Angewandte
Communications
Chemie
À
lengths (but only the shorter two out of the three decay
to AQ can occur (SI page S33). We note that the change
components are visible in Figure 3). The shortest of the three
time components is 210 ns and can be attributed unambigu-
ously to the photoinduced charge-separation reaction in
from 1:1 (v:v) CH CN/H O to neat CH CN does also increase
3 2 3
the driving force for thermal charge recombination in
a significant manner (by ca. 0.3 eV) because non-hydrogen-
+
À
À
which TAA and AQ are formed (SI pages S16, S17; the
bonded AQ is easier to oxidize than its hydrogen-bonded
3
+
MLCT excited state of III has higher extinction coefficients
analogue, and TAA reduction is more facile in neat CH CN
3
at the relevant detection wavelengths than the final charge-
separated state). The intermediate time component is 65.4 ms
and is attributable to intramolecular thermal electron transfer
(SI page S13). Thus, solvent changes do not only affect l, but
0
they also lead to significant changes of DG
in our
ET
0
compounds (SI page S36). As DGET for charge recombina-
tion increases, the driving force for photoinduced charge
separation decreases, because the energy of the initially
À
+
from AQ to TAA , i.e., to the process of main interest. The
third time component (ꢀ 400 ms; not seen in Figure 3) is
caused by intermolecular electron-transfer reactions (SI
pages S20–S23). Thus, the kinetics for intramolecular electron
transfer is clear-cut: kET increases by a factor of 8 between
compounds I and II, and then decreases by a factor of 188
3
populated MLCT excited state remains relatively constant.
In CH Cl and more apolar solvents one reaches a point at
2
2
which efficient charge separation is no longer possible in
compounds II and III. This precludes further solvent-depend-
ence studies.
0
between II and III. The reaction free energy (DGET ) for
À
+
0
intramolecular thermal electron transfer from AQ to TAA
With the l, DGET , and kET values from Table 1 for 1:1
is very similar in I, II, and III (Table 1, SI page S11).
(v:v) CH CN/H O mixtures at hand, Equation (1) can be used
3
2
The temperature dependence of k was analyzed to
to obtain estimates of HDA. We find that the electronic
ET
°
determine activation free energies (DGET ). From Arrhenius
coupling is only very weakly distance dependent with H
DA
°
À1
À1
plots the DGET values reported in Table 1 were extracted (SI
values of 0.09 Æ 0.02 cm (I), 0.10 Æ 0.02 cm (II), and 0.08 Æ
À1
page S24). In compound II electron transfer proceeds in
essentially activationless manner, whereas in compounds I
0.02 cm (III). At first glance this is a somewhat unexpected
result, particularly in view of prior studies of donor–acceptor
compounds with oligo-p-xylene bridges which have produced
distance decay constants (so-called b-values) between 0.52–
°
and III DGET is 43 Æ 2 meV and 108 Æ 9 meV, respectively.
°
0
2
[6b]
Since DGET = (l + DG ) /4·l [last term in Eq. (1)],
ET
À1
[15]
reorganization energies (l) can be determined from these
0.77
for kET.
However, the earlier studies have
0
ET
values. In the case of II, l must be equal to ÀDG (1.29 Æ
exclusively focused on photoinduced (forward) electron
transfer in the so-called normal regime. It has been noted
earlier that the kinetics of photoinduced (forward) electron
transfer and thermal (reverse) electron transfer can exhibit
significantly different distance dependences, because the
superexchange coupling pathways (determining the magni-
0
.05 eV) because the reaction is barrierless. For compounds I
°
and III, the quadratic relationship between DGET and
l yields two mathematical solutions, but in each case only
one solution is physically meaningful because l is expected to
[
4a,8,10]
increase with increasing rDA (SI pages S26, S34).
we obtain l = 0.93 Æ 0.35 eV for I, 1.29 Æ 0.05 eV for II, and
.21 Æ 0.28 eV for III in 1:1 (v:v) CH CN/H O (Table 1).
Thus
[9,16]
tude of HDA) can be fundamentally different.
We suspect
2
that the weak distance dependence of HDA in compounds I–
III is due to increasing p-conjugation between the central bpy
and adjacent p-xylene units with increasing length. This
interpretation is supported by the observation that the
3
2
The dominant contribution to l usually comes from the
outer-sphere reorganization energy (l ), while the inner-
sphere contribution is small and largely independent of rDA
o
[10]
.
3
Simple two-sphere electrostatic models fail to quantitatively
spectroscopic signature of the MLCT state of a reference
reproduce the experimentally observed increase in l for
complex lacking the AQ and TAA components but bearing p-
xylene bridging units is substantially different from the
[8,11]
compounds I–III in 1:1 (v:v) CH CN/H O,
because they
3
2
3
2+
do not take hydrogen bonding into account. Direct evidence
for the importance of hydrogen bonding in our systems comes
MLCT spectrum of [Ru(bpy)3] (SI page S19).
3
Because of the initial population of a MLCTexcited state,
À
À
+
from the AQ -related transient absorption band at 510 nm
a radical ion pair (AQ /TAA ) with triplet spin multiplicity
forms initially. Charge recombination must occur directly to
the singlet ground state, but spin effects are not expected to
play a decisive role as far as the distance dependence of kET is
concerned. Possible spin and electron-vibration coupling
effects are discussed in the Supporting Information
(
Figure 2b,c). In neat CH CN, this band appears at 565 nm for
3
compounds I–III, (SI page S27) and the shift to shorter
wavelength in 1:1 (v:v) CH CN/H O is in line with hydrogen-
3
2
[
12]
bond donation from water. Based on prior electrochemical
studies and on calculations for benzoquinone radical anion,
[6b,16b,17]
we expect that 4–5 H O molecules are involved in hydrogen
(pages S38–S41).
2
À
[13]
bonding to AQ , and this raises l significantly with respect
In conclusion, the highly unusual observation of an
electron-transfer-rate maximum at large (30.6 ) reactant
separation can be explained by a weak distance dependence
of electronic donor–acceptor coupling (HDA) combined with
a strong distance dependence of the reorganization energy
to what is predicted by a simple dielectric continuum
[
8,11]
À
model.
The further AQ is spatially separated from
2
+ +
cationic charges (Ru(bpy)3 , TAA ), the more important
the effect of hydrogen-bonding becomes. For comparison,
a study of phototriggered phenol oxidation found l = 2.0 eV
because the phenolic OÀH bond was broken in the course of
[4a]
(l), as predicted by theory more than three decades ago. As
l increases with increasing donor–acceptor distance, our
reaction systems pass through the inverted (I), barrierless (II),
and normal (III) regimes of electron transfer, and all the while
[
14]
electron transfer.
In neat CH CN the reorganization energy (l) for com-
3
0
pound III is only 1.62 Æ 0.05 eV because no hydrogen bonding
the reaction free energy (DGET ) stays essentially constant
Angew. Chem. Int. Ed. 2016, 55, 815 –819
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
817