State-selective electron transfer in an unsymmetric acceptor-Zn(II) porphyrin-acceptor triad: Toward a controlled directionality of electron transfer from the porphyrin S2 and S1 states as a basis for a molecular switch

Article abstract of DOI:10.1021/jp907824d

A series of Zn(II) poiphyrin (ZnP) compounds covalently linked to different electron acceptor units, naphthaleneimide (NI) and naphthalenediimide (NDI), are reported. The aim was to demonstrate a stateselective direction of electron transfer, where excitation to the lowest excited S₁ state of the porphyrin (Qband excitation) would give electron transfer to the NDI unit, while excitation to the higher S₂ state (Soretband excitation) would give electron transfer to the NI unit. This would constitute a basis for an opto-electronic switch in which the direction of electron transfer and the resulting dipole moment can be controlled by using light input of different color. Indeed, electron transfer from the S₁state to NDI occurred in solvents of both high and low polarity, whereas no electron transfer to NDI was observed from the S₂ state. With NI as acceptor instead, very rapid (τ= 200-400 fs) electron transfer from the S₂ state occurred in all solvents. This was followed by an ultrafast (τ≈100 fs) recombination to populate the porphyrin Si state in nearly quantitative yield. The charge-separated state ZnP^-NI^-was spectroscopically observed, and evidence was obtained that recombination occurred from a vibrationally excited ( hot ) ZnP⁺NI^-state in the more polar solvents. In these solvents, the thermally relaxedZnP ^-NI^- state lies at lower energy than the S₁ state so that further charge separation occurred from S₁ to form ZnP^-NI^-. This resulted in a highly unusual ping-pong sequence where the reaction went back and forth between locally excited ZnP states and charge-separated states: S₂ ZnP ^-NI^-hot→S₁=→ ZnP ⁺NI^-→ S₀. The electron transfer dynamics and its solvent dependence are discussed, as well as the function of the present molecules as molecular switches.

Full text of DOI:10.1021/jp907824d

J. Phys. Chem. A 2010, 114, 1709–1721
1709
State-Selective Electron Transfer in an Unsymmetric Acceptor-Zn(II)porphyrin-Acceptor
Triad: Toward a Controlled Directionality of Electron Transfer from the Porphyrin S2 and
S1 States as a Basis for a Molecular Switch
Staffan Wallin,† Cyrille Monnereau,‡ Errol Blart,‡ Jean-Richard Gankou,‡ Fabrice Odobel,*,‡
and Leif Hammarstro¨m*,†
Chemical Physics Group, Department of Photochemistry and Molecular Science, Uppsala UniVersity,
Box 523 SE-751 20 Uppsala, Sweden, and Chimie et Interdisciplinarite´: Synthe`se, Analyse, Mode´lisation
(CEISAM) UMR CNRS 6230, 2, rue de la Houssinie`re - BP 92208 - 44322 Nantes 3, France
ReceiVed: August 13, 2009; ReVised Manuscript ReceiVed: December 8, 2009
A series of Zn(II) porphyrin (ZnP) compounds covalently linked to different electron acceptor units,
naphthaleneimide (NI) and naphthalenediimide (NDI), are reported. The aim was to demonstrate a state-
selective direction of electron transfer, where excitation to the lowest excited S1 state of the porphyrin (Q-
band excitation) would give electron transfer to the NDI unit, while excitation to the higher S2 state (Soret-
band excitation) would give electron transfer to the NI unit. This would constitute a basis for an opto-electronic
switch in which the direction of electron transfer and the resulting dipole moment can be controlled by using
light input of different color. Indeed, electron transfer from the S1 state to NDI occurred in solvents of both
high and low polarity, whereas no electron transfer to NDI was observed from the S2 state. With NI as
acceptor instead, very rapid (τ ) 200-400 fs) electron transfer from the S2 state occurred in all solvents.
This was followed by an ultrafast (τ ≈ 100 fs) recombination to populate the porphyrin S1 state in nearly
quantitative yield. The charge-separated state ZnP+NI- was spectroscopically observed, and evidence was
obtained that recombination occurred from a vibrationally excited (“hot”) ZnP+NI- state in the more polar
solvents. In these solvents, the thermally relaxed ZnP+NI- state lies at lower energy than the S1 state so that
further charge separation occurred from S1 to form ZnP+NI-. This resulted in a highly unusual “ping-pong”
sequence where the reaction went back and forth between locally excited ZnP states and charge-separated
states: S2 w ZnP+NI-“hot” w S1 w ZnP+NI- w S0. The electron transfer dynamics and its solvent dependence
are discussed, as well as the function of the present molecules as molecular switches.
1. Introduction
this type, in which the direction of electron transfer is controlled
by the wavelength of the excitation light.
The concept of a molecular switch stimulates research on
both a fundamental level and in the context of, for example,
molecular sensors and logic circuits. Molecular switches may
be based on various kinds of postsynthetic control of the
function, using different types of input and output signals.1 It
is recognized as an advantage, however, if the switch can be
addressed with optical and/or electronic signals only, as this
allows for a more rapid function and avoids “wet” chemical
additions and waste products. Multicomponent molecular arrays
that exhibit a controllable light-induced electron transfer between
the components are interesting in that respect. The state
populated by light then has a different dipole moment than the
ground state, and the electrons and “holes” that are transferred
may be utilized by further transport to a circuit. Furthermore,
if the electronic response to light excitation in such an array
can be varied by using different optical signals, an opto-
electronic molecular switch with two different light-activated
states can be achieved. Instead of a more common “on-off”
switch, this would be a switch with “on1-on2-off” character
for, for example, directing signals in two different ways. In this
Article, we report attempts at constructing a molecular array of
Porphyrins have been used extensively for light-induced
electron transfer reactions, but until relatively recently only the
lowest excited singlet (S1) or triplet (T1) state has been involved,2
following a generalization of Kasha’s rule of photochemistry.3
In, for example, ZnII-tetraphenylporphyrin (ZnTPP), however,
the higher excited state denoted S2 is relatively long-lived, τ )
1-3 ps,4 and electron transfer reactivity from the S2 state to
the dichloromethane solvent was suggested.4a The relatively long
S2 lifetime was utilized to demonstrate ultrafast electron transfer
from the S2 state in porphyrin-acceptor dyads, with a covalently
bound or self-associated acceptor.5 Electron transfer studies from
the S2 state in porphyrin-acceptor dyads have more recently
been explored by a number of groups, with focus on the
dependence on reaction free energy or linking position on the
porphyrin.6
In the ZnTPP-Ru(bpy)32+ dyad of LeGourrie´rec et al., it was
found that the rate of electron transfer to the Ru(bpy)32+ acceptor
was 2 orders of magnitude faster from the porphyrin S2 state
(τET ) 1.6 ps) than from the S1 state (τET ) 100 ps).5b The S2
state was populated by excitation into the Soret band around
420 nm and the S1 state by the usual Q-band excitation around
575 nm. Thus, the electron transfer reactivity of the dyad was
very different depending on the wavelength of the excitation
light, and we have proposed to use this as a basis for a molecular
switch exhibiting different directions of the electron transfer.
* Corresponding author. E-mail: leif@fotomol.uu.se.
† Uppsala University.
‡ Chimie et Interdisciplinarite´.
10.1021/jp907824d  2010 American Chemical Society
Published on Web 01/11/2010

1710 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
SCHEME 1: (Top) Schematic State Diagram and
Reaction Properties of the Proposed Switch; (Bottom)
Cartoon Illustrating How Light of Different Wavelengths
Should Induce Electron Transfer in Different Directions
(NI) directly attached at the meso position. Naphthalene
monoimides are relatively difficult to reduce, so this unit could
act as an acceptor unit in electron transfer from S2 but not from
S1. The compounds DBZ and DPZ are dyads consisting of a
zinc porphyrin with a naphthalene diimide (NDI) attached at
the meso positions via a link. The two dyads differ in the type
of linking unit, which is either a biphenyl (DBZ) or an
ethoxyphenyl group (DPZ). The NDI is easier to reduce than
the NI, but the longer linking unit than in ZI should give a
weaker electronic coupling between the porphyrin and the NDI.
Thus, NDI is designed to act as an acceptor unit in electron
transfer from S1 but not from S2, as electron transfer would be
too slow to compete with S2 deactivation by other reactions.
Importantly, the energy of the lowest singlet excited states of
the NDI and NI units is higher than the porphyrin S2 state
energy, which prevents competing energy transfer reactions from
the S2 state. As NDI and NI show no significant absorption of
light in the visible, it is also feasible to excite the porphyrin
unit selectively. DPZI is a triad consisting of a zinc porphyrin
connected directly to a NI on one side and a NDI via a phenyl
ethoxy link on the other side. The triad should combine the
properties of the two dyads and thus have the potential to display
the desired excitation wavelength-dependent direction of electron
transfer. The energy of the different charge-separated (CS) states
can be tuned by varying the solvent polarity, which allows for
tuning of the function.
The purpose of the present Article is to investigate if the
design gave the desired function, as described above. On a more
general level, these compounds also give the possibility to
compare electron transfer from the first and the second excited
states of the porphyrin in the same donor-acceptor molecule,
which may give insight into state-dependent electron transfer
processes and how to control them. We have examined the
compounds in solvents of different polarity, and also in
hydrogen-bonding solvents, to control the electron transfer
processes. The results from emission and femtosecond transient
absorption spectroscopy are rich and informative, and also
revealed some unexpected reaction sequences. We describe and
interpret the data in the Results. In the Discussion, we discuss
the electron transfer processes, the observation of recombination
from “hot” charge-separated states, and the function of these
compounds as molecular switches of the type outlined above.
SCHEME 2: Structures of the Compounds
One way to achieve this is to link two different electron acceptor
units on opposite sides of a ZnII-porphyrin. One of the acceptors
should be strongly coupled to allow for rapid electron transfer
from the short-lived S2 state, but have a low reduction potential
so that electron transfer from the S1 state will not occur for
energetic reasons (Scheme 1). The other acceptor should be
easier to reduce to allow for electron transfer from the S1 state,
but more weakly coupled so that electron transfer from the S2
state will be too slow to compete with electron transfer to the
first acceptor. Thus, the direction of electron transfer would be
opposite with Soret- versus Q-band excitation. The suggested
switch would work with a single photon as input, which is
different from other ultrafast molecular switches that have used
a combination of two input photons.1b-e It would result in
population of two different electronic states depending on the
color of the input photon, states that have a very different spatial
location of the excess electron and dipole moments of different
directions.
2. Results
2.1. Synthesis of the Compounds. The synthesis route for
the preparation of the dyads 1 (ZI) and 2 (BDZ) is shown in
Scheme 3. The known trisaryl porphyrin 58 was nitrated with
the mixture silver nitrite and iodine with a slight modification
of the conditions reported by Osuka.9 We found that it is
essential to carry out this reaction at low temperature to limit
the oxidative dimerization of the porphyrin. The nitro group of
porphyrin 6 was then reduced by sodium borohydride in the
presence of palladium on carbon as catalyst. Finally, the
naphthalene monoimide (NI) was introduced by reacting amino
porphyrin 7 with naphthalene dianhydride in quinoline in the
presence of zinc acetate and gave the first dyad 1 in 86% yield.
The synthesis of dyad 2 involves the initial preparation of
naphthalene derivative 9, which was obtained through a Suzuki
cross-coupling reaction between naphthalene diimide 810 and
commercially available 4-bromo phenyl boronic acid. A second
Suzuki cross-coupling reaction of the latter compound 9 with
tetramethyl dioxaborolanyl porphyrin 1011 afforded the dyad 2
in a low yield (22%). Decomposition of the boronic ester of
the porphyrin 10 is certainly the main reason of this low yield,
The compounds in Scheme 2 were designed on the basis of
these ideas. The aromatic imides were chosen as acceptors
because they form stable anions with characteristic absorption
spectra, and it is possible to tune their reduction potentials by
extending the π-system or with substituents.7 Compound ZI is
a dyad consisting of a zinc porphyrin with a naphthalene imide

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1711
SCHEME 3: Synthetic Route for the Preparation of Dyads ZI (1) and DBZ (2)a
a Reagents and conditions: (i) AgNO2, I2, CH2Cl2/CH3CN, 0 °C, 47%; (ii) NaBH4, Pd/C, CH2Cl2/MeOH, room tempeature, 91%; (iii) 1,8-
naphtalic anhydride, xylene, imidazole, Zn(OAc)2, 4H2O, 130 °C, 88%; (iv) DMF/H2O, Pd(PPh3)4, K2CO3, 62%; (v) 9, DME/H2O, Pd(PPh3)4,
Ba(OH)2, 8H2O, 22%.
SCHEME 4: Synthetic Route for the Preparation of Dyad DPZ (3)a
a Reagents and conditions: (i) methane sulfonylchloride, CH2Cl2, Et3N, not isolated; (ii) LiBr, THF, reflux, overall 40%; (iii) PBr3, CH2Cl2, -20
°C, 90%; (iv) 13, K2CO3, DMF, 70 °C, 68%.
because a large amount of diarylporphyrin was formed during
this reaction.
shown in Scheme 5. Monobromoporphyrin 158 was reacted with
4-methoxy phenyl boronic acid to give trisarylporphyrin 16
following a Suzuki cross-coupling reaction in very good yield.
The resulting trisaryl porphyrin 16 was nitrated with the mixture
silver nitrite-iodine according to the conditions developed for
porphyrin 5. Reduction of the nitro group by sodium borohy-
dride and then imidation with naphthalene dianhydride afforded
porphyrin 18. Demethylation of methoxy group of 18 could be
achieved using boron tribromide at room temperature to afford
20 in 55% yield after purification. Finally, porphyrin 20 was
coupled with the naphthalene derivative 12 following a Will-
iamson reaction to give the expected triad 4.
The synthesis of dyad 3 (DPZ) was accomplished following
the route depicted in Scheme 4. The preparation of the key NDI
intermediate 12 starts from the unsymmetrical naphthalene
diimide 11,12 in which the alcohol group was first esterified by
methansulfonyl chloride and directly transformed into the bromo
derivative 12 in 40% overall yield. Porphyrin 13 was prepared
according to the Lindsey procedure13 and was subsequently
demethylated with boron tribromide at room temperature. The
phenol group of porphyrin 14 was deprotonated by potassium
carbonate in DMF to undergo a Williamson reaction with the
bromo group of the naphthalene derivative 12 to give the
expected dyad 3 in a 68% yield.
2.2. Absorption and Emission Spectroscopy. The absorp-
tion spectra in the 300-700 nm region were similar for all
compounds (Figure 1). In the visible region, the absorption
spectra in all solvents and for all compounds show characteristic
The preparation of the triad 4 (DPZI) was based on the
synthetic methodologies developed for the dyads 1-3 and is

1712 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
SCHEME 5: Synthetic Route for the Preparation of Triad DPZI (4)a
a Reagents and conditions: (i) 4-methoxybenzene boronic acid, DME, H2O, Pd2(dba)3CHCl3, PPh3, Ba(OH)2, 8H2O, 90 °C, then Zn(OAc)2,
CH2Cl2/MeOH, room temperature, 96%; (ii) AgNO2, I2, CH2Cl2/CH3CN, -20 °C, 70%; (iii) NaBH4, Pd/C, CH2Cl2/MeOH, room temperature,
94%; (iv) 1,8-naphtalic anhydride, xylene, imidazole, Zn(OAc)2, 4H2O, 180 °C, 67%; (v) PBr3, CH2Cl2, room temperature, 55%; (vi) 12, K2CO3,
DMF, 110 °C, then Zn(OAc)2, 4H2O, CH2Cl2/MeOH, room temperature, 18%.
indicates that there is no significant charge transfer character
of the excited state in DBZ or ZI. Thus, the initially formed
excited states of the dyads and the triad are the normal
porphyrin-localized states usually denoted S1 (Q-band excitation)
and S2 (Soret-band excitation).
The shape of the emission spectra also shows typical
porphyrin features with maxima at 640-650 nm and is similar
for all compounds and in all solvents. A weak S2 fluorescence
could also be seen very close to the Soret-band absorption.
The energy of the S1 and S2 states can be estimated from the
midpoint between the absorption and emission peaks for the
(0,0) transitions. For DBZ and DPZ, this gives an S1 energy
varying from of 2.11 in toluene to 2.07 eV in DMF, and an S2
energy from 2.93 to 2.92 eV. The corresponding values for ZI
and DPZI are 2.07-2.06 eV for S1 and 2.92-2.89 eV for S2.
There are some difficulties in measuring the position of the S2
emission peak due to reabsorption, and accurate measurements
Figure 1. Absorption spectra of ZnTPP (-), DBZ (- - -), and ZI
(×) in toluene. Inset: Zoom of the 520-620 nm region.
suggest that the Stokes shifts measured as done here are
overestimated by about 100 cm-1 15
. Thus, the S2 state energy
porphyrin features,2,14 that is, a Soret band located around 420
nm, which corresponds to a transition to the S2 state, and two
Q-bands between 540 and 600 nm that correspond to transitions
to the S1 state. Below 400 nm, some additional absorbance from
the imide and diimide units is observed. The absorption spectrum
of DPZ (not shown) is nearly identical to that of DBZ, and the
absorption spectrum of DPZI at λ > 400 nm (not shown) is
nearly identical to that of ZI.
As compared to ZnTPP, the Soret- and Q-band peaks are
red-shifted by ∼60 cm-1 in DBZ and DPZ and by 300-500
cm-1 in ZI and DPZI (see the Supporting Information; note
that the difference between the tri- and tetraaryl porphyrin
absorption in DBZ and DPZ is negligible). There is also a slight
increase in amplitude of the Q(0,0) transition in DBZ and DPZ
and a somewhat larger increase in ZI and DPZI. These effects
indicate a very small porphyrin-diimide interaction and a
somewhat stronger porphyrin-imide interaction. The effect on
the absorption spectra of changing solvent is small and equal
in all compounds over a large range of solvent polarities. This
could be underestimated only by about 0.01 eV, and for the
purpose of the present study this effect is not important.
The quantum yields for fluorescence varied greatly between
the compounds and with a change of solvent. Quantum yields
of fluorescence, relative to that for ZnTPP, are given in Table
1. Three different relative quantum yields are given: the S1
fluorescence yield with Q-band excitation and with Soret-band
excitation, and the S2 fluorescence yield with Soret-band
excitation. This can give important, complementary information
of the quenching pathways for the two states. The results for
DBZ show that in this compound there is little or no quenching
of the S2 fluorescence, while the S1 fluorescence is quenched
significantly in both toluene and DMF. The similar S1 fluores-
cence yields with Soret- and Q-band excitation show that S2 to
S1 internal conversion is near quantitative, as expected in the
absence of S2 quenching. The results for DPZ are very similar,
with only a somewhat stronger quenching of the S1 fluorescence.
The results for ZI are instead very different. The quantum yields

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1713
TABLE 1: Relative Fluorescence Quantum Yields with
ZnTPP as Standarda
compound solvent Ex S1, Em S1 Ex S2, Em S1 Ex S2, Em S2
DBZ
toluene
DMF
DMF
toluene
THF
butanol
ethanol
methanol
DMF
toluene
THF
butanol
DMF
0.37
0.26
0.15
0.79
0.87
0.03
0.03
0.03
0.12
0.53
0.54
0.05
0.07
0.32
0.27
0.14
0.54
0.78
0.03
0.02
0.02
0.10
0.51
0.59
0.03
0.06
∼0.8
∼1.0
DPZ
ZI
∼1.0
∼0.16
∼0.10
∼0.07
∼0.10
∼0.10
∼0.13
∼0.12
∼0.10
∼0.09
∼0.12
DPZI
a Excitation was in the maximum of either the Soret band (“Ex
S2”) or the Q-bands (“Ex S1”); emission was detected from the S2
and S1 states (“Em Ss” and “Em S1”, respectively).
Figure 2. Transient absorption spectra of DBZ in toluene after Soret-
band excitation at 422 nm at 0.2 ps (-), 20 ps (- · - · -), and 2 ns
(×).
show that the S2 fluorescence in ZI is quenched by about 90%
in all solvents, while quenching of S1 fluorescence is strongly
solvent dependent. In the alcohols, the S1 fluorescence is almost
completely quenched; in DMF, it is quenched by about 90%,
while there is very little quenching in THF and toluene. It is
interesting to note that while the S2 fluorescence is quenched,
the quantum yield for S1 fluorescence is nevertheless nearly the
same when exciting in the Soret and Q-bands. This shows that
most of the initially populated S2 state converts to the S1 state,
despite the S2 quenching reaction. The quenching of the S2 state
is therefore either because of an enhanced rate of direct internal
conversion to S1 or because it populates a state that in turn reacts
further with a high yield to form the S1 state. The transient
absorption data in section 2.4 show that the latter is the case.
For DPZI, the quenching of both the S2 and the S1
fluorescence in butanol is very similar to the results for ZI,
indicating that quenching in this solvent is mainly due to the
NI unit. In THF, the quenching of the S1 emission is somewhat
stronger in DPZI as compared to ZI, suggesting that there is
additional quenching by the NDI unit. Finally, in DMF the S1
emission yields for DPZ and ZI are quite similar, showing that
both the imide and the diimide quench the S1 excited state.
Consequently, two quenching pathways of similar importance
are expected for DPZI in this solvent. This gives the following
expected relation of the S1 fluorescence quantum yields (see
the Supporting Information):
Figure 3. Transient absorption traces of DBZ in toluene after Soret-
band excitation at 422 nm and probing at 450 nm (3), 475 nm (O),
and 645 nm (×). The lines are double-exponential fits to the data.
The S2 state has a spectrum similar to that for the S1 state,5 but
with somewhat lower amplitude. The S2 to S1 conversion is thus
seen as a general rise in the 425-600 nm region, especially in
the region of stimulated S2 emission (ca. 435 nm), and as a
development of the S1 stimulated emission band around 645
nm. The S1 state in the ZnTPP reference has a lifetime of ca. 2
ns and forms the triplet T1 state in a high yield (ca. 0.82a,14,16).
For DBZ instead the S1 state population converts to the CS state
with a time constant of 400 ps, as seen by the appearance of
the reduced diimide (NDI-) absorption peaks around 474 and
605 nm (ε ) 26 000 and 7200 cm-1 M-1, respectively)7 and
the ZnP+ absorption around 450 nm (Figure 2). These absorption
features then decay with a time constant around 10 ns, see
Figures 2 and 3, indicating charge recombination to re-form
the ground-state reactants. Nanosecond flash photolysis mea-
surements (not shown) reveal a biexponential decay of the
transient absorption around 475 nm with time constants of 12
ns and 0.35 µs. The 12 ns component is attributed to CS state
recombination, while the 0.35 µs component is attributed to the
decay of a fraction of the compounds that have undergone direct
intersystem crossing from the S1 to the T1 state, in competition
with charge separation. The 0.35 µs lifetime agrees with the
triplet lifetime of ZnTPP in aerated solution (0.42 µs). No
indication of triplet formation via charge recombination was
observed in this or any other experiment of this study.
1
1
φDPZ
1
φZI
)
+
- 1
(1)
φDPZI
Insertion of values for ΦZPD ) 0.15 and ΦZI ) 0.12 into eq 1
gives ΦDPZI ) 0.07. This is in excellent agreement with the
measured value, which indicates that the two acceptors in DPZI
react independently of each other, with the same rates as in the
corresponding dyads.
2.3. Transient Absorption Measurements on DBZ and
DPZ. Transient absorption measurements were performed to
observe the different charge-separated (CS) states and to follow
the dynamics of their formation and decay. The samples were
excited near the Soret- or Q(1,0)-band maximum, populating
the S2 and S1 state, respectively. In toluene, DBZ decays from
S2 to S1 with a time constant of 1.6 ps with no indication of a
CS state (Figure 2). These results are identical within experi-
mental uncertainty to that for the reference ZnTPP (see below).

1714 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
TABLE 2: Time Constants for Charge Separation (τET1
)
and Recombination (τCR1) after Excitation to the S1 State
compound
solvent
εsa
τET1 (ps)
τCR1 (ps)
DBZ
toluene
DMF
DMF
toluene
THF
butanol
ethanol
methanol
DMF
toluene
THF
butanol
DMF
2.4
400
350
250
b
b
28
12 × 103
37
37
2.4
8
16
24
33
37
c
DPZ
ZI
c
b
b
120
34
12
38
c
c
120
c
39
25
170
600d
400d
30e
120f
DPZI
2.4
8
16
37
a Static dielectric constant. b Not detected. c Not measurable, τCR
, τCS.
d Electron transfer to the NDI acceptor. e Electron transfer to
the NI acceptor. f Electron transfer to both acceptors in parallel.
The charge separation and recombination are much slower
than those of a very similar dyad where the NDI was linked
via a single phenyl group to the Zn-porphyrin, for which τET
) 9 ps and τCR ) 70 ps.9 The ca. 100-fold slower ET and CR
rates for DBZ are consistent with the expected rate decrease
with one extra phenyl group.17
In the more polar solvent DMF instead, the DBZ signals
decay monoexponentially from S1 to the ground state with a
time constant of about 350 ps, with no sign of a CS state. This
can be explained by forward electron transfer followed by fast
(<100 ps) recombination to the ground state. In both solvents,
the reduction in S1 lifetime as compared to the ZnTPP reference
agrees well with the reduction in fluorescence yield in Table 1.
The transient absorption data for DPZ in DMF are almost
identical to the data for DBZ. The only difference is that the S1
excited state lifetime is slightly shorter (250 versus 350 ps),
which is in agreement with the lower S1 fluorescence quantum
yield (0.15 versus 0.26). Again, we interpret the data in terms
of charge separation followed by a much faster recombination.
Rate constants for electron transfer from the S1 state are collected
in Table 2.
2.4. Transient Absorption Measurements on ZI. The
absorption spectrum of the reduced imide (NI-) has a strong
peak (ε ) 23 500 cm-1 M-1) around 420 nm.7 Although this
overlaps with the ground-state porphyrin Soret band, the latter
is narrow so it should be possible to detect the NI- on the blue
edge of the Soret-band bleach where the porphyrin S2 and S1
state transient absorption is weak. Transient absorption spectra
in methanol at 0.2, 1, and 20 ps after excitation are shown in
Figure 4 for ZnTPP exciting to the S2 state (upper panel), ZI
exciting to the S1 state (middle panel), and ZI exciting to the
S2 state (lower panel). For ZnTPP, as for DBZ above, the S2 to
S1 conversion is seen mainly as an absorption rise in the
425-600 nm region and the appearance of stimulated S1
emission at 640 nm (compare spectra at 0.2 and 20 ps). The
ZnTPP S2 lifetime is 1.5-2.3 ps in the solvents examined. In
contrast, there are important differences between the transient
absorption spectra of ZI and ZnTPP. First, the appearance of a
peak around 405 nm due to NI- for ZI is a clear sign of a
charge-separated (CS) state ZnP+NI- in this compound. This
is seen already in the 0.2 ps spectrum after S2 excitation and
on the tens of picoseconds time scale after S1 excitation. Second,
the transient signal amplitudes at 400-540 nm for ZI have
decreased significantly already after 20 ps for both S2 and S1
excitation. This is consistent with formation of the CS state
Figure 4. Transient absorption spectra in methanol at 0.2 ps (-), 1 ps
(- · - · -), and 20 ps (×). Upper panel: Soret-band excitation of
ZnTPP at 416 nm. Middle panel: Q-band excitation of ZI at 559 nm.
Lower panel: Soret-band excitation of ZI at 420 nm. The insets show
the Soret-band bleach region.
species, as the ZnP+ radical shows an absorption in this region
that is similar to but weaker than that for the S2 and S1 states.18
After S1 excitation (Figure 4, middle panel), it is also clear that
the S1 stimulated emission around 660 nm disappears as the
CS state is formed. The transient bleach in the Soret and
Q-bands has decreased only slightly after 20 ps, and for the
former band this can be explained mainly by the absorption of
ZnP+ being higher than for the excited states18 and by contribu-
tion from NI-. Thus, repopulation of the ground state occurs at
most to a minor extent on this time scale. The rapid formation
of the 405 nm peak after S2 excitation, on a 100-200 fs time
scale, is observed in all solvents except toluene, while no such

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1715
Figure 6. Transient absorption traces of ZI in methanol after Soret-
band excitation at 420 nm and probing at 405 nm (0), 428 nm (*),
449 nm (O), and 643 nm (+). The left half of the figure shows data
for the first picosecond, and the right half shows data for 1-80 ps.
Solid lines are from a global fit to the data (see text).
Figure 5. Transient absorption spectra of ZI around the NI- peak at
405 nm after excitation at 420 nm. Spectra are shown at 0.3 ps (-),
0.7 ps (- · - · -), and 3.5 ps (×). The panels show data in (from left
to right) butanol, DMF, and THF.
signal is seen during the first picosecond after S1 excitation.
This shows that there is formation of CS state intermediates
already from the S2 state in ZI.
porphyrins is very low (<2%) in these compounds. The kinetics
also agrees with the amount of CS species formed as can be
estimated by the maximum amplitude of the peak around 405
nm. The maximum amount of CS species formed in the different
solvents decreases in the order CS (butanol) > CS (ethanol) >
CS (methanol) > CS (DMF), which is what would be expected
from the recombination rates that increase throughout this series
of solvents. In toluene and THF, no electron transfer from S1
could be detected on the time scale of experiment (<6 ns). This
is consistent with the low degree of emission quenching (Table
1) and suggests that the CS state energy lies close to or even
above that of the S1 state in these less polar solvents.
The results for S2 excitation of ZI are rich and somewhat
complex. In the remainder of this section, we go through these
in some detail, but the important conclusions drawn are
summarized in the last paragraph of section 2.4. The analysis
of the reactions starting from the S2 state is complicated by the
fact that at early times there is a signal from the solvent and
also by the fact that solvent dynamics and/or vibrational
relaxation gives subpicosecond and picosecond dynamics ob-
served at many wavelengths also in ZnTPP. However, the data
at 428, 450, and 650 nm on a t < 3 ps time scale could be fairly
well fitted by a single exponential function convoluted with the
experiment response function. Data for ZI in methanol are
shown in Figure 6. A global fit at these wavelengths gave a
time constant that ranged from 0.2 ps (in methanol, ethanol,
and THF) to 0.3 ps (in butanol) and 0.4 ps (in DMF). These
time constants are assigned to S2 decay as supported by the
concomitant recovery at 428 nm (disappearance of S2 emission),
absorption increase at 450 nm (the S1 maximum), and the
appearance of S1 stimulated emission at 650 nm. These time
constants are much shorter than for S2 decay in ZnTPP, for
which we measured values ranging from 1.3 ps in DMF to 1.9
ps in methanol. The reduction in time constants is in agreement
After S2 excitation in THF, the NI- peak at 405 nm decays
rapidly, and within a few picoseconds the spectrum shows only
the S1 state with a lifetime of ca. 2 ns. The high amplitude of
the S1 absorption formed after S2 excitation, and the nearly
identical S1 fluorescence yield after S1 and S2 excitation (Table
1), show that the CS state species formed after S2 excitation
recombine to populate the S1 state in very high yield. In the
alcohols and DMF instead, the 405 nm peak formed after S2
excitation does not decay completely, but reaches a minimum
when the S1 state is populated (Figure 5). The 405 nm minimum
occurs after ca. 1 ps in methanol when the spectrum shows the
strong S1 absorption around 460 nm and the stimulated emission
at 600 nm (Figure 4, lower panel). In these more polar solvents,
charge separation occurs also from the S1 state so that the 405
nm peak increases again. The spectrum after 20 ps in methanol
shows the maximum level of charge separation in this solvent
before the CS state species recombine to the ground state. Note
that the 20 ps spectrum, and the spectral evolution kinetics after
ca. 5 ps, are very similar for S2 and S1 excitation.
The reaction kinetics starting from the S1 state was analyzed
with a global fit of the kinetic traces at four characteristic
wavelengths: 405 nm (imide radical), 428 nm (Soret bleach +
stimulated S2 emission), 450 nm (S1 maximum, also S2 and ZnP+
but with lower extinction coefficient), and 650 nm (stimulated
S1 emission). A biexponential function could be fitted very well
to the data, representing the charge separation from the S1 state
and subsequent recombination to the ground state. As S2
excitation leads to near quantitative S1 population, the results
after ca. 5 ps are very similar for S1 and S2 excitation, and the
right half of Figure 6 shows data at t > 1 ps after S2 excitation
in methanol. The time constants in all solvents are given in Table
2. The S1 lifetime relative to that of the ZnTPP reference agrees
very well with the relative S1 fluorescence yield after S1
excitation in Table 1, that is, τ/τZnTPP ) Φfl/Φfl, ZnTPP. The relative
emission yields indicate an ET time constant of around 40 ps
in the alcohols and 160 ps in DMF, assuming that the intrinsic
S1 lifetime in unaffected by the imide substitution. The good
agreement with the steady-state data gives support for simple
irreversible, single-exponential quenching kinetics and also
means that the level of impurities in the form of unquenched
with the reduction of S2 emission yield in Table 1 (τ/τZnTPP
)
Φem/Φem,ZnTPP), where the emission yield in, for example,
methanol indeed predicts a S2 lifetime in ZI of 0.2 ps, as
observed. Note that a variation of laser pump intensity by a
factor of 3 did not lead to detectable changes in the dynamics.
The kinetics for the NI- absorption at 405 nm instead shows
a rise time close to the instrumental response time, followed
by a partial decay with the same time constant (0.2-0.4 ps) as
obtained at the other wavelengths shown in Figure 6. Although

1716 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
ultrafast recombination observed after S2 excitation in the same
solvents must occur from vibrationally excited CS states. This
recombination is not quantitative in the polar solvents, however,
as a part of the 405 nm signal remains after a few picoseconds.
It thus appears that relaxation of the vibrationally excited CS
state competes with recombination and traps a fraction of the
molecules in the CS state. The long survival time of this fraction
is consistent with the fact that the relaxed CS state lies below
the S1 state in the polar solvents. In low polarity solvents (THF),
recombination from the CS to the S1 state is instead quantitative,
and no CS state is formed after S1 excitation (in THF and
toluene; see above). This is consistent with the thermalized CS
state lying above S1 in these solvents. The reactions can be
summarized by the scheme of Figure 7. For simplicity, first-
order rate constants are assumed for all reactions, but the
reactants are not thermalized and may give more complex
reaction dynamics.
The absorption of NI- at 405 nm that remains after a few
picoseconds gives directly the fraction of molecules trapped CS
in states and thus information on the relative rates of “hot”
recombination and vibrational relaxation (after correction for
the fraction of molecules in the S2 state that undergoes direct
internal conversion to S1). Figure 5 shows the 405 nm peak
region for ZI in three different solvents at 0.3, 0.7, and 3.5 ps
after time zero. In butanol, there is a small decay of the CS
state population on these time scales, in DMF a larger decay,
and in THF an almost complete decay. Note that no wavelength
shift or broadening of the 405 nm peak was observed in any of
the experiments, which could suggest that the excess vibrational
energy is distributed over modes that are not coupled to the
NI--localized transition.
The maximum concentration of CS species, relative to the
initial S2 concentration [S2]0, gives information on the rate
constants of Figure 7. These values are collected in Table 3,
estimated as described in the Experimental Section. The
concentration of CS state species is given at three different times.
First, the maximum concentration of CS species formed from
the S2 state, denoted [ZnP+NI-]max/[S2]0, is estimated from the
405 nm signal as soon as the oscillatory signals around time
zero are finished (eq E6). Second, the concentration of trapped
CS species after “hot” recombination, denoted [ZnP+NI-]int/
[S2]0, is estimated from the intermediate minimum at 1-3 ps
after excitation, before electron transfer from the S1 state
becomes significant (eq E7). Third, the maximum at long times
(20-100 ps) is given, denoted [ZnP+NI-]long/[S2]0, which is
when formation of CS species by electron transfer from the S1
state maximizes. Note that these values are not overall quantum
yields, but momentary concentrations of an intermediate in
consecutive reactions; the yield of CS from S2 is, for example,
ca. 90% in the alcohols and THF (kET ≈ 10 × kIC). There are
of course uncertainties in these estimates, especially for the first
two where the porphyrin background signal also changes on
these time scales. Also, the initial maximum appears very early
after excitation and may not have been clearly resolved in all
Figure 7. Electron transfer reaction scheme for ZI in polar solvents.
In THF and toluene, the relaxed CS state ZnP+NI- lies above the S1
state.
this seems to contradict our assignment of the 0.2-0.4 ps
component to S2 decay by electron transfer, it can be explained
by so-called inverted kinetics; the initially formed CS species
undergoes partial recombination to populate the S1 state with a
time constant that is shorter than its formation time constant,
and close to our time resolution: S2 w CS w S1. At wavelengths
where the CS absorption dominates, its apparent rise time is
given by the recombination time constant, while its decay is
given by the formation time constant.19 For a general, consecu-
tive mechanism A w I w C, with rate constants k1 and k2 for
the first and second steps, the concentration of the intermediate
I is given by:
k1
[I]
[A]0
1
2
)
(e-k t - e-k t
)
(2)
(k2 - k1)
The maximum concentration of the intermediate, [I]max, and the
time of appearance of this maximum, tmax, are given by standard
expressions for consecutive, first-order reactions (see Experi-
mental Section, eqs E4 and E5).
At wavelengths where the S2 and S1 states dominate the
signals, for example, at 428, 449, and 643 nm in Figure 6, the
intermediate CS state is not seen, and the absorbance change
in the traces reflects the S2 lifetime. Thus, the decay-associated
spectrum of the 0.2 ps component in methanol is very similar
to the corresponding spectrum for S2 to S1 conversion in the
reference ZnTPP (see the Supporting Information).
The recombination to populate the S1 state was at first a
surprising finding, as excitation of the S1 state actually leads to
charge separation on a longer time scale in the more polar
solvents (see above). Thus, the relaxed CS state ZnP+NI- must
lie below the S1 state in the alcohols and DMF. Therefore, the
TABLE 3: Relative Concentration of ZnP+NI- State Species Formed in ZI at Different Stages of Reaction after S2 Excitation,
and Rate Constants for the Ultrafast Charge Separation and Recombination Process (For Definitions, See Text and Figure 7)
solvent
[ZnP+NI-]max/[S2]0
[ZnP+NI-]int/[S2]0 (time)
[ZnP+NI-]long/[S2]0 (time)
k1 /×1012 s-1
k2 /×1012 s-1
kVR/×1012 s-1
a
b
THF
0.13
0.24
0.27
0.18
0.16
0
0
5
3.3
5
5
2.5
∼15
<0.5
∼1.5
∼2.5
∼2.5
∼0.7
buthanol
ethanol
methanol
DMF
0.18 (1.5 ps)
0.22 (1 ps)
0.14 (1 ps)
0.07 (3 ps)
0.64 (50 ps)
0.45 (25 ps)
0.29 (20 ps)
0.19 (100 ps)
∼6
∼8
∼14
∼7
a k1 ) kET + kIC; kIC ) (0.4-0.7) × 1012 s-1
.
b k2 ) kCR + kVR
.

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1717
solvents. Nevertheless, the results should be sufficiently precise
to verify the essential features of the scheme in Figure 7, and
to discuss trends in reactivity between the different solvents
(see Discussion).
To summarize, electron transfer in ZI occurs on the 0.2-0.4
ps time scale from the S2 state, and a CS state was seen in all
solvents except toluene. Very rapid recombination (τ e 0.2 ps)
from the vibrationally excited CS state follows, to form the S1
state in near-quantitative yield. This conclusion agrees with the
emission yield data in Table 1. Further electron transfer occurs
from the S1 state to the relaxed CS state on the time scale of
tens to hundreds of picoseconds in the more polar solvents, that
is, the alcohols and DMF. This is a highly unusual reaction
sequence where the reacting species go back and forth several
times between locally excited ZnP states and CS states, as
schematically shown in Figure 7.
Figure 8. Energy diagram showing the porphyrin first (S1) and second
excited (S2) states as well as the charge-separated (CS) states (see
Experimental Section for details).
2.5. Transient Absorption Measurements on DPZI. As for
the dyads, the reactions of the DPZI triad are strongly solvent
dependent. After excitation to the S1 state, the reaction kinetics
in the alcohols is very similar to that of ZI (see the Supporting
Information). This means that the electron transfer is dominated
by the NI acceptor as could be expected from the fast electron
transfer in ZI (Table 2) as compared to DPZ. In THF and
toluene, instead most of the signal (∼70%) decays monoexpo-
nentially on a subnanosecond time scale (400 ps in THF and
600 ps in toluene) with no observable signal from either the
imide or the diimide radical. The remaining signal is assigned
to the porphyrin triplet state that is formed in parallel to the
electron transfer, as expected from the moderate S1 quenching
efficiency in these solvents (τ/τZnTPP ≈ 1/3). The transient data
and S1 lifetime in toluene are identical to those for DPZ. No
electron transfer was observed from the S1 state of ZI in these
solvents. This indicates that the quenching of the S1 excited
state of DPZI in THF and toluene occurs only via the NDI
acceptor and that the absence of an observable CS state is due
to fast recombination, as for DPZ. The more flexible link to
the NDI acceptor in DPZ and DPZI as compared to DBZ
is the likely reason for rapid recombination also in toluene,
where the CS state was rather long-lived in DBZ. Finally, in
DMF the S1 state population decays monoexponentially with a
time constant of about 120 ps, which gives a decay rate that is
equal to the sum of the rates in DPZ (τ ) 250 ps) and ZI (τ )
170 ps). This shows that in this solvent the two acceptors provide
independent reaction channels, as was also suggested by the
steady-state fluorescence yield. Upon excitation to the S2 state,
the behavior of DPZI in all solvents closely resembles that of
ZI (cf., Figures 4-6; see the Supporting Information). That is,
the S2 state population decays with τ ) 0.2-0.4 ps, and there
is a CS state visible around 405 nm in butanol and DMF that
undergoes ultrafast recombination to the S1 state. On longer time
scales, further charge separation occurs from the S1 state, as
described in the previous paragraph.
recombination from “hot” CS states to populate S1, which re-
forms the same CS state but this time thermally relaxed.
3.1. Solvent Dependence of Electron Transfer. The elec-
tron transfer rates presented here vary between the solvents
employed; data are collected in Tables 2 and 3. Commonly,
the solvent dependence of the electron transfer rate constant
kET is rationalized by the effect of solvent polarity on the driving
force (-∆G0), on one hand, and the reorganization energy (λ),
on the other, in the framework of Marcus theory:17b
2π
p
1
(∆G0 + λ)2
4λkBT
kET
)
V2fi
exp -
(3)
[
]
4πλk T
√
B
where Vfi2 is the electronic coupling (the effect of nonclassical,
intramolecular acceptor modes was ignored in the following
qualitative discussion). The energies for the porphyrin S1 and
S2 states are 2.1 and 2.9 eV, respectively (see section 2.2). The
free energy of the charge transfer state relative to the ground
state can be estimated from the donor and acceptor redox
potentials and standard dielectric continuum models (see
Experimental Section).20 These procedures are approximate for
real molecules in molecular solvents and often overestimate CS
state energies, especially in low polarity solvents.6a,c,21 For
example, the ZnP+NI- state is estimated to lie above the S1
state in all solvents, although we see rapid and complete electron
transfer in the alcohols and DMF. On the basis of the observed
electron transfer reactivity, we adjusted the CS state estimates
to obtain the approximate energy diagram of Figure 8. Note
that the purpose of the study is not to investigate precise rate
versus free energy relationships; the solvent variation is rather
used as a tool to manipulate the reactivity. In this context, Figure
8 is given as a semiquantitative tool to following the discussion.
Note also that the ultrafast electron transfer from the S2 state
may occur from a reactant state that is not in thermal equilib-
rium, and could also be in the adiabatic regime, so that eq 3
may not be valid. Therefore, we discuss in the following
paragraph mainly electron transfer from the S1 state (kET1) and
subsequent recombination (kCR1). Nevertheless, the charge
3. Discussion
As was seen in the Results, the electron transfer processes
depended quite significantly on the solvent. This dependence
is then an important tool for controlling the directionality of
electron transfer from the S2 and S1 states and thus obtaining
the desired switch function described in the Introduction. First,
we discuss the solvent dependence of the electron transfer
energetics and kinetics. We then continue by discussing the state
selectivity of electron transfer in the DPZI triad, and its behavior
as a molecular switch. Finally, we discuss the unusual electron
transfer sequence starting from the S2 state, that involves
recombination rate constant following S2 electron transfer (kCR
)
shows the same trend in different solvents as does kCR1
.
The variation of CS state energy with solvent gives us a
possibility to tune the reactions. We therefore measured the
electron transfer rates in a range of solvents, in particular for
reactions with the NI unit as acceptor where tuning was

1718 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
particularly important. First, the rate of the forward electron
transfer varies only little with solvent polarity. For example,
the rate for the S1 electron transfer is almost the same for DBZ
in toluene and DMF, and the same for ZI in butanol and
methanol. In contrast, the corresponding charge recombination
rates vary greatly. This can be explained by the fact that the
reorganization energy varies with solvent polarity in a very
similar way as the driving force in eqs E2 and E3 (see
Experimental Section). For a reaction in the Marcus normal
region (-∆G0 < λ), as in the present charge separation reactions,
these effects nearly cancel each other. For a reaction in the
inverted region (-∆G0 > λ) instead, as in the present recom-
bination reactions to the ground state, the reaction becomes even
more inverted and thus slower in a nonpolar solvent. Conse-
quently, we see that the recombination reaction in ZI is
progressively slower in the series of alcohols methanol (τ )
12 ps), ethanol (τ ) 34 ps), and butanol (τ ) 120 ps); see Table
2. Likewise, recombination in DBZ is much faster in DMF
(τ < 100 ps) than in toluene (τ ) 12 ns). We note, however,
that in DMF both charge separation and recombination in ZI
are much slower than in the less polar methanol. Charge
separation in DMF is even slower than in butanol. This can
obviously not be explained by a dielectric continuum model. A
possible explanation could be the presence of hydrogen bonds
between ZI and the alcohols that accelerate electron transfer.
This has been suggested previously in a fluorescence upcon-
version study of electron transfer in similar compounds.6c
excitation led to ultrafast, near-quantitative electron transfer to
give the ZnP+NI- state, whereas S1 excitation only produced
the NDI-ZnP+ state. This agrees with the desired function of
the switch. The NDI-ZnP+ state could not be detected,
presumably because of rapid recombination, but this would be
solved by using the more rigid bridge of DBZ instead of DPZ
in a new triad, as DBZ showed a long CS lifetime in a nonpolar
solvent. More problematic, however, is the (nearly) quantitative
population of the S1 state from the ZnP+NI- state (Figure 7).
This means that on a longer time scale charge separation in
DPZI to form NDI-ZnP+ occurs irrespective of excitation
wavelength, which limits reaction selectivity. Also, only a small
amount of the ZnP+NI- intermediate could be detected in THF,
and none in toluene. DPZI in THF and toluene shows some
switching behavior in that Soret-band excitation gives a
ZnP+NI- population on very short time scales (0.1-1 ps) and
NDI-ZnP+ on a longer time scale (ca. 50 ps), while Q-band
excitation only gives the latter. If we synthesized a new triad
based on DBZ instead of DPZ, we would obtain a very similar
behavior but would also be able to directly detect both of the
CS states. This triad would still give recombination to S1 after
S2 excitation, however, and therefore limited reaction selectivity.
Instead, we hope to improve on the function by modifying the
molecular design, which is currently in progress.
3.3. Ultrafast Recombination from “Hot” CS States. The
ultrafast recombination from the unrelaxed ZnP+NI- state
formed after S2 excitation was not expected. It led to a low
maximum concentration of detectable ZnP+NI- states and to
rapid formation of the S1 state. Therefore, it also led to loss of
electron transfer selectivity upon S2 excitation, as noted above.
No porphyrin triplet formation from CS state recombination
was observed in any of the experiments.
3.2. Discussion of the Switch Function. For the switch
function described in the Introduction, that is, the control of
electron transfer direction with different excitation wavelengths,
the results were a partial success. The intrinsic properties of
the porphyrin units were perturbed very little by the attachment
of either acceptor unit, as seen by the close similarity of the
ground-state absorption spectra and by the fact that the two
porphyrin-acceptor couples of the triad molecule reacted rather
independently of each other. The latter was shown by compari-
sons of reaction rates and fluorescence yields with the corre-
sponding dyads, showing that the reaction rate of each
porphyrin-acceptor unit was the same in the DPZI triad as in
the corresponding DPZ and ZI dyads. These results strongly
suggest that attachment of the NI unit also does not significantly
reduce the intrinsic porphyrin S2 lifetime and that the ultrafast
electron transfer alone is responsible for the reduced S2 lifetime
in ZI and DPZI. The conservation of intrinsic properties and
the resulting independence of the two porphyrin-acceptor units
facilitate predictable design of a functional triad switch.
Rapid charge recombination to populate the S1 state after S2
excitation has been suggested before in other pophyrin-acceptor
systems, where the relaxed CS state energy was between that
of the S1 and S2 states. The first report was by Vauthey and
co-workers, who investigated S2 quenching of ZnTPP by added
trimethoxyebenzene or acetophenone acceptors using fluores-
cence upconversion.6d The S1 fluorescence rise matched that of
the S2 fluorescence decay, and the intensity of the former showed
that S1 population was quantitative. The results suggested for
the first time ultrafast recombination of the CS state to
repopulate S1, and, because of the short time scale, involvement
of hot CS states was proposed. Mataga, Osuka, and co-workers
have, as part of their extensive investigations of electron transfer
from the porphyrin S2 states, suggested recombination to S1 from
unrelaxed CS states in a series of porphyrin-imide, and
-diimide dyads, based on the fact that in some cases there was
no time delay between the decay of S2 fluorescence and rise of
S1 fluorescence.6a-c If the recombination rate from the relaxed
CS state is much faster than the initial electron transfer, however,
the same result can in principle be obtained with recombination
from only relaxed CS states. Ionkin and Ivanov have simulated
the experimental data of Mataga and co-workers with a model
involving “hot” CS states.22
We could indeed tune the electron transfer rates to a large
extent by changing the solvent (Tables 2 and 3). Moreover, the
more weakly coupled, but easily reduced acceptor NDI did not
quench the S2 state in any of the solvents, which was one of
the required design principles. Instead, NDI quenched the S1
state in all solvents, although it seems that in polar solvents
recombination was faster that charge separation. Thus, the
NDI-ZnP+ state was detected for DBZ in toluene but not in
DMF. The more strongly coupled acceptor NI, on the other
In the present study, we use transient absorption and directly
observe the NI- signal rise and decay during the sub-ps
reactions, also in cases where the relaxed CS state lies clearly
below the S1 state. The transient absorption data show near-
quantitative population of the S1 state on the time scale of
recombination, as judged from the magnitude of the Soret- and
Q-band bleach and the absorption at 460 nm. The S1 fluores-
cence yields are furthermore the same with Soret- and Q-band
excitation, despite strong S2 fluorescence quenching, showing
that S1 is (nearly) quantitatively generated from the quenched
S2 state. Most importantly, the further formation of the CS state
hand, quenches the S2 state efficiently in all solvents (kET
.
kIC), which fulfills another of the design criteria. In the more
polar solvents, however, the NI acceptor dominated the reaction
from both the S2 and the S1 states. This means that the desired
selectivity with Q-band excitation leading to formation of the
NDI-ZnP+ state could not be achieved in these solvents. In the
less polar solvents THF and toluene, on the other hand, S2

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1719
species from the relaxed S1 state in the more polar solvents
shows that the initial CS state populated in the S2 reaction, which
recombines to give S1, lies at higher energy than the relaxed
CS state. This gives for the first time direct and unambiguous
evidence for recombination from unrelaxed CS states to the S1
state after S2 excitation of a porphyrin-acceptor molecule. The
resulting reaction sequence is highly unusual, possibly unprec-
edented. The electron transfer reactions go back and forth
between locally excited ZnP states and charge-separated states
change-separated states in directly detectable concentrations. We
obtained direct spectroscopic evidence that charge recombination
after S2 electron transfer occurs from “hot” ZnP+NI- states, to
populate the porphyrin S1 state. In polar solvents, the S1 state
then give further electron transfer to re-form the same ZnP+NI-
charge-separated state species but this time as thermalized
products, which eventually recombine to the ground state. This
is the first direct evidence for “hot” charge recombination to
the S1 state after S2 excitation in porphyrin-acceptor systems.
The result is an unusual “ping-pong” sequence where the
reaction goes back and forth between locally excited ZnP states
of the dyad ZI: S2 w ZnP+NI-
w S1 w ZnP+NI- w S0
“hot”
(see also Figure 7).
In this context, it is interesting to compare with the recent
report of Petersson et al. on ultrafast electron transfer in a water-
soluble Zn(II) porphyrin/viologen complex.23 The CS state was
in that case ca. 0.5 eV below the S1 state. Both S2 and S1
excitation gave rapid (τ < 200 fs) formation of CS species that
recombined with τ ) 0.7 ps. In this case, recombination did
not occur to form the S1 state but a distribution of relaxed and
vibrationally excited ground states. S2 and S1 excitation resulted
in different yields of the latter product, which was evidence
that recombination occurred from unrelaxed CS states, where
the excess energy depended on the initial excited state.
and charge-separated states: S2 w ZnP+NI-
w S1 w
“hot”
ZnP+NI- w S0, as illustrated in Figure 7. Finally, the solvent
polarity has only a small effect on the rates of the charge
separation reactions, while the recombination reactions are much
more affected. For both types of reactions, however, the rate is
much higher in the alcohols than in the non-hydrogen bonding
solvents of the same polarity, as also noted in related cases.6c
Overall, the marked solvent dependence is an important tool
for control of the electron transfer switch function, but is also
a challenge if one attempts to incorporate a switch in a different
environment.
In the present study, also the rates of “hot” recombination
(kCR) and vibrational relaxation (kVR) in the ZnP+NI- state were
solvent dependent. We cannot clearly resolve the rise of the
405 nm signal to directly determine k2 (k2 ) kCR + kVR) of the
reaction scheme in Figure 7. Nevertheless, we can make some
estimates from the absorption magnitude of the initial maximum,
giving [ZnP+NI-]max/[S2]0 (Table 3), and from the rate of S2
deactivation (k1 ) kET + kIC), using eq E3. The results are
presented in Table 3, and although they are approximate we
can observe some trends. First, the rate constant k2 is dominated
by kCR that increases with polarity in the series of alcohols, but
is smaller again for DMF. This trend is parallel to that for
recombination of the thermalized ZnP+NI- state formed via S1
(Table 2). We note, however, that recombination (kCR) is
relatively rapid in the less polar solvent THF. We speculate
that this could be related to the fact that the relaxed CS state is
higher in energy in this solvent, probably even somewhat higher
than the S1 state. Second, we note that the vibrational relaxation
rate (kVR) is rather constant in the series of alcohols, but slower
in DMF. This suggests a significant solute-solvent coupling.
Vauthey and co-workers reported a faster VR of excited perylene
derivatives in alcohol solvents and attributed this to solvent-solute
hydrogen bonding and thus a larger number of vibrational
degrees of freedom.24 Kovalenko et al. found a faster cooling
of para-nitroanilines in alcohols and attributed this to solvent-
solvent vibrational energy transfer via hydrogen bonds.25 The
higher kVR in alcohols observed in the present study may be
related to the faster electron transfer from S1 in the alcohols
than in DMF (see above). Further studies with better time
resolution and smaller contributions from solvent signals in the
transient absorption are needed to give more precise information
on this ultrafast dynamics.
5. Experimental Section
5.1. Calculation of the Charge-Separated State Energies.
The free energies of the CS states relative to the ground state
were calculated from eq E5:20
∆G0CS ) e(E0(D+/0) - E0(A0/-)) + w + C
(E1)
where E0(D+/0) and E0(A0/-) are the reduction potentials of the
donor-acceptor, w is the electrostatic interaction between the
product ions (the reactants are neutral), and C is a term
correcting for the difference in ion solvation energy between
the solvents used for the reaction and for the reduction potential
determination. The work term can be estimated as the potential
energy of two spheres in a dielectric medium:
w ) -e2/(4πεsε0d)
and the correction term C by:
e2
(E2)
1
1
1
1
εr
C )
+
-
(E3)
(
)(
)
8πε0 rd
ra εs
where εs is the solvent dielectric constant, εr the dielectric
constant of the solvent in which the electrochemical potentials
were determined, d is the distance between the ions, and rd and
ra are the ionic radii of the donor and acceptor. The following
redox potentials were used: E0(NDI0/-) ) -0.48 V in buty-
ronitrile,7 E0(NI0/-) ) -1.37 V in DMF,7 and E0(ZnTPP+/0) )
+0.86 V in DCM26 (all potentials are given vs SCE). The
effective radii were estimated to 5.0 Å (ZnP unit), 3.5 Å (NDI),
and 3.0 Å (NI); cf., refs 6b, 6c. The center-to-center distances
were 11.5 Å for ZI and 20 Å for DBZ and DPZ. We thus
calculate the energy of the ZnP+NI- and NDI-ZnP+ states
ranging from 2.10 and 1.22 V, respectively, in the most polar
solvent DMF to 3.0 and 2.27 V in toluene. Clearly, this
procedure overestimates the CS state energies; although the
ZnP+NI- state is estimated to lie above the S1 state in all
solvents, we observe rapid electron transfer from the S1 state
4. Conclusions
We were able show that the porphyrin S1 and S2 states of the
unsymmetrical NDI-ZnP-NI triad DPZI gave electron transfer
in opposite directions, in the less polar solvents. This may
constitute a basis for an opto-electronic switch in which the
direction of electron transfer and the resulting dipole moment
can be controlled by using light input of different color. Further
molecular design is in progress to avoid repopulation of the S1
state after electron transfer from the S2 state, and to obtain both

1720 J. Phys. Chem. A, Vol. 114, No. 4, 2010
Wallin et al.
to produce ZnP+NI- in the alcohols and DMF. In THF and
toluene, on the other hand, recombination of the ZnP+NI- state
appears to populate the S1 state nearly quantitatively, while no
or only little quenching of the S1 state is seen in ZI. These
results suggest that eqs E5-E7 overestimate the ZnP+NI- state
energy by about 0.2-0.3 eV in most solvents and probably even
more in the very nonpolar solvent toluene. The error in the
NDI-ZnP+ energy may be somewhat smaller because the longer
distance makes at least the work term less significant. Neverthe-
less, the relaxed CS state energy in toluene is estimated to lie
at 2.27 eV, while the rapid electron transfer from the S1 state
suggests that it lies at least 0.25 eV lower. An approximate
energy diagram based on these corrections is given in Figure
8. The ZnP+NI- energies in Figure 8 are estimates down-shifted
by ca. 0.25 eV (0.5 eV in toluene) from the values calculated
with eqs 8-10, to agree with the results (see text). The
NDI-ZnP+ energies are downshifted by 0.1 eV (0.25 eV in
toluene). It is well recognized that eqs E5-E7 are only
approximations for real molecules in molecular solvents and
that they tend to overestimate CS state energies in less polar
solvents.6a,c,21
should be given by eq 2, with k1 ) kET1 and k2 ) kCR1 (Figure
7). This gives the long time maximum [ZnP+NI-]long/[S2]0, after
consideration of remaining fraction created from the S2 state
([ZnP+NI-]int). Thus, the maximum yield of the CS state
ZnP+NI- population for ZI in, for example, butanol can be
estimated to 0.64 (obtained after ca. 50 ps).
For all estimates of [ZnP+NI-], the magnitude of the 405
nm transient absorption peak was taken relative to the back-
ground in this spectral region, as the porphyrin unit in general
gave a bleach in this region.
1
5.3. General Methods. H NMR spectra were recorded on
an Avance 300 MHz or ARX 400 MHz Bruker spectrometer.
1
Chemical shifts for H NMRspectra are referenced relative to
residual protium in the deuterated solvent (CDCl3, δ ) 7.26
ppm, CD3C(O)CD3, δ ) 2.09 ppm). MALDI-TOF mass spectra
were recorded on a Bruker Biflex III MALDI-TOF mass
spectrometer. The instrument is equipped with a nitrogen laser
emitting at 337 nm, a 2-GHz sampling rate digitizer, a pulsed
ion extraction source, and a reflectron. The matrix was dithranol.
Electrospray mass spectra (ESMS) were recorded on a MS/MS
ZABSpec TOF (Micromass). The samples were injected in
chloroform solution, and the ions were accelerated with a 4000
V voltage. Polyethylene glycol (PEG) was used as an internal
reference.
5.2. Details of Kinetic Analysis for ZI in Section 2.4. For
a general, consecutive mechanism A w I w C, the maximum
concentration of the intermediate, [I]max, and the time of
appearance of this maximum, tmax, are given by standard
expressions for consecutive, first-order reactions:19
Thin-layer chromatography (TLC) was performed on alumi-
num sheets precoated with Merck 5735 Kieselgel 60F254
.
Column chromatography was carried out either with Merck 5735
Kieselgel 60F (0.040-0.063 mm mesh). Air-sensitive reactions
were carried out under argon in dry solvents and glassware,
and the solvents were distilled prior uses. Chemicals were
purchased from Aldrich and used as received. 5,15-Bis-(3,5-
di-tert-butylphenyl)-10-phenyl-porphyrinato zinc(II) 5,8 N,N′-
octyl-iodophenyl-naphthalene 1,4:5,8-tetracarboxylic diimide
8,10 tetramethyl dioxaborolanyl porphyrin 10,11 N,N′-octyl,
hydroxyethyl-1,5-dicarboxy-4,8-naphthalene dicarboximide an-
hydride 11,12 5,15-dimesityl-10,20-bis(4-methoxyphenyl) por-
phyrin 14,13 and 5-bromo-10,15-bis-(3,5-di-tert-butylphenyl)por-
phyrin 1527 were prepared according to literature methods.
5.4. Optical Spectroscopy. All spectroscopic experiments,
except femtosecond transient absorption, were performed using
1 × 1 cm quartz cells. All solvents used were spectroscopic
grade. The sample absorbance was kept around 0.1 at the
wavelength of excitation in all fluorescence measurements and
around 0.2 for nanosecond flash photolysis measurements.
UV-vis absorption spectra were recorded on a HP 8453 diode
array spectrometer, and fluorescence spectra were recorded on
a SPEX fluorolog II spectrometer. Nanosecond transient absorp-
tion measurements were done using a Brilliant B Nd:YAG laser
to pump an Opotek OPO giving ∼20 mJ, ∼5 ns laser flashes
tuned to the largest Q-band in all compounds. The transient
absorption data we obtained with an Applied Photophysics
LKK60 spectrometer, with analyzing light from a pulsed Xe-
lamp. Femtosecond transient absorption measurements were
made using an amplified 1 kHz Ti:sapphire laser system that
has been described in detail elsewhere.28 The pump energy was
kept below 1 µJ/pulse in all experiments. The sample in a 1
mm quartz cell was moved up and down with a frequency of
about 1 Hz. The sample absorbance was kept between 0.2 and
0.3 at the excitation wavelength. The experimental response
function as estimated from the stimulated Raman signals of the
solvent was about 200 fs at 400 nm.
k2/(k2-k1)
[I]max
[A]0
k1
)
(E4)
(E5)
( )
k2
k1
1
tmax
)
ln
( )
(k1 - k2)
k2
For ZI, the concentration of CS at the short time scale
maximum, [ZnP+NI-]max/[S2]0, is given by a modification of
eq E4:
[ZnP+NI-]max
[S2]0
kET
k1
k2/(k2-k1)
)
(E6)
(
)( )
(kET + kIC) k2
with k1 ) kET + kIC and k2 ) kCR + kVR. The factor kET/(kET
+
kIC), which is added as compared to eq E4, takes into account
that ca. 10% (25% in DMF) of the S2 state undergoes direct
internal conversion instead of electron transfer, as shown by
the S2 lifetime and emission yield relative to the model ZnTPP.
The absorption of NI- at 405 nm that remains after a few
picoseconds gives directly the fraction of trapped CS states (after
correction for the fraction of S2 states that undergoes direct
internal conversion to S1):
[ZnP+NI-]int
[S2]0
kET
kVR
)
(E7)
(
)(
)
(kET + kIC) kVR + kCR
where the suffix “int” stands for intermediate minimum.
To estimate [ZnP+NI-]max/[S2]0 and [ZnP+NI-]int/[S2]0 in the
different solvents, comparison was made with the maximum
height of the same 405 nm peak in the same experiment on
longer time scales following ET from S1. As the other decay
pathways of the S1 state are much slower than electron transfer,
the maximum fraction of charge separation from the S1 state
Acknowledgment. This work was supported by The Swedish
Foundation for Strategic Research, The K&A Wallenberg
Foundation, The Swedish Energy Agency, Region “Pays de la

State-Selective Electron Transfer
J. Phys. Chem. A, Vol. 114, No. 4, 2010 1721
(7) Gosztola, D.; Niemczyk, M. P.; Svec, W.; Lukas, A. S.; Wasielews-
ki, M. R. J. Phys. Chem. A 2000, 104, 6545.
Loire” (CPER photonic project), and The European Commission
(COST D35).
(8) (a) Hartnell, R. D.; Edwards, A. J.; Arnold, D. P. J. Porphyrins
Phthalocyanines 2002, 6, 695. (b) Fortage, J.; Boixel, J.; Blart, E.;
Hammarstro¨m, L.; Becker, H.-C.; Odobel, F. Chem.-Eur. J. 2008, 14, 3467.
(9) Yoshida, N.; Ishizuka, T.; Yofu, K.; Murakami, M.; Miyasaka, H.;
Okada, T.; Nagata, Y.; Itaya, A.; Cho, H. S.; Kim, D.; Osuka, A. Chem.-
Eur. J. 2003, 9, 2854.
Supporting Information Available: Description of the
synthesis of the compounds, additional absorption and
transient absorption data, and additional kinetic derivations.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(10) Redmore, N. P.; Rubtsov, I. V.; Therien, M. J. J. Am. Chem. Soc.
2003, 125, 8769.
(11) Hasobe, T.; Imahori, H.; Ohkubo, K.; Yamada, H.; Sato, T.;
Nishimura, Y.; Yamazaki, I.; Fukuzumi, S. J. Porphyrins Phthalocyanines
2003, 7, 296.
(12) Shao, X.-B.; Jiang, X.-K.; Wang, X.-Z.; Li, Z.-T.; Zhu, S.-Z.
Tetrahedron 2003, 59, 4881.
(13) Littler, B. J.; Ciringh, Y.; Lindsey, J. S. J. Org. Chem. 1999, 64,
2864.
(14) Gouterman, M. J. Mol. Spectrosc. 1961, 6, 138.
(15) Karolczak, J.; Kowalska, D.; Lukaszewicz, A.; Maciejewski, A.;
Steer, R. J. Phys. Chem. A 2004, 108, 4570.
(16) Darwent, J. R.; Douglas, P.; Harriman, A.; Porter, G.; Richoux,
M.-C. Coord. Chem. ReV. 1982, 44, 83.
(17) (a) A reduction in rate by a factor of 100 with an additional bridge
distance of 5 Å corresponds to a ꢀ value of ca. 0.9 Å-1; see ref 17b. Given
the differences in, for example, work terms and reorganization energy for
the present dyad and that of ref 9, the rate difference seems to be in a
reasonable range. (b) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985,
811, 265.
(18) (a) Fajer, J.; Borg, D. C.; Forman, A.; Dolphin, D.; Felton, R. H.
J. Am. Chem. Soc. 1970, 92, 3451. (b) Neta, P. J. Phys. Chem. 1981, 85,
3678.
(19) See any textbook on chemical kinetics, for example: Atkins, P. W.;
de Paula, J. Physical Chemistry, 7th ed.; Oxford University Press: Oxford,
2002.
(20) (a) Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259. (b) Weller,
A. Z. Phys. Chem. 1982, 133, 93.
(21) (a) Schmidt, J.; Liu, J.; Bolton, J.; Archer, M.; Gadzekpo, V.
J. Chem. Soc., Faraday Trans. 1989, 85, 1027. (b) See also: Brunschwig,
B. S.; Ehrenson, S.; Sutin, N. J. Phys. Chem. 1986, 90, 3657.
(22) Ionkin, V. N.; Ivanov, A. I. J. Phys. Chem. A 2009, 113, 103.
(23) Petersson, J.; Eklund, M.; Davidsson, J.; Hammarstro¨m, L. J. Am.
Chem. Soc. 2009, 131, 7940.
(24) Pigliucci, A.; Duvanel, G.; Daku, L. M. L.; Vauthey, E. J. Phys.
Chem. A 2007, 111, 6135.
(25) Kovalenko, S. A.; Schanz, R.; Henning, H.; Ernsting, N. P. J. Chem.
Phys. 2001, 115, 3256.
(26) Seely, G. R.; Gust, D.; Moore, T. A.; Moore, A. L. J. Phys. Chem.
References and Notes
(1) (a) Molecular Switches; Feringa, B. L., Ed.; Wiley-VCH: Wein-
heim, 2001. (b) Debreczeny, M. P.; Svec, W. A.; Wasielewski, M. R. Science
1996, 274, 584. (c) Andersson, M.; Sinks, L. E.; Hayes, R. T.; Zhao, Y. Y.;
Wasielewski, M. R. Angew. Chem., Int. Ed. 2003, 42, 3139. (d) Gust, D.;
Moore, T. A.; Moore, A. L. Chem. Commun. 2006, 1169. (e) Andre´asson,
J.; Straight, J. D.; Bandyopadhyay, S.; Mitchell, R. H.; Moore, T. A.; Moore,
A. L.; Gust, D. J. Phys. Chem. 2007, 111, 14274. (f) de Silva, A. P.;
Uchiyama, S. Nat. Nanotechnol. 2007, 2, 399. (g) de Silva, A. P.;
McClenaghan, N. D. Chem.-Eur. J. 2004, 10, 574. (h) Balzani, V.; Credi,
A.; Venturi, M. ChemPhysChem 2003, 4, 49. (i) Pischel, U. Angew. Chem.,
Int. Ed. 2007, 46, 4026. (j) Saha, S.; Stoddart, J. F. Chem. Soc. ReV. 2007,
36, 77. (k) Champin, B.; Mobian, P.; Sauvage, J.-P. Chem. Soc. ReV. 2007,
36, 358.
(2) (a) Kalyanasundaram, K. Photochemistry of Polypyridines and
Porphyrin Complexes; Academic Press: London, 1992. (b) Wasielewski,
M. Chem. ReV. 1992, 92, 435. (c) Gust, D.; Moore, T. A.; Moore, A. L.
Acc. Chem. Res. 1993, 26, 198. (d) Imahori, H. Org. Biomol. Chem 2004,
2, 1425. (e) Osuka, A.; Mataga, N.; Okada, T. Pure Appl. Chem. 1997, 69,
797.
(3) Kasha’s rule concerned radiative processes; the extension to
photochemical reactions is a generalization based on the same principle: Kasha,
M. Discuss. Faraday Soc. 1950, 9, 14.
(4) (a) Chosrowjan, H.; Tanigichi, S.; Okada, T.; Tagaki, S.; Arai, T.;
Tokumaru, K. Chem. Phys. Lett. 1995, 242, 644. (b) Gurzadyan, G. G.;
Tran-Thi, T. H.; Gustavsson, T. J. Chem. Phys. 1998, 108, 385. (c) Mataga,
N.; Shibata, Y.; Chosrowjan, H.; Yoshida, N.; Osuka, A. J. Phys. Chem. B
2000, 104, 4001. (d) Yu, H.-Z.; Baskin, J. S.; Zewail, A. H. J. Phys. Chem.
A 2002, 106, 9845. (e) Tripathy, U.; Kowalska, D.; Liu, X.; Velate, S.;
Steer, R. P. J. Phys. Chem. B 2008, 112, 5824.
(5) (a) LeGourrie´rec, D.; Andersson, M.; Davidsson, J.; Mukhtar, E.;
Sun, S.; Hammarstro¨m, L. J. Phys. Chem. A 1999, 103, 557. (b) Andersson,
M.; Davidsson, J.; Hammarstro¨m, L.; Korrpi-Tommola, J.; Peltola, T. J.
Phys. Chem. B 1999, 103, 3258.
(6) (a) Mataga, N.; Chosrowjan, H.; Taniguchi, S. J. Photochem.
Photobiol., C 2005, 6, 37. (b) Mataga, N.; Chosrowjan, H.; Taniguchi, S.;
Shibata, Y.; Yoshida, N.; Osuka, A.; Kikuzawa, T.; Okada, T. J. Phys.
Chem. A 2002, 106, 12191. (c) Mataga, N.; Chosrowjan, H.; Taniguchi,
S.; Shibata, Y.; Yoshida, N. Chem. Phys. 2003, 295, 215. (d) Morandeira,
A.; Engeli, L.; Vauthey, E. J. Phys. Chem. A 2002, 106, 4833. (e) Kubo,
M.; Mori, Y.; Otani, M.; Murakami, M.; Ishibashi, Y.; Yasuda, M.;
Hosomizu, K.; Miyasaka, H.; Imahori, H.; Nakashima, S. J. Phys. Chem.
A 2007, 111, 5136. (f) Fujitsuka, M.; Cho, D. W.; Tojo, S.; Inoue, A.;
Shiragami, T.; Yasuda, M.; Majima, T. J. Phys. Chem. A 2007, 111, 10574.
1994, 98, 10659.
(27) Odobel, F.; Suresh, S.; Blart, E.; Nicolas, Y.; Quintard, J.; Janvier,
P.; Le Questel, J.; Ilien, B.; Rondeau, D.; Richomme, P.; Ha¨upl, T.; Wallin,
S.; Hammarstro¨m, L. Chem.-Eur. J. 2002, 8, 3027.
(28) Wallin, S.; Davidsson, J.; Modin, J.; Hammarstro¨m, L. J. Phys.
Chem. A 2005, 109, 4697.
JP907824D