Received 12 July 2001; in ®nal form 10 August 2001

Abstract
Excitation-wavelength dependence of the ultraviolet ¯uorescence dynamics of 7-azaindole dimer was examined in
solution by femtosecond up-conversion method. It was found that the ¯uorescence decay of the dimer excited state
showed a signi®cant wavelength dependence. It changes from a bi-exponential decay to a single-exponential decay,
when we scanned the excitation wavelength from 280 toward 313 nm (the red-edge of the dimer absorption). This result
demonstrates that the proton-transfer dynamics itself exhibits a single exponential behavior. The obtained ¯uorescence
data deny the appearance of the intermediate species and strongly support the concerted mechanism of the double
proton transfer reaction. Ó 2001 Elsevier Science B.V. All rights reserved.

1. Introduction concerted way (without appearance of any inter-

mediate species) or the step-wise way (with ap-
Excited-state proton transfer has been receiving pearance of an intermediate in which only one
a great deal of attention since it plays crucial roles proton is transferred).
in a variety of photochemical processes. The Time-resolved measurements in the pico- and
7-azaindole dimer is one of the most important femtosecond time region are very crucial to study
systems showing the excited-state proton transfer the mechanism of this important reaction. Con-
(Fig. 1). This dimer has been studied extensively as cerning the reaction in solution, Share et al. [3]
a simpli®ed model of hydrogen-bonded base pairs carried out the ®rst femtosecond ¯uorescence
in DNA [1,2]. Moreover, the 7-azaindole dimer is measurement and found that the tautomer ¯uo-
very important as a prototypical molecular system rescence appears with a time constant of about
where we can examine the mechanism of the 1 ps. We carried out systematic ¯uorescence mea-
double proton transfer reactions. An important surements over a wide ¯uorescence wavelength
question about the double proton transfer is range from 320 to 620 nm with an improved time-
whether two protons are translocated in the resolution [4,5]. We succeeded in observing the
¯uorescence from the dimer excited state (precur-
sor of the reaction) in the ultraviolet region, and
*
Corresponding author. Fax: +81-564-54-2254. found that the dimer ¯uorescence decays in ac-
E-mail address: tahara@ims.ac.jp (T. Tahara). cordance with the appearance of the tautomer
0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 6 1 4 ( 0 1 ) 0 1 0 2 5 - 9
 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114 109

served in femtosecond time-resolved ion-detected

pump-probe spectroscopy [8] as well as femtosec-
ond time-resolved Coulomb-explosion [9].
These di€erent interpretations about the fem-
tosecond dynamics of the 7-azaindole dimer in
solution initiated very intense debates, and the
mechanism of the excited-state double proton
transfer is now a subject of discussion [10±17].
Fig. 1. Photoinduced double proton transfer reaction of Obviously, the assignment of the 0.2 and 1.1-ps
7-azaindole dimer. components is the key to elucidate the reaction
mechanism. Since the two di€erent interpretations
¯uorescence in the visible region. More interest- have been proposed on the basis of the common
ingly, a quantitative analysis revealed that the observation (existence of the two components in
decay of the dimer ¯uorescence is not a single-ex- the precursor dynamics), we need further experi-
ponential but a double exponential having time mental data in order to clarify which interpreta-
constants of 0.2 and 1.1 ps. In the molecules like tion is correct.
azaindole or indole, it is well known that there In this Letter, we report an excitation-wave-
exist two closely-lying excited states called 1 La and length dependence of the femtosecond ¯uorescence
1
Lb , and that the transitions to these two excited dynamics of the 7-azaindole dimer in a nonpolar
states overlap largely in solution to form one ab- solvent. Experiments for the excitation-wavelength
sorption band [1,6]. On the basis of the recon- dependence can, indeed, clearly distinguish the two
structed time-resolved ¯uorescence spectra, mechanisms (Fig. 2): if the concerted mechanism is
evaluated oscillator strengths, and time-resolved relevant, the lowest-energy absorption band of the
anisotropy data, we concluded that the 0.2-ps 7-azaindole dimer consists of two transitions, the
component in the dimer ¯uorescence corresponds S2 Lb † S0 and S1 La † S0 transitions. Since
to the electronic relaxation from the 1 Lb (S2 ) to the the S1 S0 transition should be located in a
1
La (S1 ) state, and the 1.1-ps component to the slightly lower energy region, varying the excita-
double proton transfer taking place from the 1 La tion-wavelength across the absorption band
state. In other words, we concluded that the pro- should change the relative S2 and S1 populations
ton-transfer dynamics itself exhibits a single ex- that are prepared by the photoexcitation. It means
ponential behavior (the concerted mechanism). that we should observe a signi®cant change in the
After our work, Fiebig et al. [7] carried out fem- relative amplitude of the 0.2 and 1.1-ps compo-
tosecond time-resolved ¯uorescence and absorp- nents because they re¯ect the population of the S2
tion measurements and con®rmed the existence and S1 states, respectively. By sharp contrast, in
of the two components in the dynamics of the the case of the step-wise mechanism, the absorp-
`precursor' excited state. However, they assigned tion band consists of a single electronic transition,
the 0.2-ps component to the ®rst proton transfer and the 0.2 and 1.1-ps components are attributed
and the 1.1-ps component to the second proton to the successive proton-transfer steps that start
transfer on the basis of the deuterium e€ect that from the single excited state. Therefore, the change
they recognized in time-resolved absorption mea- of the excitation wavelength should not a€ect
surements. They concluded that the intermediate the relative amplitude of the 0.2 and the 1.1-ps
species, in which only one proton is transferred, is components because the relative amplitude is
populated and appears during the double proton determined by the inherency of the successive
transfer reaction. In other words, they claimed proton-transfer steps.
that the proton-transfer dynamics itself shows an The ¯uorescence decay due to the precursor ex-
intrinsic bi-exponential behavior (step-wise mech- cited state(s), which we report in this Letter, showed
anism). The step-wise mechanism was originally a signi®cant change with the change of the excita-
proposed to interpret the gas-phase dynamics ob- tion wavelength. Especially, it showed a single
110 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114

Fig. 2. Two reaction mechanisms proposed for the double proton transfer of 7-azaindole dimer. (A) The concerted mechanism: the
double proton transfer proceeds from the S1 state after the S2 ! S1 electronic relaxation. No intermediate species appears during the
reaction. The lowest-energy absorption band of the 7-azaindole dimer (solid curve) consists of the S1 La † S0 and the S2 Lb † S0
absorption bands (dotted curves). (B) The step-wise mechanism: two successive proton-transfer steps start from a single excited state.
An intermediate species, in which only one proton is transferred, appears between the ®rst and the second proton transfers.

exponential decay when we excited the red-edge of verted signal is analyzed by a monochromator, and
the absorption band of the 7-azaindole dimer. is detected by a photon-counting system. The ¯u-
orescence detection at the magic angle was achieved
by rotating the excitation polarization with respect
2. Experimental to the gate polarization. The time-resolution was
evaluated as 290 fs from the measurement for in-
The experimental setup for the femtosecond stantaneous Raman signals of the solvent. This
¯uorescence up-conversion measurements is es- instrumental response time varied only slightly
sentially the same as that described previously [4,5]. (<5 %) depending on the excitation wavelength.
Brie¯y, the light source is a mode-locked Ti:sap- 7-azaindole (Aldrich) was recrystallized twice
phire laser (Spectra Physics, Tsunami) pumped by from cyclohexane, and was dried in vacuo before
an argon ion laser. The oscillator laser is tunable in use. Liquid-chromatography grade hexane (Wako
the wavelength region of 840±940 nm, and pro- Chemicals) was dried by molecular sieves. A fresh
duces a pulse train (600±800 mW) having a typical sample solution (10 2 mol dm 3 ) was prepared for
pulse duration of 75 fs. The third-harmonic (30 each time-resolved measurement.
mW) of the oscillator output was used as an exci-
tation pulse and is focused into a sample jet. The
residual fundamental pulse is used as a gate pulse 3. Results and discussion
for the up- conversion process. The emitted ¯uo-
rescence is collected into a b-BaB2 O4 crystal, and is In a concentrated solution (10 2 mol dm 3 )
up-converted with the gate pulse. The up- con- used in the present study, ca. 87% of the 7-azain-
 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114 111

dole molecules form the dimer and the rest remains observed up to several picoseconds is directly re-
as the monomer in the ground state [5]. Therefore, lated to the dynamics of the dimer in question. It
the raw absorption spectrum of the solution con- should be noted that this dimer ¯uorescence can-
tains a contribution of the dimer as well as the not be recognized in the steady-state (time-inte-
monomer. Fig. 3 depicts absorption spectra of the grated) spectra because of its very short lifetime. It
7-azaindole dimer and monomer, which were ob-
tained by the spectral decomposition procedure
described previously [5]. As seen in this ®gure, the
lowest energy absorption band of the dimer is red-
shifted by 10 nm with respect to the monomer
band. At the bottom of the ®gure, the spectra of
the excitation pulses employed in the present study
(280±313 nm) are also shown. These six wave-
lengths cover the major portion of the dimer ab-
sorption, including the red-edge of the absorption
band (313 nm). Steady-state ¯uorescence spectra
measured with all these excitation wavelengths
exhibited an identical tautomer-band in the visible
region. This assures that the proton transfer re-
action occurs regardless of the change of the ex-
citation wavelength, including the red-edge
excitation at 313 nm.
Femtosecond time-resolved ¯uorescence was
measured with these six excitation wavelengths at
room temperature, and the obtained ¯uorescence
traces are shown in Fig. 4. The ¯uorescence de-
tection wavelength is 380 nm, where the ¯uores-
cence from the dimer excited state(s) (before the
reaction) is predominant. Therefore, a rapid decay

Fig. 4. Up-converted ¯uorescence signals of 7-azaindole in

hexane 10 2 mol dm 3 † obtained with photoexcitation at 280
(a), 287 (b), 293 (c), 300 (d), 307 (e), and 313 nm (f). A typical
Fig. 3. Absorption spectra of the 7-azaindole dimer (a) and instrumental response (instantaneous Raman signals of the
monomer (b) obtained by the spectral decomposition proce- solvent) is also shown (g). The open circles connected by lines
dure. Spectra of the six excitation pulses used in the present are experimental data points, and the dashed curve represents
study are also shown at the bottom of the ®gure. the calculated monomer component (see text).
112 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114

becomes noticeable only in femtosecond time-re-

solved measurements.
A long-lived component seen in every ¯uores-
cence trace is due to the monomer that coexists in
the solution. The relative amplitude of the mono-
mer component varies with the change of the ex-
citation wavelength, re¯ecting the absorption
spectral di€erence between the dimer and mono-
mer (Fig. 3). The lifetime of the monomer ¯uo-
rescence has been determined to be as long as 1 ns
(for the 10 2 mol dm 3 solution) [5], so that it can
be considered as an almost constant o€set in the
time region of the present measurements. How-
ever, in order to focus the temporal behavior of the
dimer ¯uorescence, we subtracted the monomer
component from the raw data. In this subtraction
procedure, a temporal behavior of the monomer
component was represented by an exponential
function having the lifetime of 1 ns. It was con-
voluted with the instrumental response, and the
resultant function was ®tted to the raw data in a
late delay time region (>10 ps) where the dimer
¯uorescence is negligible. The calculated monomer
component is shown by dashed curves in Fig. 4.
In Fig. 5A, the dimer ¯uorescence obtained by
the subtraction procedure is plotted after an in-
tensity-normalization for three excitation wave-
lengths (287, 300, and 313 nm). As clearly seen, the
temporal behavior of the dimer ¯uorescence shows
remarkable excitation-wavelength dependence, es-
pecially in the early time region. In order to discuss
this dependence more quantitatively, the dimer
¯uorescence measured with all the six excitation
wavelengths are plotted in a logarithmic scale in
Fig. 5B. The dotted straight line drawn for each
curve corresponds to the 1.1-ps component
s2 ˆ 1:1  0:1 ps†, and the deviation from this Fig. 5. (A) The ¯uorescence decay of the dimer excited state(s)
straight line, which is recognized in early delay measured with the excitation wavelengths at 287 (dashed-dot-
times, is ascribed to the 0.2-ps component ted), 300 (dotted), 313 nm (solid). (B) Logarithmic plot of the
s1 ˆ 0:2  0:1 ps†. It is very clear that the ob- ¯uorescence decay of the dimer excited state(s) measured with
the excitation wavelengths at 280 (a), 287 (b), 293 (c), 300 (d),
served excitation-wavelength dependence comes
307 (e), and 313 nm (f). The dotted straight line drawn for each
from the change of the relative amplitude of the data corresponds to a 1.1-ps single-exponential decay.
0.2-ps component. In fact, the bi-exponential fea-
ture of the dimer ¯uorescence is pronounced when
the dimer is excited at short wavelengths. (It is component becomes signi®cantly smaller. Even-
consistent with our previous results obtained with tually, this 0.2-ps component vanishes when the
the 270-nm excitation [5].) However, as the exci- red-edge of the absorption band is excited: the
tation wavelength becomes longer, the 0.2-ps dimer ¯uorescence shows essentially a single
 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114 113

exponential decay. This fact implies that only one tion more eciently. Therefore, the 0.2-ps com-
excited state is generated with the red-edge pho- ponent, which is ascribed to the decay of the S2
toexcitation at 313 nm and that the proton transfer state, becomes more prominent. On the contrary,
starts directly from this state. The relative ampli- when the 7-azaindole dimer is excited with a
tude of the 0.2 and 1.1-ps components, which was longer-wavelength light, the S1 state is produced
obtained by a ®tting analysis using a bi-exponen- more eciently. Finally, by the red-edge excita-
tial function, is listed for each excitation wave- tion, only the S1 state is excited so that the dy-
length in Table 1. namics corresponding to the S2 ! S1 relaxation
In the context of the step-wise mechanism, the does not appear. The present experimental data
0.2 and 1.1-ps components are assigned to the ®rst are fully consistent with our previous conclusion
and the second proton transfer time, and hence that the 0.2-ps component arises from the S2 ! S1
these two time constants correspond to the ap- electronic relaxation before the reaction and that
pearance and disappearance of the intermediate in the 1.1-ps component is ascribed to the `concerted'
which only one proton is transferred [7] (Fig. 2B). double proton transfer dynamics [5]. Fiebig et al.
Therefore, bi-exponential nature of the time-re- [7] reported that the lifetime of the 0.2-ps compo-
solved ¯uorescence is inherent to the proton nent is a€ected by the deuterium substitution at
transfer reaction itself and the relative amplitude the NH site. However, this deuterium e€ect does
of the two components should not change signi®- not contradict our conclusion, since it has been
cantly with the change of the excitation wave- pointed out that the internal conversion rate is also
length. This does not agree with the experimental a€ected by the deuteration [18].
data showing that the decay of the dimer ¯uores- We have also examined the excitation-wave-
cence exhibits signi®cant excitation-wavelength length dependence of the visible ¯uorescence and
dependence. Furthermore, the absence of the 0.2- carried out time-resolved anisotropy measure-
ps component in the ¯uorescence decay obtained ments. The details of these systematic studies will
with the red-edge excitation directly discloses that be soon presented in a forthcoming Letter. We
the proton transfer occurs irrespective of the ap- here note that all the ¯uorescence data consistently
pearance of the 0.2-ps component. It means that deny the existence of the intermediate species
this 0.2-ps component has nothing to do with the during the double proton transfer of the 7-azain-
actual translocation of the proton. dole dimer and support the concerted mechanism
On the other hand, the present experimental in solution.
data are readily rationalized in the context of the
concerted mechanism. Since the S2 S0 absorp-
tion band is located in higher energy region than Acknowledgements
the S1 S0 absorption band, the photoexcitation
at a shorter wavelength generates the S2 popula- This work was partly supported by a Grant-in-
Aid for Scienti®c Research (B) (No. 13440183)
Table 1 from Japan Society for Promotion of Science
Relative amplitude of the 0.2 and 1.1-ps components appearing (JSPS) and a Grant-in-Aid for Scienti®c Research
in the decay of the dimer ¯uorescence on Priority Area (A) (No. 12042283) from Minis-
Excitation wavelength Relative amplitudea try of Education, Culture, Sports, Science and
(nm) Technology.
280 0.35
287 0.30
293 0.21 References
300 0.29
307 0.09 [1] C.A. Taylor, M.A. El-Bayoumi, M. Kasha, Proc. Natl.
313 <0.015 Acad. Sci. USA 63 (1969) 253.
a
The relative amplitude of the 0.2-ps component against that of [2] K.C. Ingham, M.A. El-Bayoumi, J. Am. Chem. Soc. 96
the 1.1-ps component. (1974) 1674.
114 S. Takeuchi, T. Tahara / Chemical Physics Letters 347 (2001) 108±114

[3] P. Share, M. Pereira, M. Sarisky, S. Repinec, R.M. [12] J. Catalan, J.C. del Valle, M. Kasha, Chem. Phys. Lett. 318
Hochstrasser, J. Luminescence 48/49 (1991) 204. (2000) 629.
[4] S. Takeuchi, T. Tahara, Chem. Phys. Lett. 277 (1997) 340. [13] D.E. Folmer, E.S. Wisniewski, A.W. Castleman Jr., Chem.
[5] S. Takeuchi, T. Tahara, J. Phys. Chem. A 102 (1998) 7740. Phys. Lett. 318 (2000) 637.
[6] B. Valeur, G. Weber, Photochem. Photobiol. 25 (1977) [14] A. Douhal, M. Moreno, J.M. Lluch, Chem. Phys. Lett. 324
441. (2000) 75.
[7] T. Fiebig, M. Chachisvilis, M.M. Manger, A.H. Zewail, [15] A. Douhal, M. Moreno, J.M. Lluch, Chem. Phys. Lett. 324
J. Phys. Chem. A 103 (1999) 7419. (2000) 81.
[8] A. Douhal, S.K. Kim, A.H. Zewail, Nature 378 (1995) 260. [16] J.C. del Valle, M. Kasha, J. Catalan, Int. J. Quantum
[9] D.E. Folmer, L. Poth, E.S. Wisniewski, A.W. Castleman Chem. 77 (2000) 118.
Jr., Chem. Phys. Lett. 287 (1998) 1. [17] M. Moreno, A. Douhal, J.M. Lluch, O. Castano, L.M.
[10] V. Guallar, V.S. Batista, W.H. Miller, J. Chem. Phys. 110 Frutos, J. Phys. Chem. A 105 (2001) 3887.
(1999) 9922. [18] V.J.P. Srivatsavoy, B. Venkataraman, N. Periasamy,
[11] J. Catalan, J.C. del Valle, M. Kasha, Proc. Natl. Acad. Sci. J. Photochem. Photobiol. A: Chem. 68 (1992) 169.
USA 96 (1999) 8338.