Electrocatalytic Z → E Isomerization of Azobenzenes

Article abstract of DOI:10.1021/jacs.6b10822

A variety of azobenzenes were synthesized to study the behavior of their E and Z isomers upon electrochemical reduction. Our results show that the radical anion of the Z isomer is able to rapidly isomerize to the corresponding E configured counterpart with a dramatically enhanced rate as compared to the neutral species. Due to a subsequent electron transfer from the formed E radical anion to the neutral Z starting material the overall transformation is catalytic in electrons; i.e., a substoichiometric amount of reduced species can isomerize the entire mixture. This pathway greatly increases the efficiency of (photo) switching while also allowing one to reach photostationary state compositions that are not restricted to the spectral separation of the individual azobenzene isomers and their quantum yields. In addition, activating this radical isomerization pathway with photoelectron transfer agents allows us to override the intrinsic properties of an azobenzene species by triggering the reverse isomerization direction (Z → E) by the same wavelength of light, which normally triggers E → Z isomerization. The behavior we report appears to be general, implying that the metastable isomer of a photoswitch can be isomerized to the more stable one catalytically upon reduction, permitting the optimization of azobenzene switching in new as well as indirect ways.

Full text of DOI:10.1021/jacs.6b10822

Article
pubs.acs.org/JACS
Electrocatalytic Z → E Isomerization of Azobenzenes
Alexis Goulet-Hanssens,† Manuel Utecht,‡ Dragos Mutruc,† Evgenii Titov,‡ Jutta Schwarz,†
Lutz Grubert,† David Bleger,† Peter Saalfrank,*,‡ and Stefan Hecht*,†
́
†Department of Chemistry & IRIS Adlershof, Humboldt-Universitat zu Berlin, Brook-Taylor-Strasse 2, 12489 Berlin, Germany
̈
‡Theoretische Chemie, Institut fur Chemie, Universitat Potsdam, Karl-Liebknecht-Strasse 24-25, 14476 Potsdam-Golm, Germany
̈
̈
S
* Supporting Information
ABSTRACT: A variety of azobenzenes were synthesized to
study the behavior of their E and Z isomers upon
electrochemical reduction. Our results show that the radical
anion of the Z isomer is able to rapidly isomerize to the
corresponding E configured counterpart with a dramatically
enhanced rate as compared to the neutral species. Due to a
subsequent electron transfer from the formed E radical anion
to the neutral Z starting material the overall transformation is
catalytic in electrons; i.e., a substoichiometric amount of
reduced species can isomerize the entire mixture. This pathway
greatly increases the efficiency of (photo)switching while also
allowing one to reach photostationary state compositions that are not restricted to the spectral separation of the individual
azobenzene isomers and their quantum yields. In addition, activating this radical isomerization pathway with photoelectron
transfer agents allows us to override the intrinsic properties of an azobenzene species by triggering the reverse isomerization
direction (Z → E) by the same wavelength of light, which normally triggers E → Z isomerization. The behavior we report
appears to be general, implying that the metastable isomer of a photoswitch can be isomerized to the more stable one catalytically
upon reduction, permitting the optimization of azobenzene switching in new as well as indirect ways.
rates slower than 2 V/s;14 however, higher scan rates and lower
INTRODUCTION
■
temperatures shifted the Z isomer’s cathodic peak potential by
+60 mV relative to the E isomer. In protic solvents, there is an
additional pH dependence, with identical cathodic half-wave
potentials of the E and Z isomers at pH < 11, and a +180 mV
shift in the half-wave potential of the Z isomer at pH > 11.15
The gaps between the cathodic peak potentials of the Z and E
isomers in protic environments are highly dependent on the
nature of the solvent used and range from +50 mV at pH 7 in
10% ethanol buffer16 to over +120 mV in neat ethanol17 to
+200 mV in aqueous dioxane.18 On the basis of these
principles, Fujishima and co-workers developed a memory
device using Langmuir−Blodgett films of an azobenzene
derivative on an indium tin oxide transparent electrode, at
which point after writing with light the formed Z isomer could
selectively be reduced to the corresponding hydrazo compound
due to a +400 mV shift in potential as compared to the E
isomer in the aqueous environment.19
Azobenzene continues to be one of the most widely used
photochromic molecules due to the vastly different properties
of its E and Z isomers and its ready availability, either
commercially or by facile synthesis.1−3 The isomerization
around azobenzene’s central NN double bond drastically
changes the geometry of the molecule, its polarity, and
electronic properties.4 Substitution on the aromatic rings
affects many properties, including the absorption wavelength
of both isomers and their photoisomerization quantum yield as
well as the thermal half-life of the metastable Z isomer. As with
most photoswitchable molecules, these properties are generally
intrinsic to the molecules themselves.5,6
We were motivated to exploit the difference in the electronic
levels of the two isomers of azobenzene,7 following the
successful use of other photoswitch families, such as diary-
lethenes8,9 and spiropyrans,10 in this context. Over the past half
century, during which the reversible electrochemistry of
azobenzene has been investigated,11 generally two sequential
one-electron reductions have been observed.12 Neta and
Levanon compared the radical anions derived from Z and E
azobenzene spectroscopically and concluded that the same
species were formed upon reduction, implying that there must
be a mechanism for rapid Z → E isomerization from the Z
azobenzene radical anion.13 In agreement with this finding,
Laviron and Mugnier observed no difference in the resulting
two one-electron waves between the E and Z isomers at scan
Photomodulated voltammetry (PMV) enabled Grampp and
co-workers to distinguish the reduction potentials of the two
azobenzene isomers, as they observed a cathodic shift of the
reduction potential by −60 mV for the Z isomer in DMF.20
The authors suggested that the use of PMV allowed for this
observation as the technique is better suited to measure the
Received: October 16, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
potential of short-lived species and postulated that the Z
azobenzene radical anion rapidly isomerizes to the correspond-
ing E radical anion, resulting in only the E isomer potential
being measured in conventional cyclovoltammetry (CV)
experiments.
Over the past three decades, diarylethenes (DAEs) have
emerged as complementary and very versatile photochromic
compounds.21,22 DAEs have also been switched electrochemi-
cally as demonstrated for the first time by Branda and co-
workers, reporting both oxidatively induced electrocyclic ring-
opening23 and ring-closing.24 The direction of electrochemi-
cally induced switching depends on the relative thermodynamic
stability of the two isomers, with oxidation of the metastable
species serving as an alternative to optical excitation for
triggering isomerization. Importantly, the ring-opening was
shown by Branda to be a catalytic process,23 in which the ring-
opened radical cation oxidizes a second ring-closed molecule to
propagate the reaction. This process has been initiated through
electrocatalysis, through direct oxidation, or through photo-
electrocatalysis25 using photoelectron transfer agents to oxidize
the ring-closed DAE and initiate the catalytic cycle.26 Similar
catalysis has been shown in dithiazolythiazoles, where the
current efficiency was calculated to be as high as 900%.27 In
parallel and seemingly unrelated work, increased thermal Z →
E isomerization rates of azobenzenes in the presence of
nanoparticles have been reported. Depending on the nature of
the azobenzene and the surface functionalization of the
nanoparticles, up to 40 molecules of azobenzene could be
quantitatively isomerized from the Z to the E isomer,28 and the
observed rate acceleration was quantified to be between 1 and 6
orders of magnitude.29 The exact mechanism of the enhanced
thermal Z → E isomerization of azobenzene on a (flat or
curved) gold surface was explored theoretically and exper-
imentally, and the authors suggested that an electron transfer
from the azobenzene to the nanoparticle was the preferred
reaction pathway.30 The resulting azobenzene radical cation
was postulated to have a much shorter thermal half-life.
Here we detail a comprehensive investigation of the
electrocatalytic switching of azobenzenes and show that it is a
general phenomenon. For this purpose we studied the
electrochemistry of more than a dozen different azobenzene
derivatives, which all exhibit identical cathodic peak potentials
of both their E and Z isomers due to rapid isomerization of the
Z azobenzene radical anion. To access the elusive Z isomer
radical anion, we prepared a cyclic derivative, in which Z → E
isomerization is geometrically prohibited, and observed the
expected higher cathodic peak potential relative to its E isomer.
From spectroelectrochemical studies we derived a mechanistic
model to rationalize the catalytic Z → E isomerization upon
electrochemical reduction. This reduction of the Z isomer takes
place at absolute potentials smaller than that necessary for
reduction of the E isomer, although we demonstrate that the Z
isomer’s LUMO level is in fact higher in energy than the E
isomer’s LUMO level. However, relaxed electron affinities
calculated by means of the ΔSCF approach suggest that there is
almost no difference between Z and E isomers. Anyway, the
reason for this unusual behavior is the tremendously
accelerated thermal Z → E isomerization of the radical anion
as compared to that of the neutral azobenzene. Chain
propagation via electron transfer from the E isomer radical
anion to the Z isomer allows one to rapidly isomerize a solution
of Z azobenzene with a substoichiometric amount of electrons
(Figure 1).
Figure 1. Light induces E → Z isomerization of azobenzene while an
alternative wavelength of light or heat triggers the reverse isomer-
ization. In this work we show that efficient and quantitative Z → E
isomerization can be triggered by a catalytic amount of electrons,
which can be supplied directly or indirectly with the aid of a
photoelectron transfer agent.
In addition, we demonstrate the use of photoelectron
transfer agents to induce this catalytic isomerization. The use
of this alternative pathway, as opposed to direct azobenzene
excitation, opens the door to tune the wavelength required for
isomerization and to render the quantum yield independent
from chemical functionalization of the azobenzene itself. We
show that one can even override the isomerization preference at
a given wavelength, obtaining E enriched photostationary states
(PSS) at wavelengths that would typically provide Z enriched
PSS mixtures.
RESULTS AND DISCUSSION
■
Reduction Potential of E and Z Azobenzenes. Upon
conversion of the flat and fully π-conjugated E isomer to the
twisted Z isomer, the extent of π-conjugation is reduced,
leading to a shift of both the HOMO (n) and LUMO (π*)
levels to higher energies as confirmed computationally.7,31 For
the parent azobenzene 1, the LUMO energy of the E isomer is
calculated to be −2.77 eV, being 0.27 eV lower as compared to
the Z isomer with ELUMO = −2.50 eV (see Table S7 in the
Supporting Information, calculated with B3LYP/6-311++G**,
solvent corrected for acetonitrile using the polarizable
continuum model). However, taking relaxed ΔSCF electron
affinities, we find almost no difference between Z and E isomers
in their tendency to accept an electron (see Table S7). In
agreement with these findings, our voltammetry experiments
employing a rotating ring-disc electrode (RRDE) yielded the
same results for both isomers (Figure 2), suggesting that the
cathodic peak potential as well as the kinetic stability of the
radical anion derived from the Z isomer and the E isomer are in
fact identical.
Another possible explanation for equal peak potentials could
be that a very efficient conversion of the Z to the E isomer is
induced upon reduction to the radical anion. To test this
theory, we designed the cyclic azobenzene derivative 3, in
which the Z configuration is locked (see molecular structure
determined by single crystal X-ray diffraction, Figure S25 in the
Supporting Information) and compared it to its linear analogue
2 that is able to interconvert between its E and Z isomers, E-2
and Z-2, respectively (Figure 3, top). Optical spectra in a
variety of solvents and over a temperature range from −100 to
100 °C verify that locked azobenzene Z-3 indeed does not
undergo any detectable thermal or photoinduced isomerization
(see Figure S24 in the Supporting Information). Subsequent
cyclic voltammetry experiments of azobenzene 2 (Figure 3,
bottom) resemble the ones conducted with the parent
azobenzene 1; i.e., the reduction of the E and Z isomers
B
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. Voltammetry of unsubstituted E azobenzene E-1 (top) and
Z azobenzene Z-1 (as the PSS mixture containing 83% Z-1, bottom)
with a rotating glassy-carbon ring-disc electrode, showing identical
cathodic peak potentials of the E and Z isomers and a comparable
stability of the formed radical anion species. Instrument details: r1 =
2.850 cm; r2 = 3.125 cm; r3 = 3.975 cm, N0 = 0.398, rotation = 1000
rpm, [1] = 5 × 10−4 M in 0.1 M Bu4NPF6 acetonitrile solution.
Figure 3. Chemical structures of the two isomers of azobenzene
derivative 2 (top). A Z-enriched mixture containing 81% of Z-2 is
obtained by irradiating E-2 with 365 nm light. The cyclic azobenzene
analogue 3, which is locked as the Z isomer (middle, single crystal X-
ray diffraction structure shown as ORTEP plot with 50% probability
ellipsoids), and corresponding cyclic voltammograms in acetonitrile
(bottom), showing a 250 mV higher reduction potential of Z-3 as
compared to E-2 and Z-2, which are identical.
c
occurs at the same potential of Ep = −1.73 V (all potentials are
reported vs Fc/Fc+). In the case of the cyclic azobenzene Z-3, a
reduction potential that is 250 mV more negative than the
cathodic peak potential of the linear isomers was measured.
Computations place the LUMO of the Z-3 at −2.31 eV
compared to −2.62 eV for Z-2, both of which are higher than
the LUMO level of the E-2 isomer at −2.84 eV. Therefore, we
assign the measured cathodic peak potential in the case of Z-3
to correspond to the LUMO level of the Z isomer, which in the
case of linear analogue 2, and most other azobenzene
derivatives, is not accessible experimentally. Although the
reduction Z-3 is reversible, the anodic wave in the CV at −1.4
V is indicative of a subsequent reaction of the radical anion.
Spectroelectrochemistry of E and Z Azobenzene. The
clearly distinct cathodic peak potentials of the electronically
rather similar Z azobenzenes Z-2 and Z-3 indeed support a
chemical transformation triggered during the course of the
electrochemical measurement in the case of Z-2. Using an
additional method such as optical absorption spectroscopy
during the cyclovoltammetry measurement allows for the
observation of each species generated in the course of the
overall reduction−oxidation process from the neutral azoben-
zene to the radical anion and back to the neutral azobenzene.
Azobenzene derivative 4 was selected for this study due to the
greatly enhanced thermal half-lives of ortho-fluorinated Z
azobenzenes at room temperature and the high Z isomer
content of 90% in the photostationary state (PSS) upon
irradiation above 500 nm.31,32 A slow reduction process was
initiated at a speed of 10 mV/s, and the corresponding UV−vis
spectra were recorded every 10 mV. The potential was
measured from 0.22 to −1.52 V and back to 0.22 V giving
rise to a total of 301 spectra (see Figures S33 and S34 of the
Supporting Information). Upon an increase of the (absolute
value of the) reductive potential the spectra show a strong
decrease in the absorbance of the initial π → π* transition of E-
4 and emergence of new bathochromically shifted bands at 493,
625, and 691 nm attributed to the formed radical anion of E-4.
This assignment is based on the similarity with spectra of
previously reported azobenzene anion radicals33 and the fact
that exactly 1 equiv of electrons was transferred at the
maximum absorbance and hence reflects the concentration of
the formed species as measured by coulometry. Since the
reduction is fully reversible, the final spectrum after cyclic
voltammetry is identical to the initial spectrum of E-4.
The same spectroelectrochemical experiment was carried out
with a solution of azobenzene 4, irradiated to its PSS containing
90% Z-4 (Figure 4, top). Already at a potential of −0.91 V,
clearly before reaching the onset of the reduction peak, the
spectrum undergoes a dramatic change characterized by a sharp
increase in the π → π* band with a concomitant decrease of the
n → π* transition. These spectral changes clearly show that Z
→ E isomerization is occurring before the reduction to the
radical anion is actually taking place. In contrast to the
formation of the radical anion, where 1 equiv of electrons is
C
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
In our studies, we investigated 13 different azobenzene
derivatives (Table S2), both as their pure E isomers and the Z-
rich PSS mixtures, and all compounds showed a similar Z → E
isomerization event occurring approximately 150−350 mV
before the cathodic peak potential, corresponding to the radical
anion of the E isomer (the onset of this isomerization is
tabulated in Table S5). At the employed scan rate (10 mV/s)
there is a negligible amount of isomerization before the
electrochemically induced event. Note that thermal isomer-
ization can be excluded as we studied several fluorinated
azobenzenes with thermal half-lifes of up to 2 years.31 In
addition to dramatically increasing the thermal half-life,32
fluorine substitution strongly affects the cathodic peak
potential, which was measured between −1.18 and −2.00 V
(see section 4.16 in the Supporting Information). Importantly,
in all of the investigated derivatives, the acceleration of the Z →
E isomerization was similar, indicating that it is independent of
the cathodic peak potential.
In view of our results, the reductively induced Z → E
isomerization process seems to be general, and because only a
small amount of current is flowing through the sample during
the reaction, it cannot be stoichiometric in electrons.
Integrating the region of the cyclic voltammogram where the
Z → E isomerization takes place yields a current only
approximately 5% of the value required to reduce the entire
amount of azobenzene present in solution to the radical anion.
Therefore, this experiment for the first time provides direct
evidence that the Z → E isomerization is indeed electro-
catalytic,34 as has been suggested in previous literature
reports,14,20 and as a result explains why direct measurement
of the Z isomer’s cathodic peak potential using standard
electrochemical techniques is not feasible. The fact, however,
that the isomerization event occurs anodically shifted from
where computation and our experiments with the locked Z
azobenzene 3 suggesting that the cathodic peak potential of the
Z isomer should occur can be explained with the aid of Nernst’s
equation (eq 1):
⎛
⎜
⎝
⎞
⎟
⎠
aox
RT
zF
E1/2 = E° +
ln
a
(1)
red
According to this equation, the half-wave electrode potential
(E1/2) of a given redox process (involving z electrons at a given
temperature T) depends on the standard potential (E°) and the
ratio of the activity coefficients of the two involved redox-active
species (aox and ared). In the case of a fully reversible redox
process, the second term vanishes since aox = ared and hence
E1/2 = E°. Indeed, for E azobenzene, the half-wave electrode
potential equals the standard potential. In the case of the
reduction of Z azobenzene, however, the formed Z radical
anion undergoes rapid Z → E isomerization; hence, the
concentration of this species is constantly depleted, and the
system is pulled out of equilibrium such that E1/2 ≠ E°. As a
result, positive shifts of E1/2 relative to E° are observed, which
in the investigated derivatives range between 150 and 350 mV,
corresponding to ratios of the neutral Z isomer and its
corresponding radical anion (aox:ared) in the order of 340:1 to
830 000:1. This shift explains why the formation of the Z
radical anion occurs prior to reaching the cathodic peak
potential, which is required to form the E radical anion.
Figure 4. UV−vis spectral changes during cyclovoltammetry of Z-4:
Initial and efficient Z → E isomerization in the range from −0.91 to
−1.05 V (top) followed by generation of the E-4 radical anion in the
range from −1.06 to −1.40 and back to −1.30 V (bottom). All spectra
were acquired in a 1 mm quartz cuvette, with Pt electrodes in the
presence of 0.1 M Bu4NPF6. c = 5 × 10−4 M, Pt electrode, −0.15 V →
−1.30 V → −0.15 V, dE/dt =10 mV/s.
necessary to transform the bulk solution, the initial Z → E
isomerization requires far less than 1 equiv of electrons as
judged by the negligible current of 1.4 μA (see Figure 4, top).
Subsequent further increase of the reductive potential leads to
formation of the radical anion of E-4 (Figure 4, bottom)
showing that the involved species over the entire potential
range are identical to the ones observed in the direct reduction
of E-4 as described above (see also Figure S34 in the
Supporting Information). In both cases, the species recovered
after a full reduction−oxidation cycle is the E isomer.
From these data we postulate the mechanism sketched in
Figure 5 (left), which explains why the reduction process is
catalytic once the Z azobenzene radical anion (Z•−) enters the
D
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
computational details see section 6 of the Supporting
Information, and for results see Table S8).
Photoelectrocatalytic Switching. On the basis of the
above electrocatalytic switching mechanism, we were interested
in exploiting indirect excitation to induce the Z → E
isomerization.5,40 We selected an iridium complex, fac-tris(2-
(3-p-xylyl)phenyl)pyridine iridium(III) (Ir(dmpppy)3)),
known to be an efficient photoelectron transfer agent,41 since
its LUMO energy level is higher in energy than that of
azobenzene providing the thermodynamic driving force for
efficient photoelectron transfer. We investigated the parent
azobenzene 1, which was irradiated at 365 nm to its Z-rich PSS
containing 83% of Z-1, before the iridium complex was added
and the solution was degassed. In the absence of any
Ir(dmpppy)3, upon irradiation with light of 365 nm, the PSS
remains unaffected and contains 17% E-1 (Figure 6). In
Figure 5. Proposed mechanism and reason for observed rate
acceleration: Upon initial azobenzene photoisomerization (blue
arrow), the formed Z isomer can undergo thermal Z → E
isomerization either via its neutral form (black box) or via its radical
anion (magenta box). The latter pathway involves a radical chain
reaction, which after initial reduction using catalytic amounts of
electrons to form the radical anion Z•− involving rapid isomerization
to E•− (red arrow) followed by an electron transfer equilibrium
(magenta arrows) to yield the neutral E product and regenerate Z•− to
propagate the chain. The thermoneutral electron transfer equilibrium
(KeT) is driven by the extremely rapid and irreversible isomerization
step (kiso), which in the case of azobenzene at room temperature has
been calculated to be accelerated by 13 orders of magnitude as
compared with that of neutral Z → E isomerization (right).
cycle. Upon initiation by reducing a small amount of Z isomer,
the formed Z•− isomerizes extremely rapidly to the radical
anion of the E isomer (E•−),13,14 which subsequently transfers
an electron to another neutral Z isomer to propagate the cycle.
The calculated electron affinities (see Table S7 in the
Supporting Information) suggest that the electron transfer
equilibrium (KeT) between the azobenzene isomers and
specifically from E•− to substrate Z to yield Z•− and product
E is almost thermoneutral. The subsequent removal of Z•− via
rapid and irreversible isomerization (kiso) to the thermodynami-
cally favored E isomer drives the equilibrium to the product
side.
In order to rationalize the origin of the fast thermal Z → E
isomerization taking place in the radical anion, the activation
barriers for both the charge-neutral azobenzene as well as the
corresponding radical anion were evaluated computationally
(Figure 5, right). While the thermodynamic driving force for
Z → E isomerization is nearly identical in both cases, i.e.,
ΔG(Z → E) = 13.4 kcal/mol versus ΔG(Z•− → E•−) = 13.8
kcal/mol, the energy of the transition state in the case of the
radical anion isomerization (ΔG⧧ = 19.6 kcal/mol) is much
lower as compared to the charge-neutral case (ΔG⧧ = 37.4
kcal/mol). This tremendously decreased thermal barrier in the
case of the radical anion leads to a dramatic rate enhancement
by 13 (!) orders of magnitude at 298.15 K, i.e., from kΔ = 1.9 ×
10−5 s−1 to kiso = 3.3 × 108 s−1. This calculated rate acceleration
is in qualitative agreement with the experimental 6 orders of
magnitude increase measured by Zhao and co-workers for
electric field enhanced Z → E isomerization in azobenzene
liquid crystals35 as well as the 2 orders of magnitude increase
seen by Enomoto and co-workers in Langmuir−Blodgett
films,36 with both studies conducted in the absence of an
electrolyte. In previous work by our own and the Grill group,
we could show that, at very low temperatures of 7 K and in
ultrahigh vacuum, azobenzene Z → E isomerization could only
occur in the junction of the scanning tunneling microscope,
implying a significant reduction of the isomerization barrier
under the influence of a strong electric field.37−39 Computa-
tions were performed with B3LYP/6-311++G** ((for full
Figure 6. Catalytic reductive Z → E isomerization of parent
azobenzene 1 triggered by photoinduced electron transfer from an
iridium complex (Ir(dmpppy)3): Upon addition of catalytic amounts
of Ir(dmpppy)3 to the Z-rich PSS mixture (containing 83% of Z-1)
and subsequent irradiation at 365 nm, the amount of E isomer in the
PSS is substantially improved (inset, top right) while also the
isomerization rate is dramatically increased.
contrast, with the addition of only 0.5 mol % of Ir(dmpppy)3
this yield increases to 40% E-1. Increasing the iridium to the
azobenzene ratio leads to a concomitant increase in the E
content of the PSS (Figure 6, inset) until a saturation limit is
reached, as in our experiments an increase of 6.67 to 10 mol %
Ir(dmpppy)3 led to only a slight improvement in the PSS. The
observed concentration dependence supports a photoinduced
electron transfer from the excited Ir(dmpppy)3 complex to the
Z azobenzene thereby forming the Z azobenzene radical anion
and initiating the catalytic cycle.
The extent of photochemically induced switching that can be
achieved in a photochromic system is described by the PSS at a
specific irradiation wavelength. Typically, only direct excitation
is being considered, and hence, the PSS is defined by the ratio
of the extinction coefficients of the two interconverting isomers
and the ratio of the quantum yields for the forward and
backward reactions, all at the employed irradiation wavelength.
In our case, however, we have to furthermore consider indirect
excitation via the iridium complex. The efficiency of initiating
the radical chain is governed by the kinetics of the
photoinduced electron transfer from the excited iridium
complex to the Z isomer. Therefore, the overall composition
E
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
of the PSS should be given by eq 2 (derived in section 5.1 of
the Supporting Information):
peak potential of the E isomer. This overall catalytic process is
enabled by the dramatically accelerated isomerization on the
radical anion potential energy surface; i.e., Z•− → E•− is 1013
times faster than Z → E, and driven by the buildup of the
thermodynamically more stable E isomer.
εZλφ
[E]PSS
[Z]PSS
[Ir*]PSSk2
Z→E
E→Z
=
+
εEλφ
εEλφ
(2)
E→Z
Our work clarifies the mechanism and generality of previous
observations regarding the electrochemical switching of Z
azobenzenes,13,20 and follows similar observations of electro-
chemical switching in diarylethenes,23,26,27 stilbenes,45,46 and
thioindigos.47,48 Through the use of computational methods,
we elucidate that the potential energy surface of the radical
anion allows for a dramatic increase in the rate of thermal Z →
E isomerization. As opposed to the reductive ring-closing
observed in some fulgides,49 the observed isomerization is
catalytic. We are convinced that in many families of
photoswitches this phenomenon is general and currently
underexploited. For example by initiating this electrocatalytic
switching process using an iridium complex as a photoelectron
transfer agent, azobenzenes can conveniently be isomerized via
indirect excitation. Such coupled systems provide photosta-
tionary states that are no longer solely dependent on the
isomerization quantum yields and extinction coefficients of the
individual isomers of azobenzene (or another photoswitch), but
instead can be tuned by adjusting the concentration of the
photoelectron transfer agent. Using this new blend of
photochromism and photoredoxcatalysis opens the door to
more efficient and orthogonal remote control over functional
molecular systems, materials, and devices with light.
In the case of azobenzene 1, when we irradiate with 365 nm
light, where the E isomer absorbs preferentially, the PSS
becomes enriched in Z isomer (83% Z-1) as dictated by the
first (left) term shown in black on the right side of eq 2.
However, upon addition of an appropriate photoelectron
transfer agent, i.e., the iridium complex, an additional second
(right) term on the right side of eq 2 becomes increasingly
important. In fact, this indirect excitation pathway drives the
photoreaction in the opposite direction (and can only drive it
in the Z → E direction!) of the direct excitation route and
allows one to override the intrinsic photochromic behavior of
the switch in a tunable fashion. Importantly, in such sensitized
systems the ratio of E and Z azobenzene can be constantly
varied as a function of the iridium concentration, and hence, we
could successfully overcome the boundaries of a conventional
photostationary state, i.e., not being limited by intrinsic
quantum yields and absorption spectra of the azobenzene
isomers!
Our system is closely related to photoredox catalytic systems
in the literature,42−44 where photoexcitation of a catalyst
(typically, but not necessarily a transition metal complex) drives
electron transfer to activate the substrate, which subsequently
undergoes a particular chemical transformation after which the
catalyst is regenerated by back electron transfer. While the term
photoredox catalysis relates to the catalytically active species
being present in substoichiometric amounts, the process itself is
not necessarily catalytic in electrons.34 In our case, however, the
iridium complex is present in catalytic amounts, and the
product of the initial photoinduced electron transfer, i.e., the
radical anion, itself becomes the catalyst. Formation of Z•−
followed by immediate isomerization provides E•−, from which
there is an inherent competition between the propagation of
the catalytic chain (involving electron transfer to another
neutral Z isomer) or back electron transfer to the formed
oxidized Ir(IV) complex. This competition is reflected in the
saturation behavior with increasing concentration of the iridium
complex (see inset in Figure 6).
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.6b10822.
1
Synthetic details and H and 13C NMR spectra of new
compounds, spectroelectrochemistry, and computational
details (PDF)
Crystallographic details (CIF)
AUTHOR INFORMATION
■
Corresponding Authors
Last but not least, we want to point out that our system
certainly could be improved. For example, we were not able to
identify a suitable photoelectron transfer agent with an
absorption wavelength completely separated from the azoben-
zene absorption. The overlapping spectra (see Figure S61 in the
Supporting Information) are the reason why our attainable PSS
is limited to 76% E isomer. Although the photoredox catalysis is
highly efficient, as demonstrated by more than doubling the E
content with as little as 0.5 mol % iridium complex, even at an
amount of 10 mol % iridium direct absorption of approximately
one-fifth of the photons by the azobenzene prevents
quantitative Z → E isomerization.
*petsaal@uni-potsdam.de
*sh@chemie.hu-berlin.de
ORCID
Evgenii Titov: 0000-0002-1732-1826
Stefan Hecht: 0000-0002-6124-0222
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We are grateful to our dedicated co-workers and collaborators,
who have contributed to our own research endeavor in this
field over the past years. A.G.-H. is indebted to the Alexander
von Humboldt-Foundation and the FQRNT for providing
postdoctoral fellowships. Generous support from the Daimler
Benz-Foundation (via 32-02/14) as well as the European
Research Council (via ERC-2012-StG_308117 “Light4Func-
tion”) is gratefully acknowledged. D.B. acknowledges the
German Research Foundation (DFG via BL 1269/1).
CONCLUSION
■
We have investigated the electrochemistry of a variety of
azobenzene derivatives as their E and Z isomers. In all examples
we found rapid thermal Z → E isomerization, which can be
induced reductively by a catalytic amount of electrons. This
isomerization process is independent of thermal half-life and
cathodic peak potential and occurs before reaching the cathodic
F
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(37) Alemani, M.; Peters, M. V.; Hecht, S.; Rieder, K. H.; Moresco,
F.; Grill, L. J. Am. Chem. Soc. 2006, 128, 14446−14447.
(38) Dri, C.; Peters, M. V.; Schwarz, J.; Hecht, S.; Grill, L. Nat.
Nanotechnol. 2008, 3, 649−653.
REFERENCES
■
(1) Beharry, A. A.; Woolley, G. A. Chem. Soc. Rev. 2011, 40, 4422−
4437.
(2) Merino, E. Chem. Soc. Rev. 2011, 40, 3835−3853.
(3) Mahimwalla, Z.; Yager, K. G.; Mamiya, J.; Shishido, A.; Priimagi,
A.; Barrett, C. J. Polym. Bull. 2012, 69, 967−1006.
(4) Bandara, H. M. D.; Burdette, S. C. Chem. Soc. Rev. 2012, 41,
1809−1825.
(39) Alemani, M.; Selvanathan, S.; Ample, F.; Peters, M. V.; Rieder,
K. H.; Moresco, F.; Joachim, C.; Hecht, S.; Grill, L. J. Phys. Chem. C
2008, 112, 10509−10514.
(40) Moreno, J.; Gerecke, M.; Grubert, L.; Kovalenko, S. A.; Hecht,
S. Angew. Chem., Int. Ed. 2016, 55, 1544−1547.
(5) Bleg
11349.
́
er, D.; Hecht, S. Angew. Chem., Int. Ed. 2015, 54, 11338−
(41) You, Y.; Nam, W. Chem. Soc. Rev. 2012, 41, 7061.
(42) Xuan, J.; Xiao, W. Angew. Chem., Int. Ed. 2012, 51, 6828−6838.
(43) Prier, C. K.; Rankic, D. A.; MacMillan, D. W. C. Chem. Rev.
2013, 113, 5322−5363.
(6) Molecular Switches, 2nd ed.; Feringa, B. L., Browne, W. R., Eds.;
Wiley-VCH Verlag GmbH: Weinheim, Germany, 2011.
(7) Del Valle, M.; Gutier
Nanotechnol. 2007, 2, 176−179.
(8) Gemayel, M. El; Borjesson, K.; Herder, M.; Duong, D. T.;
́
rez, R.; Tejedor, C.; Cuniberti, G. Nat.
(44) Cismesia, M. A.; Yoon, T. P. Chem. Sci. 2015, 6, 5426−5434.
(45) Majima, T.; Tojo, S.; Ishida, A.; Takamuku, S. J. Org. Chem.
1996, 61, 7793−7800.
̈
́
Hutchison, J. A.; Ruzie, C.; Schweicher, G.; Salleo, A.; Geerts, Y.;
(46) Spreitzer, H.; Scholz, M.; Gescheidt, G.; Daub, J. Liebigs Ann.
1996, 1996, 2069−2077.
Hecht, S.; Orgiu, E.; Samorì, P. Nat. Commun. 2015, 6, 6330.
(9) Leydecker, T.; Herder, M.; Pavlica, E.; Bratina, G.; Hecht, S.;
Orgiu, E.; Samorì, P. Nat. Nanotechnol. 2016, 11, 769−775.
(10) Suda, M.; Kato, R.; Yamamoto, H. M. Science 2015, 347, 743−
746.
(47) Yeh, L. S. R.; Bard, A. J. J. Electroanal. Chem. Interfacial
Electrochem. 1976, 70, 157−169.
(48) Yeh, L. R.; Bard, A. J. J. Electroanal. Chem. Interfacial Electrochem.
1977, 81, 333−338.
(11) Sadler, J. L.; Bard, A. J. J. Am. Chem. Soc. 1968, 90, 1979−1989.
(12) Stradins, J. P.; Glezer, V. T. In Encyclopedia of Electrochemistry of
the Elements, Volume XIII; Bard, A. J., Lund, H., Eds.; Dekker: New
York, 1979; p 163.
(49) Fox, M. A.; Hurst, J. R. J. Am. Chem. Soc. 1984, 106, 7626−7627.
(13) Neta, P.; Levanon, H. J. Phys. Chem. 1977, 81, 2288−2292.
(14) Laviron, E.; Mugnier, Y. J. Electroanal. Chem. Interfacial
Electrochem. 1978, 93, 69−73.
(15) Laviron, E.; Mugnier, Y. J. Electroanal. Chem. Interfacial
Electrochem. 1980, 111, 337−344.
(16) Gupta, P. N.; Raina, A. J. Indian Chem. Soc. 1988, 65, 495−497.
(17) Compton, R. G.; Wellington, R. G.; et al. J. Electroanal. Chem.
1992, 322, 183−190.
(18) Klopman, G.; Doddapaneni, N. J. Phys. Chem. 1974, 78, 1825−
1828.
(19) Liu, Z. F.; Hashimoto, K.; Fujishima, A. Nature 1990, 347, 658−
660.
(20) Grampp, G.; Muresa
2005, 582, 171−178.
̧ nu, C.; Landgraf, S. J. Electroanal. Chem.
(21) Irie, M. Chem. Rev. 2000, 100, 1685−1716.
(22) Irie, M.; Fukaminato, T.; Matsuda, K.; Kobatake, S. Chem. Rev.
2014, 114, 12174−12277.
(23) Peters, A.; Branda, N. R. J. Am. Chem. Soc. 2003, 125, 3404−
3405.
(24) Peters, A.; Branda, N. R. Chem. Commun. 2003, 954−955.
(25) Lee, S.; You, Y.; Ohkubo, K.; Fukuzumi, S.; Nam, W. Angew.
Chem., Int. Ed. 2012, 51, 13154−13158.
(26) Lee, S.; You, Y.; Ohkubo, K.; Fukuzumi, S.; Nam, W. Chem. Sci.
2014, 5, 1463−1474.
(27) Nakashima, T.; Kajiki, Y.; Fukumoto, S.; Taguchi, M.; Nagao, S.;
Hirota, S.; Kawai, T. J. Am. Chem. Soc. 2012, 134, 19877−19883.
(28) Hallett-Tapley, G. L.; D’Alfonso, C.; Pacioni, N. L.; McTiernan,
́ ́
C. D.; Gonzalez-Bejar, M.; Lanzalunga, O.; Alarcon, E. I.; Scaiano, J. C.
Chem. Commun. 2013, 49, 10073−10075.
(29) Simoncelli, S.; Aramendía, P. F. Catal. Sci. Technol. 2015, 5,
2110−2116.
(30) Titov, E.; Lysyakova, L.; Lomadze, N.; Kabashin, A. V.;
Saalfrank, P.; Santer, S. J. Phys. Chem. C 2015, 119, 17369−17377.
́
(31) Bleger, D.; Schwarz, J.; Brouwer, A. M.; Hecht, S. J. Am. Chem.
Soc. 2012, 134, 20597−20600.
(32) Knie, C.; Utecht, M.; Zhao, F.; Kulla, H.; Kovalenko, S.;
Brouwer, A. M.; Saalfrank, P.; Hecht, S.; Bleg
2014, 20, 16492−16501.
́
er, D. Chem. - Eur. J.
(33) Boto, K. G.; Thomas, F. G. Aust. J. Chem. 1973, 26, 1251−1258.
(34) Studer, A.; Curran, D. P. Nat. Chem. 2014, 6, 765−773.
(35) Tong, X.; Pelletier, M.; Lasia, A.; Zhao, Y. Angew. Chem., Int. Ed.
2008, 47, 3596−3599.
(36) Enomoto, T.; Hagiwara, H.; Tryk, D. A.; Liu, Z. F.; Hashimoto,
K.; Fujishima, A. J. Phys. Chem. B 1997, 101, 7422−7427.
G
DOI: 10.1021/jacs.6b10822
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX