Ultrafast relaxation processes of triarylpyrylium cations

Article abstract of DOI:10.1021/jp9940248

The time evolution of the fluorescence and the photoinduced absorption spectra of four triarylpyrylium cations, differing in the number of dodecyl chains attached to the chromophore, was studied by means of femtosecond fluorescence upconversion and picose

Full text of DOI:10.1021/jp9940248

J. Phys. Chem. A 2000, 104, 5181-5189
5181
Ultrafast Relaxation Processes of Triarylpyrylium Cations
‡
†
,†
‡
Vidmantas Gulbinas, Dimitra Markovitsi, Thomas Gustavsson,* Renata Karpicz, and
Mich e` le Veber§
Institute of Physics, Gostauto St. 12, 2600 Vilnius, Lithuania, CEA/Saclay, DRECAM/SPAM, CNRS-URA 331,
9
1191 Gif-sur-YVette, France, and Laboratoire de Physique des Solides, CNRS-UMR 8502, B aˆ t. 510,
UniVersit e´ de Paris Sud, 91405 Orsay, France
ReceiVed: NoVember 12, 1999; In Final Form: March 6, 2000
The time evolution of the fluorescence and the photoinduced absorption spectra of four triarylpyrylium cations,
differing in the number of dodecyl chains attached to the chromophore, was studied by means of femtosecond
fluorescence upconversion and picosecond pump-probe absorption techniques. The dependence on solvent
viscosity was also examined. The results are rationalized in terms of excited and ground-state relaxation
dynamics and may be differentiated into intramolecular conformational changes and intermolecular solute-
solvent contributions. The dynamic fluorescence shift is related to solvent relaxation, whereas the fluorescence
intensity decay is attributed to molecular twisting acting in an intricate manner. First, the geometrical relaxation
reduces rapidly the fluorescence oscillator strength, but it also greatly enhances the nonradiative S r S
0 1
internal conversion rate. Molecular back-twisting combined with vibrational relaxation in the ground state is
shown to be the last and the slowest process of the reaction.
1. Introduction
Relaxation processes of molecules undergoing an important
photoinduced change in their atomic charge distribution are
known to be quite complex, even in the absence of photochemi-
1
cal reactions or interchromophore interactions. The way that
conformational relaxation processes depend on the solvent
viscosity and the solute structure is of fundamental interest. In
particular, solvent rearrangement may be associated with or
control conformational modifications occurring both in the
excited and the ground state. A major question is whether the
solvent and the solute dynamics are coupled or if they take place
+
Figure 1. The studied triarylpyrylium tetrafluoroborates: P1-1 (R )
+
+
2
R′ ) CH
3
, R′′ ) H); P1-12 (R ) CH
3
, R′ ) C12
H
25, R′′ ) H); P12-12
, R′)R′′) C12 25).
independently.
+
(
R ) R′ ) C12
H
25, R′′ ) H); P1-12(12-12) (R) CH
3
H
The triarylpyrylium cations shown in Figure 1 are good
candidates for studies aiming to elucidate the above question.
Calculations based on the CS-INDO-MRCI method have
revealed that, in the ground state, the dihedral angle θ formed
by the pyrylium core and the phenyl group at the para position
gas phase without an energy barrier. The solution properties of
these compounds, studied by steady-state fluorescence spec-
3,6
troscopy and using the single photon counting technique, six-
3
7
wave mixing, and Kerr ellipsometry, corroborate the formation
of a nonfluorescent TICT state.
3
is 30°; a similar value (32°) was found by two-dimensional
1
4
H NMR experiments. In the first excited singlet state, rotation
A striking aspect regarding the excited-state relaxation of the
triarylpyrylium chromophores in Figure 1 is a very strong
dependence of the nonradiative decay rate on the solvent
viscosity. As a matter of fact, it increases by at least 3 orders
of magnitude going from ethanol to polymer matrixes. The
dynamic behavior has been investigated so far only in viscous
media (poly(methyl methacrylate), glycerol, ethylene glycol)
of the dialkylaminophenyl group around the y axis leads to a
decrease of the potential energy whose minimum corresponds
to θ ) 90°. In the most stable configuration of both the ground
and the first excited state, the two phenyl groups in the ortho
position are coplanar with the pyrylium core. The S0 f S1
Franck-Condon transition provokes a drastic change in the
5
atomic charge distribution, the “dipole moment” increasing
3,7
with a limited time resolution. Higher resolution experiments
from -1.6 to +14.2 D. A further increase up to +18.4 D is
induced when the aryl group in the para position becomes
perpendicular to the plane defined by the pyrylium ring and
the aryl groups in the ortho position (θ ) 90°). Thus, the results
of quantum chemistry calculations predict the formation of a
TICT (twisted intramolecular charge-transfer state) state in the
were necessary in order to get a better insight into the various
steps of relaxation by studying its dependence on the solvent
and the rotor size.
With this in mind, we have undertaken an investigation of
the relaxation dynamics using femtosecond fluorescence up-
conversion and picosecond pump-probe absorption spectro-
scopy. These techniques enabled us to extend the time domain
by 2 orders of magnitude with respect to previous studies, which
lead us to refine the understanding of the underlying processes.
We also examined a larger number of chromophores differing
*
To whom correspondence should be addressed.
†
CEA/Saclay, DRECAM/SPAM. E-mail: thomas@droopy.saclay.cea.fr.
Institute of Physics. E-mail: vidgulb@julius.ktl.mii.lt.
Universit e´ de Paris Sud. E-mail: Veber@lps.u-psud.fr.
‡
§
1
0.1021/jp9940248 CCC: $19.00 © 2000 American Chemical Society
Published on Web 05/11/2000

5
182 J. Phys. Chem. A, Vol. 104, No. 22, 2000
Gulbinas et al.
-
3
in the number of dodecyl substituents, permitting us to discern
the role of the rotor size and the solvent viscosity. The synthesis
diphenylpyrylium tetrafluoroborate (0.98 g, 1.5 × 10 mol)
-4
and N-dodecyl-N-methyl aniline (7.65 × 10 mol) were heated
together while stirring under reflux in N,N-dimethylformamide
(1.5 mL). The reaction mixture immediately turned to a very
deep purple color. The reaction was complete after 3 h. After
cooling, the reaction mixture was washed with water and
extracted with dichloromethane. The organic phase was dried
with phase-separating paper and the pyrylium salt precipitated
with ethyl acetate. The crude product was then recrystallized
in an ethanol/water mixture and extracted with heptane/CH2-
Cl2 (25/1). Finally, the desired pyrylium salt was obtained (0.16
+
+
+
4,6
of P1-1 , P1-12 , and P12-12 is described elsewhere, whereas
that of P1-12(12-12)+ is reported for the first time.
2
. Experimental Details
.1 Synthesis. The synthesis of 2,6-(4′-dodecyl)diphenyl-4-
2
(
4′-dodecylmethylaminophenyl) pyrylium tetrafluoroborate, ab-
+
breviated as P1-12(12-12) , consists of the following four steps.
Synthesis of N-Dodecyl-N-Methylaniline. This compound was
1
prepared as described in ref 6: H NMR (CDCl3, δ(ppm)): 7.2
g; 22.5%) by column chromatography (eluent: 66% CH2Cl2,
4% EtOAc, silica gel): H NMR (CDCl3, δ(ppm)): 8.2 (2
(t, 2H); 6.65 (2 superimposed d, 3H), 3.25 (t, 2H); 2.9 (s, 3H);
1
3
1
.3 (m, 20H), 0.9 (t, 3H).
superimposed d and 1s, 8H); 7.5 (d, 4H, J ) 9 Hz); 6.6 (d, 2H,
Synthesis of 4-Dodecylacetophenone. This compound was
J ) 9 Hz), 3.0 (broad t poorly resolved, 5H); 2.7 (t, 4H); 1.3
prepared by a Friedel and Craft reaction, using dried CH2Cl2
as solvent. Acetyl chloride (4.0 mL, 0.058 mol) was added
dropwise while stirring at -10 °C to a suspension of aluminum
chloride (7.78 g, 0.058 mol) in dry dichloromethane (40 mL).
The resulting solution was then left to stir for a further 10
minutes to allow for completion of generation of the pale yellow
acetyl chloride/aluminum chloride complex. Then, 1-phenyl-
dodecane (14.4 mL, 0.05 mol) was added dropwise during a
period of 30 min while maintaining the temperature at ca. -10
-
1
(
m, 60H), 0.9 (t, 9H); IR (KBr pellets); ν/cm : 2917, 2850,
1
637, 1581, 1568, 1511, 1469, 1409, 1355, 1262, 1216, 1194,
1
084.
2
.2 Apparatus and Procedure. Steady-state absorption and
emission spectra of the solutions were recorded with a CARY
E spectrophotometer and a SPEX Fluorolog 2F111A1 spec-
3
trofluorometer, respectively.
Solutions had an optical density between 0.5 and 1 at 394
nm for the femtosecond fluorescence measurements and about
0.3 at 527 nm for the transient absorption experiments. In both
cases, the optical path length was 1 mm. The corresponding
°
C. After complete addition, stirring was continued for 1.5 h in
at a temperature the range of -5 °C to 0 °C, then for a further
.5 h at room temperature until the reaction was finished, as
3
judged by thin-layer chromatography (eluent: 98% heptane, 2%
ethyl acetate, silica gel). The reaction mixture was then
quenched, with stirring, by concentrated hydrochloric acid (25
g) and ice (50 g), left for 1 h, and extracted with dichlo-
romethane. The organic phase was washed with 0.1 M hydro-
chloric acid, neutralized with saturated sodium carbonate
solution, rinsed with water, and dried over phase-separating
paper. Then, the solvent was removed under reduced pressure
and the resulting solid recrystallized in methanol to give
-4
-5
concentrations were ca. 5 × 10 M and 5 × 10 M for
fluorescence and absorption experiments, respectively.
Time-resolved emission spectra and their kinetics were
obtained by the fluorescence upconversion technique. The
experimental setup is described in detail elsewhere.9 The
femtosecond laser source was a Ti:sapphire laser (Coherent
+
MIRA 900) pumped by a continuous wave Ar laser (Coherent
INNOVA 310). The samples were excited at 394 nm by the
second harmonic of the main laser radiation. After passage
through a delay line, the residual fundamental was focused into
a 0.2-mm BBO upconversion crystal, thus serving as the gating
pulse for sum-frequency generation. The fluorescence was
collected with a parabolic mirror and focused into the upcon-
version crystal together with the gating pulse. The upconverted
light was focused onto the entrance slit of a monochromator
4
-dodecylacetophenone (11.23 g, 78%): mp ) 46.8-47.3 °C
8
1
(literature: 48 °C); H NMR (CDCl3, δ(ppm)): 7.9 (d, 2H, J )
Hz); 7.25 (d, 2H, J ) 9 Hz); 2.65 (t, 2H), 2.55 (s, 3H); 1.6
9
-
1
(m, 2H), 1.25 (s, 18H), 0.9 (t, 3H); IR (KBr pellets) ν/cm :
3
045, 2846, 1678, 1600, 1469, 1414, 1122.
Synthesis of 2,6-(4′-Dodecyl)diphenylpyrylium Tetrafluorobo-
rate. Commercial tetrafluoroboric acid (50% w/v solution; 8.72
mL, 0.07 mol) was added dropwise to a stirred solution of acetic
anhydride (4.3 mL, 0.075 mol) in glacial acetic acid (30 mL).
The addition was controlled in such a way that the temperature
did not rise above 25 °C (exothermic reaction). The anhydrous
tetrafluoroboric acid thus generated was immediately used in
the experiment.
(
Jobin-Yvon HR250), and the spectrally selected upconversion
light was detected by a photomultiplier connected to a lock-in
photon counter. Time-resolved fluorescence spectra were re-
corded directly by simultaneously changing the monochromator
wavelength, rotating the upconversion crystal, and changing the
delay line position in order to compensate for the group velocity
dispersion. The fluorescence kinetics were measured by chang-
ing only the delay line position.
4
-Dodecylacetophenone (18.0 g, 0.0625 mol) was dissolved
with stirring and warming (∼35 °C) in triethylorthoformate (104
mL, 0.625 mol). The previously prepared anhydrous tetrafluo-
roboric acid was then added dropwise during a period of 50
min at room temperature. The reaction mixture was left stirring
overnight, and after 22 h the crude product that had precipitated
was filtered off, rinsed with ether, and then recrystallized in a
The time-resolved fluorescence spectra were corrected for the
spectral response of the detection system as described in ref 9.
The fluorescence lifetimes were too short to record a “relaxed”
fluorescence spectrum. The construction of the correction factor
was therefore based on the calculated ratio between the steady-
state fluorescence spectrum and the numerically time-integrated
fluorescence spectrum recorded by the upconversion system.
In all cases, the total intensity dropped to zero before the end
of the scan. To characterize the evolution and the relaxation of
the fluorescence spectrum, the corrected time-resolved spectra
minimum amount of ethyl acetate to give a bright yellow powder
1
(
(
9
3
1
8.41 g, 41%): mp 135.1-136.36 °C; H NMR (CDCl3, δ-
ppm)): 9.1 (t, 1H); 8.55 (d, 2H, J ) 9 Hz); 8.15 (d, 4H, J )
Hz); 7.45 (d, 4H, J ) 9 Hz); 2.7 (t, 24H), 1.6 (m, 4H), 1.3 (s,
-
1
6H), 0.8 (t, 6H); IR (KBr pellets) ν/cm : 2921, 2850, 1616,
-
4
10
519, 1474, 1206, 1124, 1083.9; UV λmax/nm (10 M in
were fitted by log-normal functions. From these fits, the total
dichloromethane): 442, 332, 296, 268.
Synthesis of 2,6-(4′-Dodecyl)diphenyl-4-(4′-dodecylmethyl-
aminophenyl)pyrylium Tetrafluoroborate. 2,6-(4′-Dodecyl)-
intensity (integrated in the frequency domain for a given time
delay), the mean frequency, the bandwidth, and the asymmetry
were obtained as functions of time.

Relaxation Processes of Triarylpyrylium Cations
J. Phys. Chem. A, Vol. 104, No. 22, 2000 5183
Figure 2. Steady-state absorption and fluorescence (λex ) 527 nm)
+
spectra of P1-12(12-12) in dichloromethane.
The transient absorption study was performed using a pump-
probe spectrometer equipped with a homemade low-repetition
rate (1 Hz) Nd:glass laser, delivering pulses of 3 ps duration.
The second harmonic (527 nm) was used for excitation, and a
white light continuum generated in water was used to probe
the samples. The continuum was split into two parts that passed
through the sample at different positions, one overlapping with
the excitation pulse, the second a few millimeters beside the
first, serving as the reference. The probe and the reference pulses
were focused onto the entrance slit of a small monochromator,
and their intensities were measured by two photodiodes located
at the exit slit. Transient absorption kinetics at a selected
wavelength were measured by changing the delay time between
the excitation and the probe pulses. Transient difference
absorption spectra at fixed delay times were recorded by
scanning the monochromator wavelength and simultaneously
moving the delay line in order to compensate for the group
velocity dispersion effect.
+
Figure 3. Corrected time-resolved fluorescence spectra of P1-12 in
hexanol; λex ) 394 nm.
3
. Results
Figure 2 shows the steady-state absorption and fluorescence
+
spectra of P1-12(12-12) in dichloromethane. The absorption
spectrum is characterized by an intense band peaking at 557
nm and a weaker one at 407 nm. The maximum of the
fluorescence spectrum is located at 603 nm. The spectra of the
Figure 4. P1-12+ in hexanol: time dependence of (a) the total
fluorescence intensity and (b) the fluorescence Stokes shift; λex ) 394
nm.
three other compounds3 are similar to those of P1-12(12-12) .
-5
+
Figure 3 shows an example of the time-resolved fluorescence
spectra, a few out of forty recorded for P1-12 in hexanol. The
time evolution of the total intensity and the fluorescence Stokes
shift is presented in Figure 4.
TABLE 1: Characteristic Times (ps) of the Fluorescence
+
Stokes Shift (τ ) and the Total Intensity Decay (τ ) found
f,1
f,2
+
a
for the Triarylpyrylium Cation P1-12 in Different Solvents
solvent
viscosity (cp)
τ
f,1
τ
f,2
Two important features can be seen: first, a rapid intensity
decay, of the order of a few tens of picoseconds, and second,
an even faster red shift of the fluorescence spectra. The total
fluorescence intensity decay and the dynamic Stokes shift could
be well approximated by single exponentials. It should be noted
that the accuracy in the determination of the Stokes shift is not
high, because it was impossible to record spectra corresponding
to the fully relaxed configuration. Similar spectra and dynamics
dichloromethane
ethanol
dimethyl sulfoxide
hexanol
0.4
1.1
2.0
4.6
13.7
1.1
1.3
1.5
3.3
3.4
2.6
4.3
6.1
17
20
ethylene glycol
a
Given the difficulty of locating the “relaxed” emission spectrum,
we estimate the uncertainties in τf,1 to (25%. The uncertainties in τf,2
are (10%.
+
Figure 5 shows the differential absorption spectra of P1-12+
in hexanol recorded at different delay times after excitation.
The spectra show a negative part attributed to the bleaching of
the main absorption band and the stimulated emission. An
induced absorption is observed around 450 and 600 nm. The
induced absorption at 450 nm appears simultaneously with the
bleaching, whereas the induced absorption band at 600 nm
appears later.
were obtained for P1-12 in various solvents and also for the
other three compounds in dichloromethane. The characteristic
times of the dynamic Stokes shift (τf,1) and the total intensity
+
decay (τf,2) observed for P1-12 in different solvents and for
the other three compounds in dichloromethane are given in
Tables 1 and 2, respectively. It may be observed that both
+
characteristic times found for P1-12 increase with the solvent
viscosity. Moreover, the fluorescence decay in dichloromethane
becomes slower when the number of dodecyl substituents
increases, whereas τf,1 remains practically constant.
The dynamics of the transient absorption of P1-12+ in hexanol
can be seen more clearly in Figure 6. The induced absorption

5184 J. Phys. Chem. A, Vol. 104, No. 22, 2000
Gulbinas et al.
TABLE 2: Characteristic Times (ps) of the Fluorescence
Stokes Shift (τf,1) and the Total Intensity Decay (τf,2) found
for the Four Studied Triarylpyrylium Cations in
Dichloromethane
compound
τ
f,1
τ
f,2
+
P
1-1
0.9
1.1
1.3
1.0
1.2
2.6
3.8
4.1
+
P
P
P
1-12
+
12-12
1-12(12-12)
+
a
Given the difficulty of locating the “relaxed” emission spectrum
we estimate the uncertainties in τf,1 to (25%. The uncertainties in τf,2
are (10%.
+
Figure 7. P1-1 in dichloromethane: transient differential absorption
spectra recorded at various delay times after excitation at 527 nm.
absorbance at 600 nm has a more complex behavior; initially it
is negative, then it becomes positive and decays more slowly
than the signal at 700 nm.
To compare the various decays in a quantitative way, we fitted
them by the mono- or biexponential functions. Such a fit does
not necessarily mean that the underlying physical process can
be correctly described by such functions. The induced absorption
at 450 nm decays monoexponentially with the time constant of
3
6 ps. The bleaching at 500 nm decays biexponentially, with
+
46 ps (77%) and 101 ps (23%) time constants. The induced
absorption at 600 nm appears with a rise time (time needed for
the signal to become positive) of 19 ps and relaxes nearly
simultaneously with the minor component of the absorption
bleaching. The stimulated emission at 700 nm appears with a
rise time of 4.8 ps and decays faster (30 ps) than the absorption
bleaching at 500 nm but slower than the total fluorescence
intensity decay (17 ps).
Figure 5. P1-12 in hexanol: transient differential absorption spectra
recorded at various times after excitation at 527 nm.
Figure 7 shows the transient absorption spectra of P1-1+ in
dichloromethane for chosen delay times. The stimulated emis-
sion around 600 nm is observed only at early times. At 1 ps
delay, only the absorption bleaching and the induced absorption
around 450 nm give significant contributions, whereas only a
weak stimulated emission is observed at λ g 700 nm. The
induced absorption around 600 nm appears at 7 ps delay, when
the stimulated emission has already decayed.
The dynamics of the transient absorption obtained at various
wavelengths for the four studied compounds in dichloromethane
are shown in Figure 8. The time evolution of the absorbance is
+
qualitatively similar to that found for P1-12 in hexanol (Figure
6
). We remark that the differential absorbance (∆A) at all
wavelengths changes more slowly with time when the number
of dodecyl groups increases.
The time constants describing the signal decays observed at
various wavelengths for the four studied compounds are
summarized in Table 3. In most cases, monoexponential fits
were satisfactory for kinetics at 450, 500, and 700 nm. The
complex time dependence of the differential absorbance at 600
nm was fitted by a biexponential function. A biexponential fit
+
Figure 6.
P
1-12 in hexanol: time dependence of the differential
absorption signals observed at different wavelengths (λex ) 527 nm).
The solid lines show the fit with mono- or biexponential functions
convoluted with the instrumental function. The corresponding time
constants and their relative amplitudes are given in Table 3.
+
at 450 nm and the negative differential absorbance at 500 and
was also used for the P1-1 kinetics at 700 nm. Figure 8
7
00 nm decay monotonically. It is worth noticing that the decay
illustrates how the time constants depend on the number of
dodecyl groups. The decay times of the induced absorption at
at 500 nm is slower than those at 450 and 700 nm. The transient

Relaxation Processes of Triarylpyrylium Cations
J. Phys. Chem. A, Vol. 104, No. 22, 2000 5185
Figure 8. Time dependence of the transient absorption observed at 450 nm (squares), 500 nm (crosses), 600 nm (circles), and 700 nm (triangles)
for the four studied triarylpyrylium salts in dichloromethane (λex ) 527 nm).
TABLE 3: Characteristic Times (ps) Obtained by Fitting
the Transient Absorption Signals Recorded at Different
Wavelengths for the Four Studied Triarylpyrylium Cations
with Mono- or Biexponential Functionsa
chromophore
(solvent)
450 nm
5
500 nm
10
600 nm
700 nm
P1-1+
3
24
(-100%) (24%)
7.2 23
(-100%) (48%)
10.4 28
(-100%) (43%)
12 37
1
(
dichloromethane)
+
P1-12
6.6 16
2.5
6.8
(
dichloromethane)
+
P12-12
10
20
28
(
dichloromethane)
+
P1-12(12-12)
20
36
6.7
(
dichloromethane)
(-100%) (21%)
+
P1-12
46 (-77%) 19
104 30
(
hexanol)
101 (-23%) (-100%) (35%) 4.8 (rise)
The relative amplitude values are shown in parentheses.
50 nm, absorption bleaching at 500 nm, stimulated emission
a
4
at 700 nm, and the rise time of the induced absorption at 600
nm increase with the number of dodecyl groups. In contrast,
the decay of the induced absorption at 600 nm, which is the
slowest process, shows no clear dependence.
+
Figure 9. P1-12 solution in hexanol: (a) transient absorption decays
at 500 nm for parallel (black circles) and perpendicular (white circles)
polarization between excitation and probe pulses; (b) magic angle decay
(squares) and induced anisotropy decay (triangles).
To compare the present results to those obtained previously
7
by Kerr ellipsometry, we have explicitly studied the depolar-
+
ization dynamics for P1-12 in hexanol. Figure 9 shows the
the studied triarylpyrylium cations in dichloromethane do not
depend strongly on the number of dodecyl groups. In contrast,
they depend on the solvent and could thus be due to solvent
relaxation. Regarding the characteristic time of the total
fluorescence decay, τf,2, it increases with both the solvent
viscosity and the number of dodecyl groups. Therefore, we
associate it with a conformational relaxation involving solvent
hindered motion, which can be an increase of the dihedral angle
θ. One may now wonder to what extent the observed Stokes
shift is caused by the solvent relaxation and/or motion along
the reaction coordinate. There is an indication that the molecular
twisting during the fluorescence shift is negligible. According
to the quantum chemical calculations, the oscillator strength
transient absorption kinetics at 500 nm for parallel (∆A|) and
perpendicular (∆A⊥ ) polarization between excitation and probe
pulses. Figure 9b shows the kinetics at the magic angle (54.7°)
given by ∆A54.7° ) (∆A| + 2∆A⊥ )/3 as well as the anisotropy
decay given by r ) (∆A| - ∆A⊥ )/(3A0 + ∆A| + 2∆A⊥ ), where
A0 is the steady-state absorbance. We remark that the anisotropy
kinetics presents a long tail (≈350 ps) component that is absent
from the magic angle decay.
4
. Discussion
We have seen in Section 3 (Tables 1 and 2) that the
characteristic times of the dynamic Stokes shift, τf,1, found for

5186 J. Phys. Chem. A, Vol. 104, No. 22, 2000
Gulbinas et al.
drops down with increasing twisting angle. If the spectral shift
were related to the population motion along the twisting
coordinate, one would expect a fast fluorescence intensity decay
to occur on the same time scale as the fluorescence shift, but
such a decrease is not observed. This observation leads us to
attribute the fluorescence Stokes shift to the solvent relaxation.
Actually, it seems that solvation and twisting take place
independently, which is corroborated by the fact that both the
fluorescence shift and the decay are monoexponential within
the experimental accuracy. Characteristic times of solvation
dynamics have been reported for coumarins in various sol-
vents.11 A direct comparison between these values and those in
Tables 1 and 2 is, unfortunately, not possible because the
fluorescence lifetimes of the studied triarylpyrylium cation are
too short to follow the Stokes shift to its end.
Figure 10. Schematic diagram of the excited-state relaxation of the
We will now discuss the time evolution of the transient
absorption spectra by focusing on four typical wavelengths, each
corresponding to a particular band, as described in Section 3.
studied triarylpyrylium tetrafluoroborates. The energy levels are those
3
found by quantum chemistry calculations. The processes marked by
letters are A, solvation and population spreading; B, twisting; C, internal
conversion; D, relaxation to optically active ground state; and E, ground-
state relaxation.
The induced absorption at 450 nm may be assigned to the
first electronically excited state. As a matter of fact, such a short-
wavelength transient absorption, decaying faster than the
+
to the relaxed ground state, through all possible excited and
ground-state relaxation processes.
apparatus function, was already observed for P1-12 in polymer
7
films by nanosecond flash photolysis. The decay of the transient
In general, the transient absorption spectra presented here
show the same behavior as the photoinduced dichroic spectra
absorption found in the present work is significantly longer than
the fluorescence lifetime, which is an indication that the excited-
state population is still there but is no longer fluorescent. We
deduce that the S1 f Sn absorption transition moment is less
affected by the conformational changes responsible for the
fluorescence decay. This could be due to the S1 f Sn absorption
transition being polarized along the x axis, making it less
sensitive than the S0 r S1 transition, which is polarized along
7
recorded by Kerr ellipsometry, although the latter were obtained
in more viscous solvents and with a lower time resolution.
However, there is one notable exception, namely, the bleaching
decay. The photoinduced dichroism at 500 nm remained long
after having disappeared at 600 and 700 nm. In the present work,
the bleaching decay at 500 nm is practically as fast as the 600-
nm transient absorption decay (Figures 6 and 8). This discrep-
ancy can be explained knowing that the measured parameter in
Kerr ellipsometry is the dichroic angle δφ ) (∆A| - ∆A⊥ )(ln
3
the y axis, to a change in the angle θ. Such an explanation is
corroborated by the negative value of the transient dichroism
at 450 nm (Figure 2 in ref 7). Thus, it is reasonable to assume
that the induced absorption at 450 nm provides a better
description of the excited-state population than the emission.
1
0)/4 that contains information about both population and
polarization changes. Figure 9 shows that the anisotropy decay,
related to the photoinduced dichroism, is much slower than the
magic-angle decay. Such an effect can be provoked if the
excited, initially polarized, population undergoes a depolariza-
tion before returning to the ground state. If this excited-state
depolarization is faster than the depolarization of the photoin-
duced hole, the resulting ground-state population will be
anisotropic. The exact nature of this depolarization is outside
the scope of this paper.
The negative signal observed at 700 nm is attributed to the
stimulated emission. In general, it appears within the instru-
+
mental response function. However, for P1-12 in hexanol, which
is the most slowly relaxing solvent11 used in the present study,
the stimulated emission appears with a rise time of 4.8 ps. This
rise time is close to the characteristic time of the dynamic Stokes
shift τf,1 (3.3 ps), and it can thus be attributed to solvation.
The initial negative value of the transient absorption signal
at 600 nm is partly due to the ground-state bleaching and partly
to the stimulated emission. This is corroborated by the steady-
state spectra (Figure 2). The induced absorption at 600 nm
appears with a certain delay and decays simultaneously with
the bleaching. That this delay is real may be argued for by
studying the transient differential absorption spectra recorded
at various delay times for P1-1+ in dichloromethane, as shown
in Figure 7. At 1 ps, only bleaching is observed, the induced
absorption appearing first at 7 ps. That an accidental canceling
between stimulated emission and induced absorption should
produce the pure bleaching spectrum observed at 1 ps seems
highly improbable. These considerations lead us to attribute the
induced absorption to some nonrelaxed ground-state population.
The fact that the rise time of the signal increases with the solvent
viscosity and the number of dodecyl substituents shows that
the associated process involves solvent hindered motion.
With these observations in mind, we suggest a global scheme
for the relaxation of the photoexcited triarylpyrylium cations
treated in this work, as illustrated in Figure 10. The ground
and excited-state free energies as functions of the dihedral angle
3
θ are those calculated by the CS-INDO-MRCI method. The
characteristic times of different processes marked by letters
correspond to (a) solvation and population spreading, determined
from the fluorescence shift time; (b) twisting, determined from
the fluorescence total intensity decay time; (c) internal conver-
sion, determined from the induced absorption decay at 450 nm;
(d) relaxation to an optically active ground state, determined
from the absorption bleaching decay at 500 nm; and (e) ground-
state relaxation, determined from the induced absorption decay
at 600 nm. We will now discuss the different relaxation stages
in detail.
After the Franck-Condon transition, the molecule finds itself
on a slope of the excited-state potential curve, and its solvation
shell is not equilibrated. Solvent rearrangement is the fastest
process causing the fluorescence shift, as mentioned above. At
the same time, the system evolves along the reaction coordinate
The absorption bleaching decay time observed at 500 nm
corresponds to the ground-state recovery. It can be viewed as
the round trip going from the Franck-Condon excited state back

Relaxation Processes of Triarylpyrylium Cations
J. Phys. Chem. A, Vol. 104, No. 22, 2000 5187
leading to the fast fluorescence decay. Regarding the exact origin
of this fast decay, it may reflect excited-state depopulation via
internal conversion to the ground state or a decrease of the
fluorescence oscillator strength due to twisting. Which of the
two processes is dominant at a given time depends on many
factors such as the population distribution on the twisting
coordinate, the population drift speed, the nonradiative decay
rate, and the fluorescence oscillator strength as a function of
the twisting angle. Knowing that the excited-state population
decays more slowly than the fluorescence, we conclude that
the fluorescence decay is largely due to the loss of the S0 r S1
oscillator strength.
emission, we cannot make a more refined comparison. In
particular, we cannot determine whether the absorbing ground-
state species is that formed by the S0 r S1 Franck-Condon
transition or whether it corresponds to an already partially
relaxed state (θ0 < θ < θ2).
The nature of the nonequilibrium ground state needs some
consideration. After the internal conversion, the molecule is in
a twisted ground-state configuration, it has a nonequilibrium
solvation shell adapted to the excited state, and it has an excess
of vibrational energy. All these effects should lead to a red shift
of the absorption band. Within the framework of linear response
15,16
theory,
the ground-state solvation time should be identical
The nonradiative excited-state population continues to evolve
along the twisting coordinate. The difference between the
excited-state lifetime and the fluorescence decay time represents
the time that molecules spend in a twisted electronically excited
state. This is schematically depicted by assuming the existence
of θ-intervals defined by the limiting angles θ0 () 30°), θ1,
and θ2. In the first interval, between θ0 and θ1, the chromophores
are fluorescent, whereas in the second interval, between θ1 and
θ2, they are nonfluorescent. The internal conversion rate
increases already for angles slightly different from θ0, and it
becomes fastest at θ2. It is worth noticing that in solid matrixes,
where twisting is inhibited, internal conversion is not efficient,
with that for the excited state, that is, on the order of a few
picoseconds (τ in Table 2). Thus, solvation can hardly account
f1
for signals lasting several tens of picoseconds (Figure 8). The
relatively weak and nonmonotonic dependence of the induced
absorption relaxation rate on the number of dodecyl groups leads
us to believe that this process cannot only be due to the back-
twisting. Vibrational cooling of molecules occurs at similar times
17
of a few tens of picoseconds; however, this cannot be the only
process responsible for the induced absorption decay, since then
no clear dependence on substituents should be observed.
Moreover, the 100-ps relaxation time observed in hexanol
solution is too long for cooling. We deduce that both vibrational
cooling and back-twisting contribute to the ground-state relax-
ation. When back-twisting is very fast, vibrational cooling
constitutes the limiting rate, while back-twisting becomes
dominant for the larger solute molecules.
3
as manifested by the high fluorescence quantum yield. The
exact dependence of the internal conversion on the twisting angle
is impossible to define, but we will comment upon why it
increases with θ.
An increasing internal conversion rate may be due to a
decreasing transition energy as a function of θ. This energy
gap dependence, which is commonly used to explain very fast
internal conversion of molecules in a twisted conformation, can
hardly explain the nonradiative relaxation of the studied
triarylpyrylium cations. According to the calculated potential
curves, the energy gap between the ground and the excited states
decreases by less than 10% going from 30° to 90°. This is too
small to account for an increase of the internal conversion rate
by 3 orders of magnitude, as deduced from the fluorescence
lifetimes measured for the chromophores in dichloromethane
The qualitative model proposed above for the relaxation of
the studied triarylpyrylium cations, involving a twist in the
excited state, has been invoked in the case of many other
1
8,19
compounds,
even if alternative explanations have been put
2
0
forward. We will, however, compare our findings to those
reported for a particular class of molecules, namely, the
triphenylmethane (TPM) dyes, because their photophysical
properties are similar to those of the studied triarylpyrylium
cations. After photoexcitation, TPM dyes in nonviscous solvents
undergo a very rapid (on the order of a few picoseconds) internal
1
2,21
conversion to the ground state.
A detailed account of the
(Tables 1 and 2) and in solid films (3.2 ns as reported in ref 3).
TPM photophysics is far outside the scope of the present work,
but a very complete review article appeared recently.22 Excited-
state relaxation is explained in terms of rapid rotation of all
three phenyl groups, hindered by the solvent, bringing the system
to a configuration with much enhanced probability for internal
conversion. After having returned to the ground state, a slower
back-rotation to the equilibrium configuration takes place. The
fluorescence decays much faster than the ground-state recovery.
The system spends some time in a transient nonfluorescent state,
as evidenced by a residual transient absorption, which is slightly
red-shifted with regards to the ground-state absorption. A central
question, still not resolved, is the exact nature of this transient
state; is it a relaxed excited state or a nonrelaxed ground state?
Even if the energy gap is insufficient to explain the data, the
fast nonradiative decay may be rationalized in terms of a
coordinate dependent sink.12 In this case, one would expect a
delay in the bleaching decay, as reported for 1,1-diethyl-4,4-
cyanine.13 However, no such delay has been observed for the
studied triarylpyrylium cations, indicating that the internal
conversion rate is substantial even for small twisting angles. In
a recent paper, a similar conclusion was drawn regarding the
twisting process of N,N-dimethylaminobenzylidene-1,3-indan-
dione.14 It was proposed that fast molecular conformational
changes during the twisting time might cause the rapid internal
conversion. Since the twisting mode is strongly coupled with
other vibrational modes, it causes the excitation to spread over
a large number of molecular vibrations. As a result, the overlap
of the ground- and excited-state vibrational wave functions will
increase, leading to fast internal conversion.
23
Several authors favor a ground-state configuration, whereas
recent work, based on the idea that an increase in the twist angle
reduces the π-electron conjugation and thus induces a blue shift
24
in the absorption band, opts for an excited state. In the case
After internal conversion, chromophores find themselves in
a nonequilibrated ground state, which we identify as responsible
for the transient absorption signal observed at 600 nm. This
assignment is based on the observation that the decay at 450
nm, which is a good measure of the excited-state population, is
roughly similar to the rise time of the absorption at 600 nm.
However, knowing that the signal at 600 nm also contains
contributions from the ground-state bleaching and the stimulated
of the studied triarylpyrylium cations, quantum chemistry
calculations have shown that the S0 f S1 transition energy of
a twisted configuration (θ > 30°) is lower than that for the
3
equilibrium configuration (θ ) 30°). This is due to the fact
that, with increasing twisting angle θ, the ground state is more
destabilized than the excited state. Consequently, the attribution
of the 600-nm transient absorption peak to a nonequilibrated
ground state is reasonable.

5188 J. Phys. Chem. A, Vol. 104, No. 22, 2000
Gulbinas et al.
3
TABLE 4: Van der Waals Volumes (Å ) Corresponding to
the Two Rotating Parts of the Four Studied Triarylpyrylium
a
Cations (Figure 1)
chromophore
V
1
V
2
V
TOT
V
RED
+
P
P
P
P
1-1
227.8
227.8
227.8
636.6
133
320
507
320
360.8
547.8
734.8
956.6
84
+
1-12
133
157
213
+
12-12
1-12(12-12)
+
a
V
1
Is the volume of the 2,6-diarylpyrylium moiety, V
of the 4-dialkylaminophenyl moiety, VTOT ) V + V
/(V +V ).
2
is the volume
1
2
, and VRED ) V
1
V
2
1
2
Figure 12. Characteristic times of the dynamic Stokes shift (a, open
+
circles) and the fluorescence intensity decay (b, closed circles) of P1-12
in dichloromethane, ethanol, dimethyl sulfoxide, hexanol, and ethylene
glycol as a function of solvent viscosity. The data are fitted with the
functions (a) τ ) 1.5η0 and (b) τ ) 5.6η
.35
0.5
.
due to a size effect of the rotating parts. Even if this simple
picture is very attractive, one should keep in mind that the
dodecyl substituents cannot be considered as rigid objects, but
rather as flexible, randomly curled chains. Molecular twisting,
thus, may involve only parts of them. Furthermore, as noted in
Section 3, the characteristic times were obtained by mono- or
biexponential fits, which are not necessarily reflecting the time
dependence in a correct way.
+
In our study of P1-12 in various solvents, we observe
sublinear dependencies on the solvent viscosity (η) for both the
Stokes shift and the total fluorescence intensity decay (Table
1
/3
1
). In the former case, a η dependence seems to be the case,
whereas in the latter, the characteristic times are consistent with
Figure 11. Characteristic times corresponding to the decay of the
bleaching at 500 nm as a function of the van der Waals total (a) and
reduced (b) volumes of the studied triarylpyrylium cations (Table 4).
1
/2
η
(Figure 12). A sublinear dependence on the solvent
2
/3
viscosity, proportional to η , in line with the F o¨ rster-Hoffman
model, has been observed for different di- and triphenyl-
methane dyes.
found in a recent study of crystal violet. It would therefore
be very interesting to correlate these phenomenological observa-
tions to a more rational interpretation based on the electronic
structure of both the solvent and the solute.
2
1b
2
7,28
In contrast, an exponent larger than 1 was
Further comparisons between the dynamical behavior of the
studied triarylpyrylium cations and the TPM dyes can be made
regarding the effect of the rotor size and the influence of the
solvent viscosity. We have seen in Section 3 that all time
constants determined in the present work, with perhaps the
exception of the dynamic Stokes shift and the 600-nm transient
absorption decay, show a clear dependence on the number of
the dodecyl substituents. The intramolecular rotation responsible
for the observed dynamics is best described as a large amplitude
motion of the two moieties of the cation, rotating with respect
to each other and thus shuffling about considerable volumes of
the surrounding solvent. In a very simple picture given by the
Stokes-Einstein-Debye model, the rotational time, and thus
the excited-state lifetime, is directly proportional to the rotor
2
4
References and Notes
(1) Lippert, E.; Rettig, W.; Bonacic-Koutecky, V.; Heisel, F.; Mieh e´ ,
J. A. In Prigogine, I., Rice, S. A., Eds.; Advances in Chemical Physics
John Wiley & Sons: New York, 1987; Vol. 68, pp 1-173.
(
2) (a) Fonseca, T.; Kim, H. J.; Hynes, J. T. J. Mol. Liq. 1994, 60,
61. (b) Fonseca, T.; Kim, H. J.; Hynes, J. T. J. Photochem. Photobiol., A
1994, 82, 67. (c) Kim, H. J.; Hynes, J. T. J. Photochem. Photobiol., A 1997,
05, 337.
3) Markovitsi, D.; Sigal, H.; Ecoffet, C.; Milli e´ , P.; Charra, F.; Fiorini,
1
1
(
C.; Nunzi, J.-M.; Strzelecka, H.; Veber, M.; Jallabert, C. Chem. Phys. 1994,
182, 69.
(4) Lampre, I.; Markovitsi, D.; Birlirakis, N.; Veber, M. Chem. Phys.
1996, 202, 107.
2
5
size. The volumes of the two parts of the studied compounds
can easily be estimated by adding up the van der Waals
increments of the constituent atoms.2 The resulting values are
given in Table 4 and, as can be seen, the volumes of the two
parts are comparable. For this reason, neither is a good measure
of the moving volume since both will contribute to the relative
motion. We will use both the total volume VTOT and the reduced
volume defined by VRED ) V1V2/(V1 + V2).
6
(5) The dipole moment of a charged species, like a triarylpyrylium
cation, is not coordinate-independent and needs special consideration in
quantum chemistry calculations. The way these values were obtained is
described in ref 3.
(
6) Markovitsi, D.; Jallabert, C.; Strzelecka, H.; Veber, M. J. Chem.
Soc., Faraday Trans. 1990, 86, 2819.
7) Lampre, I.; Marguet, S.; Markovitsi, D.; Delysse, S.; Nunzi, J.-M.
Chem. Phys. Lett. 1997, 272, 496.
(
Figure 11 shows that the characteristic time of the bleaching
decay at 500 nm, corresponding to the overall relaxation process,
is proportional to both VTOT and VRED. The characteristic times
of the total fluorescence intensity decay, the 450-nm transient
absorption decay, the 600-nm transient absorption rise, and the
(8) Farbenind, I. G. U.S. Patent 2,195,198, 1938.
(9) Gustavsson, T.; Cassara, L.; Gulbinas, V.; Gurzadyan, G.; Mialocq,
J.-C.; Pommeret, S.; Sorgius, M.; van der Meulen, P. J. Phys. Chem. A
1998, 102, 4229.
(
(
10) Siano, D. B.; Metzler, D. E. J. Chem Phys. 1969, 51, 1856.
11) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. J.
700-nm stimulated emission decay also increase monotonically
Phys. Chem. 1995, 99, 17311.
12) Bagchi, B.; Fleming, G. R.; Oxtoby, D. W. J. Chem. Phys. 1983,
78, 7375.
but not linearly with the volume. Some caution should thus be
taken when interpreting the change in relaxation rate as only
(

Relaxation Processes of Triarylpyrylium Cations
J. Phys. Chem. A, Vol. 104, No. 22, 2000 5189
(
13) A° berg, U.; A° kesson, E.; Alvarez, J.-L.; Fedchenia, I.; Sundstr o¨ m,
V. Chem. Phys. 1994, 183, 269.
14) Gulbinas, V.; Kodis, G.; Jursenas, S.; Valkunas, L.; Gruodis, A.;
Mialocq, J.-C.; Pommeret, S.; Gustavsson, T. J. Phys. Chem. A 1999, 103,
(21) (a) Oster, G.; Nishijima, Y. J. Am. Chem. Soc. 1956, 78, 1581. (b)
F o¨ rster, Th.; Hoffman, G. Z. Phys. Chem. NF 1971, 75, 63. (c) Cremers,
C. A.; Windsor, M. W. Chem. Phys. Lett. 1980, 71, 27.
(22) Duxbury, D. F. Chem. ReV. 1993, 93, 381.
(23) Sundstr o¨ m, V.; Gillbro, T.; Bergstr o¨ m, H. Chem. Phys. 1982, 73,
439.
(24) Jurczok, M.; Plaza, P.; Martin, M. M.; Rettig, W. J. Phys. Chem.
A 1999, 103, 3372.
(25) Vogel, M.; Rettig, W. Ber. Bunsen-Ges. Phys. Chem. 1987, 91,
1241.
(
3
969.
(
(
(
(
(
15) Carter, E. A.; Hynes, J. T. J. Chem Phys. 1991, 94, 5961.
16) Brown, R. J. Chem Phys. 1995, 102, 9059.
17) Elsaesser, T.; Kaiser, W. Annu. ReV. Phys. Chem. 1991, 42, 83.
18) Rettig, W. Angew. Chem., Int. Ed. Engl. 1986, 25, 971.
19) Rettig, W. In Topics of Current Chemistry; Springer-Verlag: Berlin,
1
994, p 254.
20) Zachariasse, K. A.; Grobys, M.; Haar, T.; Hebecker, A.; Il’ichev,
Y. V.; Jiang, Y.-B.; Morawski, O.; Kuhnle, W. J. Photochem. Photobiol.
996, 102, 59.
(26) Edward, J. T. J. Chem. Educ. 1970, 47, 261.
(27) Yu, W.; Pellegrino, F.; Grant, M.; Alfano, R. R. J. Chem. Phys.
1977, 67, 1766.
(
1
(28) Hirsch, M. D.; Mahr, H. Chem. Phys. Lett. 1979, 60, 299.