Electronic decoupling approach to quantitative photoswitching in linear multiazobenzene architectures

Article abstract of DOI:10.1021/jp2044114

A strategy to optimize the photoswitching efficiency of rigid, linear multiazobenzene constructs is presented. It consists of introducing large dihedral angles between azobenzene moieties linked via aryl - aryl connections in their para positions. Four bi

Full text of DOI:10.1021/jp2044114

ARTICLE
pubs.acs.org/JPCB
Electronic Decoupling Approach to Quantitative Photoswitching
in Linear Multiazobenzene Architectures
David Blꢀeger,‡ Jadranka Dokiꢀc,† Maike V. Peters,‡ Lutz Grubert,‡ Peter Saalfrank,*,† and Stefan Hecht*,‡
†Theoretische Chemie, Institut f€ur Chemie, Universit€at Potsdam, Karl-Liebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany
‡Department of Chemistry, Humboldt-Universit€at zu Berlin, Brook-Taylor-Str. 2, 12489 Berlin, Germany
S Supporting Information
b
ABSTRACT: A strategy to optimize the photoswitching effi-
ciency of rigid, linear multiazobenzene constructs is presented.
It consists of introducing large dihedral angles between azo-
benzene moieties linked via arylꢀaryl connections in their para
positions. Four bisazobenzenes exhibiting different dihedral
angles as well as three single azobenzene reference compounds
have been synthesized, and their switching behavior has been
studied as well as experimentally and theoretically analyzed. As
the dihedral angle between the two azobenzene units increases
and consequently the electronic conjugation decreases, the photochromic characteristics improve, finally leading to individual
azobenzene switches operating independently in the case of the perpendicular ortho,ortho,ortho0,ortho0-tetramethyl biphenyl linker.
The electronic decoupling leads to efficient separation of the absorption spectra of the involved switching states and hence by
choosing the appropriate irradiation wavelength, an almost quantitative E f Z photoisomerization up to 97% overall Z-content can
be achieved. In addition, thermal Z f E isomerization processes become independent of each other with increasing decoupling. The
electronic decoupling could furthermore be proven by electrochemistry. The experimental data are supported by theory, and
calculations additionally provide mechanistic insight into the preferred pathway for the thermal Z,Z f Z,E f E,E isomerization via
inversion on the inner N-atoms. Our decoupling approach outlined herein provides the basis for constructing rigid rod architectures
composed of multiple azobenzene photochromes, which display practically quantitative photoswitching properties, a necessary
prerequisite to achieve highly efficient transduction of light energy directly into motion.
’ INTRODUCTION
the small geometrical difference between the close and open
forms, diarylethenes have been successfully utilized for the photo-
controlled deformation of molecular crystals,11 but their use for
the deformation of soft materials has thus far been rather limited.
In order to exploit this feature, we have recently investigated
the utilization of azobenzene-containing rigid rods as candidates
for light-gated molecular actuators undergoing substantial and
reversible structural changes upon UV and visible light irradia-
tion.12 The construction of such photoresponsive rigid rods
requires the linking of several azobenzene units via π-conjugated
para connections to ensure both rigidity and linearity, while
retaining the photoswitching efficiency. However, direct para
connection extends electronic conjugation and leads to a dra-
matically decreased photoreactivity of azobenzenes.13 One solu-
tion to attenuate the electronic coupling consists of linking
azobenzenes through meta connections.13 However, such kinked
meta connections are not suitable for the design of rod-type
macromolecules, which constitute linear objects.
Molecular switches1 have lately attracted a tremendous inter-
est in the area of nanochemistry as bistable objects capable of
performing tasks on demand, for example store data2 or perform
mechanical work,3 at the molecular level. This class of responsive
molecules additionally allows, when ingeniously incorporated
into an adequate matrix, for reversibly inducing dramatic changes
in various physicochemical properties of bulk materials and
surfaces (through self-assembled monolayers4 or thin films5)
transferring effects from the molecular level to the macroscopic
scale.6 Light, as an attractive stimulus offering high spatial
resolution and good compatibility with most materials, has been
widely used to trigger the isomerization of molecular switches,
azobenzene photochromes being among the most frequently
employed.7 This success can mainly be explained by the large
geometrical change that azobenzenes undergo, from an extended
and flat E isomer to a three-dimensional and compact Z isomer,
upon irradiation with UV-light. Azobenzenes are therefore
suitable for the photocontrolled deformation of a vast range of
soft materials8 such as liquid crystals, liquid crystalline elasto-
mers, colloidal spheres, or polymer gels, whereas their use in the
deformation of molecular crystals has been reported only
recently.9 It is interesting to note that diarylethenes,10 another
class of popular photochromes, exhibit the opposite trend: due to
Herein, we present a straightforward strategy to realize elec-
tronically decoupled and hence photoswitchable azobenzenes
Received: May 11, 2011
Revised:
June 29, 2011
Published: July 12, 2011
r
2011 American Chemical Society
9930
dx.doi.org/10.1021/jp2044114 J. Phys. Chem. B 2011, 115, 9930–9940
|

The Journal of Physical Chemistry B
ARTICLE
Figure 1. Molecular structures of the bisazobenzenes compounds 1ꢀ4 connected by increasing torsion angles (ϕ).
Table 1. Composition of the Photostationary State (PSS) in
Cyclohexane upon Irradiation with UV Light (at 368 nm for 1
and 2 and 357 nm for a, b, c, 3, and 4)
PSS composition (%)a
dihedral angle
ϕ (°)b
E,E
Z,E
Z,Z
total Z
a, b, cc
97
16
64
93
97
1
2
3
4
72
53
2
25
22
10
4
3
25
88
95
36.7
89.9
90.0
1
a As determined by UPLC analyses using integration of UV signal at
wavelengths of the isosbestic points. b DFT-calculated dihedral angles
(see Table S2). c The PSS compositions are identical ((1%) for
reference compounds aꢀc.
Figure 2. Molecular structures of the single azobenzene reference
compounds aꢀc (detailed synthesis in the Supporting Information).
The photophysical, electronic, and photochemical properties
have been investigated by means of UVꢀvis spectroscopy, cyclic
voltammetry, and kinetic analyses. For the sake of comparison,
the properties of the bisazobenzenes have been systematically
compared to the properties of single azobenzene reference
compounds (aꢀc, see Figure 2). The experimental observations
for compounds 1ꢀ4 and aꢀc were complemented by quantum
chemical calculations based on hybrid density functional theory.
connected in a linear fashion, providing the structural basis for
our photoresponsive rigid rods.12 Our approach relies on con-
nected azobenzenes joined via ortho substituted biphenyl lin-
kages to enforce large dihedral angles and therefore electronic
decoupling between the photochromes.14 A series of bisazoben-
zenes connected by a single benzene ring (compound 1) or
biphenyls (compounds 2ꢀ4), exhibiting different dihedral an-
gles (ϕ), have been synthesized (Figure 1).15 As the dihedral
angle increases and consequently the electronic conjugation
decreases,16 the photoswitching characteristics (amount of iso-
merized moieties in the photostationary state (PSS), thermal
stability of the Z isomer) improve, finally leading to azobenzenes
operating independently for the essentially perpendicular biphe-
nyl linker in bisazobenzene 4 (Table 1).
’ METHODS
Synthetic Methodology. Bisazobenzene 1 has been synthe-
sized by a three-step procedure starting from 1,4-diiodobenzene
and involving a Pd-catalyzed N-arylation protocol.17 In a differ-
ent approach, bisazobenzenes 2ꢀ4 have been prepared either by
assembling suitably functionalized azobenzenes through Suzuki
coupling (compounds 2 and 3) or by forming the NdN double
9931
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
Scheme 1. Synthetic Routes to Bisazobenzenes 1ꢀ4
bonds on the corresponding diaminobiphenyl moiety (com-
pound 4). Scheme 1 summarizes the detailed syntheses of
bisazobenzenes 1ꢀ4.
method was preferred to standard halogen/lithium exchange
with n-BuLi followed by transmetalation with alkoxyboron
compounds, which mainly results in the addition of the organo-
lithium compound to the azo group, as previously reported in the
literature.22 Bisazobenzenes 2 and 3 were finally obtained via
Suzuki coupling between 8 and 10 and 9 and 11, respectively.
The sterically hindered biaryl 3 was formed in good yield thanks
to the use of Buchwald’s 2-(20,60-dimethoxybiphenyl)-dicyclo-
hexylphosphine (SPhos) ligand.23
Another synthetic pathway, allowing for larger scale synthesis,
has been used for the preparation of compound 4. The diami-
nobiphenyl 13 was obtained with 55% yield over two steps
involving first the formation of the hydrazine 12 by reductive
homocoupling of 3,5-dimethylnitrobenzene, followed by benzidine
For the preparation of compound 1, 1,4-diiodobenzene was
converted to the corresponding NBoc-protected phenyl hydra-
zide 518 that was subsequently coupled to 3,5-di-tert-Bu-bromo-
benzene via a Pd-catalyzed cross-coupling reaction to yield 6.
After subsequent deprotection/oxidation, 1 was obtained in 27%
yield over three steps. For bisazobenzenes 2 and 3, the NdN azo
linkage was formed by condensation between nitrosoarene and
aniline derivatives.19 Starting from the nitroso 7,20 azobenzenes 8
and 9 functionalized at the para positions by halogen atoms were
obtained in good yield and subsequently borylated with pina-
colborane PinBH via a Pd(0)-catalyzed coupling reaction.21 This
9932
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
Figure 3. Time evolution of the absorption spectra of 2 ꢁ 10ꢀ5 M solutions of E,E-1 (ε = 46 070 Mꢀ1 cmꢀ1), E,E-2 (ε = 67 280 Mꢀ1 cmꢀ1), E,E-3 (ε =
65 260 Mꢀ1 cmꢀ1), and E,E-4 (ε = 67 010 Mꢀ1 cmꢀ1) in acetonitrile at 25 °C upon irradiation at 368 nm for 1 and 2 and 357 nm for 3 and 4, until the
photostationary state (PSS) is reached (in 40 s for 1, 110 s for 2, 50 s for 3, and 40 s for 4). Molar extinction coefficients (ε) were measured at the
λmax(πꢀπ*) of each of the compounds.
rearrangement of 12. In a last step, 12 was converted to the
bisazobenzene 13 via Mills condensation reaction with the nitroso 7.
Three single azobenzenes (aꢀc in Figure 2) have also been
prepared as reference compounds (a and b as references for
bisazobenzene 2 and c for bisazobenzene 4) following similar
synthetic strategies (see Supporting Information).
with experiment and to account for line broadening, the stick
spectrum was broadened by normalized Gaussians. Accordingly,
the extinction coefficients were calculated as
ꢀ
ꢁ
2
~ν ꢀ ν~
i
fi
1
ꢀ1=2
σ
pffiffiffiffiffi
σ 2π
εðν~Þ ¼
e
ð2Þ
∑
k
i
Computational Methodology. All calculations were carried
out with the Gaussian 03 program package,24 using hybrid
density functional theory with the B3LYP exchange-correlation
functional.25 For neutral and doubly negatively charges species,
singlets were calculated. Doublets were considered in the case of
singly negatively charged species with unrestricted DFT (UB3LYP).
In most cases, the 6-311G* basis set was used, in selected cases
(kinetics), a smaller basis set, 6-31G*, was used.26 Differences
between both basis sets are small (see, for example, Table S2).
Reactant (e.g., E or E,E) and product minima (e.g., Z or E,Z and
Z,Z) for compounds 1ꢀ4 and aꢀc, were determined and the
minimum character was verified by normal-mode analysis.
Transition states connecting reactants and products were
found by the synchronous transit-guided quasi-Newton method
(QST2 and QST3).27,28 From reactant and transition states and
normal-mode analyses of stationary points, activation free en-
ergies ΔG‡ = ΔH‡ ꢀ TΔS‡ were determined. From the ΔG‡
values, reaction rate constants k(T) were computed with Eyring’s
transition state theory as
Here, ν~ = 1/λ is the wavenumber, k = 4.319 ꢁ 10ꢀ10 mol mꢀ1, fi
is the oscillator strength for transition i, and υ~i is the correspond-
ing transition wavenumber, and σ is an empirical broadening
factor.
All calculations were carried out with gas phase models. For
similar azobenzenes,30 also solvent effects on computed kinetic
and spectral data were considered using polarizable continuum
models. It was found that rates and optical absorption behavior
do somewhat depend on a polarizable solvent. However, the
differences between both theoretical treatments are at most in
the order of a few percent. Larger, remaining differences to
experiment for certain properties (e.g., activation entropies), can
probably only be reduced by using explicit solvation models.
’ RESULTS AND DISCUSSION
E f Z Photoisomerization. The absorption spectra of pure
E,E-1 to E,E-4 compounds were measured in acetonitrile
(Figure 3). As the electronic conjugation decreases, the absorp-
tion maxima (πꢀπ* transition) are continuously blue-shifted,
from λmax(πꢀπ*) = 368 nm for E,E-1 to λmax(πꢀπ*) = 348 nm
for E,E-4. Upon UV irradiation at either 357 or 368 nm
(depending on the absorption maximum: PSSs with larger
Z-content were obtained by irradiating at the absorption max-
imum), the intense band corresponding to the πꢀπ* transition
decreases and the band attributed to the nꢀπ* transition
kBT
h
‡
kðTÞ ¼
eꢀΔG =RT
ð1Þ
Further details of the rate calculations are outlined below.
Vertical UV/vis absorption spectra for reactant and product
molecules were calculated from linear-response time-dependent
density functional theory (TD-B3LYP).29 To make closer contact
9933
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
(centered at around 450 nm) increases for the four compounds.
This typical photochromic behavior is the consequence of the
stepwise E,E f Z,E f Z,Z isomerization process. However, the
decrease in the πꢀπ* transition band of 1 is unusually small,
indicative of a Z-poor PSS. The relative decrease of the πꢀπ*
transition band becomes continuously stronger when going from
1 to 4, consistent with an increasing amount of Z forms in the
PSS (as confirmed by analysis of the PSS using ultraperformance
liquid chromatography coupled to a mass spectrometry detector,
UPLC-MS, see Table 1).
The presence of well-defined isosbestic points at 290 nm in the
absorption spectra of 3 and 4 as well as the possible reconstruc-
tion of the intermediate E,Z isomer’s spectrum by simple
addition of the spectra of the E,E and Z,Z isomers both indicate
that the two para-connected azobenzene chromophores are
decoupled. However, in the absorption spectra of 1 and 2, such
well-defined isosbestic points do not exist, most clearly seen in
the crossing region around 320ꢀ330 nm for 2. The absence of
sharp isosbestic points for 1 and 2 hence points to a high degree
of electronic coupling between the two azobenzene moieties in
these compounds. To obtain insight into the composition of the
solution in the PSS, the ratios of the three isomers (Z,Z, Z,E, andE,E)
were analyzed by UPLC of the irradiated solutions (Table 1).
As indicated by the change in the UV/vis spectra of 1ꢀ4 upon
UV irradiation, the total amount of Z isomer in the PSS
continuously increases going from 1 to 4 as the dihedral angle
increases. It is worth noting the dramatic difference between the
PSS composition of 1 (16% of total Z isomer) and 4 (97% of total
Z isomer), the latter exhibiting the same PSS composition
compared to single azobenzene reference compounds aꢀc. To
the best of our knowledge, such a high content of Z form in the
PSS of multiazobenzene chromophores has not been reported
thus far.31
The experimental findings detailed above are generally sup-
ported by theory. Initial geometry optimization for the E,E
isomers on the B3LYP/6-311G* level of theory provides the
dihedral angles ϕ, which amount to 36.7° for compound 2, 89.9°
for compound 3, and 90.0° for compound 4. These dihedral
angles do not vary substantially if one or two azo units have been
switched from the E to the Z isomer, giving E,Z and Z,Z isomers,
respectively (see Table S2). Furthermore, the barriers for rota-
tion about ϕ were calculated by determining the energy differ-
ence between fully geometry-optimized structures and those in
which the molecules are forced to coplanarity (i.e., ϕ = 0°, while
reoptimizing all other parameters). For rotation in compound 2,
0.07 eV (6.8 kJ/mol) are required, whereas 0.64 eV (61.7 kJ/
mol) are necessary in 3 and 1.65 eV (159 kJ/mol) in 4. Thus, the
two azobenzene units of 3 and 4 are expected to be well separated
electronically and coupling by twisting about ϕ is associated with
high energetic penalty, while in compound 2 the azobenzene
units are significantly coupled and flexible. Although the com-
puted ϕ angles are very similar and close to 90° for both
compounds 3 and 4, the former will appear less decoupled than
the latter in finite-temperature experiments, since the energy to
planarize 3 (0.64 eV) is much lower than the energy to planarize
4 (1.65 eV). Also, in the nonbiphenylic compound 1 with para-
connected diazo moieties, the electronic coupling is expected to
be strong, because the molecule is flat, and the frontier orbitals
have large coefficients at para positions of the benzene rings.30,32
All of these expectations based on geometry considerations are
born out by the calculations, which fully support the optical
spectroscopydata. For instance in compoundE,E-1 the theoretical
Figure 4. Comparison of computed UV/vis spectra of species 1ꢀ4 in
their optimized E,E (a), E,Z (b), and Z,Z (c) geometries. In all cases, TD-
B3LYP/6-311G* calculations were performed, and the stick spectrum
was broadened with Gaussians of 1300 cmꢀ1 in width to obtain
extinction coefficients.
UV/vis spectrum (Figure 4a) is dominated by a π f π*
transition at λmax(πꢀπ*) = 422 nm. Similarly we find λmax
-
(πꢀπ*) = 419 nm for compound E,E-2. Both values are strongly
red-shifted in comparison to the signals of compounds E,E-3 and
E,E-4, with peaks at λmax(πꢀπ*) = 359 nm and λmax(πꢀπ*) =
361 nm, respectively. The latter signals are close to the absorp-
tion maximum of reference compound a with a single azo unit
only (λmax(πꢀπ*) = 345 nm, see Table S3). Thus, the two
azobenzene subunits of 1 and 2 are π-conjugated and red-shifted,
while the two π-subsystems of each of the compounds 3 and 4
are electronically decoupled.
When switching one unit in the compounds 1ꢀ4 to form the
E,Z species, we note that, for compounds 3 and 4, an n f π*
band appears at around 491 nm, and a high-energy shoulder at
around 300 nm, which overlaps with the lowest-energy π f π*
transition (Figure 4b). The new signals of 3 and 4 are at identical
positions to those of reference compound a, which has, for
example, its n f π* peak at 493 nm (see Table S3). We also find
that the π f π* signal around 350 nm becomes weaker by a
factor of about two relative to the E,E case, but does not shift
much. This nicely shows conversion of one-half of the E-
azobenzene chromophores and indicates again that here the
two azo units are independent of each other. In contrast for
compounds 1 and 2, we observe a large blue shift of the π f π*
transition at the E-portion of the molecule, when the other azo
unit is switched to Z.
When switching both units to Z, calculated spectra of the Z,Z
species were obtained (Figure 4c). These reflect, for compounds
3 and 4, in very good approximation the added spectra of two
independent molecules a in their Z form (see Figure S11). In fact,
also the E,Z and E,E spectra of 3 and 4 can be very well
represented by adding the E and Z spectra of a, and two E
spectra of a, respectively (see Figure S11). In contrast the Z,Z
absorption spectra of compounds 1 and 2 are more complicated
and shifted and cannot be reconstructed by simply adding the
spectra of independent subunits. Due to the spectral shifts of
signals of individual subunits in the cases of 1 and 2, but not for 3
and 4, the former cannot show a sharp isosbestic point during
9934
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
also seen that switching of each single azo unit from E to Z costs
about 0.7 eV of energy, largely independent of the connection
geometry, i.e., 1ꢀ4.
In summary, the high Z-content in the PSS can be correlated
with the degree of decoupling leading to largely altered absorp-
tion spectra of the Z,Z isomer. As the extent of switching, i.e., the
PSS composition, is governed by the ratio of the extinction
coefficients at the irradiation wavelength (assuming constant
quantum yields), spectral separation of both isomers is the
essential parameter to attain a Z-rich PSS. As it can be seen from
the calculated UV/vis spectra (Figure 4), such spectral separation
of both isomers at the irradiation wavelength can be achieved for
compound 3 and 4, but not for compound 1 and 2. Please note
that in addition, the coupling of an electron-accepting group,
such as the NdN moiety, in general seems to cause lower Z-
content in the PSS as observed in our laboratory for various
azobenzene derivatives.13b
Electrochemical Behavior. The degree of electronic commu-
nication between the two azobenzenes in compounds 1ꢀ4 was
also investigated by cyclic voltammetry (Figure 5). While
electrochemical behavior of E-azobenzenes is well described in
the literature,33 here we focus on the reduction processes
consisting of the addition of one electron per azobenzene
unit.34 Compound E,E-1 exhibits a first reversible reduction
process at ꢀ1.1 V, corresponding to the addition of one electron
in one of the two NdN azo fragments,35 and a second quasi-
reversible process at ꢀ1.6 V relative to the addition of one
electron in the second azo unit. Hence, there is a difference of
potentials ΔE(1) = 0.5 V between the two reduction processes.
Compounds E,E-2 and E,E-3 also exhibit two distinguishable
reversible reduction processes associated with the values
ΔE(2) = 0.2 V and ΔE(3) = 0.1 V, respectively. Finally, the
two reduction processes occur at roughly the same potential in
E,E-4, i.e., ΔE(4) ≈ 0 V.
As the conjugation between the two azobenzenes decreases,
the two reduction processes occur at closer potential (ΔE
decreases in the order 1 f 4). This behavior illustrates the
strong electronic coupling between the two azobenzenes of E,E-
1, whereas in E,E-4 the two azobenzenes are mostly decoupled.
The increasing decoupling of the two azobenzenes going from 1
to 4 is additionally illustrated by the value of the reduction
potentials. Indeed, the first reduction peak of 1 appears at
significantly less negative potential (ꢀ1.1 V) relative to the
reference single azobenzene a (ꢀ1.5 V) (see Table S1), as a
result of the extended π-system, which lowers the energy of the
LUMO. On the other hand, the reduction potentials of com-
pound 4 and reference compound a are identical (ꢀ1.5 V).
The observed electrochemical behavior can also be rationa-
lized by quantum chemical results. The tendency of species A to
attach a first electron is given by the electron affinity
Figure 5. Cyclic voltammograms of bisazobenzenes 1ꢀ4; conditions:
1 mM in dimethylformamide (DMF) containing 0.1 M (n-Bu)4NPF6,
1 V/s, vs ferrocene/ferrocenium (fc/fc+) couple. The difference of
potential between the two reduction processes (ΔE) decreases from 1 to
4 as a result of the increasing electronic decoupling.
EAðAÞ ¼ EðAÞ ꢀ EðA•ꢀÞ
ð3Þ
where E(A) is the energy of the neutral species and E(A•ꢀ) that
of the radical anion. To determine cyclic voltammogram peaks,
free reaction energy differences, which take vibrational and finite
temperature contributions and solution effects into account are
needed, all of which is neglected here. Instead, nonvertical
electron affinities from (U)B3LYP/6-311G* energies of the
optimized structures of neutral and singly negatively charged
bisazobenzenes 1ꢀ4 were calculated with this Δ-SCF method.
Positive electron affinities of 1.54 eV for species E,E-1, 1.48 eV
switching in contrast to the latter. This is in agreement with the
experimental observations. Note that the computed spectra of all
azobenzenes considered here are shifted somewhat to the red
when compared to experiment, but overall features are repro-
duced. Furthermore, upon switching, the dipole moments of
compounds 1ꢀ4 change, from zero or low values for E,E isomers,
to values of around 5 D for Z,Z isomers (see Table S2). There it is
9935
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
Figure 6. Thermal back isomerization monitored by UPLC starting from the PSS (here for compound 4 heated at 60 °C in cyclohexane); (a) evolution
of the UPLC traces upon heating (monitored at 290 nm) and (b) ratios of the three isomers Z,Z and Z,E as well as E,E as a function of heating time.
Table 2. Kinetic Data and Thermodynamic Parameters @ 298 K for Thermal Z f E Isomerization of Single Azobenzene
Reference Compounds aꢀc and Bisazobenzenes 2 and 4 in Cyclohexanea
‡
‡
‡
‡
‡
‡
|k1|
k2
|τ1|
τ2
|ΔH1
|
ΔH2
|ΔS1
|
ΔS2
|ΔG1
|
ΔG2
(sꢀ1
)
(sꢀ1
)
(h)
(h)
(kJ molꢀ1
)
(kJ molꢀ1
)
(J molꢀ1 Kꢀ1
)
(J molꢀ1 Kꢀ1
)
(kJ molꢀ1
)
(kJ molꢀ1
)
a
b
c
7.2 ꢁ 10ꢀ6
1.4 ꢁ 10ꢀ5
3.3 ꢁ 10ꢀ6
1.6 ꢁ 10ꢀ5
4.6 ꢁ 10ꢀ6
27
14
57
12
42
91
90
95
86
85
ꢀ37
ꢀ36
ꢀ32
ꢀ48
ꢀ61
102
101
104
100
103
2
4
1.8 ꢁ 10ꢀ5
4.6 ꢁ 10ꢀ6
10
42
89
85
ꢀ37
ꢀ61
100
103
a k1 and k2 refer to the Z,Z f Z,E and Z,E f E,E isomerizations, respectively. Normalized data |x|, equal to the corresponding x value divided by two,
were used in order to consider the statistical character of the first isomerization.
for E,E-2, 1.30 eV for E,E-3, and 1.14 eV for E,E-4 were obtained
(see Table S4). The trend reflects the decreasing tendency to
accept an electron when going from 1 to 4 as observed
experimentally in Figure 5. In fact, the electron affinity differ-
ences relative to 1, are 0.06 eV for compound 2, 0.24 eV for
compound 3, and 0.40 eV for compound 4, which compares
reasonably well with the experimental voltage shifts of the first
reduction peak relative to compound 1, i.e. ∼ 0.1 V for 2, ∼ 0.2 V
for 3, and ∼0.4 V for 4. Similar trends are obtained by estimating
electron affinities more crudely from Koopmans’ theorem as
geometry changes, in particular in the ϕ dihedral angles, which all
become smaller the more electrons are added (see Table S4).
That indicates that the degree of conjugation between the two
azobenzene units increases in anionic species, with possible
implications for electrochemical or electron-induced switching.
Thermal Z f E Isomerization. Bisazobenzenes 1ꢀ4 ther-
mally revert from the Z,Z to the E,E isomer via the Z,E form. The
process has been followed for compounds 2 and 4 by UPLC-MS
monitoring. Figure 6 shows the typical evolution of the ratios of
the three isomers upon heating, in that case for 4 at 60 °C,
starting from the PSS (Z,Z = 95%, Z,E = 4%, E,E = 1%).
The three evolution curves (Figure 6b) perfectly fit with
exponential decays of the first order for the Z,Z isomer, and of
the second order for the Z,E and E,E isomers, in agreement with
an apparent first order cascade reaction Z,Z f Z,E f E,E.36 The
rate constants k1 and k2 associated with the Z,Z f Z,E and Z,E f
E,E reactions, respectively, were obtained by using the Z,Z and Z,
E ratios vs time curves, respectively (see Figures S6ꢀS9).
Activation parameters for the thermal isomerization of com-
pounds 2 and 4 (Table 2) were consequently determined via
Eyring analysis37 by measuring the rate constants at different
temperatures (see Figure S10). In order to accurately compare
the kinetics of the two consecutive thermal isomerizations, the
statistical character of the first isomerization (two different
azobenzene units per molecule are available for the first isomer-
ization, Z,Z f Z,E or Z,Z f E,Z) must be considered. Thus,
normalized data, denoted by |x|, were introduced for the Z,Z f
Z,E process. The normalized rate constant |k1| is therefore equal
to the corresponding k1 value divided by two. The normalized
thermodynamic parameters (that is the parameters correspond-
ing to the first Z,Z f Z,E or E,Z isomerization and taking into
account statistics) were subsequently determined by using |k1|
values determined at different temperatures. For comparison,
negative LUMO KohnꢀSham energies, i.e., EA(A) ≈ ꢀεLUMO
.
We find that from 1 to 4, εLUMO becomes continuously less
negative (i.e., the electron affinity decreases), with energy
shifts relative to compound 1 of 0.15 eV for 2, 0.46 eV for 3,
and 0.50 eV for 4.
It is more difficult to describe the second reduction step, i.e.
the attachment of a second electron. The electron affinity of the
anion, EA(A•ꢀ) = E(A•ꢀ) ꢀ E(A••2ꢀ) is negative when gas-
phase molecular structures are used (see Table S4), and no clear
trends emerge on this level of theory. Instead we may take the
energy differences of the LUMO and LUMO+1 of the neutral
molecules as a rough measure for the quantity ΔE as defined in
Figure 5. In this way we get ΔE(1) = 1.06 eV, ΔE(2) = 0.48 eV,
ΔE(3) = 0.07 eV, and ΔE(4) = 0.03 eV, compared to the
experimental values of ΔE(1) = 0.5 (e)V, ΔE(2) = 0.2 (e)V,
ΔE(3) = 0.1 (e)V, and ΔE(4) = 0.0 (e)V (see Table S4). Thus,
despite the meaning and value of KohnꢀSham orbital energies
should not be overemphasized, the computed trend shows that
the LUMO π* orbitals of individual azobenzene units remain
almost intact and uncoupled in the case of compound 4 and even
3, while in 1 and 2 they are coupled and form bonding and
antibonding combinations with lower and higher energies,
respectively. Interestingly, the attachment of electrons leads to
9936
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
Table 3. Kinetic Data and Thermodynamic Parameters Derived from B3LYP/6-31G* Eyring Transition State Calculations for
Compounds aꢀc As Well As 2 and 4 in the Gas Phase @ 298.15 Ka
‡
‡
‡
‡
‡
‡
|k1|
k2
|ΔH1
|
ΔH2
|ΔS1
|
ΔS2
|ΔG1
|
ΔG2
(sꢀ1
)
(sꢀ1
)
(kJ molꢀ1
)
(kJ molꢀ1
)
(J molꢀ1 Kꢀ1
)
(J molꢀ1 Kꢀ1
)
(kJ molꢀ1
)
(kJ molꢀ1
)
a
4.6 ꢁ 10ꢀ5
9.4 ꢁ 10ꢀ5
9.0 ꢁ 10ꢀ5
1.1 ꢁ 10ꢀ3
3.9 ꢁ 10ꢀ4
1.5 ꢁ 10ꢀ5
8.4 ꢁ 10ꢀ5
101.5
97.2
12.6
4.1
97.8
96.0
96.1
89.9
92.5
100.5
96.3
b
c
99.9
12.6
12.5
22.3
8.3
2
6.3 ꢁ 10ꢀ4
1.6 ꢁ 10ꢀ4
1.3 ꢁ 10ꢀ5
6.6 ꢁ 10ꢀ6
94.1
98.9
93.6
9.5
14.2
5.9
91.3
94.6
4
99.1
2*
102.6
103.3
103.0
101.4
100.8
102.6
4*
ꢀ0.7
17.1
a Rates k1 and k2 refer to the Z,Z f Z,E and Z,E f E,E isomerizations, respectively. Normalized data |x|, equal to the corresponding x value divided by
two, were used in order to consider the statistical character of the first isomerization. For the latter two compounds, two different entries indicate low-
activation pathways (without *) and high-activation pathways (with *) respectively, for Z,Z f Z,E (index 1) and Z,E f E,E (index 2) isomerizations.
kinetic data and thermodynamic parameters of the Z f E
isomerization have also been determined for the reference
compounds aꢀc employing UV/vis spectroscopy (see the
Supporting Information).
First, it is interesting to note the difference of stability between
the single azobenzenes aꢀc, which, at room temperature (298
K), display thermal half-lives (τ) of 27 h, 14 h, and 57 h,
respectively. From these values, the two following rules of thumb
can be deduced: (i) extension of the conjugation is detrimental to
the stability of the Z form and (ii) meta substitution is beneficial
to the stability of the Z form.38
i.e., ΔS‡ = ΔS‡(T0) was assumed, where T0 is a reference
temperature, chosen as 298.15 K. The temperature dependence
of ΔH‡ was approximated as ΔH‡ = ΔH‡(0) + TΔCv(T0) from
the zero-point corrected enthalpy ΔH‡(0) = ΔE‡ + ΔE‡
at
ZPE
T = 0 K (ΔE is a difference of SCF energies), and the heat
capacity at constant volume Cv.
In analogy to Table 2, reaction rates and kinetic parameters as
obtained from gas phase B3LYP/6-31G* Eyring transition state
theory are compiled in Table 3. The meaningfulness of the
calculated values with regard to experiment becomes apparent
from the reference compounds aꢀc, for which the computed
Z f E isomerization rates k2 are in the order of the experimental
Both Z,Z and Z,E isomers of compound 2 exhibit similar
stabilities, with half-lives of 10 and 12 h, respectively, the Z,E-2
isomer being slightly more stable than Z,Z-2. Both isomers
display similar thermal stability as their reference isomer, i.e.,
Z-b (half-life of 14 h). Remarkably, both Z,Z-4 and Z,E-4 isomers
were observed to thermally revert at exactly the same rate (|τ1| =
τ2 = 42 h), indicating again that the two azobenzenes in 4 are
electronically and geometrically decoupled. Interestingly, when
two azobenzenes are connected into a cyclophane39 or by a
simple methylene CH2 linker,31a it was found that the rate
constants of the Z,Z f Z,E and Z,E f E,E isomerizations are
very different (sometimes up to 4 orders of magnitude39a), as a
consequence of either ring strain or π-stacking tendency, i.e.
geometrical or electronic coupling, respectively. Half-lives of Z,Z-
4 and Z,E-4 are somewhat shorter compared to the reference
isomer Z-c, which observed to revert with a half-life of 57 h.
The activation enthalpies ΔH‡, entropies ΔS‡, and free
energies ΔG‡ of reference compounds aꢀc are in good agree-
ment with parameters determined for other azobenzene
derivatives.7b,30,40 The activation free energies ΔG‡ of bisazo-
benzenes 2 and 4 are practically identical to their reference
compounds (around 100 kJ molꢀ1), but ΔH‡ and ΔS‡ of
bisazobenzenes 2 and 4 are lower and higher, respectively, when
compared to their references b and c. According to our previous
work,30 electronically active substituents in para-positions of
azobenzenes generally lead to lower barriers for thermal Z f E
isomerization. This effect is particularly large for electron-accept-
ing groups, i.e., displaying a (-M) effect. Because the azo moiety
constitutes such an acceptor group, connecting two azobenzenes
in para-positions therefore reduces the thermal Z f E isomer-
ization barriers, as observed in our experiments.
values of ∼10ꢀ5
s
ꢀ1. It was also confirmed that compound b
displays the highest isomerization rate. In contrast to experiment,
however, the activation entropies are positive in theory, an
observation which already has been made.30 This failure is due
to the fact that activation entropies for larger molecules are often
inaccurate when calculated under the harmonic approximation,
and most of all also probably due to the neglect of an explicit,
viscous solvent in the present theory. However, absolute rates are
in pretty good agreement because gas phase B3LYP activation
enthalpies ΔH‡ are by about 5ꢀ10 kJ/mol too large when
compared to experiment (see Table 2) and thus compensating
the positive activation entropies.
Returning to the thermal isomerization of bisazobenzenes,
theory nicely corroborates experimental findings. The rows in
Table 3 labeled 2 and 4 (without *) indicate that rate constants
for individual isomerization steps Z,Z f E,Z and E,Z f E,E differ
by not much more than a factor of 2 (for both compounds). This
is a rather small difference given the exponential sensitivity of rate
calculations. Also activation free energies differ by approximately
2 kJ/mol only. Further, the activation free energies for 2 are 2ꢀ3
kJ/mol smaller than those of 4, which is in good agreement with
experiment and consequently rates are faster for 2 than for 4, and
half-lives shorter. Absolute activation free energies and entropies,
as well as rates are not in quantitative agreement with experiment,
as expected: rates are too fast by a factor of up to about 100.
Again, this has to do with a too large (entropy) prefactor, and
with the fact that a viscous solvent can hinder the isomerization
reactions, even more so for large molecules such as 2 and 4.
The detailed analysis of the origin of these results is compli-
cated by the fact that an azobenzene unit with two nonequivalent
ligands at the N atoms has two nonequivalent transition states
with (almost) linear NꢀNꢀC units. This is indicated in Figure 7,
where the isomerization from Z,Z to E,Z and further to E,E for
The thermal isomerization behavior was also studied by
quantum chemistry using Eyring’s transition state theory accord-
ing to eq 1.37 The temperature dependence of ΔS‡ was neglected,
9937
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
Figure 7. Schematic representation of the thermal isomerization Z,Z f Z,E f E,E for molecule 4. As shown, we can define four different TSs with
almost linear NꢀNꢀC units. Characteristic rates and electronic energy differences (without ZPE and temperature corrections), ΔE (in eV), between
stationary states are also shown. Calculations on B3LYP/6-31G* level of theory.
compound 4 is demonstrated, with two distinct transition states
for each partial reaction (labeled TS1a/TS1b and TS2a/TS2b). It
can be seen that the Z f E isomerizations have their lowest
activation energies when isomerizing at one of the inner N atoms.
The associated activation energies are reduced by up to 10 kJ/
mol as compared to isomerizations taking place at the outer N
atoms. As a consequence, the isomerization rates are faster for
these lower-energy paths. In Table 3, the entries 2/4 and 2*/4*
refer to the low- and high-activation energy pathways, respec-
tively, although in experiment only the lower energy path should
be relevant.
It could be shown that as the dihedral angle increases, in the
order from 1 to 4, and consequently the electronic coupling
decreases, the photoswitching characteristics dramatically im-
prove, finally leading to azobenzenes operating independently of
each other for the practically perpendicular (ϕ = 90°) ortho,ortho,
ortho0,ortho0-tetramethyl biphenyl linker in bisazobenzene 4. Due
to the electronic decoupling, the absorption spectra in 3 and 4 are
well separated, and therefore by choosing the appropriate
irradiation wavelength almost quantitative photoisomerization
giving rise to a total amount of 97% Z isomer in the PSS for the
latter compound. Furthermore, decoupling in bisazobenzene 4
leads to the same thermal stability of both Z,Z and Z,E isomers
(half-life of 42 h), also comparable to single azobenzene refer-
ence compound a. These two findings are in stark contrast to
strongly coupled azobenzenes. For example compound 1, in-
corporating a direct para-phenylene bridge, displays a very low Z
content in the PSS (16% Z form overall). In the same compound
(1) or when two azobenzenes are connected into a cyclophane or
by a methylene CH2 linker, the two rate constants of the Z,Z f
Z,E and Z,E f E,E isomerizations (k1 and k2, respectively) can
differ dramatically (sometimes up to 4 orders of magnitude).13,39a
The general conclusion that large dihedral angles allow for
efficiently decoupling azobenzene moieties (compound 4) was
furthermore confirmed by the inspection of the UVꢀvis spectra,
revealing well-defined isosbestic points, and the cyclic voltam-
mograms, displaying a single reduction wave corresponding to
the simultaneous addition of two electrons per molecule, one in
each NdN double bond.
’ CONCLUSIONS AND OUTLOOK
Herein, we have detailed a molecular design based strategy to
optimize the photoswitching efficiency of para-connected bis-
azobenzenes, constituting the key structural motif to realize
photoresponsive rigid rod architectures. Our approach is based
on introducing large dihedral angles (ϕ) between adjacent
azobenzene moieties linearly linked via highly twisted arylꢀaryl
connections to electronically decouple the individual photo-
chromes. In order to validate our approach, four bisazobenzenes
(1ꢀ4) exhibiting different dihedral angles as well as three single
azobenzene reference compounds (aꢀc) have been synthesized
and were investigated with regard to their photochromic and
thermochromic behavior as well as electrochemical properties.
Experimental findings have been systematically complemented
by DFT calculations.
9938
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
In most of this work, experiment and theory are in good
agreement regarding the spectroscopic, electronic, and thermo-
dynamic data. The theorically found additivity of fragment UV/
spectra for compounds 3 and 4, and nonadditivity for 1 and 2, is a
direct proof of the decoupling of fragment π-systems in the
former species, a prerequisite for restoring the switching ability.
Moreover, calculations provide mechanistic insight, indicating a
preferred pathway for the thermal back Z,Z f Z,E f E,E
isomerization via inversion on the inner N-atoms.
The above-described strategy for achieving almost quantita-
tive photoswitching of para-connected azobenzenes can be used
to build complex photoresponsive systems.6,7 For instance,
connecting several electronically and geometrically decoupled
azobenzenes in their para-position results in linear rigid rod
oligomers and polymers, displaying significant changes in size
and shape upon UV irradiation.12 Integration of such shrinkable
rigid rod architectures into self-assembled nanostructured soft
materials to design systems, capable of efficiently converting light
into molecular and eventually macroscopic motion, is currently
being explored in our laboratories.
(b) Kausar, A.; Nagano, H.; Ogata, T.; Nonaka, T.; Kurihara, S. Angew.
Chem., Int. Ed. 2009, 48, 2144–2147. For a general overview on surface
relief gratings, see: (c) Smart Light-Responsive Materials: Azobenzene-
Containing Polymers and Liquid-Crystals; Zhao, Y., Ikeda, T., Eds.; John
Wiley and Sons: Hoboken, NJ, 2009; Chapter 11.
(6) Russew, M. M.; Hecht, S. Adv. Mater. 2010, 22, 3348–3360 and
references therein.
(7) For comprehensive reviews on azobenzene photochromes, see:
(a) Tamai, N.; Miyasaka, O. H. Chem. Rev. 2000, 100, 1875–1890.
(b) Organic Photochromes; El’tsov, A. V., Ed.; Consultants Bureau: New
York, 1990; Chapter 3. (c) Photochromism: Molecules and Systems; D€urr,
H., Bouas-Laurent, H., Eds.; Elsevier: Amsterdam, 2003; Chapter 4.
(8) For selected examples, see: (a) Yu, Y. L.; Nakano, M.; Ikeda, T.
Nature 2003, 425, 145. (b) Hosono, N.; Kajitani, T.; Fukushima, T.; Ito,
K.; Sasaki, S.; Takata, M.; Aida, T. Science 2010, 330, 808–811. For a
general overview, see: (c) Smart Light-Responsive Materials: Azobenzene-
Containing Polymers and Liquid-Crystals; Zhao, Y., Ikeda, T., Eds.; John
Wiley and Sons: Hoboken, NJ, 2009; Chapter 3.
(9) Koshima, H.; Ojima, N.; Uchimoto, H. J. Am. Chem. Soc. 2009,
131, 6890–6891.
(10) For a comprehensive review on diarylethenes, see: Irie, M.
Chem. Rev. 2000, 100, 1685–1716.
(11) For a representative example, see: Kobatake, S.; Takami, S.;
Muto, H.; Ishikawa, T.; Irie, M. Nature 2007, 446, 778–781.
(12) To be published elsewhere.
’ ASSOCIATED CONTENT
(13) (a) Cisnetti, F.; Ballardini, R.; Credi, A.; Gandolfi, M. T.;
Masiero, S.; Negri, F.; Pieraccini, S.; Spada, G. P. Chem.—Eur. J. 2004,
10, 2011–2021. (b) Peters, M. PhD Thesis; Humboldt-Universit€at zu
Berlin: Berlin, 2008.
(14) This strategy has already been reported for modulating the
absorption spectra of para-connected bisazobenzene derivatives; how-
ever, their isomerization ability has not been investigated, see: Morris,
R. J.; Brode, W. R. J. Am. Chem. Soc. 1948, 70, 2485–2488.
(15) The tert-Bu groups were placed at different positions in
compound 1 when compared to compounds 2ꢀ4 as in yet unpublished
work, the 2D self-assembly and photoswitching behavior of compound 1
at metallic surfaces was investigated (see also ref 4e).
S
Supporting Information. Synthetic details and com-
b
pound characterization, photochemistry, electrochemistry, spec-
troscopy, kinetic analyses, and computational methods. This
material is available free of charge via the Internet at http://
pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: sh@chemie.hu-berlin.de; petsaal@rz.uni-potsdam.de.
(16) For studies on the conductance of single-biphenyl junctions, see:
(a) Venkataraman, L.; Klare, J. E.; Nuckolls, C.; Hybertsen, M. S.;
Steigerwald, M. L. Nature 2006, 442, 904–907. (b) Vonlanthen, D.;
Mishchenko, A.; Elbing, M.; Neuburger, M.; Wandlowski, T.; Mayor, M.
Angew. Chem., Int. Ed. 2009, 48, 8886–8890. (c) Mishchenko, A.;
Vonlanthen, D.; Meded, V.; Burkle, M.; Li, C.; Pobelov, I. V.; Bagrets,
A.; Viljas, J. K.; Pauly, F.; Evers, F.; Mayor, M.; Wandlowski, T. Nano Lett.
2010, 10, 156–163.
’ ACKNOWLEDGMENT
Generous support by the German Research Foundation
(DFG via SFB 658, subprojects B8 and C2) is gratefully
acknowledged. Wacker Chemie AG, BASF AG, Bayer Industry
Services, and Sasol Germany are thanked for generous donations
of chemicals.
(17) Lim, Y. K.; Lee, K. S.; Cho, C. G. Org. Lett. 2003, 5, 979–982.
(18) Protocol adapted from:Kwong, F. Y.; Klapars, A.; Buchwald,
S. L. Org. Lett. 2002, 4, 581–584.
(19) It should be noted that the formation of azobenzenes from
condensation of nitrosobenzenes with anilines in acetic acid, frequently
referred to as the Mills reaction: (a) Mills, C. J. Chem. Soc. 1895,
67, 925–933. was first described by Baeyer: (b) Baeyer, A. Chem. Ber
1874, 7, 1638–1640.
(20) Nitrosoarenes are smoothly synthesized via Oxone mediated
oxidation of electron-rich aniline derivatives, see: Priewisch, B; R€uck-
Braun, K. J. Org. Chem. 2005, 70, 2350–2352.
(21) Murata, M.; Watanabe, S.; Masuda, Y. J. Org. Chem. 1997, 62,
6458–6459.
(22) Garlichs-Zschoche, F. A.; D€otz, K. H. Organometallics 2007, 26,
4535–4540.
(23) Barder, T. E.; Walker, S. D.; Martinelli, J. R.; Buchwald, S. L.
J. Am. Chem. Soc. 2005, 127, 4685–4696.
(24) Frisch, M. J. et al. Gaussian 03, revision C.02; Gaussian, Inc.:
Wallingford, CT, 2004.
(25) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(26) Ditschfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1971,
54, 724–728.
’ REFERENCES
(1) Molecular Switches; Feringa, B. L., Ed.; Wiley-VCH: Weinheim,
Germany, 2001.
(2) Green, J. E.; Choi, J. W.; Boukai, A.; Bunimovich, Y.; Johnston-
Halperin, E.; DeIonno, E.; Luo, Y.; Sheriff, B. A.; Xu, K.; Shin, Y. S.;
Tseng, H. R.; Stoddart, J. F.; Heath, J. R. Nature 2007, 445, 414–417.
(3) Hugel, T.; Holland, N. B.; Cattani, A.; Moroder, L.; Seitz, M.;
Gaub, H. E. Science 2002, 296, 1103–1106.
(4) For a general review, see: (a) Klajn, R. Pure Appl. Chem. 2010,
82, 2247–2279. For selected examples, see: (b) Katsonis, N.; Kudernac,
T.; Walko, M.; van der Molen, S. J.; van Wees, B. J.; Feringa, B. L. Adv.
Mater. 2006, 18, 1397–1400. (c) Pace, G.; Ferri, V.; Grave, C.; Elbing,
ꢁ
M.; von Hanisch, C.; Zharnikov, M.; Mayor, M.; Rampi, M. A.; Samori,
P. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 9937–9942. (d) Ferri, V.;
ꢁ
Elbing, M.; Pace, G.; Dickey, M. D.; Zharnikov, M.; Samori, P.; Mayor,
M.; Rampi, M. A. Angew. Chem., Int. Ed. 2008, 47, 3407–3409. (e)
Alemani, M.; Peters, M. V.; Hecht, S.; Rieder, K. H.; Moresco, F.; Grill, L.
J. Am. Chem. Soc. 2006, 128, 14446–14447. (f) Dri, C.; Peters, M. V.;
Schwarz, J.; Hecht, S.; Grill, L. Nat. Nanotechnol. 2008, 3, 649–653.
(5) For selected examples utilizing liquid-crystalline films, see: (a)
Eelkema, R.; Pollard, M. M.; Vicario, J.; Katsonis, N.; Ramon, B. S.;
Bastiaansen, C. W. M.; Broer, D. J.; Feringa, B. L. Nature 2006, 440, 163.
(27) Peng, C. Y.; Schlegel, H. B. Isr. J. Chem. 1993, 33, 449–454.
9939
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940

The Journal of Physical Chemistry B
ARTICLE
(28) Peng, C. Y.; Ayala, P. Y.; Schlegel, H. B.; Frisch, M. J. J. Comput.
Chem. 1996, 17, 49–56.
(29) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R.
J. Chem. Phys. 1998, 108, 4439–4449.
(30) Dokic, J.; Gothe, M.; Wirth, J.; Peters, M. V.; Schwarz, J.; Hecht,
S.; Saalfrank, P. J. Phys. Chem. A 2009, 113, 6763–6773.
(31) The highest values (Z-content in the PSS) thus far reported
are: 92% in(a) Norikane, Y.; Tamaoki, N. Eur. J. Org. Chem. 2006,
1296ꢀ1302; 87% in (b) Norikane, Y.; Kitamoto, K.; Tamaoki, N. J. Org.
Chem. 2003, 68, 8291–8304. 79% in ref 13a; 76% in(c) van Delden,
R. A.; Mecca, T.; Rosini, C.; Feringa, B. L. Chem.—Eur. J. 2004,
10, 61–70.
(32) In contrast, if a compound analogous to 1 with meta-connec-
tion is considered (called meta-BAB in Table S3), the two subunits are
largely independent from each other, because the frontier orbitals of
azobenzene are nodeless in meta-positions.
(33) Sadler, J. L.; Bard, A. J. J. Am. Chem. Soc. 1968, 90, 1979–1989.
(34) Note that by applying larger voltages the addition of two
electrons per azobenzene unit occurs.
(35) The LUMO π* is largely localized in the NdN bridge.
(36) Applying kinetic equations for two step cascade reactions, i.e.,
[Z,Z] = [Z,Z]0 eꢀk1t and [Z,E] = k1/(k2 ꢀ k1)[Z,E]0(eꢀk t ꢀ eꢀk t) +
1
2
[Z,E]0eꢀk t, see: Chemical Kinetics; Connors, K. A.; Wiley-VCH:
2
Weinheim Germany, 1990.
(37) Eyring, H. Chem. Rev. 1935, 17, 65–77.
(38) This last point has already been reported in ref 30.
(39) See ref 31b and(a) Tamaoki, N.; Koseki, K.; Yamaoka, T.
Angew. Chem., Int. Ed. 1990, 29, 105–106. (b) Norikane, Y.; Katoh, R.;
Tamaoki, N. Chem. Commun. 2008, 1898–1900.
(40) Stoll, R. S.; Peters, M. V.; Kuhn, A.; Heiles, S.; Goddard, R.;
B€uhl, M.; Thiele, C. M.; Hecht, S. J. Am. Chem. Soc. 2009, 131, 357–367.
9940
dx.doi.org/10.1021/jp2044114 |J. Phys. Chem. B 2011, 115, 9930–9940