An optical and theoretical investigation of the ultrafast dynamics of a bisthienylethene-based photochromic switch

Article abstract of DOI:10.1021/jp020903+

The switching behavior of 1,2-bis(5-phenyl-2-methylthien-3-yl)cyclopentene is studied by means of polarization selective nonlinear optical spectroscopy and time-dependent density functional theory. The combined information from the observed population and orientational dynamics together with the results of theoretical calculations show that on a subpicosecond time scale rapid mixing and relaxation of electronic states occur, before switching takes place. Such preswitching dynamics was not studied in detail in these systems before. Then, the switching process itself occurs by the formation of a C-C bond in the central cyclopentene ring with a time constant of 4.2 ps. Driven by the ring closure, the side groups of the switch molecules rotate to a nearly coplanar conformation with a time constant of about 8 ps. The switching process is completed by relaxation of the vibrationally hot ground state of the closed form of the molecule to thermal equilibrium.

Full text of DOI:10.1021/jp020903+

8498
J. Phys. Chem. A 2002, 106, 8498-8507
ARTICLES
An Optical and Theoretical Investigation of the Ultrafast Dynamics of a
Bisthienylethene-Based Photochromic Switch
P. R. Hania,† R. Telesca,‡ L. N. Lucas,§ A. Pugzlys,† J. van Esch,§ B. L. Feringa,§
J. G. Snijders,‡ and K. Duppen*,†
Ultrafast Laser and Spectroscopy Laboratory, Theoretical Chemistry, Organic Chemistry and Catalysis,
Materials Science Center, RijksuniVersiteit Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands
ReceiVed: April 4, 2002; In Final Form: July 5, 2002
The switching behavior of 1,2-bis(5-phenyl-2-methylthien-3-yl)cyclopentene is studied by means of polarization
selective nonlinear optical spectroscopy and time-dependent density functional theory. The combined
information from the observed population and orientational dynamics together with the results of theoretical
calculations show that on a subpicosecond time scale rapid mixing and relaxation of electronic states occur,
before switching takes place. Such preswitching dynamics was not studied in detail in these systems before.
Then, the switching process itself occurs by the formation of a C-C bond in the central cyclopentene ring
with a time constant of 4.2 ps. Driven by the ring closure, the side groups of the switch molecules rotate to
a nearly coplanar conformation with a time constant of about 8 ps. The switching process is completed by
relaxation of the vibrationally hot ground state of the closed form of the molecule to thermal equilibrium.
I. Introduction
In recent years, there has been increasing interest in the
synthesis, investigation, and application of organic photochromic
materials as a possible basis for optoelectronic and photooptical
devices. Diarylethenes with heterocyclic aryl groups belong to
a new class of thermally irreversible photochromic switches.
Besides low fatigue and high thermal stability, diarylethenes
and, in particular, bisthienylethene (BTE)-based compounds
feature remarkable switching sensitivity (high quantum yield)
and rapid response.1,2
Figure 1. Chemical structure of the B-DTCP molecule in the open
(1A) and closed (1B) forms. Interconversion is possible by irradiation
with UV (1A f 1B) and with visible (1B f 1A) light.
Photochromism is defined as a reversible photoinduced
chemical reaction during which optical, dielectric, and confor-
mational properties of molecules change. In Figure 1 the
chemical structures of 1,2-bis(5-phenyl-2-methylthien-3-yl)-
cyclopentene (B-DTCP) in the open- (1A) and closed-ring (1B)
configurations are presented. The ground-state interconversion
between the isomers is prohibited but upon UV irradiation 1A
converts to 1B. As a consequence of the near-coplanar geometry
of the thiophene moieties of 1B, the π-conjugation spreads
throughout the molecule, resulting in the appearance of a new
red-shifted absorption band. Upon visible irradiation this
conjugation is broken again by opening the ring to yield the
initial open form 1A again. The steady-state absorption spectra
of 1A and 1B are presented in Figure 2 (for more details see
section III).
Figure 2. Steady-state absorption spectra of B-DTCP in the open form
(solid line), B-DTCP in the closed form (dotted line), and BT-
chromophore (dashed line) that constitutes the side groups of B-DTCP.
The solvent is cyclohexane.
In the past decade many experimental investigations on the
switching dynamics of various photochromic compounds were
performed including both ring-closure3-6 and ring-opening7-9
reactions. The time constants reported in the literature for ring-
opening reactions vary from 2.2 ps7 and <10 ps3 to 325 ps,8
whereas ring-closure times in most cases are substantially shorter
and much less spread. For instance, pump-probe experiments
* To whom correspondence should be addressed. Tel: +31.50.3634560.
Fax: +31.50.3634441. E-mail: K.Duppen@chem.rug.nl.
† Ultrafast Laser and Spectroscopy Laboratory.
‡ Theoretical Chemistry.
§ Organic Chemistry and Catalysis.
10.1021/jp020903+ CCC: $22.00 © 2002 American Chemical Society
Published on Web 08/27/2002

Bisthienylethene-Based Photochromic Switch
J. Phys. Chem. A, Vol. 106, No. 37, 2002 8499
on several dithienylperfluorocyclopentene switches have re-
vealed ring-closure time constants from 1 to 3 ps.4-6,10 The
different switching time constants presumably originate from
differences in the chemical structure of the investigated com-
pounds.
laser system (Hurricane, Spectra Physics) and optical parametric
amplifiers (OPA). The laser system produces 120-fs, 800-µJ
pulses at 1 kHz, centered at 800 nm. About 300 µJ of that energy
is used to pump a travelling wave optical parametric amplifier
of superfluorescence (TOPAS, Light Conversion LTD), which
is employed to generate the pump pulses. The sample is excited
with 70-nJ pulses centered at 310 nm (fourth harmonics of the
signal wave of the TOPAS at 1240 nm). Before the generation
of 310 nm light, the second harmonics of the output of the
TOPAS (620 nm) is compressed in a double-pass compressor
based on two BK7 prisms. In addition to the pulse shortening,
the compressor allows the spatial separation of different spectral
components of the parametric light (signal and idler beams).
The polarization of the pump beam is changed by using a λ/2
plate.
Woodward-Hoffmann rules for electrocyclic reactions11
predict in the case of 6π-electron ring-closure a concerted one-
step mechanism through a conrotatory path in the excited state.12
Indeed, no intermediates in the pathway of the ring-closure
reaction were found by Ern et al.10 in the case of 1,2-bis(5-
formyl-2-methylthien-3-yl)perfluorocyclopentene, by Ohtaka et
al.5 in the case of diarylethene derivatives with terthiophene,
and by Owrutsky et al.6 in the case of 1,2-bis(5-pyridyl-2-
methylthien-3-yl)perfluorocyclopentene. In contrast, for a dia-
rylethene structure with thiophene oligomers as side groups,
Tamai et al.4 reported the rapid (∼100 fs) formation of an
intermediate state, followed by ring closure with a time constant
of about 1.1 ps. In that study the nature of the intermediate
precursor states was not discussed. Electronic-conformational
relaxation to a precursor state on a time scale of about 0.9 ps,
followed by ring closure with a time constant of about 10 ps,
was also reported by Ern et al.13 in the case of 1,2-bis(5-anthryl-
2-methylthien-3-yl)perfluorocyclopentene. Recently the exist-
ence of a precursor to the ring closure that is formed within
less than 1 ps was reported by Bens et al.14 for di-bpe
dithienylethene.
All the measurements reported so far were performed with a
time resolution of a few hundred of femtoseconds. In this paper
we present results of spectroscopic studies on the switching
behavior of B-DTCP with a time resolution below 100 fs. In
this way the presence or absence of precursor states to the ring-
closure reaction itself can be unequivocally established. To our
knowledge, the work presented here is the first study of
dithienylcyclopentene switches possessing nonfluorinated cy-
clopentene moieties. Because the ring closure by definition is
related to conformational changes in the molecule, a second
goal of our studies is to trace the orientational dynamics in the
switch molecule during the ring-closure reaction. This should
provide important information on the mechanism of the switch-
ing process. The nature of the electronic states, involved in the
ring-closure reaction, is established by comparing the experi-
mental results to electronic structure calculations, using density
functional theory.
The probe pulses are generated in a home-built noncollinearly
pumped optical parametric amplifier (NOPA17). Briefly, the
1-mm type I BBO crystal of the NOPA is pumped by pulses
with an energy of about 10 µJ centered at 400 nm. The pulses
are tilted by passing them through a 45° fused silica prism and
subsequently focused onto the crystal by a spherical mirror of
r ) 40 cm. A white light continuum generated in a 2 mm
sapphire plate is used as a seed for the NOPA. Before
amplification, the white light is precompressed in a compressor,
based on a 1200 mm-1 grating and a curved mirror (Light
Conversion LTD) designed to compensate for the enhanced
angular dispersion of the grating at the red side of the spectrum.
A mask placed in the compressor allows spectral selection and
shaping of the probe pulse. The NOPA produces pulses with
an energy of about 0.5 µJ that are tunable in the spectral region
450-900 nm. By generating second harmonics in a 150-µm
BBO crystal, the tunability can be extended to the UV spectral
region (230-450 nm).
The time resolution of the experiments is probe wavelength
dependent. In the spectral region 460-750 nm the duration of
the probe pulses is about 30 fs and the time resolution is mainly
determined by the duration of the pump pulse. The cross-
correlation function of the pump pulse centered at 310 nm and
the probe pulse centered at 600 nm, as measured by monitoring
two-color two-photon absorption in a 100-µm glass plate,18 has
a width of about 90 fs fwhm. The probe pulses, produced by
generating second harmonics from the NOPA output, are slightly
stretched because of the spectral filtering in the BBO doubling
crystal. For the wavelength range of 380-450 nm (second
harmonics from 760 to 900 nm light) the time resolution is
degraded to about 110 fs because of strong spectral phase
distortion of the white light around the carrier wavelength 800
nm.
The paper is organized as follows: in section II the
experimental setup and main concepts of the experiments are
described. The results of femtosecond pump-probe and transient
anisotropy experiments are presented in section III. These results
are discussed and related to a theoretical analysis of the
electronic structure in section IV. The conclusions are sum-
marized in section V.
The transients of the photoinduced absorption are measured
in a standard pump-probe geometry. The probe pulse before
the sample is split into two to serve as both probe and reference.
The probe pulse is delayed with respect to the pump pulse by
using a computer-controlled delay stage, whereas the reference
pulse is set to reach the sample before the pump and probe
pulses arrive. The pump, probe, and reference pulses after
attenuation are focused by a spherical mirror of r ) 25 cm into
a 200-µm flow cell containing the sample. They are all
monitored by silicon photodiodes connected to a sample-and-
hold device. During the measurements, every second pump pulse
is blocked using a chopper that is synchronized with the laser.
Optical density changes (∆OD) in absolute units are calculated
for every single pair of probe pulses and then averaged over 50
measurements. To monitor zero delay between pump and probe
II. Experimental Section
Compound 1A (see Figure 1) was synthesized starting from
1,2-bis(5′-chloro-2′-methylthien-3′-yl)cyclopentene15 via
a
Suzuki reaction according to a slightly modified procedure
described by Lehn et al.16 For the optical experiments, 1A was
dissolved in cyclohexane (p.a., Merck) at a concentration of
0.7 mM. The solution was pumped at a speed of about 3 mL/s
to ensure that fresh sample was available for every single
measurement. The optical density of the sample did not exceed
0.3, ensuring homogeneous excitation. The experiments were
performed at room temperature (T ) 294 K).
Femtosecond polarization selective frequency resolved pump-
probe experiments were performed by using a 1 kHz Ti:sapphire

8500 J. Phys. Chem. A, Vol. 106, No. 37, 2002
Hania et al.
pulses, a cross-correlation function of pump and probe pulses
is measured for every probe wavelength.
This means that in our experiments only about half of the
molecules (those in the C2 form) participate in the switching
process.
In a separate experiment the dynamics of the whole pump-
probe spectrum is recorded by using white light probe pulses,
generated in a standing 1 cm doubly distilled water cell, and
monitored via a polychromator in a double-array OMA system
(Princeton Instruments). The pump beam is opened/closed every
2 s to obtain difference spectra (∆OD). Because these white
light pulses have a large temporal dispersion, which is hard to
correct accurately, the early time ∆OD spectra reported in the
next section are constructed from single color pump-probe
measurements as described above. The construction is possible
because the ∆OD values are measured in absolute units and
because the zero delay position between the pump and probe
pulses is known.
To determine the quantum yield of ring closure, a 0.14 mM
solution of B-DTCP in cyclohexane was irradiated by 300 µW
UV light at a wavelength of 300 nm while the buildup of the
lowest energy absorption band of the closed form was monitored
with a weak probe at 550 nm. The known ratio between the
number of B-DTCP molecules in open-ring and closed-ring
forms allows us to calculate the extinction coefficient of the
closed form ꢀ1B,550 ) (16.6 ( 0.3) × 103 L mol-1 cm-1 and to
convert the observed optical density into a number of absorbing
molecules. The slope of the absorption curve, thus obtained,
gives the value of 0.59 ( 0.1 switched molecule per absorbed
photon (the error is mainly determined by the accuracy of the
measurement of the power of the excitation light). By taking
into account that there are two different conformations in the
open-ring form of B-DTCP molecules at a close to 1:1 ratio19,20
and only one conformer is able to switch, we can conclude that
the quantum yield of ring closure in the case of switchable
conformers is very close to 1.
III. Results
3.1. Photoconversion Pathways. The steady-state absorption
spectrum of the open form of B-DTCP (1A), dissolved in
cyclohexane, is presented in Figure 2. In addition, the steady-
state absorption spectrum of the 2-phenyl-5-methylthiophene
(BT) chromophore (0.1 mM solution in a 1 mm quartz cell) is
given. The fact that there is only a small difference between
the shapes of the steady-state absorption spectra of B-DTCP
and the BT chromophore indicates that the electronic interaction
between the two BT groups in the 1A molecule is rather weak.
During the time-resolved experiments on the ring-closure
process, molecules in the open form are continuously being
converted into the closed form. This means that finally both
types of molecules will be excited by the pump pulses, which
considerably complicates the interpretation of the observed
transients. However, at low irradiation doses and not too small
sample volume, the net effect of this buildup of photoproduct
is limited and we can assume that only the forward reaction is
probed. The average laser power in the pump-probe experi-
ments is about 70 µW, which is equivalent to 1 × 1014 photons/
s. With an optical density of 0.3 at the excitation wavelength
and a switching probability of 0.59, this leads to the creation
of ∼3 × 1013 molecules in the closed form per second. Initially,
the flow system contains a 30 mL solution of the molecules in
the open form at a concentration of 0.7 mM, meaning that the
total amount of optically active molecules is about 1.3 × 1019.
If we take as the acceptable limit for contamination of the
solution the situation that at most 5% of the molecules are in
the closed form, the sample can be irradiated for about 2.2 ×
104 s (more than 6 h) before replacement is necessary.
Upon photoexcitation by UV light (extinction coefficient
ꢀ1A,310 ) (21.2 ( 0.3) × 103 L mol-1 cm-1), the open form 1A
is converted into the closed form 1B via a cyclization (ring-
closure) reaction. This results in a spread of π-conjugation
throughout the molecule and, consequently, in the appearance
of a new red-shifted absorption band. Because of nonzero
absorption of both the open and closed forms in the UV spectral
region, simultaneously ring closure and ring opening take place
upon UV irradiation. This leads to an equilibrium situation
determined by the relative quantum yields of the processes and
results in an absorption spectrum that is determined by both
the open-ring and closed-ring conformers. The ratio between
the number of B-DTCP molecules in open-ring and closed-ring
forms can be determined by proton NMR spectroscopy. Then
the steady-state absorption spectrum of the closed form (Figure
2) is obtained by subtracting the open-form absorption spectrum
from the absorption spectrum of the mixture of closed-ring and
open-ring molecules.
In Figure 3, four typical transients are shown for excitation
of the open form at 310 nm, and probing at the wavelengths
cited in the figure. At all probe wavelengths, the dynamics can
be separated into three regimes: First, ultrafast dynamics
(0-0.5 ps) is observed over the whole 360-700 nm spectral
region, then a 4.2 ps decay occurs with varying amplitude
throughout the investigated spectral region, and finally some
slower absorption changes occur on a few tens of picoseconds
time scale. In the following sections, we will refer to these three
regimes in the observed transients as preswitching dynamics,
ring closure, and postswitching dynamics. The arguments for
these assigments will be presented below and more extensively
in the Theory and Discussion (section IV).
It is important to realize that the analysis of the ring-closure
reaction is complicated because of the coexistence of two
different conformations of B-DTCP molecules in the open form.
In one of them the two central CH3 groups point in the same
direction (Cs symmetry), in the other in opposite direction (C2
symmetry). The central cyclopentene ring in principle breaks
these symmetries, but it is useful to discuss the level structure
and dynamics of B-DTCP in terms of the (approximate)
symmetries Cs and C2. Only the conformer with the C2
symmetry can undergo the ring-closure reaction in a conrotatory
fashion, according to the Woodward-Hoffmann rules.11 In
addition, the central methyl groups in the Cs symmetry cause
substantial steric hindrance for the ring-closure reaction to occur.
NMR data19,20 on BTE compounds in solution indicate a close
to 1:1 ratio between the C2 and Cs forms. For most BTE-type
switches, including the one studied here, interconversion
between the two forms by rotation of the sidearms is a fast
process on the NMR time scale at room temperature (milli-
seconds), but a slow process on the time scale of switching.19,20
3.2. Preswitching Dynamics. The experimentally observed
ultrafast part of the pump-probe response is summarized in
the contour plot of Figure 4. In Figures 5 and 6 slices of this
contour plot are shown along the frequency and time axes,
yielding transient absorption spectra at particular delays and
pump-probe transients at particular wavelengths, respectively.
Following the spectra in time, a number of important observa-
tions can be made. The initial photoinduced absorption spectrum
(at a pump-probe delay of 0 fs) is situated in the spectral region
430-480 nm. As shown in the top left frame of Figure 5, this

Bisthienylethene-Based Photochromic Switch
J. Phys. Chem. A, Vol. 106, No. 37, 2002 8501
changes. As the initially excited band rapidly decays, a red-
shifted photoinduced absorption band in the 550-700 nm
spectral region builds up on the same time scale. This precur-
sor-successor relation is modified by a rapid blue spectral shift
of the newly formed band from about 650 to 500 nm, during
which also line broadening and a small drop in intensity occur.
After a few hundred femtoseconds, no spectral changes occur
anymore, as can, for instance, be seen from the right middle
and bottom panels in Figure 5, where it is shown that the 500-
fs and 2.5-ps transient spectra are practically identical. Only an
overall decay of the spectrum on a picosecond time scale is
observed, which, as we will argue below, is related to the process
of ring closure.
To parametrize the dynamics, a simple phenomenological
model was adopted, and the pump-probe transients, such as
those displayed in Figure 6, were fitted globally. According to
the model, the measured signal at a probe frequency ω during
the first few picoseconds (when the postswitching dynamics can
be neglected; see section 3.4) is expressed as
pst -kclt
SSW(ω,t) ) gpre(ω)e-k t + gsuc(ω,t)(1 - e-k )e
+
ps
gcl(ω)(1 - e-k t) (1)
cl
Here, gpre(ω) is the photoinduced absorption spectrum of the
initially excited state (the precursor), gsuc(ω,t) is the photo-
induced absorption spectrum of the state that appears with a
small delay (the successor), and gcl(ω) is the absorption spectrum
of the closed form of the molecule that appears on a picosecond
time scale (vide infra). The precursor goes over to the successor
with rate constant kps, whereas the ring-closure rate constant is
kcl.
The spectra g(ω) can, in principle, be obtained from global
fitting of the data. For the precursor and closed form of the
molecule this is not difficult, because they dominate the
spectrum at extremely short and long delay times, respectively.
For the successor state the uncertainty is larger, although there
are spectral regions (e.g., 550-700 nm) where the induced
absorption is dominated by this state.
Figure 3. Normalized pump-probe transients for B-DTCP in cyclo-
hexane at different probe wavelengths. Probe wavelengths are indicated
in the panels. Areas corresponding to preswitching, switching and
postswitching dynamics are marked by (xxx), (\\\), and (///), respectively.
The spectrum gsuc(ω,t) is explicitly time dependent, due to
the rapid blue shift and line broadening that is observed in
Figures 4 and 5. Both effects were modeled in an exponential
way, by writing for the central frequency ωc ) ω0 + ωbl(1 -
exp(-kblt)) and for the line width ∆ω ) ∆ω0 + ∆ωbl(1 -exp-
(-kblt)). Here ωbl and ∆ωbl are the amount of blue shift and
the line broadening, respectively, and kbl is the rate constant of
both processes. A small drop in intensity of the signal was
modeled with the same time constant kbl. In the fitting procedure,
the kinetic parameters were found to be kps-1 ) 70 ( 15 fs for
the conversion from the initially excited state to the successor
-1
state, kbl ) 150 ( 30 fs for the spectral dynamics of the
-1
successor state, and kcl ) 4.2 ( 0.6 ps for the ring closure.
The model, just described, gives good results on the red side
of the pump-probe spectrum, but yet another dynamical process
has to be added to explain the transients to the blue of 500 nm.
From 410 to 490 nm, a 325 fs decay component was observed,
whereas from 350 to 380 nm a delayed growth of the signal
with the same time scale occurs. This can be modeled by adding
to eq 1 two more terms:
Figure 4. Contour plot representing the subpicosecond dynamics of
B-DTCP in cyclohexane. The change in OD is plotted as a function of
probe wavelength (vertical axis) for pump-probe delays up to 1 ps
(horizontal axis).
spectrum closely resembles the photoinduced spectrum of the
individual side groups of the switch, again confirming that the
different parts of the molecule are only weakly coupled when
the switch is in the open form. Similar observations were made
by Tamai et al.4 and Owrutsky et al.6 in the cases of bithiophene-
and bisthienylethene-based photochromic switches, respectively.
Whereas the photoinduced spectrum of the individual BT-
chromophore side groups does not display any subpicosecond
dynamics, the spectrum of the B-DTCP switch molecules rapidly
pnt
SNONSW(ω,t) ) gpre(ω)e-k t + gnsw(ω)(1 - e-k
)
(2)
pn
We believe this process to be totally unrelated to the switching
from the open form to the closed form of the molecule. One
reason for this conjecture is that in these wavelength regions

8502 J. Phys. Chem. A, Vol. 106, No. 37, 2002
Hania et al.
Figure 5. Transient absorption spectra of B-DTCP in cyclohexane at different delay times between pump and probe pulses. The delay times are
indicated in the panels. The squares represent experimental data, and the solid lines are added as a guide for the eye. The dashed line in the upper
left panel represents the transient absorption spectrum of the BT chromophore.
absorption bands appear that do not display any dynamics at
the time scale of a few picoseconds. Therefore, the ring-closure
rate constant kcl does not appear in eq 2. This indicates that we
are observing transient contributions from nonswitching con-
formers of B-DTCP. As discussed above, when the geometry
of the side chains is such that the central CH3 groups provide
too much steric hindrance (Cs symmetry), ring closure does not
occur.
likely candidate. The relative weight of this contribution to the
observed transients as a function of frequency is displayed in
Figure 7. It is clear that the dynamics occur predominantly from
the successor state, defined above, whereas around 430 nm a
band appears that relaxes much more slowly.
Similar time constants, in the range from 1 to 3 ps, were
reported by Tamai,4 Ern,10 and Owrutsky6 in the case of ring
closure of dithienylperfluorocyclopentene derivatives. The main
difference with these systems is the fluorination of the central
cyclopentene bridge. The difference in time constants indicates
that the fluorination speeds up the C-C bond formation. A
detailed comparison between the dynamics of fluorinated and
nonfluorinated switches will be published elsewhere.21
In Figure 6, representative fitting curves are shown that
reproduce all details of the dynamics and clearly demonstrate
the success of the model. Both the processes of the switching
molecules (eq 1) and those of the nonswitching molecules (eq
2) were included in the global fitting routine. Because the time
scale of the ultrafast dynamics is very close to the time resolution
of the experiments (for details see the experimental chapter),
the response functions were deconvoluted from the experimental
traces by using their cross-correlation function, as measured by
sum frequency generation. The final result is as described
above: both the switching and the nonswitching conformers
The absorption band at 430 nm, which does not display
dynamics at a few picosecond time scale, suggests that there
are species present that do not participate in the switching
process. As discussed before, this is very likely due to molecules
in the open-ring form with a conformation that prevents ring
closure.1,19,20 As discussed above, the photoexcitation of these
molecules leads to rapid dynamics that is observable in the blue
part of the pump-probe spectrum, until a state is reached that
apparently has it absorption maximum at 430 nm.
-1
display dynamics at ultrafast time scales (kps-1 ) 70 fs and kbl
-1
) 150 fs for the switching ones and kpn ) 325 fs for the
nonswitching ones). However, on the time scale of a few
picoseconds, only the switching conformers show further
spectral changes. These are related to the ring-closure process
At the probe wavelength of 700 nm, where only the successor
state absorbs, the pump-probe signal relaxes to zero in a single-
exponential way with the time constant of 4.2 ps. Therefore, it
can be concluded that the switching efficiency of the conformers
that are able to do so is very close to 1. This agrees with the
overall value of the quantum yield of ring closure of 0.59 (see
previous section), when the ground-state equilibrium ratio
between the switching and nonswitching conformations of the
open form of the B-DTCP molecules is about 3:2.
-1
(kcl ) 4.2 ps).
3.3. Ring Closure. The ultrafast transients, reported above,
are too fast to be associated with substantial movements of parts
of the molecule. It is therefore unlikely that they reflect the
time scale of the ring closure, which involves movements in
both the central ring of the molecule and in the side chains.
Instead, the time constant of τcl ) 4.2 ps that is present at all
probe wavelengths, but with a varying amplitude, is a more

Bisthienylethene-Based Photochromic Switch
J. Phys. Chem. A, Vol. 106, No. 37, 2002 8503
Figure 6. Pump-probe transients of B-DTCP in cyclohexane after excitation of the open form at 310 nm. The probe wavelength is shown in each
panel. Dots represent experimental data, the solid lines are fits according to the model presented by eqs 1 and 2. An indication of the time resolution
is given by dashed lines.
Figure 7. Transient spectra of B-DTCP recorded at delay times of 3
Figure 8. Transient spectra at a delay of 10 ps (bold line) and 100 ps
(regular line). The spectra are normalized in the spectral region 500-
550 nm. The dashed and dotted arrows indicate the decay of the band
centered around 430 nm and the matching red shift of the band at 530
nm, respectively. The solid arrow points to the spectral narrowing as
a result of vibrational cooling. The steady-state absorption spectrum
of the closed form of B-DTCP (1B) (dashed line) is added for
comparison.
(solid line) and 9 (dashed line) ps. The circles with error bars show
the relative contribution of the 4.2-ps decay process in the observed
transients.
3.4. Postswitching Dynamics. Figure 8 displays photo-
induced absorption spectra in the time window 10-100 ps. To
help visualize the spectral dynamics, the spectra are normalized
in the region 500-550 nm. The main features are a decay of
the absorption band at 430 nm, a narrowing on the red side of
the spectrum (550-620 nm), and an apparent red shift of the
band at about 520 nm. Although these effects are due to different
phenomena, they occur on similar time scales. At the longest
recorded pump probe delay of 100 ps, the shape of the
photoinduced absorption spectrum begins to approach that of
the ground-state absorption of the closed form.
The band, centered at 430 nm and attributed to the photo-
induced absorption of the nonswitchable conformers, has
partially decayed but is still clearly recognizable at a delay of
100 ps. This implies that the lifetime of the excited state of the

8504 J. Phys. Chem. A, Vol. 106, No. 37, 2002
Hania et al.
nonswitchable conformers is in the order of a few hundred
picoseconds.
The narrowing on the red side of the spectrum is indicative
of vibrational cooling, which apparently also is not completed,
yet, at a delay of 100 ps. The switching process involves the
radiationless transition of the successor state of the open form
to the ground state of the closed form (see section IV).
Consequently, there is a substantial amount of excess vibrational
energy in the hot ground state of the closed form, directly after
the switching. The decay constant, associated with the cooling
process, is about 50 ps, when inferred from the data of Figure
8. This relatively low speed of cooling might originate from
the large amount of vibrational energy that needs to be
dissipated. In this case the cooling process becomes associated
not only with the excitation of rotational modes of solvent
molecules in the closest surrounding, but also with the distribu-
tion of the energy between the solvent molecules via transla-
tional motion.
Finally, the apparent shift of the maximum of the absorption
band at 520 nm is due to the combination of both effects, just
discussed. Both the decay of the 430 nm band of the nonswit-
chable conformers and the narrowing of the spectrum due to
vibrational cooling of the hot ground state of the closed form
molecules contribute to this effect.
3.5. Orientational Dynamics. More information about the
ring-closure process and the pre- and postswitching behavior
can be obtained from polarization sensitive measurements. In
particular the dynamics of the photoinduced anisotropy can
reflect changes in the orientation of the probed transition
moments, for instance, due to conformational changes in the
molecule when the switching takes place. The photoinduced
anisotropy is defined as follows:
Figure 9. Early-time dynamics of the photoinduced anisotropy at
different probe wavelengths (indicated in the panels). Dots represent
experimental data; the solid lines are two-exponential fits with time
constants of 70 and 325 fs. An indication of the time resolution in
each case is given by the dashed lines.
∆OD|(t) - ∆OD (t)
r(t) )
(3)
∆OD|(t) + 2∆OD (t)
Here, ∆OD|(t) and ∆OD (t) are the optical density changes
measured with parallel and orthogonal polarizations of the pump
and probe beams, respectively. The time behavior of the
photoinduced anisotropy r(t) is not sensitive to the dynamics
of population of the states during the dynamical processes but
only traces the changes in the orientation of the transition dipole
moment. In a liquid, where the molecules are randomly oriented
and rapidly moving, the anisotropy can also be written as
B-DTCP displays ultrafast dynamics over the entire probe
spectrum. Determination of the initial value of the photoinduced
anisotropy is difficult because of the limited time resolution of
these experiments and also because of possible coherent
coupling effects that may take place during the first tens of
femtoseconds, within the dephasing times of the optical transi-
tions involved.25 Nevertheless, it is clear that in the spectral
region 460-510 nm the initial value of the photoinduced
anisotropy is positive and close to the theoretical maximum of
0.4, whereas on the blue side it is negative and possibly close
to its theoretical minimum of -0.2. Apparently, two optical
transitions are probed that are polarized perpendicular to each
other.
The kinetic parameters, found for the orientational dynamics
on femtosecond time scales, agree with those of the pump-
probe transients discussed above. To the red of 500 nm only a
70 fs component is present, whereas to the blue both 70 and
325 fs transients are observed. The fits support the notion that
these time constants are related to the kinetics of two different
species.
3 cos2 θ(t) - 1
r(t) ) r(0)
(4)
2
In eq 4, r(0) is the initially photoinduced anisotropy and θ(t) is
the angle between the orientations of the transition dipole
moments of the excitation at time t ) 0 and of the probed
transition at time t. When a single transition is probed in a
sample of randomly oriented molecules, such as in isotropic
solutions, the initial value of the photoinduced anisotropy r(0)
can range from +0.4 to -0.2, depending on whether the
transition dipole moments involved in the excitation and probing
are parallel (+0.4) or perpendicular (-0.2) in the molecular
frame. When more than one transition is probed (or excited), a
summation over all possible combinations of transitions is
needed for evaluation of the total anisotropy.
In Figure 10, the photoinduced anisotropy is followed on a
picosecond time scale. Over most of the spectral range the
anisotropy rises exponentially with a time constant of 8 ps,
before relaxing via rotational diffusion with a time constant of
about 100 ps. In the upper left panel it is shown that around
410 nm, where no switching dynamics was observed in the case
of the pump-probe experiments (see Figure 7), the dynamical
process with time constant 8 ps is absent. This indicates that
The photoinduced anisotropy relaxes to zero both through
electronic and nuclear rearrangements that alter the direction
of the transition dipole moment, as well as through rotational
diffusion.22-24 In Figure 9 it is shown that the anisotropy of

Bisthienylethene-Based Photochromic Switch
J. Phys. Chem. A, Vol. 106, No. 37, 2002 8505
Figure 11. Potential energy curves for the lowest electronic energy
levels, calculated for the B-DTCP molecule in C2 symmetry. A and B
denote the two possible different symmetries. For a definition of the
reaction coordinate see text.
studied in detail. Very fast femtosecond kinetics, including rapid
rotations of transition moments, is followed by picosecond ring
closure, which, in turn, leads to inertial forced conformational
dynamics and vibrational cooling on much longer time scales.
The femtosecond processes are so fast that nuclear displacements
during that time can only occur over a very limited excursion
range. In particular, the very fast changes in the induced
anisotropy indicate that transitions between electronic states are
involved, rather than movements of part of the molecule.
To interpret these experimental findings, we calculated the
geometries, energy levels, and excitation spectra using density
functional theory and time-dependent density functional theory.
For this purpose, the Amsterdam Density Functional Program
Package (ADF)26-28 was used, which is able to provide accurate
solutions of the Kohn-Sham (KS) equations even for fairly
large systems. For the ground-state calculations we used the
local density approximation, based on the parametrization of
the electron gas data given by Vosko, Wilk, and Nusair.29 The
basis sets used were of triple ú plus polarization Slater type
function (STO) quality (basis IV in ADF).
Excited states were calculated using time-dependent density
functional theory (TDDFT), as implemented in the RESPONSE
part of ADF. Because TDDFT describes (in principle, exactly)
how the electron density changes in time under the influence
of a time-dependent perturbation, and because this time-
dependent density will resonate at the excitation energies of
the system, linear response theory based on TDDFT is able to
provide both these excitation energies as well as the corre-
sponding oscillator strengths. As discussed before, we will use
the nomenclature of C2 symmetry for the switching conformer
where the central CH3 groups point in opposite directions, and
Cs symmetry is assumed for the nonswitching conformers with
the central CH3 groups pointing toward each other.
In Figure 11, the lowest electronic energy levels are depicted
as a function of a reaction coordinate. The reaction coordinate
is defined as the distance between the two C atoms in the central
ring, which become bonded in the closed form. All other
coordinates are optimized for the lowest ground-state energy
at each position along this reaction coordinate. Note that this
path conserves the original C2 symmetry. Clearly, a high barrier
separates the open and closed form in the ground state, which
explains why both are thermally stable and can only be switched
into each other by excited-state pathways.
Figure 10. Transients of the photoinduced anisotropy, recorded at
different probe wavelengths in the time interval 0-100 ps (probe
wavelengths are indicated in the panels). Experimental results are given
by the points; the solid lines are two-exponential fits with time constants
of 8 and 100 ps.
the 8 ps rise at the other wavelengths is related to the ring
closure.
We tentatively attribute the 8 ps component in the transients
of the photoinduced anisotropy to the motion of the thienyl
wings of the B-DTCP molecule toward coplanarity as a result
of the ring closure. The lower limit for the time constant of
this “forced” movement towards coplanarity is of course the
switching time itself, i.e., 4.2 ps. However, due to the moment
of inertia of these thienyl side groups, some larger time constant
may be expected. An indication for the size of this effect may
be obtained from the reorientation times of the individual BT
chromophores, which we measured to be τor1 ) 2 ps and τor2
)
18 ps. The combination of torque induced by the ring closure
and moment of inertia of the side groups may very well lead to
the observed value of 8 ps in the orientational dynamics.
In the open form of the molecule, the low-lying electronically
excited states come in almost degenerate pairs, each containing
one state of the two possible symmetries A and B. The first
two excited states arise from charge transfer like excitations
IV. Theory and Discussion
The experimental results of the previous section indicate that
the switching dynamics of B-DTCP is quite complicated, when

8506 J. Phys. Chem. A, Vol. 106, No. 37, 2002
Hania et al.
from the central bond HOMO π-orbital to the LUMO’s on the
chromophore side chains. Because the two chromophore
π-systems have very little interaction in the open form (they
are almost orthogonal to each other), the two chromophore
LUMO’s combine to one orbital of a symmetry and one orbital
of b symmetry, leading to the two nearly degenerate excited
states 2A and 1B. Both transitions from the ground state have
very low oscillator strengths, because the HOMO and the
LUMO’s are located in different parts of the molecule. Note
that the corresponding transition is, of course, absent in the
isolated chromophores, because the central bond is then not
present.
The second pair of excited states are of an entirely different
character. They turn out to correspond to the HOMO-LUMO
transitions in the chromophores themselves. Again, both transi-
tions combine to A and B states that are almost degenerate,
due to the small interaction between the chromophore π-systems.
In this case we calculate the 2B state to have considerable
oscillator strength, whereas the 3A state is essentially dark. This
is easily understood when one realizes that 2B corresponds to
an excited state that has the approximate form |L f L* +
|R f R* , where L and L* are the left chromophore HOMO
and LUMO orbitals and R and R* are the corresponding orbitals
on the right side of the molecule. For this state, the monomer
transition dipole moments interfere constructively, leading to
large oscillator strength, whereas for the 3A state with character
|L f L* - |R f R* , the monomer transition moments
interfere destructively. The lowest energy optically allowed
excitation of the open form of B-DTCP therefore is the 1A f
2B transition. Due to the weak coupling between the left and
right part of the molecule, this transition closely resembles the
first excitation of the isolated chromophores. The pump-probe
spectrum at zero delay therefore is essentially the same as in
the isolated chromophores, in accordance with the experimental
findings above (see Figure 5).
Figure 12. Potential energy curves for the 1A and 1B electronic states,
but optimized for the 1B excited state instead of the ground state (as
in Figure 11).
contain the same barrier. It is therefore this state that gives rise
to the actual ring closure. As we will argue now, it is the
relaxation of the 2B/3A states to the 1B state that is the rate
determining step in going from the open to the closed form.
From Figure 11, it is clear how relaxation may occur: the 2B
and 2A states cross a little way along the reaction coordinate.
When the C2 symmetry is broken, the crossing will in fact be
avoided. This means that there will be a conical intersection
between the two surfaces, leading to a feasible relaxation
mechanism.30 We believe that this process gives rise to the 4.2
ps transients, observed in the pump-probe data. The subsequent
mixing of the near degenerate states 2A and 1B is supposed to
be a fast process again, and also the downhill motion from the
open to the closed form in the state 1B will be very rapid. This
model is supported by the fact that the anisotropy in the
wavelength range 460-700 nm after a femtosecond decay
recovers again on a picosecond time scale: a B state is populated
again.
The final step in the switching process, i.e., the motion along
the 1B potential to the closed form, is actually not depicted
correctly in Figure 11. The energy levels are plotted along the
minimum energy reaction path in the ground state, which is
not the minimum (electronic) energy path in the 1B excited state.
For the excited-state minimum energy path, the calculations give
the potential energy curves shown in Figure 12. The energy
barrier in the ground state has of course increased, whereas the
1B excited state has come down in energy. Because the motion
along the reaction coordinate is rather fast, it is possible that
the system in the 1B state does not have enough time to find
its minimum energy configuration everywhere along this curve.
In that case the potential energy surface will be somewhat
different.
The rapid fluctuations in the immediate environment of the
switch lead to symmetry breaking and therefore coupling of
the A and B states. Hence, the originally excited 2B state relaxes
to a thermal equilibrium with the almost degenerate 3A state.
The important point is that this relaxation process does not
involve any change of geometry of the switch and therefore it
can be expected to proceed extremely rapidly. We therefore
identify it with the fastest time experimentally observed, i.e.,
the 70 fs precursor-successor sequence. The 150 fs spectral
dynamics that follows is due to motion on the new potential
energy surface. The change of character from 2B to 3A is
experimentally confirmed not only by the rapid changes in the
pump-probe spectra, but also by the fast loss of anisotropy,
which is indicative of a rapid change in the direction of the
probed dipole moment.
The fact that the two states in Figure 12 cross at two points
means that at slightly deformed nonsymmetric geometries there
will be avoided crossings; hence two conical intersections
appear. These conical intersections provide a mechanism for
the relaxation of the excited 1B state to the 1A ground state
and, though the left crossing in the graph would lead back to
the original open form ground state, the right crossing point
leads to the ground state of the closed switch, thus completing
the switching process. The large angle of the two potential
energy curves at the first crossing point explains why hardly
any back-transfer to the ground state of the open form
occurs.31,32
For the nonswitching molecules a similar coupling scheme
holds. Due to the different geometry, the relative position of
the 2B and 3A states is somewhat different and also the coupling
between these two states might be slightly modified. As a result,
the decay occurs with a time constant of 325 fs, instead of 70
fs. This was observed both in the pump-probe and the transient
anisotropy measurements. The major difference occurs at later
times, however. Where one conformer displays ring closure,
the other cannot perform this process due to steric hindrance of
the two central CH3 groups, which prevent the relevant C atoms
from approaching each other close enough.
V. Summary and Conclusions
According to Figure 11, both the 2B and 3A states of the
switching molecules exhibit a high barrier along the reaction
path. However, there is a lower lying state, 1B, that does not
The switching behavior of B-DTCP, which is based on ring
closing, was studied in detail with femtosecond nonlinear optical

Bisthienylethene-Based Photochromic Switch
J. Phys. Chem. A, Vol. 106, No. 37, 2002 8507
(4) Tamai, N.; Saika, T.; Shimidzu, T.; Irie, M. J. Phys. Chem. 1996,
100, 4689.
(5) Ohtaka, N.; Hase, Y.; Uchida, K.; Irie, M.; Tamai, N. Mol. Cryst.
Liq. Cryst. 2000, 344, 83.
(6) 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.
experiments and density functional calculations. The overall
dynamics can be separated into preswitching dynamics, the ring
closure proper, and postswitching dynamics. After optical
excitation, the system propagates through a rather complicated
sequence of electronical states. The coupling between these
states, governed by geometrical distortions, determines the
reaction rate of the overall process. After the ring closure, forced
geometrical changes and vibrational cooling determine the
dynamics of the optical response.
The overall picture of the ring-closure reaction is as follows:
first, the initially excited state rapidly, with a time constant of
70 fs, equilibrates with an electronic state of different symmetry
(in terms of the energy levels of Figure 11: 1A + hν f 2B T
3A). Second, population relaxation takes place with a time
constant of 4.2 ps, to a state that experiences energy lowering
when ring closure occurs (2B f 2A f 1B). Third, the formation
of the central C-C bond induces rotation of the side groups of
the molecule to a nearly coplanar conformation, with a time
constant of 8 ps (1B f 1A*). Fourth, the vibrationally hot
ground state of the closed conformer relaxes to thermal
equilibrium with a time constant of about 50 ps (1A* f1A).
This completes the switching process.
(7) Ern, J.; Bens, A. T.; Bock, A.; Martin, H.-D.; Kryschi, C. J. Lumin.
1998, 76&77, 90.
(8) Ern, J.; Bens, A. T.; Martin, H.-D.; Mukamel, S.; Schmid, D.;
Tretiak, S.; Tsiper, E.; Kryschi, C. Chem. Phys. 1999, 246, 115.
(9) Miyasaka, H.; Murakami, M.; Itaya, A.; Guillaumont, D.; Naka-
mura, S.; Irie, M. J. Am. Chem. Soc. 2001, 123, 753.
(10) Ern, J.; Bens, A.; Martin, H.-D.; Mukamel, S.; Schmid, D.; Tretiak,
S.; Tsiper, E.; Kryschi, C. J. Lumin. 2000, 87-89, 742.
(11) Woodward, R. B.; Hoffman, R. The conserVation of orbital
symmetry; Verlag Chemie/Academic Press: Weinheim, 1970.
(12) Nakamura, S.; Irie, M. J. Org. Chem. 1988, 53, 6136.
(13) Ern, J.; Bens, A. T.; Martin, H.-D.; Mukamel, S.; Tretiak, S.;
Tsyganenko, K.; Kuldova, K.; Trommsdorff, H. P.; Kryschi, C. J. Phys.
Chem. A 2001, 105, 1741.
(14) Bens, A. T.; Ern, J.; Kuldova, K.; Trommsdorff, H. P.; Kryschi,
C. J. Lumin. 2001, 94-95, 51.
(15) Lucas, L. N.; Esch, J. v.; Feringa, B. L.; Kellogg, R. M. Chem.
Commun. 1998, 2313.
(16) Tsivgoulis, G. M.; Lehn, J.-M. Chem. Eur. J. 1996, 2, 1399.
(17) Wilhelm, T.; Piel, J.; Riedle, E. Opt. Lett. 1997, 22, 1494.
(18) Reuther, A.; Laubereau, A.; Nikogosyan. Opt. Commun. 1997, 141,
180.
(19) Uchida, K.; Nakayama, Y.; Irie, M. Bull. Chem. Soc. Jpn. 1990,
63, 1311.
The combination of time-resolved optical experiments with
theoretical analysis of the electronic structure allows us to obtain
this detailed picture of both the population and the conforma-
tional dynamics of the switching process. We expect that the
insight, thus obtained, will provide important clues for the design
of new switches with characteristics that can be predicted from
the theory with a fair level of confidence.
(20) Irie, M.; Sakemura, K.; Okinaka, M.; Uchida, K. J. Org. Chem.
1995, 60, 8305.
(21) Hania, P. R.; Lucas, L. N.; Pugzlys, A.; Esch, J. v.; Feringa, B. L.;
Duppen, K. To be published.
(22) Lakowicz, J. R. Principles of fluorescence spectroscopy, 2nd ed.;
Kluwer Academic/Plenum Publishers: New York, 1999.
(23) Fleming, G. R. Chemical applications of ultrafast spectroscopy;
Oxford University Press: New York, 1986; Vol. 13.
(24) Waldeck, D.; Jr., A. J. C.; McDonald, D. B.; Fleming, G. R. J.
Chem. Phys. 1981, 74, 3381.
(25) Wynne, K.; Hochstrasser, R. M. Chem. Phys. 1993, 171, 179.
(26) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41.
(27) Guerra, C. F.; Visser, O.; Snijders, J. G.; Velde, G. t.; Baerends,
E. J. Methods and Techniques in Computational Chemistry; STEF: Cagliari,
Italy, 1995.
Acknowledgment. This work was supported by the “Ned-
erlandse Organisatie voor Wetenschappelijk onderzoek” (NWO).
R.T. is grateful for the support by EU FULPROP PROGRAM
“FMRX-CT97-0155”. The Royal Netherlands Academy of
Sciences is gratefully acknowledged for a fellowship for J.v.E..
(28) Velde, G. t.; Baerends, E. J. J. Comput. Phys. 1992, 99, 84.
(29) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
(30) Turro, N. J. Modern molecular photochemistry; University Science
Books: Mill Valley, CA, 1991.
(31) Landau, L. Phys. Z. USSR 1932, 46, 2.
(32) Zener, C. Proc. R. Soc. London A 1932, 137, 696.
References and Notes
(1) Irie, M. Chem. ReV. 2000, 100, 1685.
(2) Tamai, N.; Miyasaka, H. Chem. ReV. 2000, 100, 1875.
(3) Miyasaka, H.; Araki, S.; Tabata, A.; Nobuto, T.; Mataga, N.; Irie,
M. Chem. Phys. Lett. 1994, 230, 249.