Picosecond time-resolved Stokes and anti-Stokes Raman studies on the photochromic reactions of diarylethene derivatives

Article abstract of DOI:10.1021/jp035055c

The cyclization and cycloreversion reactions of diarylethene derivatives have been studied with picosecond time-resolved Stokes and anti-Stokes Raman spectroscopies. The cyclization reaction of 1,2-bis(2,5-dimethyl-3-thienyl)perfluorocyclopentene (DMTF) is found to occur within 4 ps to produce the vibrationally excited closed forms in the ground electronic (S₀) state. The time constant of the vibrational relaxation toward a thermal equilibrium with solvent molecules is estimated to be about 10 ps. The cycloreversion reaction of 1,2-bis(3,4-dimethyl-5-phenyl-2-thienyl)perfluorocyclopentene (DMPTF) also generates the vibrationally excited open forms in the So state within 4 ps, which decay on a picosecond time scale. The picosecond time-resolved anti-Stokes Raman spectra of DMPTF show two vibrational bands assignable to the C=C stretching modes of the cyclopentene and thiophene moieties of the generated open forms. The Raman intensity arising from the cyclopentene moiety relative to that from the thiophene moiety becomes smaller with the delay time, indicating that part of the excess energy generated via the cycloreversion reaction is localized on the C=C stretching mode of the cyclopentene moiety. This result suggests that the C=C stretching mode of the cyclopentene moiety is one of the promoting or the accepting modes in the cycloreversion reaction.

Full text of DOI:10.1021/jp035055c

5384
J. Phys. Chem. A 2003, 107, 5384-5390
Picosecond Time-Resolved Stokes and Anti-Stokes Raman Studies on the Photochromic
Reactions of Diarylethene Derivatives
Chie Okabe,† Takakazu Nakabayashi,‡ Nobuyuki Nishi,§ Tuyoshi Fukaminato,|
Tsuyoshi Kawai,| Masahiro Irie,| and Hiroshi Sekiya*,†
Department of Chemistry, Faculty of Science, Kyushu UniVersity, Hakozaki 6-10-1,
Higashi-ku, Fukuoka 812-8581, Japan, Research Institute for Electronic Science, Hokkaido UniVersity,
Sapporo 060-0812, Japan, Institute for Molecular Science, Myodaiji, Okazaki 444-8585, Japan, and
Department of Chemistry and Biochemistry, Graduate School of Engineering, Kyushu UniVersity,
6-10-1 Hakozaki, Higashiku, Fukuoka 812-8581, Japan
ReceiVed: April 19, 2003
The cyclization and cycloreversion reactions of diarylethene derivatives have been studied with picosecond
time-resolved Stokes and anti-Stokes Raman spectroscopies. The cyclization reaction of 1,2-bis(2,5-dimethyl-
3-thienyl)perfluorocyclopentene (DMTF) is found to occur within 4 ps to produce the vibrationally excited
closed forms in the ground electronic (S0) state. The time constant of the vibrational relaxation toward a
thermal equilibrium with solvent molecules is estimated to be about 10 ps. The cycloreversion reaction of
1,2-bis(3,4-dimethyl-5-phenyl-2-thienyl)perfluorocyclopentene (DMPTF) also generates the vibrationally excited
open forms in the S0 state within 4 ps, which decay on a picosecond time scale. The picosecond time-
resolved anti-Stokes Raman spectra of DMPTF show two vibrational bands assignable to the CdC stretching
modes of the cyclopentene and thiophene moieties of the generated open forms. The Raman intensity arising
from the cyclopentene moiety relative to that from the thiophene moiety becomes smaller with the delay
time, indicating that part of the excess energy generated via the cycloreversion reaction is localized on the
CdC stretching mode of the cyclopentene moiety. This result suggests that the CdC stretching mode of the
cyclopentene moiety is one of the promoting or the accepting modes in the cycloreversion reaction.
1. Introduction
SCHEME 1
Photochromism is characterized by a light-induced reversible
reaction of a chemical species. This reaction gives rise to the
formation of photoisomers whose electronic absorption spectra
are distinguishably different from that of the reactant molecule,
which results in the dramatic color change. Changes in other
molecular properties such as reflective and dielectric constants
can also occur during the photochromic reaction. The back
reaction is induced mostly by a thermal mechanism,1,2 whereas
some photochromic molecules yield thermally stable photo-
products.3-9 For such systems, back reactions are photochemical.
These phenomena have received much attention in recent years
because of their potential applications in optoelectronic devices.
Knowledge on the vital factors governing photochromic reac-
tions is essential for designing new photochromic molecules
for optical applications.
Diarylethene derivatives are promising photochromic com-
pounds for optical applications because of their thermal
irreversibility,5-9 fatigue-resistance,6,7,9 rapid response,9-19 and
high sensitivity.20 The photochromic reaction scheme of di-
arylethene derivatives is shown in Scheme 1. The open forms
of diarylethenes are mostly colorless and turn to the closed forms
by UV irradiation (cyclization reaction). The generated closed
forms are thermally stable and exhibit absorption spectra in the
visible region. Upon irradiation of visible light, the closed forms
revert back to the original open forms (cycloreversion reaction).
From NMR6,7,21 and absorption measurements,21 it is proved
that the open form in solution has at least two rotational
conformers: antiparallel and parallel conformers depicted in
Scheme 1. The conrotary cyclization reaction cannot proceed
from the parallel conformer. Thus, it can be considered that the
* To whom correspondence should be addressed.
† Department of Chemistry, Kyushu University.
‡ Hokkaido University.
§ Institute for Molecular Science.
| Department of Chemistry and Biochemistry, Kyushu University.
10.1021/jp035055c CCC: $25.00 © 2003 American Chemical Society
Published on Web 06/24/2003

Photochromic Reactions of Diarylethene Derivatives
J. Phys. Chem. A, Vol. 107, No. 28, 2003 5385
cyclization yield depends on the ratio of the antiparallel and
parallel conformers.
dimethyl-5-phenyl-2-thienyl)perfluorocyclopentene (DMPTF) in
this study. These chemical structures are depicted in Scheme
1.
Recently, theoretical calculations have been carried out for a
detailed understanding of the reaction dynamics of diarylethene
derivatives.22,23 Ab initio MCSCF calculations have suggested
that the internuclear distance between the reactant carbon atoms
in the lowest excited singlet (S1) state is one of the vital factors
governing the cyclization quantum yield. This means that the
cyclization yield of diarylethenes depends on the internuclear
distance as well as the population of the antiparallel conformer.
This result is consistent with the experimental fact that the
cyclization yield does not correlate with the population of the
antiparallel conformer in some cases.22
2. Experimental Section
The spectrometer for time-resolved Raman measurements was
already described elsewhere.25,26 A beam from a regenerative
amplifier (Quantronix, Taitan, wavelength 790 nm, repetition
rate 1 kHz, pulse energy 2.8 mJ, pulse duration 4 ps) was used
to excite two optical parametric amplifier systems (Light
Conversion, TOPAS). By using sum- and difference frequency
mixings, we obtained the continuous tuning of light between
189 and 11 200 nm with keeping microjoule pulse energy. The
outputs from the two OPA systems were used as the pump beam
for exciting photochromic molecules and the probe beams for
resonance Raman observation. After passing through fixed (for
the pump beam) and variable (for the probe beam) optical delay
lines, the pump and probe beams were focused noncollinearly
onto the sample with an f ) 100 mm lens. The sample solution
was flowed through a squeezed stainless steel nozzle to form a
thin (∼0.3 mm) liquid jet. The polarization of the pump beam
was rotated by 54.7° relative to that of the probe beam to ensure
the observed kinetics free from the effects of molecular rotations.
The instrumental function was determined by a sum-frequency
mixing between the outputs from the two OPA systems. It was
estimated to be a Gaussian function with a full width at half-
maximum of 4 ps.
The 90°-scattered Raman signal was collected and dispersed
by a single spectrograph (JOBIN YVON-SPEX, 500MS) and
detected by a liquid-nitrogen-cooled CCD camera (Princeton
Instruments, LN/CCD-1340PB, 1340 × 400 pixels). The un-
shifted scattered light was rejected with a holographic notch
filter (Kaiser Optical Systems) and/or a color glass filter
(SIGMA KOKI) placed in front of the entrance slit of the
spectrograph. The spectral slit width was ∼10 cm-1. In a series
of measurements, the order of the delay time was set to be
random to minimize effects of long-term drifts of the laser
system on the observed data. The typical exposure time was
200 s. The spectra reported here were taken by the average of
the data for 10-30 cycles. The pump-only and probe-only
spectra were subtracted from the pump-probe spectra, providing
the spectra due to transients. Emission lines from a neon lamp
and solvent Raman bands were used to calibrate the transient
Raman spectra.
The photoexcitation processes of diarylethene derivatives have
been studied with time-resolved techniques.10-19 Miyasaka and
co-workers have applied picosecond transient absorption and
fluorescence techniques to the diarylethene photochromic reac-
tions. The cyclization and cycloreversion reactions of 1,2-bis-
(2,4,5-trimethyl-3-thienyl)maleic anhydride (TMTMA) were
found to proceed within 10 ps.10 The cyclization reaction was
strongly suppressed in acetonitrile, which was ascribed to the
stabilization of nonreactive intermediate states in polar solvents.
Recently, the solvent viscosity effect on the photochromism of
TMTMA has been investigated.13 The yield for the cyclization
decreased with increasing solvent viscosity, whereas that for
the cycloreversion was almost independent of the solvent
viscosity. The rapid (<12 ps) reaction channel of the cyclization
of 1,2-bis(2,5-dimethyl-3-thienyl)perfluorocyclopentene (DMTF)
has been observed in solution and crystalline phases.11 Tamai
et al. have studied the cyclization dynamics of a thiophene
oligomer with a diarylethene structure by femtosecond time-
resolved absorption spectroscopy.14 They observed the absorp-
tion due to the intermediate state immediately after photoexci-
tation (<100 fs) and estimated the time constant of the
cyclization reaction from this intermediate state to be 1.1 ps.
Ern and co-workers have reported that the cyclization reaction
of 1,2-bis(2-methyl-3-thienyl)perfluorocyclopentene proceeds
within 1.9 ps.15
Vibrationally resolved spectra have the potential for distin-
guishing coexisting species from each other in solvents.
Recently, we have reported the FT-Raman spectra of 1,2-bis-
(3-methyl-2-thienyl)perfluorocyclopentene in acetonitrile.24 The
Raman bands in the 1400-1600 cm-1 region of the open forms
are clearly distinguished from those of the closed forms. The
vibrational assignments were also performed by density func-
tional theory (DFT) calculations at the B3LYP/6-31G** level.
The FT-Raman spectroscopic study suggests that time-resolved
vibrational spectroscopy is a promising method for investigating
structural changes of the molecule during the photochromic
reaction process. To our knowledge, however, no one has
reported time-resolved Raman spectra of diarylethenes, primarily
because of the difficulty in obtaining two broadly tunable
picosecond pulses applicable to time-resolved Raman spectros-
copy.
DMTF was synthesized and purified as reported previously.8
Synthesis method of DMPTF is described in the Appendix.
Solvents (Wako Pure Chemical Industries, special reagent grade)
were used as received. The concentration of the sample solution
was 1 × 10-3∼5 × 10-4 mol dm-3. All experiments were
carried out at room temperature (295 K).
3. Results
3.1. Time-Resolved Stokes Raman Spectra of DMTF.
Figure 1 shows the electronic absorption spectra of the open
and closed forms of DMTF in 1-butanol. The open forms are
almost colorless, and their electronic absorption intensities start
from the UV region (<360 nm), whereas the closed forms
generated by UV irradiation show the absorption in the visible
region. Figure 2A shows the picosecond time-resolved Stokes
Raman spectra on excitation of the open forms of DMTF in
1-butanol. Raman bands of the solvent and fluorescence from
the open forms are subtracted in these spectra. The pump and
probe wavelengths are 310 and 568 nm, respectively. The probe
wavelength of 568 nm is in resonance with the absorption of
We have recently constructed a two-color independently
tunable Raman spectrometer with a time resolution of 4 ps.25,26
By using this Raman spectrometer, we have studied the
cyclization and cycloreversion reactions of diarylethene deriva-
tives in solutions. The resonance Stokes and anti-Stokes Raman
spectra of the open and closed forms generated during the
photochromism has been observed. These vibrational spectra
reflect configurational changes and vibrational relaxation of the
generated photoisomers. We have used DMTF and 1,2-bis(3,4-

5386 J. Phys. Chem. A, Vol. 107, No. 28, 2003
Okabe et al.
Figure 1. Electronic absorption spectra of open and closed forms of
DMTF in 1-butanol solution. Solid line, open form; dotted line, closed
form.
Figure 3. (A) Time dependence of the Stokes Raman bands of the
generated closed forms of DMTF in 1-butanol. The intensity scale is
normalized for the peak intensities of the respective spectra. Solid trace,
3 ps; dotted trace, 13 ps; dashed trace, 65 ps. Pump, 310 nm; probe,
568 nm. (B) Temporal change in the center wavenumber of the Raman
band of the closed forms of DMTF at 1551 cm-1. The solid curve
represents the best fit to the data by the convolution of the instrumental
function with a single-exponential function. The obtained time constant
is 10 ps.
behavior. These Raman results lead us to a conclusion that the
cyclization reaction of DMTF occurs within 4 ps. The instru-
mental-limited rise is also observed in ethylene glycol with a
high viscosity coefficient. Fluorescence arising from the excited
open forms seriously interferes with the observation in the anti-
Stokes Raman region. With the probe wavelengths of 568, 632,
670, and 790 nm, we have detected no Raman bands assignable
to the transient species in electronically excited states. It has
been known that the transient species of DMTF show only weak
absorption intensities in the visible and NIR region.11 Because
the resonance Raman intensity is proportional to the electronic
transition moment to the fourth power, the resonance Raman
intensities due to the transient species are not so emphasized
as to be detected within our experimental uncertainty.
Figure 2. (A) Picosecond time-resolved Stokes Raman spectra of
DMTF in 1-butanol. Pump, 310 nm; probe, 568 nm. The delay time is
given on the right-hand side of each spectrum. (B) Temporal intensity
change of the transient Raman band at 1551 cm-1. The solid curve is
the fit to the data by the convolution of the instrumental function with
a nanosecond decay function.
In Figure 2A, the peak positions and bandwidths of the Raman
bands also change with the delay time. Figure 3A shows the
time-resolved Raman spectra at delay times 3, 13, and 65 ps.
The intensity of each spectrum is roughly normalized by
reference to the intensity of the band at 1551 cm-1. The peak
positions of the transient Raman bands shift to higher-wave-
numbers, and their bandwidths become narrower as the delay
time is increased. The center wavenumber of the 1551 cm-1
band is plotted against the delay time in Figure 3B. We adopt
the first moment analysis to estimate the center wavenumber.25
The peak position at 0 ps delay time is found to be about 16
cm-1 lower than that at 65 ps. The observed time profile is
satisfactorily reproduced by the convolution of the instrument
function with a single-exponential function. The obtained time
constant is 10 ps.
the closed forms in Figure 1. The peak positions and the intensity
ratios of the two bands at 1551 and 1592 cm-1 in Figure 2A
are almost the same as those of the bands observed in cw
resonance Raman spectra of the closed forms. These two bands
disappear with the probe wavelengths off resonance with the
visible absorption of the closed forms. Thus, the observed two
Raman bands in Figure 2A can be ascribed to the closed forms
in the S0 state generated via the cyclization.
Figure 2B shows the time-development of the integrated
intensity of the Raman band at 1551 cm-1. The time-resolved
curve rises within the experimental temporal resolution (<4 ps)
and then remains almost unchanged for delay times >10 ps.
The intensity of the 1592 cm-1 band also shows a very similar

Photochromic Reactions of Diarylethene Derivatives
J. Phys. Chem. A, Vol. 107, No. 28, 2003 5387
with the absorption due to the open forms. The peak positions
of the two bands at 1545 and 1599 cm-1 in Figure 5A are almost
the same as those of the vibrational bands of the open forms
observed by FT-Raman spectroscopy. Thus, the observed two
anti-Stokes Raman bands in Figure 5A can be assigned to the
vibrationally excited open forms in the S0 state generated by
the photoexcitation of the closed forms. In Figure 5A, the
intensities of the anti-Stokes Raman bands rise within the
experimental temporal resolution (<4 ps), and then these bands
completely disappear at 16 ps delay time. This result indicates
that the cycloreversion of DMPTF proceeds within 4 ps and
the vibrationally excited open forms are generated as the
intermediate state. It is difficult to obtain the reliable Stokes
Raman data because of interference by fluorescence from the
excited open forms.
Figure 4. Electronic absorption spectra of open and closed forms of
DMPTF in acetonitrile solution. Solid line, open form; dotted line,
closed form.
As shown in Figure 5A, the relative intensity of the two anti-
Stokes Raman bands also changes with the delay time. Figure
5B shows the anti-Stokes Raman spectra at delay times 2 and
8.5 ps. Each spectrum is roughly normalized to the intensity of
the 1599 cm-1 band. The intensity of the 1545 cm-1 band
relative to that of the 1599 cm-1 band becomes larger as the
delay time increases.
4. Discussion
4.1. Cyclization Reaction of DMTF. We first discuss the
dynamics of the cyclization reaction of DMTF. As mentioned
before, the Stokes Raman intensities due to the photogenerated
closed forms show the instrumental-limited rise (<4 ps),
indicating the cyclization of DMTF occurs within 4 ps. The
changes in the band shape as well as in the band intensity are
also important to understand the cyclization process. As shown
in Figure 3, the center wavenumbers of the Stokes bands shift
upward with a time constant of 10 ps, being accompanied by
band narrowing. The high-wavenumber shift and the band
narrowing of vibrational bands, which occur on a picosecond
time scale, have been observed in other polyatomic molecules
in solution. All of the dynamics have been considered to be
associated with intermolecular vibrational relaxation of hot
solutes toward thermal equilibria with surrounding solvent
molecules.25-31 Iwata and Hamaguchi have observed the
picosecond high-wavenumber shift and the band narrowing of
the Raman bands of trans-stilbene in the S1 state.30 The temporal
shift of the olefinic CdC stretching band was found to be a
good indicator of the temperature change in the solute. For
picosecond Raman spectra of nickel octaethylporphyrin, Mi-
zutani et al. have shown that the time constant of the high-
wavenumber shift is close to that of vibrational relaxation of
the solute toward a thermal equilibrium with solvent mol-
ecules.31 The peak positions and bandwidths of vibrational bands
depend on the temperature of solute molecules; in general, the
vibrational bands due to skeletal modes broaden and shift to
lower-wavenumbers with increasing temperature. This behavior
has been considered to be mainly caused by anharmonic
couplings with the bath of low-wavenumber modes.32
Figure 5. (A) Picosecond time-resolved anti-Stokes Raman spectra
of DMPTF in acetonitrile. Pump, 480 nm; probe, 395 nm. Dotted lines
indicate the peak positions of the anti-Stokes Raman bands of the open
forms of DMPTF in the S0 state. The delay time is given on the right-
hand side of each spectrum. (B) Time dependence of the anti-Stokes
Raman bands of the generated open forms of DMPTF in acetonitrile.
The intensity scale is normalized for the peak intensities of the
respective spectra. Solid trace, 2 ps; dotted trace, 8.5 ps.
3.2. Time-Resolved Anti-Stokes Raman Spectra of DMPTF.
The electronic absorption spectra of the open and closed forms
of DMPTF in acetonitrile are shown in Figure 4. The open forms
of DMPTF are pale yellow and have the absorption maximum
at ≈370 nm. The closed forms photogenerated from the open
forms are orange and show the visible absorption around 470
nm. Figure 5A shows the picosecond time-resolved anti-Stokes
Raman spectra on excitation of the closed forms of DMPTF in
acetonitrile. These anti-Stokes Raman spectra arise from only
the vibrationally excited transients. Fluorescence from the
sample and solvent Raman bands are subtracted. The pump
wavelength of 480 nm is near the absorption maximum of the
closed forms. The probe wavelength of 395 nm is in resonance
Immediately after the cyclization reaction, the vibrationally
excited closed forms in the S0 state should be generated as an
intermediate state. This is because the ultrafast cyclization
deposits a large amount of excess energy [in this case, the
difference between electronic energies of the photoexcited open
forms and the closed forms in the S0 state] into the vibrational
manifold of the photoproducts. The relaxation process from the
vibrationally excited states to a thermodynamically favorable
state may be hence observed spectroscopically. Generally
speaking, it takes several to tens of picoseconds for the

5388 J. Phys. Chem. A, Vol. 107, No. 28, 2003
Okabe et al.
vibrationally excited large molecules to dissipate excess energy
into surrounding solvent molecules through intermolecular
interactions.25-31,33 Thus, the high-wavenumber shift and the
band narrowing observed in the present study can be ascribed
to the intermolecular vibrational relaxation of the photogenerated
closed forms in the S0 state toward the thermal equilibrium with
solvent molecules. The anti-Stokes/Stokes intensity ratios will
provide information on vibrational relaxation, however, the
difference between the resonance effects of the Stokes and anti-
Stokes Raman spectra must be considered in their analysis.
parallel and antiparallel conformers exist in almost equal
amounts in solution.6,7,9 It may be necessary to clarify the large
intensity difference between the subnanosecond and nanosecond
components, if we assign the subnanosecond fluorescence to
the antiparallel conformers and the nanosecond fluorescence to
the parallel conformers.
Both the parallel and antiparallel conformers existing in
solution are photoexcited on the UV irradiation. Because there
is no chance for the parallel conformers to be converted to the
closed forms, it is conceivable that the photoexcited parallel
conformers have subnanosecond and nanosecond lifetimes and
show the observed fluorescence. This means that both the
subnanosecond and nanosecond fluorescence signals are due
to the photoexcited parallel conformers.
The above result indicates that the cyclization of DMTF
occurs within 4 ps and the vibrationally excited closed forms
in the S0 state are generated as the intermediate. The energy
dissipation into surrounding solvent molecules occurs with a
time constant of 10 ps, generating the thermally equilibrated
S0 closed forms at room temperature.
In the present study, we cannot determine which of the two
mechanisms proceeds after the photoexcitation of the open forms
of DMTF. Measurements with higher temporal resolution are
needed to reach a more definitive conclusion on this point.
The activation energy of the cyclization reaction of DMTF
should be very small because the cyclization occurs within 4
ps. This result is consistent with the fact that the ultrafast
cyclization (<4 ps) is also observed in ethylene glycol with a
high viscosity coefficient. However, two fluorescence compo-
nents with lifetimes of subnanoseconds and a few nanoseconds
were observed on the photoexcitation of the open forms of
DMTF.11 These components do not arise from the generated
closed forms, which are known to show no fluorescence in the
nanosecond time range.9 Thus, it is considered that the
nanosecond fluorescence signals are due to the open forms,
although the very rapid cyclization channel (<4 ps) exists in
their photoexcited states. Two possible mechanisms can be
considered to explain the existence of both the ultrafast and
the nanosecond components in the photoexcited open forms as
follows.
4.2. Cycloreversion Reaction of DMPTF. In Figure 5A, the
anti-Stokes Raman intensities of the photogenerated open forms
show the instrumental-limited rise, indicating that the cyclo-
reversion of DMPTF occurs within 4 ps. The excess energy
generated via the ultrafast cycloreversion must be initially
localized on Franck-Condon active vibrations of the S0 open
forms; the vibrationally excited populations must be greatly
different from those under the thermal equilibrium at room
temperature. The localized excess energy is then distributed
among all intramolecular vibrational modes according to the
Boltzmann distribution (intramolecular vibrational relaxation).
This intramolecular relaxation creates a quasi-equilibrium among
all intramolecular modes, resulting in a single, mode-indepen-
dent vibrational temperature. As discussed in subsection 4.1,
this randomized vibrational energy dissipates to solvent mol-
ecules in tens of picoseconds (intermolecular vibrational
relaxation).
The optical transition between the S0 and S1 states is forbidden
by symmetry.21-23 Under irradiation of UV light, the open forms
are excited to optically allowed, excited singlet state (S2). Ab
initio molecular orbital calculations have suggested that the S1
molecules formed via the S2 state have two potential minima:
one is a reactive site from which the very rapid cyclization
occurs and the other is a nonreactive site. We have observed
the fluorescence excitation and dispersed fluorescence spectra
of the open forms of 1,2-bis(3-methyl-2-thienyl)perfluoro-
cyclopentene in a supersonic free jet.34 The dispersed fluores-
cence spectrum was strongly Stokes-shifted with respect to the
excitation wavelength, indicating that the optically allowed state
is not the S1 state. The S1 state may be generated via very fast
internal conversion from the S2 state, providing fluorescence.
Thus, the experimental results are consistent with the theoretical
ones. Hence, it is considered that the observed subnanosecond
and nanosecond fluorescence signals arise from the open forms
in the nonreactive site with subnanosecond and nanosecond
lifetimes. This means that the cyclization takes place mainly
from the reactive site in the S1 state, whereas the reaction from
the nonreactive site may also occur with a low quantum yield.
This model is consistent with the work of Miyasaka et al., who
have suggested that both the very rapid and slow reaction
pathways exist in the cyclization of TMTMA.13 They observed
the subnanosecond and nanosecond fluorescence signals fol-
lowing the excitation of the open forms of TMTMA and ascribed
the subnanosecond component to the intermediate for the slow
cyclization and the nanosecond component to the photoexcited
parallel conformers which cannot convert to the closed forms.
The fluorescence intensity of the nanosecond component was
observed to be rather smaller than that of the subnanosecond
component, although NMR spectroscopy has suggested that the
Time-resolved anti-Stokes Raman spectroscopy is one of the
powerful methods for studying vibrational relaxation in solu-
tion.29,31,34,35 It probes only those molecules that are populated
in excited vibrational states, and hence provides state-specific
dynamical information that is essential for a detailed under-
standing of vibrational relaxation. In general, anti-Stokes Raman
intensities for high-wavenumber modes are very weak, because
of small populations on their vibrationally excited levels at room
temperature. In Figure 5A, on the contrary, the anti-Stokes
Raman bands for the high-wavenumber modes of the open forms
are clearly observed at short delay times. These anti-Stokes
Raman intensities decrease with increasing delay time and
disappear at 16 ps delay time within the experimental accuracy.
The large intensity decrease with the delay time in Figure 5A
clearly indicates that the anti-Stokes Raman spectra at short
delay times are attributed to the vibrationally excited open forms
generated via the cycloreversion. At 16 ps delay time, the
photogenerated open forms have almost reached the thermal
equilibrium at room temperature. Thus, no transient anti-Stokes
Raman signal is observed at 16 ps within our experimental
uncertainty.
The relative intensity between the two anti-Stokes bands also
changes with the delay time (Figure 5B). It is hard to explain
the observed relative intensity change on the basis of the
Boltzmann distribution because the energy difference between
these bands is only ≈54 cm-1. If the Boltzmann distribution of
the excess energy is assumed, the single intramolecular vibra-
tional temperature must be higher than 1500 K, which is
apparently an unrealistic value. Thus, the observed relative

Photochromic Reactions of Diarylethene Derivatives
J. Phys. Chem. A, Vol. 107, No. 28, 2003 5389
intensity change reflects the nonstatistical distribution of the
excess energy in the generated open forms at short delay times.
In view of the temporal resolution of the present experiment, it
is concluded that it takes at least several picoseconds for the
open forms of DMPTF in the S0 state to reach the intramolecular
thermal equilibrium. In other words, the intramolecular vibra-
tional relaxation does not complete within a few picoseconds.
For medium-size to large molecules in solution, an empirical
rule that the time scale of intramolecular vibrational relaxation
is in the subpicosecond range and that of intramolecular
relaxation in tens of picoseconds has been proposed from a
number of data on vibrational relaxation.33 However, recent
experiments have suggested that intramolecular relaxation occurs
in the time range of several picoseconds in some cases.29,31,35-37
At short delay times, the intensity of the 1599 cm-1 band is
stronger than that of the 1545 cm-1 band. With increasing delay
time, the intensity difference between these two bands becomes
smaller. This indicates that the excess energy is localized on
the 1599 cm-1 mode rather than on the 1545 cm-1 mode at
short delay times. Ab initio DFT calculations have assigned the
observed 1545 and 1599 cm-1 bands to the CdC stretching
modes of the thiophene and the cyclopentene moiety, respec-
tively.24 Thus, it is concluded that the excess energy generated
via the cycloreversion is localized on the CdC stretching mode
of the cyclopentene moiety of the S0 open forms of DMPTF.
This result indicates that the CdC stretching mode of the
cyclopentene moiety is a promoting or an accepting mode in
the cycloreversion reaction. This mode can be regarded as one
of the vital factors governing the cycloreversion rate.
methylethylenediamine (TMEDA; 6.0 mL, 40 mmol) in anhy-
drous ether (40 mL) was added dropwise n-BuLi (1.6 M in
hexane, 25 mL, 41 mmol) and stirred for 1 h at 0 °C under
argon atmosphere. After cooling the solution to -78 °C, tri-n-
butyl borate (11.5 mL, 43 mmol) was gradually added and
stirred for 2 h. After warming to room temperature, the reaction
mixture diluted with anhydrous THF (120 mL). To the solution
was added 20 wt % Na2CO3 aqueous solution (48 mL),
iodobenzene (7.2 g, 35 mmol), Pd(PPh3)4 (2.15 g). The solution
was refluxed for 5 h at 70 °C. After the heating was over, the
reaction mixture was poured into the water, and the reaction
was extracted with ether. The combined organic layer was dried
with MgSO4, filtered, and evaporated in vacuo. The residue was
purified by silica gel column chromatograph (hexane) to give
4.63 g of 3,4-dimethyl-2-phenylthiophene in 68% yield as a
1
colorless liquid: H NMR (200 MHz, CDCl3) δ ) 2.19 (s, 3H),
2.21 (s, 3H), 6.90 (s, 1H), 7.2-7.5 (m, 5H); MS m/z 188 (M+).
Anal. Calcd for C12H12S: C, 76.55; H, 6.42. Found: C, 76.38;
H, 6.51.
1,2-Bis(3,4-dimethyl-5-phenyl-2-thienyl)perfluorocyclo-
pentene. To a solution containing 4.63 g (24.6 mmol) of 3,4-
dimethyl-2-phenylthiophene in 30 mL anhydrous ether (150 mL)
was added 16.2 mL (26.6 mmol) of n-BuLi (1.6 N in hexane
solution) at -5 °C under an argon atmosphere. After the mixture
was heated under reflux for 45 min, the reaction mixture was
cooled to -78 °C. Perfluorocyclopentene (1.62 mL, 12.1 mmol)
was added gradually and stirred for 1 h at this temperature.
Methanol was added to the reaction mixture and extracted with
ether, dried with MgSO4, filtered, and evaporated in vacuo. The
residue was purified by silica gel column chromatography
(hexane) followed by recrystallization from hexane afforded
yellow crystals of 1,2-bis(3,4-dimethyl-5-phenyl-2-thienyl)per-
fluorocyclopentene (2.0 g, 30%): mp 150.1-150.8 °C; 1H NMR
(200 MHz, CDCl3) δ ) 1.75 (s, 3H), 2.14 (s, 3H), 7.3-7.5 (m,
5H); MS m/z 548 (M+). Anal. Calcd for C29H22F6S2: C, 63.49;
H, 4.04. Found: C, 63.27; H, 4.14.
5. Conclusion
Photochromic reactions of diarylethene derivatives have been
studied by picosecond time-resolved Stokes and anti-Stokes
Raman spectroscopies. Two common features of the cyclization
and cycloreversion reactions are found: the photochromic
reactions proceed within 4 ps and the vibrationally excited
photoisomers in the S0 state are generated as the intermediate
state. Two anti-Stokes Raman bands at 1545 and 1599 cm-1
are observed on photoexcitation of the closed forms of DMPTF.
They are assignable to the CdC stretching modes of the
thiophene and the cyclopentene moiety of the generated S0 open
forms, respectively. The intensity of the cyclopentene moiety
relative to that of the thiophene moiety becomes smaller with
the delay time, indicating that part of the excess energy
generated via the cycloreversion is localized on the CdC
stretching mode of the cyclopentene moiety. This means that
the CdC stretching mode of the cyclopentene moiety is one of
the promoting or the accepting modes in the cycloreversion
reaction. The CdC stretching mode of the cyclopentene moiety
is found to be one of the vital factors governing the cyclo-
reversion yield. Finally, we note that time-resolved vibrational
spectroscopy provides important information on vibrational
relaxation and reaction coordinates, which is difficult to obtain
from electronic spectroscopic techniques in condensed phases.
This method will give new scope for elucidating the dynamical
features of diarylethene photochromism.
References and Notes
(1) Hirano, M.; Miyashita, A.; Nohira, H. Chem. Lett. 1991, 209.
(2) Sachman, E. J. Am. Chem. Soc. 1971, 93, 7088.
(3) Heller, H. G.; Oliver, S. J. Chem. Soc., Perkin Trans. 1 1981, 197.
(4) Darcy, P. J.; Heller, H. G.; Strydom, P. J.; Whittall, J. J. Chem.
Soc., Perkin Trans. 1 1981, 202.
(5) Irie, M.; Mohri, M. J. Org. Chem. 1988, 53, 803.
(6) Uchida, K.; Nakayama, Y.; Irie, M. Bull. Chem. Soc. Jpn. 1990,
63, 1311.
(7) Hanazawa, M.; Sumiya, R.; Horikawa, Y.; Irie, M. J. Chem. Soc.,
Chem. Commun. 1992, 206.
(8) Kobatake, S.; Yamada, T.; Uchida, K.; Kato, N.; Irie, M. J. Am.
Chem. Soc. 1999, 121, 2380.
(9) Irie, M. Chem. ReV. 2000, 100, 1685.
(10) Miyasaka, H.; Araki, S.; Tabata, A.; Nobuto, T.; Mataga, N.; Irie,
M. Chem. Phys. Lett. 1994, 230, 249.
(11) Miyasaka, H.; Nobuto, T.; Itaya, A.; Tamai, N.; Irie, M. Chem.
Phys. Lett. 1997, 269, 281.
(12) Miyasaka, H.; Murakami, M.; Itaya, A.; Guillaumont, D. Nakamura,
S.; Irie, M. J. Am. Chem. Soc. 2001, 123, 753.
(13) Miyasaka, H.; Nobuto, T.; Murakami, M.; Itaya, A.; Tamai, N.;
Irie, M. J. Phys. Chem. A 2002, 106, 8096.
(14) Tamai, N.; Saika, T.; Shimidzu, T.; Irie, M. J. Phys. Chem. 1996,
100, 4689.
(15) Ern, J.; Bens, A. T.; Bock, A.; Martin, H.-D.; Kryschi, C. J. Lumin.
1998, 76/77, 90.
(16) Ern, J.; Bens, A. T.; Martin, H.-D.; Mukamel, S.; Schmid, D.;
Tretiak, S.; Tsiper, E.; Kryschi, C. Chem. Phys. 1999, 246, 115.
(17) Ern, J.; Bens, A. T.; Martin, H.-D.; Mukamel, S.; Schmid, D.;
Tretiak, S.; Tsiper, E.; Kryschi, C. J. Lumin. 2000, 87/79, 742.
(18) Ern, J.; Bens, A. T.; Martin, H.-D.; Kuldova, K.; Trommsdorff, H.
P.; Kryschi, C. J. Phys. Chem. A 2002, 106, 1654.
(19) Owrutsky, J. C.; Nelson, H. H.; Baronavski, A. P.; Kim, O.-K.;
Tsivgoulis, G. M.; Gilat, S. L.; Lehn, J.-M. Chem. Phys. Lett. 1998, 293,
555.
Acknowledgment. This work was supported by the Joint
Studies Program (2002-2003) of the Institute for Molecular
Science.
Appendix
3,4-Dimethyl-2-phenylthiophene. To a solution of 3,4-
dimethylthiophene (4.1 g, 36.3 mmol) and N,N,N′,N′-tetra-

5390 J. Phys. Chem. A, Vol. 107, No. 28, 2003
Okabe et al.
(20) Kawai, T.; Fukuda, N.; Gro¨schl, D.; Kobatake, S.; Irie, M. Jpn. J.
Appl. Phys. 1999, 38, L1194.
(21) Goldberg, A.; Murakami, A.; Kanda, K.; Kobayashi, T.; Nakamura,
S.; Uchida, K.; Sekiya, H.; Fukaminato, T.; Kawai, T.; Kobatake, S.; Irie,
M. to be published.
(22) Uchida, K.; Guillaumont, D.; Tsuchida, E.; Mochizuki, G.; Irie,
M.; Murakami, A.; Nakamura, S. J. Mol. Struct. (THEOCHEM) 2002, 579,
115.
(23) Guillaumont, D.; Kobayashi, T.; Kanda, K.; Miyasaka, H.; Uchida,
K.; Kobatake, S.; Shibata, K.; Nakamura, S.; Irie, M. J. Phys. Chem. A
2002, 106, 7222.
(24) Okabe, C.; Tanaka, N.; Fukaminato, T.; Kawai, T.; Irie, M.; Nibu,
Y.; Shimada, H.; Goldberg, A.; Nakamura, S.; Sekiya, H. Chem. Phys. Lett.
2002, 357, 113.
(25) Nakabayashi, T.; Kamo, S.; Sakuragi, H.; Nishi, N. J. Phys. Chem.
A 2001, 105, 8605.
(26) Nakabayashi, T.; Kamo, S.; Watanabe, K.; Sakuragi, H.; Nishi, N.
Chem. Phys. Lett. 2002, 355, 241.
(27) Butler, R. M.; Lynn, M. A.; Gustafson, T. L. J. Phys. Chem. 1993,
97, 2609.
(28) Hester, R. E.; Matousek, P.; Moore, J. N.; Parker, A. W.; Toner,
W. T.; Towrie, M. Chem. Phys. Lett. 1993, 208, 471.
(29) Qian, J.; Schultz, S. L.; Jean, J. M. Chem. Phys. Lett. 1995, 233,
9.
(30) Iwata, K.; Hamaguchi, H. J. Phys. Chem. A 1997, 101, 632.
(31) Mizutani, Y.; Uesugi, Y.; Kitagawa, T. J. Chem. Phys. 1999, 111,
8950.
(32) Asher, S. A.; Murtaugh, J. J. Am. Chem. Soc. 1983, 105, 7244.
(33) Elsaesser, T.; Kaiser, W. Annu. ReV. Phys. Chem. 1989, 40, 143.
(34) Tanaka, N.; Okabe, C.; Sakota, K.; Fukaminato, T.; Kawai, T.; Irie,
M.; Goldberg, A.; Nakamura, S.; Sekiya, H. J. Mol. Struct. 2002, 616, 113.
(35) Nakabayashi, T.; Okamoto, H.; Tasumi, M. J. Phys. Chem. A 1998,
102, 9686.
(36) Shreve, A. P.; Mathies, R. A. J. Phys. Chem. 1995, 99, 7285.
(37) Lenz, K.; Pfeiffer, A.; Lau, A.; Elsaesser, T. Chem. Phys. Lett.
1994, 229, 340.