Dynamics of the cyclization reaction in photochromic furyl fulgides

Article abstract of DOI:10.1021/jp961905v

The dynamics of the photoinduced cyclization reaction of furyl fulgides has been studied applying femtosecond time-resolved transient absorption spectroscopy. For furyl fulgide 1 and adamantylidene-substituted furyl fulgide 2 in different solvents, a scheme of the reaction mechanism is set up and the reaction rates are determined by fitting individual transients. Two parallel reaction pathways can be distinguished, one of them including an intermediate state. Reaction times in the picosecond range support the application of fulgides in optical data storage media and as molecular photoswitches.

Full text of DOI:10.1021/jp961905v

502
J. Phys. Chem. A 1997, 101, 502-506
Dynamics of the Cyclization Reaction in Photochromic Furyl Fulgides
Martin Handschuh, Martin Seibold, Helmut Port,* and Hans Christoph Wolf
3. Physikalisches Institut, UniVersita¨t Stuttgart, D-70550 Stuttgart, Germany
ReceiVed: June 27, 1996X
The dynamics of the photoinduced cyclization reaction of furyl fulgides has been studied applying femtosecond
time-resolved transient absorption spectroscopy. For furyl fulgide 1 and adamantylidene-substituted furyl
fulgide 2 in different solvents, a scheme of the reaction mechanism is set up and the reaction rates are determined
by fitting individual transients. Two parallel reaction pathways can be distinguished, one of them including
an intermediate state. Reaction times in the picosecond range support the application of fulgides in optical
data storage media and as molecular photoswitches.
Introduction
SCHEME 1: Molecular Structures and Photochromic
Reactions E h C of Fulgides 1 and 2; Absorption
Spectra of 2E and 2C in Toluene (295 K)
Fulgides are photochromic compounds suitable for use in
optical data storage media1-3 and as molecular functional units
such as optical switches.4 Both fields of application require
optical bistability, which is provided by several heteroaromatic
fulgides and fulgimides. Irradiation with UV or visible light
induces electrocyclic ring-closure/ring-opening reactions, leading
to the production of closed C or open E isomers according to
Scheme 1. Generation of the so-called Z isomer by UV
irradiation of E is neglected in the context of this paper, as it
can be minimized by proper substitution,5,6 and does not affect
the processes discussed in this paper.
The good photochromic performance of fulgides is based on
a combination of the following properties: thermal stability of
both isomers at room temperature;7,8 full reversibility of the
photoreactions without photochemical fatigue;9-11 high conver-
sion efficiencies due to both clearly separated absorption bands
with high extinction coefficients12-14 and high reaction quantum
yields;15,16 and minor environmental effects in liquid solution
and rigid polymer matrix.17,18 Different isomers can be distin-
guished by means of UV/vis absorption as well as IR spectros-
copy, which has been proved to be a totally nondestructive
method for detecting the isomerization states.19
model for the reaction mechanism and to give quantitative results
for the time constants involved.
Experimental Section
Time-resolved transient absorption measurements were per-
formed with a pump and probe technique. Pulses are generated
using a passively mode-locked Ti:sapphire laser oscillator
(Coherent Mira900) pumped by an Ar2+ ion laser (Coherent
Innova310). A Q-switched, intracavity frequency-doubled Nd:
YLF laser (Quantronix 4820) pumps a regenerative Ti:sapphire
amplifier (Quantronix 4880) which, after recompression of the
previously stretched pulses, supplies 200-fs pulses at 784 nm
with a repetition rate of 1 kHz. After splitting this beam in a
20:80 ratio, the lower intensity beam is frequency-doubled in a
LBO crystal and used as the pump beam for pulse excitation of
the sample at 392 nm (25 500 cm-1; 150 cm-1 FWHM; pulse
energy 5 µJ). The 80% intensity beam runs through a delay
line before being focused on a rotating quartz plate to generate
a white light continuum between 10 000 and 24 000 cm-1. The
white light is then split into a probe beam, which is focused
into the sample such as to achieve maximum overlap with the
pump beam, and a reference beam, which is focused into a
sample spot not excited by the pump beam. Both transmitted
beams are analyzed by a spectrograph and detected by a double-
diode array (Hamamatsu S4801-512Q).
Although being of crucial importance for the feasibility of
optical switches and memories, little is known about the reaction
dynamics and the reaction mechanisms of fulgides. Several
studies reporting the transient absorption of fulgides after pulse
excitation of the respective E isomer suffer from insufficient
time resolution. Hence, they can only give upper limits for the
E f C reaction times: 20 ps for 1E in toluene20 and 10 ps for
a furyl fulgide in polymer matrix21 and 6 ps for a phenyl-
substituted fulgide in toluene.22 As for similar photochromic
molecules, 1.1 ps was recently reported for the ring-closure
reaction of a diarylethene compound.23 Following Woodward-
Hoffmann’s rules,24 the E f C isomerization of fulgides is often
described as a conrotatory ring-closure reaction. However,
experimental data on the relaxation pathway via the ground-
and excited-state potential surfaces are missing. Only qualitative
models have been proposed for phenyl fulgides22 and furyl
fulgides,21 implying barrierless relaxation in the excited E state.
Here we present transient absorption measurements with
femtosecond time resolution of the E f C isomerization of
fulgides 1 and 2 (cf. Scheme 1) which allow us to set up a
The system provides a response time of 300 fs with delay
times up to 900 ps. Due to temporal dispersion of the white
X Abstract published in AdVance ACS Abstracts, December 15, 1996.
S1089-5639(96)01905-6 CCC: $14.00 © 1997 American Chemical Society

Cyclization Reaction in Photochromic Furyl Fulgides
J. Phys. Chem. A, Vol. 101, No. 4, 1997 503
Figure 2. Absorption spectra of 2, 1.5 ps (s) and 900 ps (- - -) after
pulse excitation of 2E and cw absorption of 2C (- -), in toluene (295
K).
Figure 1. Transient absorption of 2 after pulse excitation of 2E in
acetonitrile (νj ) 25 500 cm-1; T ) 295 K); ordinate values T/T*:
relative transmission intensities with (T*) and without (T) excitation.
Linear spectral shift of band A with increasing delay time due to
temporal white light dispersion of probe pulse (chirp). See text for
explanation of bands A and B.
light (chirp), blue spectral components of the probe beam arrive
at the sample later than red components, causing the absorption
signal at higher energies to appear earlier than at lower energies.
Although this effect can numerically be eliminated from the
results, we present here uncorrected spectra to give an impres-
sion of the original data.
Compounds 1E and 2E were purchased from ABERCHRO-
MICS LTD., Cardiff, U.K., and, without further purification,
dissolved in toluene or acetonitrile (Merck Uvasol grade). For
each pulse, a fresh sample portion was supplied by pumping
the solution through a 1-mm flow cell. In order to avoid
accumulation of colored C isomer in the solution, a 500-mL
reservoir was bleached with frequency-doubled Nd:YLF leakage
at 527 nm.
For time-resolved fluorescence measurements, a time-cor-
related single-photon-counting system was used with a frequency-
doubled, mode-locked, Nd:YAG laser (Quantronix 416) for
pulse excitation at 532 nm; a time resolution of 30 ps was
obtained.
Figure 3. Time evolution of the transient absorption at νj ) 19 200
cm-1 after pulse excitation of 2E in toluene (295 K): experimental
data (b) and fitted model curve (s); notice stretched abscissa in insert.
approximately 1 ps and approaching a constant maximum value
at 200 ps. According to Figure 1, the first intensity peak can
be attributed to absorption band A, whereas for long delay times
the absorption signal is due to band B. Taking the minimum
intensity at 1 ps as 100%, the overall signal recovery is 100%
for 2 in toluene; for 1 in acetonitrile and toluene, the corre-
sponding values are 30% and 40%. At very long delay times
of several hundred picoseconds, a slight decrease of absorption
intensity was observed for 1 and 2 in toluene; however, since
we did not find uniform behavior and the effect was not
significant at given signal-to-noise ratio, it is not considered in
the interpretation of the experimental results.
Results
Immediately after pulse excitation of 2E at 25 500 cm-1 (392
nm), a strong, rather broad absorption band A rises as shown
in Figure 1, where the spectral shift of the signal toward lower
wavenumbers, proceeding linearly with delay time, is due to
white light dispersion (chirp) of the probe pulse. After fast
decay of absorption A within 600 fs, the growth of a narrower
band B centered at 17 700 cm-1 becomes observable. The
intensity of B increases on a 100-ps time scale, while its spectral
shape remains unchanged after 1.5 ps (Figure 2). By compari-
son with the cw absorption of 2C, band B can be identified as
the C0 ground-state absorption, thus indicating the production
of C isomer. The qualitative features described here are
common for all samples investigated. The rise times of
absorption band B are 60 and 70 ps for 1 in acetonitrile and
toluene, respectively, and 200 ps for 2 in toluene.
Model Description
In order to carry out a quantitative analysis of the experi-
mental results, a model for the mechanism of an optically
induced reaction E f C between two thermally stable isomers
is set up (Figure 4). Since the model is defined as the general
excited state diagram, no assumptions on the specific reaction
mechanism are implied. In particular, only as many states are
included as are necessary to describe the time evolution of the
absorption signal as shown in Figure 3. Thus, it is sufficient
to consider the ground and excited states E0 and E* of the E
isomer and C0 and C* of the C isomer as well as an intermediate
For 2 in toluene, the time evolution of the absorption intensity
at 19 200 cm-1 can be seen in Figure 3. After an initial increase
within the response time of the system, the signal decays with
a time constant of 600 fs before slowly increasing after

504 J. Phys. Chem. A, Vol. 101, No. 4, 1997
Handschuh et al.
k3
0k - kh
NI(t) ) N
(e-kht - e-kt
)
(7)
and
NC (t) )
0
k2
k3k4
1
1
N0 (1 - e-kt) +
- (1 - e-kt) + (1 - e-kht
)
(8)
{
}
[
]
k
k - kh
k
kh
!
It follows from (8) and N (∞)
φ
ECN0, with φEC being the E
)
C0
f C reaction quantum yield, that
k2kh + k3k4
φEC
)
(9)
kkh
If only states E*, I, and C0 contribute to the white light
absorption, the optical density OD of a sample at wavenumber
νj at time t is given by
ODνj(t) ) R[NE*(t)ꢀE*,νj + NI(t)ꢀI,νj + NC (t)ꢀC ,νj
] (10)
0
0
where R is a factor representing thickness and molar concentra-
tion of the sample and ꢀS,νj are the molar extinction coefficients
of states S at wavenumber νj. If we set φCE ) k5/kh and define
Figure 4. Model diagram for the E f C reaction mechanism of furyl
fulgides 1 and 2.
ꢀE*,νj
ꢀI,νj
ꢀ1 ≡
ꢀ2 ≡
(11)
state I on the reaction pathway E* f C0 whose nature needs
not to be specified; nevertheless, I is indispensable for explaining
why the rise time of B differs from the decay time of A.
All possible transitions between participating states are taken
into account. Starting from E0, pulse excitation creates popula-
tion of E* which relaxes radiatively and/or nonradiatively to
the ground state E0 (rate constant k1); alternatively, transitions
from E* to C0 with rate constant k2 and to I with k3 take place.
From I relaxation both to C0 (k4) and to E0 (k5) is possible.
Because of the thermal stability of the E and C isomers, no
transistions between the E0 and C0 ground states are considered.
While stimulated emission is neglected, all states involved can,
in principle, contribute to the white light absorption of the probe
pulse.
ꢀ
ꢀ
C0,νj
C0,νj
inserting (4), (7), (8), and (9) in (10) yields
k3
k3(1 - φCE)
ODνj(t) ) Rνj ꢀ1 - ꢀ2
+
- φEC e-kt
+
(
[
)
k - kh
k - kh
k3
(ꢀ2 + φCE - 1)
e
-kht + φEC (12)
]
k - kh
(12) serves as model expression for the transient absorption of
the sample after pulse excitation at t ) 0. After convolution
with the system response function of the setup, the model
function can numerically be fitted to the experimental data, i.e.,
the temporal evolution of the transient absorption at a cer-
tain wavenumber νj, thus providing a set of parameters
{Rνj,k,kh,k3,ꢀ1,νj,ꢀ2,νj} from which ki and ꢀS,νj can be obtained.
With the model set up according to Figure 4, the reaction
kinetics can completely be described by a set of differential
equations for the time derivatives of the populations NE*(t), NI(t),
and NC (t) of states E*, I, and C0 at time t after excitation:
0
Discussion
N˙ E*(t) ) -(k1 + k2 + k3)NE*(t)
N˙ I(t) ) k3NE*(t) - (k4 + k5)NI(t)
(1)
(2)
(3)
Identification of the absorption signals observed in Figure 1
is facilitated by the lack of E* r E0 absorption in the visible
range. Because of its quasi-instantaneous appearance, absorp-
tion band A must be attributed to the transient absorption of
the E isomer, i.e., an excitation from E* to higher excited states
by the white light probe pulse. Upon closer inspection, band
A turns out to be a superposition of a slowly increasing
contribution of band B as well as two separate absorption bands
centered at 15 000 and 19 000 cm-1 and with slightly different
rise times. The nature of these absorptions can not yet be
explained satisfactorily; however, vibrational relaxation in the
E* state seems to play a role. Changing the solvent polarity µ
affects the shape of band A; for both 1 and 2 in polar acetonitrile
(µ ) 3.4 D), the contributions of the absorptions at 15 000 and
19 000 cm-1 are about equal, leading to the plateaulike shape
of A, whereas in less polar toluene (µ ) 0.35 D), the absorption
at higher wavenumber is significantly stronger, causing the shape
of A to become asymmetric. Hence, the excited-state geometry
of E is assumed to be polar.
N˙ C (t) ) k2NE*(t) + k4NI(t)
0
(1) directly yields
NE*(t) ) N0e-kt
(4)
with N0 ≡ NE*(0) being the initial population of E* after pulse
excitation and
k ≡ k1 + k2 + k3
(5)
If we further define
kh ≡ k4 + k5
(6)
!
!
and consider the boundary conditions NI(∞) ) 0 and NI(0) ) 0,
we obtain
No significant absorption of the intermediate state I can be
found. Absorption band B can be identified as the C0 ground-

Cyclization Reaction in Photochromic Furyl Fulgides
J. Phys. Chem. A, Vol. 101, No. 4, 1997 505
TABLE 1: Model Parameters Obtained for Reaction Rates
ki (ps-1), Time Constants τi (ps), and Relative Extinction
Coefficients Ej,νj of 1 and 2 in Toluene (tol) and Acetonitrile
(acn) at Maxima of Band Ba
via P2 on a 100-ps time scale. We have no evidence for the
nature of I nor for its energetic position relative to C*; it seems
to be very unlikely that I is an excited C state C* because upon
excitation of E, no fluorescence of C can be detected. In
addition, the time constant of the fluorescence decay of 2C upon
excitation at 18 800 cm-1 is larger than 100 ps and therefore
much larger than the time constant for the relaxation of I.
Switching Times. The photoisomerization E h C of fulgides
can be used for optical data storage and imaging; it can also be
regarded as a reversible optically induced switching process.
For the photoinduced cyclization E f C examined, two
“switching times” can be defined on the basis of our time-
resolved measurements:
(1) The first switching time is the time delay between the
excitation and the detection of C0 population; this is determined
by time constants τ1 and τ2 and corresponds to the closing of a
photoswitch or to the minimum write f read delay in a
recording device.
(2) The second switching time is the time delay between the
excitation and the completion of the photoreaction, i.e., maxi-
mum population of C0 without measurable change; this is
governed by the largest of the time constants τ3, τ4, and τ5 and
corresponds to the minimum write f erase delay in a recording
device.
The existence of C0 population according to case 1 is
indicated by the observation of C* r C0 absorption and thus
by band B. Following the definition given above, our experi-
ments yield switching times of approximately 1 ps for 1 and 2,
depending mainly on the time constants of the direct reaction
pathway P1.
The photoreaction is completed in the sense of definition 2
when band B reaches its maximum intensity. Thus, the
corresponding switching times are 70 (60) ps for 1 in toluene
(acetonitrile) and 200 ps for 2 in toluene. They are determined
by the time constants of the indirect reaction pathway P2.
νj, cm-1 k1 k2 k3 k4
k5
τ1 τ2 τ3 τ4 τ5 ꢀ1,νj ꢀ2,νj
1 tol 19 600 2.9 0.3 0.4 0.1 0.004 0.3 3.3 2.7 16 230 0.3 ≈0
acn 18 800 3.7 0.4 0.4 0.1 0.004 0.3 2.8 2.3 12 250 0.3 ≈0
2 tol 19 200 4.0 0.4 0.7 0.1 0.02 0.3 2.6 1.4 32 62 0.4 ≈0
acn 17 700
a As the measurements on 2 in acetonitrile were limited to short
delay times, numerical analysis was not possible.
state absorption of the closed C isomer since the center
wavenumber and spectral shape agree with those of the cw
absorption of C (cf. Figure 2). As C0 is the final state of the
reaction process, the appearance of band B indicates the
photoreaction E f C of a certain number of molecules; in
particular, when B reaches its maximum intensity, the conver-
sion of a fraction φEC of the excited molecules is complete.
The convolution of model function (12) with the Gaussian
system response function can be fitted to the experimental data,
i.e., the time evolution of the transient absorption at a certain
wavenumber νj, as exemplified in Figure 3; this procedure yields
a set of reaction rates ki, corresponding time constants τi ) 1/ki,
and relative molar extinction coefficients ꢀj,νj (cf. eq 11), with
an uncertainty of (30% inherent to the numerical analysis. Table
1 summarizes the results obtained by this method for 1 and 2
in toluene and acetonitrile. Reaction quantum yields are φEC
) 18% and φCE ) 8.4% for 1 in toluene, φEC ) 17% and φCE
) 4.8% for 1 in acetonitrile, and φEC ) 17% and φCE ) 34%
for 2 in toluene.
No E* fluorescence could be detected; thus, since k1 is larger
than k2 and k3 by 1 order of magnitude, the fast decay of the
E* transient absorption (band A) is caused by nonradiative
relaxation E* f E0. Transitions from E* to I and C0 are of the
same order of magnitude. I decays much more slowly than
E*; in the case of 1, relaxation from I is mainly to C0 as reflected
by the respective k4/k5 ratios. Independent of solvent, substitu-
tion, and wavenumber, we find the general relation ꢀ1,νj . ꢀ2,νj
≈ 0 within the error limits of the numerical analysis. We
therefore do not consider the transient absorption of state I in
the interpretation of the absorption spectra obtained from our
time-resolved measurements (e.g. Figure 1).
For 1, the relaxation from I to C0 is faster than for 2, as
indicated by rate constants k4 and relaxation times τ4. This is
in agreement with the observation of shorter rise times (60, 70
ps) of absorption band B for 1 as compared to 2 (200 ps). On
the cyclization reaction coordinate, I might be shifted toward
the E isomer by substitution of a bulky adamantylidene group,
thus reducing the I f C0 transition rate.
Reaction Pathways. After excitation of E, the ring-closure
reaction E f C proceeds via two different pathways: firstly,
on a fast direct pathway P1, E* f C0; secondly, on a slow
indirect pathway P2, E* f I f C0. The relative weight of P1
and P2 is given by the ratio P1/P2 ) k2/(k3[k4/(k4 + k5)]). For
compound 1, both pathways are about equally important: P1/
P2 ≈ 0.9; i.e., 53% of the molecules undergoing ring closure
use the indirect pathway P2. On the other hand, for 2 in toluene,
investigation yields P1/P2 ≈ 0.8; therefore, the contribution of
P2 is somewhat larger, namely 56%.
Conclusions
The transient absorption of two furyl fulgides was investigated
with a subpicosecond time resolution. Based on the temporal
evolution of the absorption spectra, a model for the mechanism
of the cyclization reaction can be set up; this includes the ground
and excited states of E and C isomers as well as an intermediate
state whose nature is not evident. According to our model,
cyclization proceeds via two distinct pathways: a direct pathway
from the E* excited state to the C0 ground state and an indirect
pathway via the intermediate state I. The time demand of the
photocyclization depends on the process considered: while for
optical switching or recording of data the time constant is
approximately 1 ps, it is on a 100-ps time scale for the less
time-critical write f erase cycle in an optical storage device.
Acknowledgment. Support from the Deutsche Forschungs-
gemeinschaft (SFB 329) and Fonds der Chemischen Industrie
is gratefully acknowledged.
References and Notes
(1) Hirshberg, Y. J. Am. Chem. Soc. 1956, 78, 2304.
(2) Psaltis, D. Byte 1992, 17, 179.
(3) Matsui, F.; Taniguchi, H.; Yokoyama, Y.; Sugiyama, K.; Kurita,
Y. Chem. Lett. 1994, 1869.
(4) Walz, J.; Ulrich, K.; Port, H.; Wolf, H. C.; Wonner, J.; Effenberger,
F. Chem. Phys. Lett. 1993, 213, 321.
(5) Ulrich, K.; Port, H.; Wolf, H. C.; Wonner, J.; Effenberger, F.; Ilge,
H.-D. Chem. Phys. 1991, 154, 311.
In both cases, the intermediate state I works as a bottleneck,
delaying the transition from E* to C0 on account of its rather
long lifetime (τ4 > 10 ps) as compared to E* (τ1 e 0.3 ps). As
a consequence, the absorption band B does not increase
monoexponentially; after fast buildup of about 50% via P1 with
τ2 ≈ 3 ps, the absorption approaches its maximum intensity
(6) Kiji, J.; Okano, T.; Kitamura, H.; Yokoyama, Y.; Kubota, S.; Kurita,
Y. Bull. Chem. Soc. Jpn. 1995, 68, 616.

506 J. Phys. Chem. A, Vol. 101, No. 4, 1997
Handschuh et al.
(7) Heller, H. G. In Electronic Materials. From Silicon to Organics;
Miller, L. S., Mullin, J. B., Eds.; Plenum: New York, 1991; p 471.
(8) Suzuki, H.; Tomoda, A.; Ishizuka, M.; Kaneko, A.; Furui, M.;
Matsushima, R. Bull. Chem. Soc. Jpn. 1989, 62, 3968.
(9) Heller, H. G. Spec. Publ.sR. Soc. Chem. 1986, 60 (Fine Chemicals
for the Electronic Industry), 120.
(10) Kaneko, A.; Tomoda, A.; Ishizuka, M.; Suzuki, H.; Matsushima,
R. Bull. Chem. Soc. Jpn. 1988, 61, 3569.
(11) Ulrich, K.; Port, H. J. Mol. Struct. 1990, 218, 45.
(12) Tomoda, A.; Kaneko, A.; Tsuboi, H.; Matsushima, R. Bull. Chem.
Soc. Jpn. 1992, 65, 1262.
(16) Yokoyama, Y.; Inoue, T.; Yokoyama, M.; Goto, T.; Iwai, T.; Kera,
N.; Hitomi, I.; Kurita, Y. Bull. Chem. Soc. Jpn. 1994, 67, 3297.
(17) Yokoyama, Y.; Hayata, H.; Ito, H.; Kurita, Y. Bull. Chem. Soc.
Jpn. 1990, 63, 1607.
(18) Seibold, M.; Port, H.; Wolf, H. C. Mol. Cryst. Liq. Cryst., in press.
(19) Seibold, M.; Port, H. Chem. Phys. Lett. 1996, 252, 135.
(20) Parthenopoulos, D. A.; Rentzepis, P. M. J. Mol. Struct. 1990, 224,
297.
(21) Kurita, S.; Kashiwagi, A.; Kurita, Y.; Miyasaka, H.; Mataga, N.
Chem. Phys. Lett. 1990, 171, 553.
(22) Ilge, H.-D.; Su¨hnel, J.; Khechinashvili, D.; Kaschke, M. J.
Photochem. 1987, 38, 189.
(13) Heller, H. G.; Hughes, D. S.; Hursthouse, M. B.; Koh, K. V. S. J.
Chem. Soc., Chem. Commun. 1994, 2713.
(14) Sun, Z.; Hosmane, R. S.; Tadros, M. Tetrahedron Lett. 1995, 36,
3453.
(23) Tamai, N.; Saika, T.; Shimidzu, T.; Irie, M. J. Phys. Chem. 1996,
100, 4689.
(15) Glaze, A. P.; Heller, H. G.; Whittall, J. J. Chem. Soc., Perkin Trans.
1992, 2, 591.
(24) Woodward, R. B.; Hoffmann, R. J. Am. Chem. Soc. 1965, 87, 395,
2046, 2511.