Charge transfer through cross-hyperconjugated versus cross-π-conjugated bridges: An intervalence charge transfer study

Article abstract of DOI:10.1039/c3sc50844g

Recently there has been much interest in electron transfer and transport through cross-conjugated molecules as interesting test cases for the interplay between molecular and electronic structure as well as potential motifs in the design of new compounds f

Full text of DOI:10.1039/c3sc50844g

Chemical Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Charge transfer through cross-hyperconjugated versus
cross-p-conjugated bridges: an intervalence charge
transfer study†
Cite this: Chem. Sci., 2013, 4, 3522
a
b
b
a
Erik G o¨ ransson, Rikard Emanuelsson, Kjell Jorner, Todd F. Markle,
a
b
Leif Hammarstr o¨ m* and Henrik Ottosson*
Recently there has been much interest in electron transfer and transport through cross-conjugated molecules
as interesting test cases for the interplay between molecular and electronic structure as well as potential
motifs in the design of new compounds for molecular electronics. Herein we expand on this concept and
present the synthesis and characterization of a series of four organic mixed-valence dyads to probe the
effect of the bridge structure on the electronic coupling. The electronic coupling between two triarylamine
units could be mediated either by cross-hyperconjugation through a saturated ER
2
bridge (E ¼ C or Si, R ¼
alkyl or silyl group), or via a cross-conjugated p-system. The aim of the study is to compare the electron
transfer through the various saturated bridges to that of a cross-p-conjugated bridge. The electronic
coupling in these mixed-valence compounds was determined by analysis of intervalence charge transfer
ꢀ1
bands, and was found to be in the range of 100–400 cm . A complementary DFT and TD-DFT study
indicated that the electronic coupling in the dyads with saturated ER segments is highly conformer
dependant. Furthermore, the calculations showed that two types of interactions contribute to the
electronic coupling; through-bond cross-(hyper)conjugation mechanism and through-space
2
Received 28th March 2013
Accepted 17th June 2013
a
a
mechanism. Taken together, these findings suggest the possibility for new architectures for molecular
electronics applications utilizing cross-hyperconjugation through properly selected saturated segments
which have comparable electron transfer characteristics as regular cross-p-conjugated molecules.
DOI: 10.1039/c3sc50844g
www.rsc.org/chemicalscience
benets that are not achievable with purely p-conjugated
molecules.
Introduction
Recent years have witnessed substantial progress within the
We have recently reported a study which compares cross-p-
conjugation with cross-hyperconjugation of various strengths.
1,2
14

eld of single molecule electronics. The majority of the
molecules utilized in the eld are of regular p-conjugated type, Two simple p-conjugated phenylethynyl groups were covalently
although during the last few years there has been a growing linked by either a single saturated moiety (CR or SiR ) or via a
2
2
interest in molecular structures with alternative conjugation single cross-p-conjugated unit (C(]CMe )). The study indi-
2
3–10
topologies such as cross-p-conjugation and s-conjugation.
cated that changing the substituents, R, on the saturated center
Hyperconjugation, i.e., the interaction between a local p- markedly affects the strength of the cross-hyperconjugation,
symmetric orbital at an unsaturated molecular segment and a
and for that with R ¼ SiMe
3
the cross-hyperconjugation was of
similar magnitude as cross-p-conjugation on the basis of UV
11,12
local p-symmetric orbital at a saturated segment,
esting both as a mediator for charge transport and as a mode of
is inter-
absorption spectroscopy, electrochemistry and density func-
13
linking several p-conjugated systems. Such saturated hyper- tional theory (DFT) computations.
conjugated molecular segments could allow for a range of
In the present work we investigate if strong cross-hyper-
conjugation also translates into a similar mediation of charge
transfer and charge transport as that across a cross-p-conju-
gated segment. I.e., the study reported herein focuses on the
impact of cross-hyperconjugation on a chemical process, while
our earlier study primarily concerned the static electronic
a
˚
Department of Chemistry, Angstr ¨o m Laboratory, Uppsala University, Box 523, 751 20
Uppsala, Sweden. E-mail: leif.hammarstrom@kemi.uu.se
b
Department of Chemistry, BMC, Uppsala University, Box 576, 751 23 Uppsala,
Sweden. E-mail: henrik.ottosson@kemi.uu.se
†
Electronic supplementary information (ESI) available: Details on the synthetic structures of molecules which are cross-hyperconjugated to
procedure, the analysis of spectroscopic data, the benchmarking procedure of
the computational methods, the deconvolution of absorption spectra of the
singly oxidized compounds, full data from the conformational scan, full
reference for Frisch et al. [ref. 46], and absolute electronic energies and
Cartesian coordinates. See DOI: 10.1039/c3sc50844g
various extents in comparison with a strongly cross-p-conju-
gated one. With this aim we designed and prepared a series of
four symmetric bis-triarylamine dyads, which are bridged with
2
units that either allow for cross-p-conjugation (D-C(]CMe )-D)
14
3
522 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
or for cross-hyperconjugation to varying degrees (D-Si(TMS)
2
-D, describes the cases for which there is a moderate electronic
D-Si(Me) -D and D-C(Me) -D), see Chart 1. One-electron oxida- coupling that causes a mixing of the two diabatic states result-
2
2
tion of these dyads will result in organic mixed-valence ing in two new adiabatic states; |ai and |bi (Fig. 1). Whether a
15
compounds, which allows us to study the intervalence charge Class II compound is best described by the adiabatic or diabatic
transfer (IVCT) as a function of the bridging unit. Such inter- states depends on which aspects one wants to discuss. For
valence charge transfer allows direct and more detailed insight optical IVCT the observed transition is between the adiabatic
into electronic coupling through these structures than has been states |ai and |bi. As seen in Fig. 1 the energy of this transition
previously possible. Triarylamines have previously been used as ( ꢀn ab) can be used to approximate the Marcus reorganization
25
26
electron donor moieties in organic mixed valence compounds, energy (l) between the two diabatic states. The magnitude of
because of their synthetic accessibility and the stability of the the optical IVCT band can also be used to calculate the elec-
1
5–19
radical cation oxidized forms.
In this paper we will present a tronic coupling, vide infra.
Class III of the Robin–Day classication describes the cases
set of new dyads and the spectroscopic and theoretical charac-
terization of their mixed-valence behavior. We will also discuss where the electronic coupling is so strong that the excess charge
how the electronic coupling in these dyads varies with different is delocalized over all sites. For these cases the diabatic
conformations that can be adopted. Our vision is that the description is clearly inappropriate and the system should be

ndings reported herein could enable the identication and described by the two adiabatic states (Fig. 1). Optical transition
design of completely new compound classes to be used for between the adiabatic states |ai and |bi can oen be observed
charge transfer and transport.
and since the energy of this transition corresponds to 2V12 one
can use the position of the absorption band as a direct
measurement of the electronic coupling. The border between
the different Robin–Day classes is not clearly dened and there
are many interesting studies on mixed-valence compounds that
Theoretical background
IVCT and mixed-valence compounds
Mixed-valence compounds are characterized by having an are on the Class II/III border.27–31 However, the cross-conjugated
excess charge that can be localized on two or more (equal) sites bridges used in the dyads in this study should result in relatively
within the molecule. Having the excess charge on a specic site weak couplings and are expected to show Class II behavior.
corresponds to a diabatic state |ii, and IVCT processes corre-
spond to transitions between these states. Studies of symmetric
mixed-valence compounds, notably di-ruthenium and di- Generalized Mulliken–Hush theory
osmium complexes, such as the Creutz–Taube ion, have
The electronic coupling between the diabatic states in a Class II
provided many fundamental insights into the understanding of
compound can be extracted from the strength of the IVCT band
20–23
electron transfer processes.
In the dyads in Chart 1 the
15,22,32–34
using generalized Mulliken–Hush (GMH) theory.
For the
possible sites for the excess charge are formally on either the
le” or “right” triarylamine, corresponding to diabatic states
case where there are only two relevant states, the relationship
can be summarized by eqn (1).
“
+
+
|N ,Ni and |N,N i, which are denoted |1i and |2i, respectively.
The behavior of mixed-valence compounds is closely linked to
the strength of the electronic coupling (V ) between the two
1
2
diabatic states. Robin and Day used this to categorize mixed-
24
valence compounds into three general classes. Class I repre-
sents the cases where the electronic coupling is very weak and
no thermal or optical charge transfer can be observed. Class II
Fig. 1 Schematic description of the states in a symmetric mixed-valence
compound. On top Class II and below Class III. The diabatic states, |1i and |2i are
the dashed, parabolic potential energy surfaces. The adiabatic states, |ai and |bi,
that form through mixing of |1i and |2i are marked with solid lines. The splitting
between the adiabatic states corresponds to 2V12. ꢀv ab is the energy of the IVCT
absorption band, indicated by the vertical arrow, and l is the Marcus reorgani-
zation energy.
Chart 1 The investigated donor–bridge–donor systems and the 4-(tert-butyl-
0
ethynyl)-N,N-bis(4 -methoxyphenyl)aniline reference.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3522–3532 | 3523

View Article Online
Chemical Science
Edge Article
jm jvab
Spectroscopic and electrochemical characterization
ab
V
12
¼
(1)
jDm j
2
1
Cyclic voltammetry on the dyads and the reference compound
bearing a single triarylamine group were performed in aceto-
6
nitrile (0.1 M TBAPF , MeCN). Although each dyad bears two
Here V12 is the electronic coupling between the diabatic
states. mab is the transition dipole moment and ꢀn ab is the
transition energy for the excitation between the two adiabatic
states. |Dm | corresponds to the difference in static dipole
triarylamines, the cyclic voltammograms display a single wave
corresponding to a two-electron oxidation (ESI Fig. S8†). These
two-electron waves are situated at similar potentials as the
reference compound (D-tBu), but with larger peak separation
2
1
moment when an unpaired charge moves between the two
between cathodic and anodic waves (DE ), Table 1. These vol-
adiabatic states. The transition energy ( ꢀn ) can readily be
p
ab
tammograms suggest that the two triarylamine moieties are
oxidized sequentially, but at potentials that are too close to be
resolved by CV (this conclusion is consistent with chemical
extracted from the peak position of the observed IVCT band
and the transition dipole moment (|mab|) can be calculated
35
from eqn (2a) or (2b).
ð
ð
oxidation experiments described below). From the DE
p
values
3
hc3
0
ln 10 3ðnÞ
2
jm j ¼
ab
d n
d n
(2a)
(2b)
2
we estimate that the two oxidation potentials are separated by
2p N
A
n
39
4
0–80 mV. This small degree of electrochemical splitting is an
3
ðnÞ
indication that the interaction between the two triarylamines in
these cross-conjugated dyads is fairly weak.
2
ꢀ4
jm j ¼ 3:98 ꢁ 10
ab
n
The absence of a distinct potential window where the dyads
would be oxidized only once complicates the use of spec-
troelectrochemistry as a method to detect and characterize the
singly oxidized species. Instead, titration with a chemical
oxidant was used to afford the spectra of the singly oxidized
In eqn (2b) the constants have been evaluated and the
resulting transition dipole moment will have the unit [e A˚ ],
providing one expresses the extinction coefficient in
ꢀ1
ꢀ1 36
[M
cm ]. The change in static dipole moment is most
4 2
species. Cu(ClO ) proved to be an efficient and convenient
oen evaluated as the movement of a point charge over a
distance: |Dm | ¼ |r | ꢁ e, where |r | is the charge transfer
oxidant for these experiments (Caution: perchlorate salts are
2
1
21
21
2+
explosion hazards!). The moderate potential of Cu in MeCN
distance and e is the elementary charge. However, an exact
charge transfer distance can be difficult to dene, particularly
with the relatively delocalized ions oen found in organic
+/0
(
ca. +0.55 V vs. Fc ) is sufficiently strong to rapidly oxidize each
2+
+
triarylamine moiety once, but not more. The Cu and Cu salts
are relatively transparent in the region of interest, and the
triarylaminium radical cations are stable in the presence of Cu
40
18
systems, such as the radical cation of a triarylamine. For
+
these cases it is better to evaluate |Dm21| from the static dipole
4 2
product. Alternatively, use of Cu(BF ) salts resulted in some-
what increased decomposition of the radical product, and
37
moments of the two diabatic states, m
1
and m
2
.
These static
dipole moments are usually most readily obtained from
quantum chemical calculations, which give the values for ma
NOBF oxidations were slow under these conditions. Addition
4
of Cu(ClO ) to MeCN solutions of the colorless triarylamines
4
2
(
ground state) and m (excited state) necessary to calculate m
b
2
1
yielded deep blue solutions, characteristic of the radical cation
product. Fig. 2 and Table 1 show the results from the titration
experiments. The procedure for extracting the spectra of the
neutral, singly oxidized and doubly oxidized forms of the dyads
is detailed in the ESI (ESI Fig. S6†).
34
using eqn (3).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
2
jDm j ¼ jm ꢀ m j ¼ ðm ꢀ m Þ þ 4m
(3)
2
1
2
1
b
a
ab
2 2 2
The electronic spectra of the Si(TMS) , Si(Me) and C(Me)
Results and discussion
Synthesis
bridged dyads are very similar in all three forms, with only small
+
shis of the peak positions (Fig. 2). However, for D -C(Me) -D
2
ꢀ
1
there is an additional broad peak around 8500 cm . The cross-
p-conjugated dyad, D-C(]CMe )-D, differs from the other
The donor–bridge–donor compounds were assembled in the
nal step from the separate donor and bridge segments
Scheme 1). The donor fragments were synthesized from a tri-
2

dyads by having broader radical cation absorption peaks in both
of its oxidized forms. This could be due to increased length of
the conjugated system, or increased interaction between the
two triarylamine moieties. The spectra of the singly and doubly
(
arylamine moiety and following iodination to 1, and subse-
quent Sonogashira chemistry, 2, 3 or D-tBu could be achieved in
accordance with the method by Lambert et al. We have
previously shown how the different bridge segments can be
38
oxidized species are dominated by the intense and relatively
ꢀ1
sharp band at ꢂ13 500 cm
from the triarylamine radical
14
synthesized,
D-C(]CMe )-D from 4 using Sonogashira chemistry with in situ
deprotection of the trimethylsilyl (TMS) groups. Similarly,
compounds D-Si(TMS) -D and D-C(Me) -D were constructed
and we assembled the cross-p-conjugated
cation, and the spectra of the doubly oxidized dyads are similar
in shape to those of the singly oxidized species, but with twice
the molar absorptivity. This is another indication of weak (Class
II) electronic coupling between the triarylamine moieties in
these dyads.
2
2
2
from the iodotriarylamine 1 and bis-alkynes 5 and 6, respec-
tively. The SiMe bridged D-Si(Me) -D, on the other hand, was
2
2
However, close examination of the spectral region between
assembled via a substitution reaction between 3 and dichloro-
dimethylsilane (see ESI for synthetic procedures†).
ꢀ1
5000 and 15 000 cm
reveals that there are differences
3
524 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
Scheme 1 Synthesis of the four dyads and the reference compound.
between the singly and doubly oxidized forms of the dyads order to get a reliable t of the IVCT shoulder the whole
Fig. 3). The singly oxidized mixed-valence complexes have an spectrum was deconvoluted by tting it with a series of
(
additional absorption feature in the near-IR ascribed to an Gaussian peaks. Gaussian deconvolution of the band ascribed
IVCT band. The band is most clearly seen for D-C(Me) -D, to the IVCT band is shown in Fig. 3. The parameters for the
2
where the singly oxidized species has a distinct peak at ca. IVCT peaks are summarized in Table 2. Details of the proce-
ꢀ
1
8
000 cm . For the other three dyads the IVCT band appears dure and results of the spectral deconvolution can be found in
as a shoulder on the low energy side of the radical band. In the ESI (ESI Fig. S9–S14†).
Table 1 Summary of absorption and electrochemical data measured in acetonitrile
a
3
ꢀ1
ꢀ1
Absorption lmax/nm (3/10 M cm
)
b
+/0 c
+
d
Neutral
Singly oxidized
Doubly oxidized
K
disp
E
½
p
(TPA ) /mV vs. Fc /Fc DE /mV
D-tBu
D-Si(TMS)
D-Si(Me)
D-C(Me)
315 (17) 381 (22), 413 (12), 748 (35)
324 (49) 340 (33), 370 (30), 411 (12), 752 (34)
330 (52) 342 (34), 376 (29), 418 (14), 753 (34)
3e 29
67
e
2
-D
-D
-D
374, (46), 412 (24), 754 (68) 0.05
2
380 (47), 420 (29), 754 (69)
0.1
0.1
348
315
92
80
2
325 (54) 320 (34), 379 (31), 413 (17), 749 (39), 379 (50), 414 (31), 751 (84)
180 (1.4)
)-D 294 (47) 294 (43), 356 (48)
46 (76) 418 (19), 751 (35)
1
D-C(]CMe
2
276 (34), 377 (42)
450 (21), 756 (67)
0.03
313
76
3
a
b
Position and extinction coefficient for observed peaks. Equilibrium constant for the disproportionation reaction of singly oxidized species,
c
determined from the titration data (relative uncertainties are ꢂ30%). Half-wave potential for the rst oxidation wave. In the dyad systems, the
d
e

rst and second oxidation merged into a two-electron wave. Peak separation between the cathodic and anodic waves. Not measured.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3522–3532 | 3525

View Article Online
Chemical Science
Edge Article
Computational results
Geometry optimizations of the singly oxidized forms of the four
dyads, using PBE0/6-31G(d) and a polarizable continuum model
for the acetonitrile solvent, resulted in geometries where the
electron hole is localized to one of the two triarylamines. This
agrees with the electronic spectra (Fig. 2) and our expectation
that the dyads would fall into Robin–Day Class II. Preliminary
geometry optimizations also resulted in a wide distribution in
the geometries of the conformers adopted by the four dyads. The
most notable differences were in the degree of rotation of the two
inner phenylene groups relative to the ethynylene bridge unit.
Since the barrier for these rotations about acetylene units can be
expected to be small, we wanted to investigate to what extent the
geometry and conformational freedom could affect the elec-
6,41
tronic coupling. We performed relaxed conformational scans
on the rotation of each of the inner phenylene groups, by inde-
ꢃ
pendently varying the two relevant dihedral angles in steps of 30
ꢃ
ꢃ
between 0 and 150 (see experimental section and ESI Fig. S15†).
The results indicated that only D -C(]CMe )-D has a signicant
2
+
energy barrier preventing rotation, stemming from the p-
conjugation between theamine andthe bridge unit, buteven this
dyad is expected to have a substantial population of all confor-
mations at room temperature (Fig. 4 and ESI Fig. S16†).
TD-DFT was used to calculate the three lowest excitations at
each of the 36 geometries from the relaxed scan. The lowest
excitation corresponds to the IVCT band and is characterized as
a pure (ꢂ99%) b-HOMO / b-LUMO transition, with the b-
HOMO localized on the neutral triarylamine and b-LUMO on
the oxidized triarylamine. The second and third lowest excita-
tions are transitions localized on the oxidized triarylamine and
they are similar to those calculated for a single oxidized tri-
arylamine. The calculated energies for the IVCT transitions are
Fig. 2 Absorption spectra in MeCN for the neutral (dashed), singly (line) and
doubly oxidized forms (dotted) of the four dyads. Method for extracting the
extinction coefficients are described in the ESI.†
ꢀ
1
on the order of 4000–4500 cm (Table 3), which is about a
factor of two lower than the experimental values. The failure to
accurately reproduce the correct transition energies for charge
42
transfer transitions is a well-known weakness of TD-DFT, and
for our systems it became evident in a benchmarking study in
which the CAM-B3LYP, LC-uPBE and PBE0 functionals were
compared against experimental values (see ESI†). Yet, the main
purpose of the TD-DFT calculations, and this work in general,
was to study differences in electronic coupling within a series of
Class II mixed-valence compounds. This makes the oscillator
strength a more interesting parameter than the transition
energy since it relates to the probability of a transition between
the initial and nal states (|1i and |2i). Comparing the oscil-
lator strengths of the four dyads we note that they all display a
strong dependence on the rotational conformation (Fig. 4 and
ESI Fig. S17†). For example, in all four dyads the oscillator
strength is zero when the pair of dihedral angles is at either
ꢃ ꢃ ꢃ ꢃ
[
0 ,90 ] or at [90 ,0 ], i.e., when the donor and acceptor p-
Fig. 3 Expanded view of the absorption features in the NIR region for the singly
oxidized (line) and doubly oxidized (dotted) forms of the four dyads. The doubly systems are orthogonal. However, there are also notable
oxidized species have been scaled down by a factor of two so that the localized
differences in the conformational dependence of IVCT within
radical band is at ꢂ13 500 cmꢀ has the same strength as in the singly oxidized
species. The dashed-dotted lines represent the IVCT bands and come from a
spectral deconvolution of the absorption spectra. (See ESI for a complete set of
Gaussian peaks derived from the spectral deconvolution.†) The sharp features at
ꢂ5200 and ꢂ7500 cmꢀ are due to incomplete background subtraction.
1
+
the series. The oscillator strength for D -Si(TMS)
2
-D and
+
D -C(]CMe
2
)-D reaches a maximum when the two phenylene
groups are in the same plane as the plane spanned by the two
1
ꢃ
ꢃ
ethynylene groups, [0 ,0 ]. In contrast, the oscillator strength
3
526 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
Table 2 Summary of the experimental and computed IVCT parameters used to evaluate the electronic coupling
a
3
ꢀ1
a
3
ꢀ1
a
3
ꢀ1
ꢀ1
b
c
d
e
ꢀ1
˚
˚
˚
ꢀ
n
max /10 cm
ꢀn FWHM /10 cm
3
max /10 M cm
r
N–N /A
Dm21 /eA
|mab| /eA
V12 /cm
D-Si(TMS)
D-Si(Me) -D
D-C(Me) -D
2
-D
9.94
3.37
4.25
4.64
2.66
0.15
0.17
1.37
1.85
14.4
13.7
13.6
13.7
13.1
13.3
13.0
12.1
0.15
0.18
0.59
0.47
111
133
374
375
2
10.06
8.24
9.75
2
2
D-C(]CMe )-D
a
b
Parameters obtained from the Gaussian t to the IVCT band. Calculated distance between the two nitrogen atoms for the conformer which
c
contributes most to the electronic coupling. Change in static dipole moment using eqn (3) and values from TD-DFT calculations. Parameters
for the static dipole moments can be found in ESI Table S7.† Transition dipole moments calculated from eqn (2b). The integral was
numerically evaluated using the parameters from the Gaussian t. The electronic coupling calculated from eqn (1) using the parameters in
d
e
Table 2.
+
+
ꢃ
ꢃ
for D -Si(Me)
2
-D and D -C(Me)
2
-D at the [0 ,0 ] conformation is
The TD-DFT calculations were also used to evaluate the
in both cases negligible, indicating that cross-hyperconjugation change in static dipole moment (Dm21) between the initial and
in these two compounds is very weak. Instead, the latter two nal state using eqn (3) (see ESI for details†). A comparison of
species have their strongest transitions at the conformations the values for rN–N and Dm21 in Table 2 shows that the effective
with the phenylene groups perpendicularly oriented to the movement of electron density is somewhat shorter than the
ꢃ
ꢃ
plane of the two ethynylene groups, [90 ,90 ]. The implications nitrogen–nitrogen distance. When the conformational freedom
of these differences will be discussed in more detail below. is taken into account, the overall agreement between the DFT-
To account for the conformational freedom in these systems derived V12 values (which span 144–392 cm ) and those
ꢀ
1
ꢀ1
and to compare the TD-DFT results with experimental spectra calculated using experimental IVCT bands (111–375 cm ) is
we calculated the Boltzmann weighted oscillator strength and quite striking. Nevertheless, there are discrepancies that will be
Boltzmann and oscillator strength weighted transition energies discussed in more detail below.
(Table 3). Using the weighted averages and eqn (4) allowed us to
deduce transition dipole moments based on the TD-DFT
calculations.
32
Trends in electronic coupling
2
Using GMH theory (eqn (1)) we evaluated the electronic coupling
fromtheexperimentalresults(Table2). Asexpectedthemagnitude
of the electronic coupling elements for these cross-(hyper)conju-
gated dyads are decreased relative to analogous linearly conju-
3
2
he
f
2
jm j ¼
ab
(4)
8
p m
e
c nmax
In eqn (4) h is Planck's constant, e is the elemental charge, me
gated systems. For comparison, using 1,4-diethynyl-phenylene as
is the mass of an electron and c is the speed of light. The oscil-
lator strength is denoted by f and the transition energy by ꢀn max
ꢀ1 28,43
the bridging unit results in a coupling element of ꢂ1000 cm
.
.
However, the coupling is still signicant indicating that these or
Fig. 4 Summary of the results from the conformational scan at the PBE0/6-31 G(d,p)//PBE0/6-31G(d) level. The top four panels show the energy barrier for rotation of
the inner phenylene group on the oxidized triarylamine with the inner phenylene group on the neutral side either in-plane, or perpendicular to the plane of the
ꢃ
ꢃ
ethynylene bridge; i.e., 0 (red squares and curve) or 90 (blue circles and curve). Changes along the x-axis correspond to rotation of the phenylethynyl group on the
oxidized side. The dashed horizontal line marks the thermal energy at 298 K. The lower four panels show the conformational dependence of the oscillator strength for
ꢃ
ꢃ
ꢃ
ꢃ
+
the IVCT transition. Note that the oscillator strength is zero for all dyads when the pair of dihedral angles is either [0 ,90 ] or [90 ,0 ]. Also note that D -Si(TMS)
2
-D
+
ꢃ
ꢃ
+
+
and D -C(]CMe
2
ꢃ
)-D have the largest oscillator strength when the dihedral angles are [0 ,0 ], while D -C(Me)
2
-D and D -Si(Me)
2
-D have the highest transition
ꢃ
probability at [90 ,90 ].
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3522–3532 | 3527

View Article Online
Chemical Science
Edge Article
Table 3 Summary of values obtained from the TD-DFT calculations
a
3
ꢀ1
b
E
00 /10 cm
Osc. str., ( f )
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
c
˚
d
˚
e
ꢀ1
0
,0
90 ,90
BW
0 ,0
90 ,90
BW
Dm21 /eA
|mab
|
DFT /eA
V
12,DFT /cm
D-Si(TMS)
D-Si(Me) -D
D-C(Me) -D
D-C(]CMe
D-C(TMS) -D
2
-D
4.16
4.45
4.08
3.76
3.45
4.12
4.41
4.54
4.17
4.27
4.02
4.14
4.39
4.50
4.09
3.81
3.79
4.15
0.0246
0.0012
0.0019
0.1097
0.2383
0.0101
0.0143
0.0301
0.0596
0.0532
0.0189
0.0368
0.0106
0.0090
0.0191
0.0657
0.0974
0.0115
13.2
13.3
13.0
12.3
13.5
13.2
0.42
0.43
0.66
1.32
1.54
0.51
156
144
206
392
432
159
2
2
2
)-D
f
2
f
2
D-CH -D
a
ꢃ
ꢃ
ꢃ
ꢃ
Calculated excitation energy for the IVCT transition for the [0 ,0 ] and [90 ,90 ] conformer. BW corresponds to a Boltzmann and oscillator
b
ꢃ
ꢃ
ꢃ
ꢃ
strength ( f ) weighted average for all 36 conformers. Calculated oscillator strengths for the IVCT transition for the [0 ,0 ] and the [90 ,90 ]
conformer. BW corresponds to a Boltzmann weighted average for all 36 conformers. Change in static dipole moment calculated from eqn (3).
Parameters for static dipole moments are reported in ESI Table S2.† Transition dipole moment for the IVCT excitation calculated from the
Boltzmann weighted oscillator strengths. Eqn (4) has been used to link the oscillator strength ( f ) to |mab|. Electronic coupling calculated from
c
d
e
f
eqn (1) using the parameters in Table 3. This dyad was only studied computationally.
similar architectures are perhaps viable for molecular electronics
applications. Among the four dyads it is rather surprising that the
experimentally derived coupling for the cross-hyperconjugated
D-C(Me) -D approaches that of the dyad with a cross-p-conjugated
2
bridge. Furthermore, the trend in the coupling among the three
ER bridged dyads (C(Me) > Si(Me) z Si(TMS) ) was not as
2
2
2
2
14
2 2 2
expected from our previous results (Si(TMS) > C(Me) > Si(Me) ).
However, when conformational freedom is taken into account the
theoretical investigation (Table 3) reproduces the experimental
trend. The calculations do, however, predict a marked increase in
2
V12 for D-C(]CMe )-D relative to the compounds with saturated
ER -bridges. The agreement between experimental and computed
2
values for the electronic coupling is striking, particularly because
the calculated transition energies nmax are underestimated by
roughly a factor two. This suggests that the calculated oscillator
strengths are then overestimated, resulting in a cancellation of
errors. As the precise computation of oscillator strengths is
strongly basis set dependent, particularly for transitions with
charge transfer character, the overestimation of the present
oscillator strengths is tentatively a result of the small basis set
(6-31G(d,p)) used in our calculations.
Further insights into the various mechanisms of the elec-
tronic coupling in this series of compounds can be gained by
examination of the trends in oscillator strength ( f ) calculated at
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
the [0 ,0 ] and [90 ,90 ] conformations (as noted above, the
[
ꢃ
ꢃ
0 ,90 ] and [90 ,0 ] conformations have computed oscillator
strengths of zero in all cases because the donor and acceptor p-
41
ꢃ
ꢃ
systems are orthogonal). In the planar [0 ,0 ] conformations
one observes the trend C(]CMe )> Si(TMS) >C(Me) z Si(Me) ,
2
2
2
2
which is what is expected for this series of cross-p-conjugated
and gradually enhanced cross-hyperconjugated dyads based on
14
our previous study. The calculated spin densities give a quali-
tative illustration of this trend (Fig. 5). Here one can see that the
Fig. 5 Spin densities (isovalue ¼ 0.0004) of the singly oxidized state of the four
dyads (hydrogen atoms are not displayed to enhance clarity). The numbers under
2
stronger coupling through the C(]CMe ) bridge is due to a each conformation correspond to the calculated oscillator strengths (f). The left
relatively large presence of spin density on the “neutral” side, column shows the conformations where the inner phenylene group is in the plane
spanned by the two ethynylene groups (2ꢁ(C^C)). This is the conformation
which indicates for a strong interaction between the two states.
+
where cross-hyperconjugation is maximal. The right column shows the confor-
Similarly, D -Si(TMS) -D also shows a rather large spin density in
2
ꢃ
mations where the two inner phenylene groups have been rotated 90 out of the
the center (on the two Si–Si s-bonds), while spin densities for
2
ꢁ(C^C) plane. In this conformation the coupling between the two sides occurs
+
+
D -Si(Me)
2
-D and D -C(Me)
2
-D are much lower in the center via direct through-space p–p-interaction between the ethynylene units. The
(
Si–Me and C–Me s-bonds, respectively). These trends show that cartoons at the bottom illustrate schematically the two types of interactions.
3
528 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
the electronic coupling mediated through cross-hyper-
conjugation can be tuned via the R-groups.
ꢃ
ꢃ
In the [90 ,90 ] conformation, a considerably different
trend for the IVCT oscillator strengths is observed; C(Me) z
2
C(]CMe
coupling for D -C(Me)
2
) > Si(Me)
2
> Si(TMS)
2
. The oscillator strength and
+
2
-D is unexpectedly high, but inspection
ꢃ
ꢃ
of the spin density provides an explanation. At the [90 ,90 ]
conformation the close proximity of the two ethynylene groups
allows for direct p–p-overlap of the in-plane orbital lobes that
protrude into the space near the central atom. That the through-
space p–p-overlap is generally lower in the Si-bridged dyads
can be explained by the long Si–C bonds resulting in longer
ethynylene–ethynylene distances. Yet, other electronic and
geometric effects are also clearly contributing to the strength of
the electronic coupling at this conformation.
Given the general success of our (TD-)DFT calculations in
reproducing the electronic couplings determined experimen-
tally, and in order to further probe the cross-hyperconjugation
effect, we have expanded upon the series of dyads in Chart 1
computationally to include dyads bridged with ER ¼ C(TMS)
2
2
2
Fig. 6 Summary of the results from the conformational scans of D-C(TMS) -D
and CH
found that ER
conjugation when compared to ER
D-C(TMS) -D displays remarkable features (Fig. 6). First, DFT
calculations predict that the IVCT band for D -C(TMS)
2
(D-C(TMS)
2
-D and D-CH
2
2
-D). In our previous report we and D-CH -D at PBE0/6-31G(d,p)//PBE0/6-31G(d) level. The top two panels show
2
¼
CR
2
provided increased cross-hyper- the energy barrier for rotation of the inner phenylene group on the oxidized
14
triarylamine with the inner phenylene group on the neutral side either in-plane,
2
¼
SiR
2
.
Indeed,
ꢃ
or perpendicular to the plane of the ethynylene bridge; i.e., 0 (red squares and
2
ꢃ
curve) or 90 (blue circles and curve). Changes along the x-axis correspond to
+
2
-D would
rotation of the phenylethynyl group on the oxidized side. The dashed horizontal
be the lowest in energy and signicantly more intense than any line marks the thermal energy at 298 K. The lower two panels show the confor-
of the other dimers in this series, more intense even than the mational dependence of the oscillator strengths for the IVCT transition.
cross-p-conjugated D-C(]CMe )-D (Table 3). Secondly, the
2
variation in conformational energy has a stronger resemblance
with that of D-C(]CMe
dyads. Lastly, the oscillator strength at the [0 ,0 ] conformation
of D-C(TMS)
-D is the largest among all the dyads (more than With this series of four dyads we have used IVCT bands to
twice that of D-C(]CMe
)-D). Each of these point to a remark- evaluate the electronic communication through cross-p-conju-
ably strong cross-hyperconjugative coupling, and suggest that gated and saturated ER -bridging units, which have the poten-
2 2
)-D than for any of the ER segmented
Conclusions
ꢃ
ꢃ
2
2
2
with appropriate substituents cross-hyperconjugation can tial to be cross-hyperconjugated. From the analysis of the IVCT
provide stronger coupling than cross-p-conjugation. The CH2 bands we may rst conclude that either cross-p-conjugation or
group, on the other hand, provides for an electronic coupling cross-hyperconjugation results in relatively weak electronic
which is of intermediate strength to those of C(TMS)
C(Me)
, indicating that the coupling strength reects the elec- electronic coupling through the bridge unit of a dyad is strongly
tronegativity of the R group and the relative energy of the local dependent on conformation. This is another illustration of the
p(ER
) orbital. The opposite trend is found for the through- importance of conformational freedom when one models
space p–p interaction because for the three CR2 segmented charge transfer systems.
2
and coupling. Quantum chemical calculations indicate that the
2
2
6,41
The computational results also
dyads this interaction is strongest for R ¼ Me and weakest for indicate that there are two viable pathways for charge transfer.
R ¼ TMS (see Table 3). Taken together, the TD-DFT computa- The rst pathway is based on the cross-(hyper)conjugation and
tions reveal that these ER
2
linkers provide a structural motif this mechanism dominates when the inner phenylene groups
that allows for tuning between weak and moderately strong are in the plane spanned by the ethynylene groups. Calculations
couplings via cross-hyperconjugation, similar to what was indicate that electron releasing R-groups are required for cross-
6
earlier found for cross-p-conjugated bridges.
hyperconjugation in dyads with central saturated ER
-D computational results suggest
charge transfer in these dyads (see cartoon at bottom of Fig. 5). that cross-hyperconjugation can result in stronger coupling
The rst is cross-p-conjugation and cross-hyperconjugation than in the cross-p-conjugated D-C(]CMe )-D system. The
2
segments,
In summary, there are twoparallel mechanisms for mediating and in the case for D-C(TMS)
2
2
ꢃ
ꢃ
which is most pronounced at the [0 ,0 ] conformations. The second pathway for charge mediation is via a through-space p–
second is a through-space p–p interaction between the p-interaction between the two ethynylene groups. This pathway
ethynylene units, and this is most active at the [90 ,90 ] confor- dominates when the inner phenylene groups are rotated out of
mation. The orbital arrangements in these two pathways for the ethynylene–ethynylene plane. The strength of this mode
mediating charge transfer are orthogonal, and consequently, the should have a strong dependance on the distance and angle
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
coupling is zero at the [0 ,90 ] and [90 ,0 ] conformations.
between the two ethynylene groups. Indeed, the access to two
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3522–3532 | 3529

View Article Online
Chemical Science
Edge Article
2
+
different pathways for charge mediation should provide for was prepared and the exact concentration of Cu was deter-
design of molecules with similarly strong coupling strengths at mined from the absorption spectrum (3750 nm ¼ 20 Mꢀ cm
1
ꢀ1
several different conformations. Such compounds could display for Cu(ClO ) in MeCN). A ꢂ1 mM stock solution was prepared
4
2
charge transfer and transport characteristics which are only for the analyte and the exact concentration was determined
modestly conformer dependent, a feature which could prove from titration with the oxidant. A slow degradation of the tri-
useful in molecular electronics applications.
arylamminium radicals could be observed and thus the rela-
2
+
tively fast reaction using Cu as oxidant (reaction complete in
4
4,45
less than 5 s)
was preferred compared to the slow reaction of
Experimental section
Synthesis
+
28
NO (5–15 min for complete reaction). The oxygen or humidity
level did not seem to accelerate the degradation of the oxidized
All reactions were carried out using dry conditions using stan- triarylamines, but the quality of the acetonitrile was of great
dard Schlenk techniques under an atmosphere of argon. importance. Only spectroscopic grade acetonitrile resulted in
ꢃ
Glassware was dried by the use of a heat gun at 650 C under stable aminium radicals. This may be due to reactions with
high vacuum. Dry solvents were obtained from a Pure Solv MD- amylenes, which oen are added as a stabilizer to other grades
4-EN solvent purication system. In the following, the writing of acetonitrile. The analysis of the recorded spectra was per-
“9 : 1 to 4 : 1” for eluent composition means that the eluent was formed with Microso Excel and Igor Pro 6, where an example
changed immediately to the new proportions without interme- of the analytic procedure can be seen in ESI Fig. S6.†
diate steps (no gradient). Microwave-assisted chemistry was
carried out in a Biotage Initiator Sixty scientic microwave
1
Computational protocol
reactor. The H spectra were recorded at 300 or 400 MHz, and
1
3
the C spectra were recorded at 75.4 or 100.6 MHz. Referencing All calculations in this work have been performed with the
46
was made to tetramethylsilane via the residual solvent peaks Gaussian 09 program package where the computational
47
1
13
(
H, CDCl at 7.26 ppm, C D at 7.16 ppm; C, CDCl at 77.16 protocol is an adaptation to the procedure used by Renz et al.
3 6 6 3
ppm, C D at 128.0 ppm). N,N-Bis(4-methoxyphenyl)aniline (1), However, instead of using their custom hybrid functional we
6
6
N,N-bis(4-methoxyphenyl)-4-[2-(trimethylsilyl)ethynyl]aniline (2) decided to use PBE0/6-31G(d) for geometry optimization and
4
8–50
and 4-ethynyl-N,N-bis(4-methoxyphenyl)aniline (3) have been PBE0/6-31G(d,p) for TDDFT.
Two other functionals were also
38
51–54
55
prepared as reported previously. Compounds 3, 4, 5 and 6 tested, LC-uPBE
and CAM-B3LYP, but a benchmark test
were prepared as reported by Emanuelsson et al. Details on the showed that these functionals were ill-suited for characteriza-
14
synthetic procedure can be found in the ESI.†
tion of IVCT transitions. Details on the benchmarking procedure
can be found in the ESI.† Each dyad was optimized in the singly
oxidized form (charge: +1, multiplicity: doublet) using a solvent
Electrochemistry
56,57
continuum model to stabilize the charge (IEFPCM,
acetoni-
Cyclic voltammetry was recorded with a potentiostat-galvano-
stat (AutoLab) controlled by the GPES 4.9 soware (General
Purpose Electrochemical System) using a three electrode setup.
A 2 mm platinum disc electrode was employed as the working
electrode, the counter electrode was a platinum wire and an
trile). At the optimized geometry we also performed TD-DFT to
simulate the IVCT band. Furthermore, to probe the conforma-
tional dependence of the IVCT band potential energy surface
scans on the rotation of the triarylamine moiety about the
ethynylene groups were also carried out. The dihedral angle was
dened as the plane spanned by the ethynylene groups and the
plane created by the phenylene group bound to the ethynylene
group. There are two relevant dihedral angles, one for each
triarylamine, and they were scanned independent of each
other in steps of 30 , between 0 and 150 (0 being when the
phenylene group is in the same plane as the ethynylene groups,
see ESI Fig. S15†). All other coordinates were allowed to relax at
each point along the conformational scan. The conformational
scan was followed by a TD-DFT calculation at each of the 16
geometries. Lastly, we also tested if expanding the basis set with
diffuse function (6-31+G(d,p)) would affect the TD-DFT results,
but no signicant difference (see Table S1, ESI†) was found.
3
AgNO (10 mM, MeCN)/Ag(s) electrode was used as reference.
Spectroscopic grade acetonitrile (Sigma Aldrich, Sweden), dried
˚
with 3 A molecular sieves, was used as solvent and 0.1 M TBAPF
6
(Electrochemical grade, Sigma Aldrich) was added as support-
ing electrolyte. The analyte concentration was ꢂ1 mM and aer
initial scans ꢂ1 mM of ferrocene was added as an internal
reference. The analyte solution was deaerated with Ar (g) for 10
min before each experiment. The potential was allowed to vary
between ꢀ0.3 and +0.9 vs. the reference electrode with a scan
ꢃ
ꢃ
ꢃ
ꢃ
ꢀ1
+
rate of 0.1 V s . All potentials are reported against Fc /Fc.
Spectroscopy
A Cary 5000 spectrometer was used to record absorption spectra
in the visible and near IR region (300–2200 nm). The double
beam setup was employed and 1 cm quartz cuvettes were used
Acknowledgements
in both the reference and sample chamber. The spectra of the We are grateful for nancial support from the Swedish Research
singly oxidized and doubly oxidized dyads was acquired via a Council, Knut and Alice Wallenberg foundation and Uppsala
3
series of titration experiments using copper(II) perchlorate, University through the U MEC KoF07 initiative on molecular
Cu(ClO
4 2
) (Sigma-Aldrich, Sweden), as a chemical oxidant. A electronics. The National Supercomputer Center (NSC) in Link-
stock solution with ꢂ2 mM Cu(ClO
4
)
2
in MeCN (spec. grade) ¨o ping, Sweden, and the Uppsala Multidisciplinary Center for
3
530 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
Advanced Computational Science (UPPMAX), Uppsala, Sweden 26 For cases where the two different adiabatic states are not
are acknowledged for generous allotment of computer time.
isoenergtic the position of the IVCT band corresponds to
the sum of the reorganization energy (l) and the driving
force (DG).
References
2
7 K. D. Demadis, C. M. Hartshorn and T. J. Meyer, Chem. Rev.,
1
2
3
4
A. Aviram and M. A. Ratner, Chem. Phys. Lett., 1974, 29, 277–
83.
H. Song, M. A. Reed and T. Lee, Adv. Mater., 2011, 23, 1583–
608.
G. C. Solomon, D. Q. Andrews, R. P. Van Duyne and
M. A. Ratner, J. Am. Chem. Soc., 2008, 130, 7788–7789.
R. R. Tykwinski, U. Gubler, R. E. Martin, F. Diederich,
2001, 101, 2655–2686.
28 C. Lambert, S. Amthor and J. Schelter, J. Phys. Chem. A, 2004,
108, 6474–6486.
29 C. Lambert and G. N ¨o ll, J. Am. Chem. Soc., 1999, 121, 8434–
2
1
8442.
30 B. S. Brunschwig, C. Creutz and N. Sutin, Chem. Soc. Rev.,
2002, 31, 168–184.
C. Bosshard and P. G u¨ nter, J. Phys. Chem. B, 1998, 102, 31 S. D. Glover and C. P. Kubiak, J. Am. Chem. Soc., 2011, 133,
4451–4465.
8721–8731.
5
6
R. R. Tykwinski and Y. Zhao, Synlett, 2002, 1939–1953.
A. B. Ricks, G. C. Solomon, M. T. Colvin, A. M. Scott, K. Chen,
32 C. Creutz, M. D. Newton and N. Sutin, J. Photochem.
Photobiol., A, 1994, 82, 47–59.
M. A. Ratner and M. R. Wasielewski, J. Am. Chem. Soc., 2010, 33 R. J. Cave and M. D. Newton, Chem. Phys. Lett., 1996, 249, 15–
32, 15427–15434.
19.
R. S. Klausen, J. R. Widawsky, M. L. Steigerwald, 34 B. S. Brunschwig, C. Creutz and N. Sutin, Coord. Chem. Rev.,
1
7
L. Venkataraman and C. Nuckolls, J. Am. Chem. Soc., 2012,
34, 4541–4544.
S. McDermott, C. B. George, G. Fagas, J. C. Greer and
1998, 177, 61–79.
1
35 W. W. Parson, Modern Optical Spectroscopy [electronic
resource]: With Examples from Biophysics and Biochemistry,
Springer-Verlag, Berlin, Heidelberg, 2007.
8
9
M. A. Ratner, J. Phys. Chem. C, 2009, 113, 744–750.
D. Q. Andrews, G. C. Solomon, R. H. Goldsmith, T. Hansen, 36 Debye (D) is a more common unit for dipole moments, but
˚
M. R. Wasielewski, R. P. V. Duyne and M. A. Ratner, J. Phys.
Chem. C, 2008, 112, 16991–16998.
we chose to use [eA] as the unit, since it intuitively relates
the movement of one electron over a certain distance
˚
˚
1
0 G. C. Solomon, D. Q. Andrews, R. H. Goldsmith, T. Hansen,
(in A). 1 eA ¼ 4.802 D.
M. R. Wasielewski, R. P. Van Duyne and M. A. Ratner, J. Am. 37 S. F. Nelsen and M. D. Newton, J. Phys. Chem. A, 2000, 104,
Chem. Soc., 2008, 130, 17301–17308.
10023–10031.
1
1
1 R. S. Mulliken, J. Chem. Phys., 1939, 7, 339–352.
2 R. S. Mulliken, C. A. Rieke and W. G. Brown, J. Am. Chem.
Soc., 1941, 63, 41–56.
38 C. Lambert, G. N ¨o ll, E. Schm ¨a lzlin, K. Meerholz and
C. Br ¨a uchle, Chem.–Eur. J., 1998, 4, 2129–2135.
39 A. J. Bard and L. J. Faulkner, in Electrochemical methods:
Fundamentals and applications, Wiley, 2nd edn, 2001,
ch. 12.3.6, pp. 505–509.
13 I. V. Alabugin, K. M. Gilmore and P. W. Peterson, WIREs:
Comp. Mol. Sci., 2011, vol. 1, pp. 109–141.
2
+
ꢀ1
ꢀ1
1
4 R. Emanuelsson, A. Wallner, E. A. M. Ng, J. R. Smith, 40 Cu in acetonitrile has 3 ¼ 20 M cm at lmax ¼ 750. This
D. Nauroozi, S. Ott and H. Ottosson, Angew. Chem., Int.
Ed., 2013, 52, 983.
makes this
absorption.
a negligible contribution to the overall
1
5 A. Heckmann and C. Lambert, Angew. Chem., Int. Ed., 2012, 41 M. P. Eng and B. Albinsson, Chem. Phys., 2009, 357, 132–
1, 326–392. 139.
6 D. R. Kattnig, B. Mladenova, G. N. Grampp, C. Kaiser, 42 A. Dreuw and M. Head-Gordon, Chem. Rev., 2005, 105, 4009–
A. Heckmann and C. Lambert, J. Phys. Chem. C, 2009, 113, 4037.
983–2995.
43 The coupling value of 1000 cm is for CH
5
1
ꢀ1
2
Cl as solvent and
2 2
1
1
7 C. Lambert, G. Noll and J. Schelter, Nat. Mater., 2002, 1, 69–73.
8 S. F. Nelsen, A. E. Konradsson, M. N. Weaver and J. P. Telo, J.
Am. Chem. Soc., 2003, 125, 12493–12501.
is based on a three state model of the GMH theory, which the
authors prefer for systems with stronger electronic coupling.
44 S. Sumalekshmy and K. R. Gopidas, Chem. Phys. Lett., 2005,
413, 294–299.
45 K. Sreenath, C. V. Suneesh, K. R. Gopidas and R. A. Flowers,
J. Phys. Chem. A, 2009, 113, 6477–6483.
46 Frisch, M. J. and Trucks, G. W., et al., Gaussian 09, Revision
A.02, Wallingford CT, 2009, see ESI for complete citation.†
47 M. Renz, K. Theilacker, C. Lambert and M. Kaupp, J. Am.
Chem. Soc., 2009, 131, 16292–16302.
1
2
2
9 S. Amthor and C. Lambert, J. Phys. Chem. A, 2005, 110, 1177–
1
189.
0 C. Creutz and H. Taube, J. Am. Chem. Soc., 1969, 91, 3988–
989.
1 C. Creutz and H. Taube, J. Am. Chem. Soc., 1973, 95, 1086–
094.
2 N. S. Hush, Coord. Chem. Rev., 1985, 64, 135–157.
3
1
2
2
3 D. E. Richardson and H. Taube, Coord. Chem. Rev., 1984, 60, 48 C. Adamo and V. Barone, J. Chem. Phys., 1999, 110, 6158–
07–129. 6170.
4 M. B. Robin and P. Day, Adv. Inorg. Chem., 1968, 10, 247–422. 49 M. M. Francl, W. J. Pietro, W. J. Hehre, J. S. Binkley,
1
2
2
5 R. A. Marcus and N. Sutin, Biochim. Biophys. Acta, Rev.
Bioenerg., 1985, 811, 265–322.
M. S. Gordon, D. J. DeFrees and J. A. Pople, J. Chem. Phys.,
1982, 77, 3654–3665.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3522–3532 | 3531

View Article Online
Chemical Science
Edge Article
5
5
5
5
0 P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, 1973, 28, 54 O. A. Vydrov, G. E. Scuseria and J. P. Perdew, J. Chem. Phys.,
13–222. 2007, 126, 154109.
1 Y. Tawada, T. Tsuneda, S. Yanagisawa, T. Yanai and K. Hirao, 55 T. Yanai, D. P. Tew and N. C. Handy, Chem. Phys. Lett., 2004,
J. Chem. Phys., 2004, 120, 8425–8433. 393, 51–57.
2 O. A. Vydrov, J. Heyd, A. V. Krukau and G. E. Scuseria, 56 J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005,
J. Chem. Phys., 2006, 125, 074106. 105, 2999–3094.
3 O. A. Vydrov and G. E. Scuseria, J. Chem. Phys., 2006, 125, 57 G. Scalmani and M. J. Frisch, J. Chem. Phys., 2010, 132,
34109. 114110.
2
2
3
532 | Chem. Sci., 2013, 4, 3522–3532
This journal is ª The Royal Society of Chemistry 2013