Effect of Positional Substitution on the Optical Response of Symmetrically Disubstituted Azobenzene Derivatives

Article abstract of DOI:10.1021/jp037436w

We have synthesized a series of symmetrically substituted p-diamidoazobenzenes, and we compare their steady-state and time-resolved responses to those of p-diaminoazobenzene and azobenzene. Our experimental and computational data indicate that the identity of the azobenzene ring substituent plays a significant role in determining the ordering of excited electronic states and the oscillator strengths for the various S_n a?? S ₀So transitions. We find experimentally that the symmetric disubstitution of azobenzenes also influences the So isomerization surface, with the largest So barrier being found for the unsubstituted compound and the smallest barrier for p-diaminoazobenzene. We understand these data in the context of the relative electron-donating character of the para side groups.

Full text of DOI:10.1021/jp037436w

4962
J. Phys. Chem. B 2004, 108, 4962-4968
Effect of Positional Substitution on the Optical Response of Symmetrically Disubstituted
Azobenzene Derivatives
A. A. Blevins and G. J. Blanchard*
Department of Chemistry, Michigan State UniVersity, East Lansing, Michigan 48824-1322
ReceiVed: NoVember 11, 2003; In Final Form: February 10, 2004
We have synthesized a series of symmetrically substituted p-diamidoazobenzenes, and we compare their
steady-state and time-resolved responses to those of p-diaminoazobenzene and azobenzene. Our experimental
and computational data indicate that the identity of the azobenzene ring substituent plays a significant role in
determining the ordering of excited electronic states and the oscillator strengths for the various S
transitions. We find experimentally that the symmetric disubstitution of azobenzenes also influences the S
isomerization surface, with the largest S
n
r S
0
0
0
barrier being found for the unsubstituted compound and the smallest
barrier for p-diaminoazobenzene. We understand these data in the context of the relative electron-donating
character of the para side groups.
Introduction
Azobenzenes are a family of compounds that has found wide
use because of their facile photoisomerization and relatively high
barrier to thermal (ground state) isomerization. These properties
are of potential utility in the development of molecular-scale
information storage and optical switching strategies.1-11 Among
the key issues under investigation for the azobenzenes is the
ability of substitution about their aromatic rings to mediate their
optical properties and isomerization behavior. While it is
typically assumed that the linear optical response and isomer-
ization behavior of these chromophores are linked, this relation-
ship has not been established. The purpose of this work is to
examine how the presence of substituents on the azobenzene
phenyl rings influences the spectroscopic and isomerization
behavior of these compounds. We have synthesized a family
of symmetrically para-disubstituted azobenzenes and have
studied their steady-state optical properties. We have also
evaluated the branching ratio for isomerization and the rate of
trans isomer recovery for these species. Our findings indicate
that the addition of substituents to the azobenzene chromophore
can influence the steady-state and time-resolved optical proper-
ties of the chromophore significantly, and the isomerization
surfaces for these molecules are affected as well. This is not a
surprising result because of the relationship between electronic
structure and state ordering and the electron density distribution
for the bond(s) that dominate the isomerization coordinate. We
understand this finding in the context of the NdN bond
dominating the isomerization behavior of these molecules, and
the effective bond order of this moiety is influenced by the
presence of electron-donating or electron-withdrawing substit-
uents on the phenyl rings. Our finding that the isomerization
barrier is reduced when electron-donating substituents are placed
on the rings indicates that it is the π* state that is influenced
most strongly, and these findings are supported by the steady-
state spectra. Our data also point to the potential importance of
Figure 1. Normalized absorption spectra of azobenzene, p-diaminoa-
zobenzene, and the series of p-diamidoazobenzenes 1-5, as indicated.
Molar absorptivities of these compounds are given in Table 1.
Experimental Section
Steady-State Absorption Spectroscopy. The absorbance
spectra of the various azobenzenes reported here were obtained
on a Varian/Cary model 300 UV/vis absorption spectrometer.
Spectral resolution for all measurements was 1 nm, and the
normalized absorption spectra are shown in Figure 1. The molar
absorptivities and absorption maxima of the compounds are
given in Table 1.
Calculated Results. Semiempirical calculations were per-
formed on a Windows-based PC using Hyperchem v. 6.0.
Dihedral angles were incremented for isomerization barrier
21
calculations using macros written in Excel. For these calcula-
tions, initial geometry optimization was performed using PM3
parametrization, and then single point calculations were made
for the molecule at each incremented angle without additional
geometry optimization.
asymmetric substitution of the azobenzene rings in structurally
mediating isomerization.12
*
To whom correspondence should be addressed. E-mail: blanchard@
chemistry.msu.edu.
1
0.1021/jp037436w CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/25/2004

Symmetrically Disubstituted Azobenzene Derivatives
J. Phys. Chem. B, Vol. 108, No. 16, 2004 4963
TABLE 1: Absorption Maxima and Molar Absorptivities of
of Azobenzene, p-Diaminoazobenzene, and the
p-Diamidoazobenzenes 1-5
the solvent required a methanol washing of the recovered solid
to remove unreacted starting materials but eliminated the need
for recrystallization.
absorption
max (nm)
ꢀ
max
ꢀ
450
Compound 1 (N-[4-(4-(Acetylamino)phenylazo)phenyl]-
acetamide). This compound was produced with a crude yield
of 80% and recrystallized yield of 53% using CH2Cl2 solvent.
The resulting brown flakes showed proton NMR peaks (300
MHz, d-methanol) at δ 2.14 ppm (s, 6H alkyl protons), 7.70
ppm (m, 4H on benzene ring), and 7.90 ppm (m, 4H on benzene
ring). The mass spectrum (GC/MS) of the compound showed
peaks at m/z 43, 65, 92, 134, 162, and 296.
compound
azobenzene
(L/(mol cm)) (L/(mol cm))
313
394
364
364
364
364
364
22 400
27 800
17 900
23 750
24 200
24 900
25 300
500
1630
1300
2400
2530
2500
2450
p-diaminoazobenzene
1
2
3
4
5
Compound 2 (N-[4-(4-(Propionylamino)phenylazo)phenyl]-
propionamide). This compound was produced with a crude
Chemical Synthesis and Sample Preparation. The series
of p-diamidoazobenzenes we report on here were synthesized
using a modification of a standard polyamido polymerization
route. We have replaced the dicarboxylic acid with an acid
chloride, precluding the possibility of polymerization and
eliminating the need for an initiator. The reaction is illustrated
in Scheme 1. 4,4′-Azodianiline was purchased from Sigma-
yield of 73% and recrystallized yield of 54% using CH Cl2
2
solvent. The resulting brown flakes showed proton NMR peaks
1
3
(300 MHz, d-methanol) at δ 1.22 ppm (m, 6H â to carbonyl),
2.44 ppm (m, 4H R to carbonyl), 7.77 ppm (m, 4H on benzene
ring), and 7.87 ppm (m, 4H on benzene ring). The mass
spectrum (GC/MS) of the compound showed peaks at m/z 57,
9
2, 119, 148, 149, 176, 324, and 325.
Compound 3 (N-[4-(4-(Butyrylamino)phenylazo)phenyl]-
SCHEME 1: (top) Structures of Compounds Investigated
in This Work; (bottom) Reaction Scheme for Synthesis of
p-Diamidoazobenzenesa
butyramide). This compound was produced with a crude yield
of 46% and recrystallized yield of 44% using CH2Cl2 solvent.
The resulting orange needles showed proton NMR peaks (300
MHz, d-methanol) at δ 0.92 ppm (m, 6H γ to carbonyl), 1.63
ppm (m, 4H â to carbonyl), 2.29 ppm (m, 4H R to carbonyl),
7
.70 ppm (m, 4H on benzene ring), and 7.90 ppm (m, 4H on
benzene ring). The mass spectrum (GC/MS) of the compound
showed peaks at m/z 43, 71, 92, 120, 162, 190, 211, 282, 352,
and 353.
Compound 4 (Pentanoic Acid [4-(4-(Pentanoylamino)-
phenylazo)phenyl]amide). This compound was produced with
a crude yield of 34% and recrystallized yield of 30% using CH2-
Cl2 solvent. The resulting orange powder showed proton NMR
peaks (300 MHz, d-methanol) at δ 0.89 ppm (m, 6H δ to
carbonyl), 1.32 ppm (m, 4H γ to carbonyl), 1.61 ppm (m, 4H
â to carbonyl), 2.33 ppm (m, 4H R to carbonyl), 7.67 ppm (m,
4H on benzene ring), and 7.78 ppm (m, 4H on benzene ring).
The mass spectrum (GC/MS) of the compound showed peaks
at m/z 57, 92, 120, 176, 211, 296, 380, and 381.
Compound 5 (Nonanoic Acid [4-(4-(Nonanoylamino)-
phenylazo)phenyl]amide). This compound was produced with
a crude yield of 69% using THF solvent, and no recrystallization
was performed. The resulting yellow powder showed proton
NMR peaks (300 MHz, d-dimethyl sulfoxide) at δ 0.85 ppm
a
1
: R ) -CH
3
; 2: R ) -CH
2
CH
3 2 2 3
; 3: R ) -(CH ) CH ; 4: R )
-
(CH
2
)
3
CH ; 5: R ) -(CH
3
2
)
7
CH .
3
Aldrich (CAS 538-41-0). The reactions were carried out under
an inert atmosphere to limit hydrolysis of the acid chloride.
Methylene chloride, acetone, and tetrahydrofuran were pur-
chased from Sigma-Aldrich and used as received. N-Methyl-
morpholine was used to scavenge hydrochloric acid produced
during the reaction. Reaction times of 30 min at room temper-
ature were typical, with little or no increase in yield for longer
reaction times. Typical reaction conditions were for 2 mmol of
(
d, 6H), 1.25 ppm (m; 20H η, φ, ꢀ, δ, and γ to carbonyl), 1.55
ppm (m, 4H â to carbonyl), 2.33 ppm (m, 4H R to carbonyl),
.81 ppm (d, 4H on benzene ring), 7.52 ppm (d, 4H on benzene
6
4
,4′-azodianiline to be combined with 8 mmol of N-methyl-
ring), and 7.81 ppm (s, amide proton). The mass spectrum (GC/
morpholine and 8 mmol of the appropriate acid chloride in 150
mL of CH2Cl2. Following completion of the reaction, the
solution was washed with 150 mL of 2 M HCl. The diamido-
azobenzene precipitated from solution and was collected by
filtration. The resulting solid was dissolved in acetone and
washed with an equal volume of brine solution to separate the
derivatized azobenzene from the acid reaction byproducts, and
the crude product was then separated from the acetone. The
solid was purified by recrystallization from methanol/water (80:
MS) of the compound showed peaks at m/z 43, 108, 120, 211,
2
32, 352, 490, and 492.
Results and Discussion
The azobenzenes find significant use in work aimed at optical
information storage and switching because of their somewhat
unusual isomerization properties. Until recently, the standard
picture for the isomerization of azobenzenes was that there is a
significant barrier to isomerization in the S0 state, with the
magnitude of the barrier being a consequence of the azo (NdN)
moiety. We are not aware of an accurate determination of the
thermal isomerization barrier for azobenzenes, either experi-
mental or computational, but it has been thought to be on the
order of 50-60 kcal/mol. Because of this substantial energy,
the modeling of this barrier cannot be done using the usual
2
0). As the alkyl substituent of the acid chloride increased in
length, the purity of the p-diamidoazobenzene product could
be increased significantly by additional washing with brine
solution. The reaction to synthesize 5 produced lower yields
due to the nonpolar nature of the product. For this reaction THF
was used in place of CH2Cl2 as the solvent due to the limited
solubility of nonanoyl chloride in CH2Cl2. The use of THF as

4964 J. Phys. Chem. B, Vol. 108, No. 16, 2004
Blevins and Blanchard
assumption of a one-dimensional isomerization coordinate; i.e.,
both isomerization and ring rotation must be coordinated in some
unresolved manner. When excited to the S1 state, the azobenzene
chromophore is thought to isomerize by an inversion, or
rehybridization mechanism, and excitation to the S2 state is
thought to cause isomerization to proceed by rotation about the
azo bond, which is taken to be substantially single bond in
1
4
character. This widely accepted picture has continued to stir
debate, owing to the relative time scales of large-amplitude
molecular motion and internal conversion from the S2 to the S1
manifold.
Recent elegant work by both the Tahara14 and Kobayashi15
groups has cast doubt on the “standard” picture. Using femto-
second optical techniques, Tahara’s group has shown that
excitation to the S2 state of azobenzene undergoes rapid
14
relaxation to the S1 state, where isomerization proceeds. When
viewed in the context of the characteristically short lifetime of
higher excited electronic states in organic molecules, and the
time frame required for inertial motion of the phenyl ring(s),
these results are fully consistent with the behavior of most
organic chromophores. The Kobayashi group, using chirped
femtosecond pulses, has demonstrated clearly that vibronic
coupling is substantial in an excited azobenzene derivative and
that the redistribution of energy within the vibrational manifold
of the S1 state serves to mediate the isomerization behavior of
15
that species. These findings on the earliest stages of relaxation
within azobenzenes not only shed significant light on the factors
mediating isomerization but also underscore the fact that a full
understanding of isomerization in azobenzenes remains to be
achieved.
With this background, we consider whether it is possible to
mediate the isomerization behavior of azobenzenes by synthetic
means. We are interested in being able to incorporate para-
disubstituted azobenzenes into layered assemblies for the
purposes of probing local structure using isomerization and for
determining whether a layered structural motif is capable of
storing information via optical read/write means. In an attempt
to address these issues, we have focused on the synthesis and
characterization of a family of simple para-disubstituted azoben-
zenes, with particular interest in the influence of the para
disubstitution on the spectroscopic and isomerization behavior
of the resulting chromophores. The compounds studied are
structurally simplified models for disubstituted species capable
of layer incorporation. We consider the experimental spectro-
scopic and calculated isomerization results separately.
Figure 2. Calculated electronic state energies and oscillator strengths
for trans (l) and cis (r) azobenzene (top), trans (l) and cis (r)
p-diaminoazobenzene (center), and trans (l) and cis (r) p-diamido-
azobenzene (bottom). Calculations were performed using the PM3
parametrization for geometry-optimized ground-state structures.
change in the electronic excited state(s), and this is seen in the
results of semiempirical calculations for both cis and trans
conformers (Figure 2), where the dominant absorption transition
-1
(
S2 r S0) is seen to red-shift by ∼7000 cm upon addition of
the p-amino functionalities. Reaction of the diamino function-
alities to produce the p-diamido azobenzene derivatives produces
a blue-shifted absorption band relative to the p-diamino
compound. The experimental spectra can be reconciled with
the results of semiempirical calculations for both conformers
(Figure 2), and for all of the azobenzenes, it appears that multiple
S r S transitions contribute significantly to the observed linear
Steady-State Spectroscopy. The steady-state absorption
spectroscopy of azobenzene is well understood. The S1 r S0
transitions for both the trans and cis forms are characterized by
weak absorption bands ( ꢀ ∼ 500 L/(mol cm) for cis, ꢀ ∼ 2000
L/(mol cm) for trans), and these bands are centered at ∼434
nm. The S2 r S0 absorption bands for both isomers are stronger,
with the trans band at 313 nm (ꢀ ∼ 22 400 L/(mol cm)) and the
n
0
optical response. While the spectral red-shift seen for diami-
doazobenzenes relative to that seen for diaminoazobenzenes may
appear to violate one’s intuition, we believe that these findings
are the result of changes in transition oscillator strength of low-
lying transitions with the addition of amido substituents rather
than actual band shifts. Both experimental data and semiem-
pirical calculations point to the significant substituent depen-
dence of the electronic state energies of the azobenzenes.
The azobenzene derivatives we study here exhibit remarkable
substituent-dependent variation in their steady-state optical
responses, and these findings are mirrored by substantial changes
in the electronic state energies and singlet transition oscillator
strengths calculated at the semiempirical level. It appears, on
the basis of a comparison of experimental and calculated results
for these compounds, that the transition energies are predicted
relatively well, but the oscillator strengths for these transitions
1
6-18
cis band at 254 nm (ꢀ ∼ 12 000 L/(mol cm)).
The addition
of amino groups at the para positions causes a substantial change
in the absorption spectrum. For p-diaminoazobenzene, the
dominant absorption band is at ∼420 nm (ꢀ ) 22 800 L/(mol
cm)), with weaker bands seen at 310 and 250 nm. For the
solution phase spectra we recorded, we did not attempt to isolate
cis and trans conformers, so it is not clear on the basis of the
experimental data whether the bands at 310 and 250 nm
correspond to trans and cis forms, respectively, or whether these
represent two different excited electronic states of the same
(presumably trans) conformer. This remarkable change in the
absorption spectra resulting from para disubstitution reflects a

Symmetrically Disubstituted Azobenzene Derivatives
J. Phys. Chem. B, Vol. 108, No. 16, 2004 4965
are not. Given these findings, intuition may suggest that the
isomerization behavior of these molecules would likewise vary
with substitution, and we examine this issue next.
To understand the isomerization behavior of azobenzene and
the symmetrically disubstituted azobenzenes, we have evaluated
the branching ratio for isomer formation and the recovery time
required for the samples to return to an equilibrium ground-
state population distribution. The branching ratio data are
reflective of the location of the ground-state surface maximum
and the excited electronic state surface minimum. If these two
electronic state features coexist at the same point in conforma-
tional space, we would expect a branching ratio of 1; i.e., 50%
of the excited molecules relax to form a S0 trans conformer,
and 50% relax to form a S0 cis conformer. Deviations from
this ideal behavior are indicative of either a noncorrespondence
between the electronic state surface minima and maxima or a
ground-state potential energy surface that is not well character-
ized in the context of Rulliere’s model.19
To characterize the branching ratio, we irradiate a sample
until the ratio of trans to cis remains constant, as measured by
absorption spectroscopy. Because the extinction coefficients of
the cis and trans bands will, in general, differ, we need to correct
the data either through band ratio measurements or by direct
conversion of the absorbance data to concentration data using
Beer’s law. Upon UV irradiation, the ratio of trans to cis
conformers reaches a steady state, which recovers to the
equilibrium ratio once UV irradiation of the sample is stopped.
The steady-state [trans]/[cis] ratio, prior to recovery, is 0.19 for
azobenzene, 0.51 for diamidoazobenzene, and 4.5 for diami-
noazobenzene. We note the similarity of azobenzene and
diamidobenzene and the contrasting behavior of diaminoazoben-
zene.
Figure 3. (a) Time-resolved cis f trans absorbance recovery of
azobenzene following photoisomerization. The time constant for the
recovery is τ ) 10 900 ( 150 min. (b) Time-resolved cis f trans
absorbance recovery of p-diaminoazobenzene following photoisomer-
ization. The time constant for the recovery is τ ) 4.7 ( 0.6 min. (c)
Time-resolved cis f trans absorbance recovery of p-diamidoazobenzene
following photoisomerization. The time constant for the recovery is τ
)
317 ( 2 min.
We note that there is an inverse correlation between barrier
Of perhaps greater significance than the branching ratio data
are those for the isomerization recovery. We have measured
the isomerization recovery time constants for azobenzene,
p-diaminoazobenzene, and p-diamidoazobenzene. The time
constants are for cis f trans conversion and vary enormously
with substitution and scale qualitatively with the energy of the
dominant electronic transition of the chromophores. For azoben-
zene, we find that the recovery time after photoisomerization
is 10 900 min (Figure 3a), for the p-diamidoazobenzenes, the
recovery time is 317 min (Figure 3c), and for p-diaminoazoben-
zene, the recovery time is 4.7 min (Figure 3b). These remarkably
different time constants for trans recovery are due to variations
in the isomerization barrier height, which are related to the
electron density distribution about the azo bonds in these
compounds. While the back-isomerization time constants are
remarkably different for these compounds, it is important to
keep in mind that there is a logarithmic relationship between
the measured time constant and the barrier height. We have
calculated the barrier heights consistent with the experimental
back-isomerization data for the substituted azobenzenes, as-
suming a prefactor for the activated process of 1013 Hz. For
p-diaminoazobenzene, with a characteristic recovery time
constant of 4.7 min, we calculate a barrier height of 21.2 kcal/
mol, for p-diamidoazobenzene, with a recovery time constant
of 317 min, we calculate a barrier height of 23.7 kcal/mol, and
for azobenzene, with a recovery time constant of 10 900 min,
we calculate a barrier height of 25.8 kcal/mol. These barrier
heights for S0 cis f trans back-conversion vary substantially
less than the recovery time data would seem to imply, owing
to the logarithmic relationship between the recovery rate constant
and the barrier height. These results are in qualitative agreement
with semiempirical calculations (vide infra).
height and electron donor strength of the para substituents. This
correlation is the result of an effective increase in the average
electron density of the π* (antibonding) state, reducing the
effective bond order of the NdN bond and thereby reducing
the isomerization barrier height. We seek to understand the
detailed basis for this relationship and find that the linear optical
response of these compounds is useful for this purpose. The
ground state (thermal) barrier for isomerization is correlated with
the transition cross section of the first allowed electronic
transition for these azobenzenes. All of the azobenzenes studied
here have transitions in the 400-450 nm region (Figure 1), so
it is not the energy of the S r S transition that is related to
1
0
the thermal back-isomerization barrier height. Rather, it is the
transition cross section of the lowest energy transition that is
correlated with the barrier height. We rationalize this relationship
by noting that a transition cross section is reflective of large
overlap integrals for the transitions. The isomerization barrier
height is related to the bond order of the NdN bond, and the
S r S transition coordinate is thought to lie along the long
1
0
axis of the azobenzene molecule. We postulate that the
isomerization and electronic transition coordinates are nearly
the same. The ground electronic states (HOMOs) of the
azobenzenes are characterized by a relatively high bond order
of ∼2 for the NdN bond. The first excited singlet states
(LUMOs) of these compounds are characterized by a substantial
reduction in electron density and thus bond order on excitation.
Because the activation barrier for ground-state back-isomeriza-
tion of the azobenzenes involves a high-energy intermediate
state, mixing with states in the S1 manifold is possible, and the
strength of this coupling is reflected in the transition cross
sections of the S1 r S0 transitions for both the cis and trans
forms. Thus, the more strongly allowed the S1 r S0 transition
20

4966 J. Phys. Chem. B, Vol. 108, No. 16, 2004
Blevins and Blanchard
Figure 4. Calculated isomerization surfaces for azobenzene isomerization by rehybridization of an azo nitrogen (x-axis) as a function of ring
rotation angle (y-axis). Top pane is the S surface, and bottom pane is the S surface. Color scale spans 70 kcal/mol.
0
1
(
Table 1), the more single-bond character there will be for the
electron-donating amine groups at the para positions. For 4,4′-
diamidoazobenzene, the groups at the para positions are electron-
withdrawing, and the lone pair of the amido nitrogen will interact
most strongly with the carbonyl moiety, allowing contributions
from zwitterionic species (Scheme 2 bottom, structure ii),
precluding significant contributions from structures iii and iv.
The result of the presence of the amido groups is that the azo
NdN moiety will have substantially more double-bond character
than is seen for diaminoazobenzene. On the basis of these
qualitative arguments, we expect diaminoazobenzene to have
the lowest thermal isomerization barrier, azobenzene the highest,
transition state on the S0 back-isomerization surface, giving rise
to a lower barrier height. This explanation provides qualitative
agreement with the experimental data and does not invoke
coupling of higher excited electronic states.
We can also understand these data in the context of simple
resonance structure arguments. The unique behavior and low
back-isomerization barrier of 4,4′-diaminoazobenzene can be
accounted for in terms of contributions from the several
resonance structures possessing a N-N bond (Scheme 2). Such
resonance structures are stabilized by the presence of the

Symmetrically Disubstituted Azobenzene Derivatives
J. Phys. Chem. B, Vol. 108, No. 16, 2004 4967
Figure 5. Calculated isomerization surfaces for azobenzene isomerization by rotation of the azo NdN bond (x-axis) as a function of ring rotation
angle (y-axis). Top pane is the S
0
surface, and bottom pane is the S
1
surface. Color scale spans 70 kcal/mol.
and diamidoazobenzene to be intermediate, in agreement with
our experimental findings.
mechanism of ground-state (thermal) isomerization for this class
of molecules. Our calculations for azobenzene, presented in
Figures 4 and 5, show the calculated energy surfaces for the S0
(a) and S1 (b) states, for isomerization by rehybridization of
one N (Figure 4) or rotation about the NdN bond (Figure 5).
These calculated results predict that (thermal) isomerization in
the ground state proceeds by rehybridization and not by -Nd
N- bond rotation. We calculate a cis f trans barrier height of
33 kcal/mol for S0 rehybridization, compared to a 48 kcal/mol
barrier for S0 ring rotation. When comparing the calculated
results for isomerization on the S0 and S1 surfaces, a relatively
We have attempted the calculation of the isomerization
surfaces for the substituted azobenzenes, and we find that
semiempirical calculations do not reflect the experimental
substituent dependence on back-isomerization barrier height.
This is not surprising given the parametrization used in
semiempirical calculations and the multidimensional nature of
the isomerization coordinate for azobenzenes. Even with the
limitations of such calculations, we can extract useful informa-
tion from them. One issue that has not been addressed is the

4968 J. Phys. Chem. B, Vol. 108, No. 16, 2004
Blevins and Blanchard
SCHEME 2: (top) Resonance Structures for
azobenzene being different than diamidoazobenzene and azoben-
zene, both in terms of ground-state isomerization barrier and
branching ratio. We find that the experimental S0 barrier to back-
isomerization is correlated with the S1 r S0 transition cross
section, and we attribute this relationship to the strength of
vibronic contributions to the isomerization transition state. These
experimental results combined with calculated isomerization
surfaces indicate that the ground-state isomerization of azoben-
zenes proceeds by rehybridization of one of the azo nitrogens
rather than by rotation about the -NdN- bond. These
calculations provide no analogous clear indication of the
dominant isomerization process on the S1 surface. We anticipate,
on the basis of these results, that we can, through synthetic
means, demonstrate even greater control over the isomerization
barrier height for azobenzenes, and this control will be of use
in controlling the dynamics of this family of molecules in
restricted environments.
a
4
,4′-Diaminoazobenzene; (bottom) Resonance Structures
for a 4,4′-Diamidoazobenzene
b
Acknowledgment. We are grateful to the National Science
Foundation for support of this work through Grant 0090864.
References and Notes
(
1) Saremi, F.; Tieke, B. AdV. Mater. 1998, 10, 388.
(2) Wang, C.; Fei, H.; Xia, J.; Yang, Y.; Wei, Z.; Yang, Q.; Sun, G.
Appl. Phys. B: Laser Opt. 1999, 68, 1117.
3) Howe, L. A.; Jaycox, G. D. J. Polym. Sci., Part A: Polym. Chem.
998, 36, 2827.
4) Yoshii, K.; Machida, S.; Horie, K. J. Polym. Sci., Part B: Polym.
Phys. 2000, 38, 3098.
5) Aoshima, Y.; Egami, C.; Kawata, Y.; Sugihara, O.; Tsuchimori,
M.; Watanabe, O.; Fujimara, H.; Okamoto, N. Polym. AdV. Technol. 2000,
1, 575.
6) Zilker, S. J.; Bieringer, T.; Haarer, D.; Stein, R. S.; van Egmond,
(
1
(
(
1
a
All structures are expected to contribute because of the presence
(
b
of the electron-donating amine groups. Structures iii and iv are expected
to not contribute significantly because of the electron-withdrawing
character of the amide groups.
J. W.; Kostromine, S. G. AdV. Mater. 1998, 10, 855.
(
7) Eichler, H. J.; Orlic, S.; Schulz, R.; Rubner, J. Opt. Lett. 2001, 26,
581.
(
8) Hagen, R.; Bieringer, T. AdV. Mater. 2001, 13, 1805.
(9) Pedersen, T. G.; Johansen, P. M.; Pedersen, H. C. J. Opt. A: Pure
small S1 barrier is indicated for rehybridization (∼12 kcal/mol),
and the S1 surface is calculated to be barrierless for ring rotation.
While it is tempting to use this information to infer the
dominance of NdN rotation as the dominant isomerization
coordinate in the S1, we do not feel justified in making this
conclusion because an almost barrierless isomerization pathway
can be traced out on either surface, provided the ring rotation
coordinate has sufficient time to locate the minimum on the S1
surface prior to relaxation back to the S0 surface. These
calculated results should be taken as an indication of the
substantially more facile nature of isomerization on the S1
surface than is possible on the S0 surface for azobenzene.
Appl. Opt. 2000, 2, 272.
(10) Patane, S.; Arena, A.; Allegrini, M.; Andreozzi, L.; Faetti, M.;
Giordano, M. Opt. Commun. 2002, 210, 37.
(
11) Astrand, P.-O.; Ramanujam, P. S.; Hvilsted, S.; Bak, K. L.; Suauer,
S. P. A. J. Am. Chem. Soc. 2000, 122, 3482.
12) Tamada, K.; Akiyama, H.; Wei, T.-X.; Kim, S.-A. Langmuir 2003,
19, in press.
(
(13) Misra, G. S. Introductory Polymer Chemistry; Wiley Eastern
Limited: New Delhi, India, 1993.
(14) Fujino, T.; Arzhantsev, S. Y.; Tahara, T. J. Phys. Chem. A 2001,
105, 8123.
(
(
15) Saito, T.; Kobayashi, T. J. Phys. Chem. A 2002, 106, 9436.
16) Lednev, I. K.; Ye, T.-Q.; Matousek, P.; Towrie, M.; Foggi, P.;
Neuwahl, F. V. R.; Umapathy, S.; Hester, R. E.; Moore, J. N. Chem. Phys.
Lett. 1998, 290, 68.
(
17) Lednev, I. K.; Ye, T. Q.; Abbott, L. C.; Hester, R. E.; Moore, J. N.
J. Phys. Chem. A 1998, 102, 9161.
18) Zimmerman, G.; Chow, L. Y.; Paik, U. J. J. Am. Chem. Soc. 1958,
80, 3528.
Conclusions
(
Our investigation of a family of symmetrically disubstituted
azobenzenes has shown that significant control over the linear
optical and isomerization properties of these molecules can be
exerted synthetically. The experimental spectra point to diamino-
(
(
(
19) Rulliere, C. Chem. Phys. Lett. 1976, 43, 303.
20) Andersson, J. A. J. Photochem. 1983, 22, 245.
21) These macros are available from the corresponding author upon
request.