N-doping of organic electronic materials using air-stable organometallics: A mechanistic study of reduction by dimeric sandwich compounds

Article abstract of DOI:10.1002/chem.201202591

Several 19-electron sandwich compounds are known to exist as "2×18-electron" dimers. Recently it has been shown that, despite their air stability in the solid state, some of these dimers act as powerful reductants when co-deposited from either the gas phase or from solution and that this behavior can be useful in n-doping materials for organic electronics, including compounds with moderate electron affinities, such as 6,13-bis[tri(isopropyl)silylethynyl]pentacene (3). This paper addresses the mechanisms by which the dimers of 1,2,3,4,5-pentamethylrhodocene (1 b ₂), (pentamethylcyclopentadienyl)(1,3,5-trialkylbenzene)ruthenium (alkyl=Me, 2 a₂; alkyl=Et, 2 b₂), and (pentamethylcyclopentadienyl)(benzene)iron (2 c₂) react with 3 in solution. Vis/NIR and NMR spectroscopy, and X-ray crystallography indicate that the products of these solution reactions are 3^.- salts of the monomeric sandwich cations. Vis/NIR kinetic studies for the Group 8 dimers are consistent with a mechanism whereby an endergonic electron transfer from the dimer to 3 is followed by rapid cleavage of the dimer cation. NMR crossover experiments with partially deuterated derivatives suggest that the C-C bond in the 1 b₂ dimer is much more readily broken than that in 2 a ₂; consistent with this observation, Vis/NIR kinetic measurements suggest that the solution reduction of 3 by 1 b₂ can occur by both the mechanism established for the Group 8 species and by a mechanism in which an endergonic dissociation of the dimer is followed by rapid electron transfer from monomeric 1 b to 3. Doped up: Air-stable dimers of pentamethylrhodocene and pentamethylcyclopentadienyl arene ruthenium and iron can be used to n-dope acceptors such as bis[tri(isopropyl)silylethynyl] pentacene. NMR crossover experiments and variable-temperature Vis/NIR kinetic measurements indicate that, depending on the reaction conditions and the choice of dimer and acceptor, this doping can take place by two different mechanisms (see scheme). Copyright

Full text of DOI:10.1002/chem.201202591

DOI: 10.1002/chem.201202591
n-Doping of Organic Electronic Materials Using Air-Stable Organometallics:
A Mechanistic Study of Reduction by Dimeric Sandwich Compounds
Song Guo,[a] Swagat K. Mohapatra,[a] Alexander Romanov,[b] Tatiana V. Timofeeva,[b]
Kenneth I. Hardcastle,[c] Kada Yesudas,[a] Chad Risko,[a] Jean-Luc Brꢀdas,[a]
Seth R. Marder,*[a] and Stephen Barlow*[a]
Abstract: Several 19-electron sandwich
compounds are known to exist as “2ꢀ
18-electron” dimers. Recently it has
been shown that, despite their air sta-
bility in the solid state, some of these
dimers act as powerful reductants
when co-deposited from either the gas
phase or from solution and that this be-
havior can be useful in n-doping mate-
rials for organic electronics, including
compounds with moderate electron af-
N
G
G
E
gonic electron transfer from the dimer
to 3 is followed by rapid cleavage of
the dimer cation. NMR crossover ex-
periments with partially deuterated de-
ꢀ
rivatives suggest that the C C bond in
the 1b2 dimer is much more readily
broken than that in 2a2; consistent
with this observation, Vis/NIR kinetic
measurements suggest that the solution
reduction of 3 by 1b2 can occur by
both the mechanism established for the
Group 8 species and by a mechanism
in which an endergonic dissociation of
the dimer is followed by rapid electron
transfer from monomeric 1b to 3.
finities, such as 6,13-bis
ACHTUNGTREN[UNNG triACHTUNRTGENN(GUN isoACHTNUGERTNpNUGN ropyl)-
ACHTUNGTRENNUNG
addresses the mechanisms by which the
dimers of 1,2,3,4,5-pentamethylrhodo-
cene (1b2), (pentamethylcycloACTHNUTRGNEpUNG enta-
Introduction
barriers to charge-carrier injection or collection at electro-
des.[1] Molecular n-dopants[3] offer the possibility of circum-
venting some of the drawbacks of n-doping with the widely
used alkali metals, which can include the tendency of the
small ions to diffuse within devices[4] and perhaps to act as
electrostatic traps for charge carriers, and, when introduced
as LiF/Al, the possibility of side-reactions between the reac-
tive dopant system and the electron-transport material.[5]
However, in order to dope materials of interest for many or-
ganic electronics applications, a molecular n-dopant operat-
ing by a simple electron transfer (ET) must have a low ioni-
zation potential (IP) and, therefore, cannot be handled in
air, complicating its use.[1] Accordingly, several approaches
have been developed in which air-stable precursors are used
as sources of reactive dopants, either by thermolysis of air-
stable precursors on sublimation or by thermal or photo-
chemical activation subsequent to deposition.[6]
Controlled and spatially localized n- and p-type doping of
organic electron- and hole-transport materials, respectively,
is increasingly used to control the performance of devices,[1]
such as organic light-emitting diodes, organic photovoltaic
cells, and, more recently, organic field-effect transistors,[2] by
increasing the conductivity of materials and/or by lowering
[a] Dr. S. Guo,+ Dr. S. K. Mohapatra,+ Dr. K. Yesudas, Dr. C. Risko,
Prof. J.-L. Brꢁdas, Prof. S. R. Marder, Dr. S. Barlow
School of Chemistry and Biochemistry and
Center for Organic Photonics and Electronics
Georgia Institute of Technology
Atlanta, GA 30332-0400 (USA)
Fax : (+1)404-894-5909
E-mail: seth.marder@chemistry.gatech.edu
stephen.barlow@chemistry.gatech.edu
We have recently shown that the dimers formed by rhodo-
cene, 1a2 (Figure 1),[7] and the Group 8 mixed-ring sandwich
compounds, 2a2–2c2 (Figure 1),[8] which are air stable in the
solid state and moderately air stable in solution, can be used
[b] A. Romanov, Prof. T. V. Timofeeva
Department of Chemistry
New Mexico Highlands University
Las Vegas, NM 87701 (USA)
[c] Prof. K. I. Hardcastle
Department of Chemistry
Emory University
Atlanta, GA 30322 (USA)
[+] These authors contributed equally.
to n-dope materials such as 6,13-bisACTHUNGTREN[NGUN triACHTUNTRGEG(NNUN isoACHTUNGTRENNUNGpropyl)-
AHCTUNGTREGsNNUN ilylethynyl]pentacene (3, Figure 1) and copper phthalocya-
nine,[9] which have much lower electron affinities (EA,
values of 3.0 and 3.1 eV, respectively, in the solid state, as
measured by inverse photoelectron spectroscopy) than other
materials n-doped to date through “air-stable” precursor ap-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201202591.
14760
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

FULL PAPER
ing an understanding of the fundamental organometallic
chemistry of these reactions, further mechanistic informa-
tion may be useful in better establishing the limits of doping
with these compounds and may help with the development
of new dopants that are more powerful dopants and/or
which can be spin-coated in the presence of acceptors in air,
with subsequent thermal or photochemical activation. More-
over, if the mechanisms by which these dimers act as reduc-
tants are understood, they may be useful for in situ genera-
tion of reduced species for spectroscopy, for example, in
studies of organic or inorganic mixed-valence compounds,
or to help in the assignment of features observed in transi-
ent-absorption spectra.
In our previous report we presented preliminary data that
suggest that the solution reaction of 3 with 2b2 or 2c2 does
not in fact proceed by mechanism I, but which are consistent
with the alternative mechanism II (Figure 2).[9a] Here we
present a more thorough mechanistic study of the Group 8
dimers and a comparison to the behavior of the Group 9
dimer 1b2.
Results
Figure 1. Structures of some dimeric n-dopants (12, 22), the corresponding
+
+
C
C
monomeric and neutral (1 , 2 ) and cationic species (1 , 2 ), and 6,13-bis-
ACHTUNGTRENNUNG[tri(isopropyl)silylethynyl]pentacene (3).
Compounds and mechanisms under consideration: As noted
in the introduction, there are two main classes of “2ꢀ18e”
sandwich compound dimers known to be formed by dimeri-
zation of the corresponding 19e species: the dimers of vari-
ous Group 8 mixed cyclopentadienyl/arene complexes[8] and
those formed by several rhodocenes and iridocenes.[7,10a,12]
We were interested in whether there were fundamental dif-
ferences in the reactivity of the two classes. Thus, in addition
to the previously studied Ru and Fe dimers, 2a2–2c2, we ex-
amined [{RhCp*Cp}2] (1b2; Cp*=pentamethylcyclopenta-
dienyl; Cp=cyclopentadienyl)[10a] as a representative of the
various Group 9 metallocene dimers. This particular exam-
ple is more well-suited to the specific mechanistic studies
described below than the dimer of rhodocene itself (1a2),
owing to its greater solubility and the inequivalence of the
two cyclopentadienyl rings, which is convenient for NMR-
based crossover experiments (see below). As an acceptor,
proaches (EA>ca. 3.8 eV). Based on previous evidence
that the rhodocene dimer 1a2 dissociates to the correspond-
ing monomer in the gas-phase,[7a] one might anticipate gas-
phase doping to take place via formation of the 19-electron
monomers (mechanism I, Figure 2), which are suggested by
Figure 2. Two possible mechanisms by which dimers (M2) formed by
we used 6,13-bisACTHUNTGRNEUNG[tri(isopropyl)silylethynyl]pentacene (3) for
C
highly reducing species (M ) can reduce an acceptor (A) even though ET
several reasons. It is one of the less readily reduced accept-
ors that we have successfully doped with these dimeric dop-
ants and, therefore, an example in which a straightforward
exergonic ET from dimer to acceptor can be ruled out. As
shown in the following sections, the corresponding radical
anion has a readily identifiable UV/Vis/NIR spectrum and
the organometallic salts of the anion are stable and soluble
in certain suitably dry and deoxygenated solvents. Electro-
chemical data for the compounds under consideration are
summarized in Table 1, along with DFT estimates of the IPs
of the monomeric and dimeric species.
C
from M2 to A is not thermodynamically favorable. The superscript indi-
cates species with an unpaired electron.[11]
electrochemical data[7b,8–10] to be comparably reducing to
(1a ), or more strongly reducing than (1b , 2a –2c ), deca-
C
C
C
C
A
ACHTUNGTRENNUNG
spin-coated (under nitrogen) from solutions containing
dimers and acceptors (such as 3) are doped without any
thermal treatment and that this doping reaction takes place
in solution. This raises the question of how these dimers can
effect reduction of species, such as 3, to the corresponding
highly air-sensitive radical anions,[11] despite their own air
stability and despite NMR evidence that they retain their di-
amagnetic dimeric structures in solution. In addition to gain-
Figure 2 shows the two pathways that we consider most
probable. The first step of mechanism I is dissociation to the
monomeric 19e species; this is presumably an endergonic
process at room temperature, since, according to NMR spec-
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14761

S. R. Marder, S. Barlow et al.
Table 1. Electrochemical data, estimated free energies of reaction with 3, and calculat-
ed ionization potentials for dimeric and monomeric sandwich compounds.
ceptor providing that the first step remains rate-de-
termining, whereas that of mechanism II will
depend on DGET for the reaction between the
dimer and acceptor. The entropies of activation,
Potentials
DGET [eV, (kJmolꢀ1)][b]
IP [eV][c]
+/0
vs. FeCp2 [V][a]
+/0 [d]
[e]
+
+
(M+/0
)
M2 +3!M2
3
M +3!M 3
M2
M
C
Cꢀ
C
Cꢀ
C
E
(M2
)
E
¼
DS , are expected to be positive and negative for
1a
ꢀ0.75
ꢀ0.95
ꢀ1.09
ꢀ1.06
ꢀ1.85
ꢀ2.06
ꢀ2.70
ꢀ2.13
+0.70 (+68)
+0.50 (+48)
+0.36 (+35)
+0.39 (+38)
ꢀ0.40 (ꢀ39)
5.66
5.23
4.98
5.23
4.82
4.23
3.33
4.23
the dissociative first step of I and the associative
first step of II, respectively. Finally, since the first
step of mechanism II is an ET reaction, its rate is
expected to vary with the polarity of the solvent.
1b
2b[f]
2c
ꢀ0.61 (ꢀ59)
ꢀ1.25 (ꢀ121)
ꢀ0.68 (ꢀ66)
[a] In THF/0.1m nBu4NPF6. [b] Free-energy changes for the ET reaction of 3 with
dimer and monomer, estimated using a value of E1/20/ꢀ =ꢀ1.45 V vs. FeCp2
for 3 in
+/0
the same solvent (reduction of 3 to 32ꢀ is seen at ꢀ1.93 V). [c] Gas-phase adiabatic
Cꢀ
[10a]
[8,9]
Synthesis: The dimers 1b2
and 2a2–2c2
were
ionization potentials obtained from DFT calculations (IP values calculated in the pres-
synthesized by sodium amalgam reduction of the
hexafluorophosphate salts of the corresponding cat-
ions,[8b,9a,15] essentially as described in the literature.
For crossover experiments (see below), selectively
deuterated analogues of 1b2 and 2a2 were synthe-
sized. As shown in Scheme 1, base-catalyzed per-
ence of an implicit solvent continuum representing THF are given in the Supporting
+/0
Information). [d] Peak potential vs. FeCp2
for the irreversible oxidation of the
dimer measured at 50 mVsꢀ1. [e] Half-wave potential vs. FeCp2
for the partially re-
+/0
versible reduction of the cation. [f] Essentially identical electrochemistry is seen for
the 2a system.[9a]
troscopy, all the dopants under consideration exist predomi-
nantly in solution as the diamagnetic dimer species, with in-
sufficient paramagnetic monomer present to be detectable.
When 3 is the acceptor, the subsequent ET step would be
exergonic, according to the electrochemical potentials
(Table 1), and presumably rapid. The first step of mecha-
ACHTUNGTRENNUNGnism II—electron transfer from the dimer to the acceptor—
will also be endergonic in the case of reduction of 3
(Table 1). The second step—cleavage of the dimer cation—
is likely to be rapid, given the irreversibility of the dimer ox-
idation, which, in the case of 1a2, has previously been shown
to be coupled to the formation of the corresponding mono-
meric cation.[13]
Table 2 compares some characteristics of the two mecha-
nisms. In general, both mechanisms will give the same final
products, although, as discussed in the following section, dif-
ferences may occur with excess dimer when the acceptor
can be doubly reduced. Mechanism I is only feasible if the
dimer is indeed in equilibrium with a small concentration of
the corresponding free 19-electron monomer, in which case
a mixture of two closely related dimers is expected to
evolve over time to contain a near-statistical mixture of the
homodimers and the corresponding mixed dimer; according-
ly, such crossover experiments provide a valuable test for
the feasibility of this mechanism. If the first step of each
mechanism is rate-determining, the rates of reaction will be
zero or first order, respectively, with respect to the acceptor,
while somewhat more complicated kinetic scenarios will be
found if the second step of either pathway is rate determin-
ing.[14] The activation barrier for mechanism I will be inde-
pendent of the identity and reduction potential of the ac-
Scheme 1. Synthesis of selectively deuterated RhCp*Cp dimers.
deuteration of [{RhCp*Cl2}2][16] according to the literature[17]
was followed by treatment with freshly sublimed TlCp;[18]
ꢀ
subsequent addition of aqueous NH4+PF6 gave the Cp*-
ꢀ
deuterated salt 1b’+PF6 . An analogue with cyclopentadien-
yl deuteration, 1b’’+, was obtained in an analogous fashion
using [{RhCp*Cl2}2][16] and Tl
ACTHNUGTRNEG(UN C5D5), which was prepared
from deuterated cyclopentadiene[19] using a modification of
literature procedures.[20] The hexafluorophosphate salts were
reduced to the corresponding deuterated dimers, 1b’2 and
1b’’2, in a similar fashion to the
non-deuterated parent com-
Table 2. Comparison of proposed mechanisms for n-doping with organometallic dimers.
pounds.
A mixture of 1b’2,
¼
¼
Rate law[a]
DG depen-
DS
Influence
of solvent e
1b’’2, and 1b’1b’’, for which two
isomers are possible depending
on whether the central C C
bond of the dimer is formed be-
tween deuterated Cp* and deu-
Mech- Products
anism
Crossover
dence on A[a]
Cꢀ
M+
M+
A
A
(+A2ꢀ?) yes
d[M2]/dt=ꢀk[M2]
no
>0
<0
Weak
potentially strong
ꢀ
I
II
Cꢀ
not necessarily d[M2]/dt=ꢀk[M2][A] yes
[a] Assuming that in each case the first step is rate-determining.[14]
14762
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

Organic Electronic Materials
FULL PAPER
terated Cp (isomer 1, Scheme 1) or between non-deuterated
Cp* and non-deuterated Cp (isomer 2), was also obtained
3 and the appearance of new features. We attribute these
new features to 3 based on the similarity of the spectra to
Cꢀ
ꢀ
by reduction of a mixture of the two salts, 1b’+PF6 and
that obtained by brief exposure of 3 in THF to Na–K alloy.
Moreover, the appearance of the spectra, which consist of
severally vibronically structured bands significantly batho-
chromically shifted from those of the corresponding neutral
compound, is typical of acene radical anions,[25] and has
strong similarities to that of pentacene radical anion.[25,26]
Thirdly, crystals have been isolated from the stoichiometric
reactions of 3 and 1b2 or 2b2 in benzene; although in both
cases the crystals were of insufficient quality to justify a de-
tailed discussion of bond length and angles, the unit cells
were found to contain equal numbers of monomeric sand-
wich moieties and 3-like molecules, consistent with forma-
ꢀ
1b’’+PF6 , as a control for the crossover experiments de-
ACHTUNGTRENNUNGscribed below.
ꢀ[15b]
As shown in Scheme 2, 2a+PF6
was converted to an
analogue with deuterated mesitylene methyl groups, 2a’+
ꢀ
PF6 , using NaOD in D2O. Although use of these conditions
+
+
Cꢀ
Cꢀ
tion of the salts 1b 3 (Figure 3) and 2b 3 (see Support-
Scheme 2. Synthesis of selectively deuterated RuCp*(mesitylene) dimers.
has previously been found to lead to complete deuteration
of both Cp* and arene methyl groups of various
[FeCp*(C6MenH6ꢀn)]+ derivatives over 12 h,[21] deuteration
of 2a+ takes place almost exclusively on the mesitylene
methyl groups (only 1–5% deuteration of Cp* positions,
even after 3 d at reflux). The analogue with deuterated me-
+
Cꢀ
Figure 3. View of the crystal structure of 1b 3 . The asymmetric unit
contains two crystallographically independent 1b+ molecules, one of
Cꢀ
which is shown here, two half-molecules of 3 , one of which corresponds
Cꢀ
to half of the centrosymmetric anion shown here, and one complete 3
anion. Atoms are shown as spheres of arbitrary size and hydrogen atoms
are excluded for clarity.
sitylene aromatic positions, 2a’’2+, was obtained from the re-
ꢀ[22]
action of [RuCp*
(NCCH3)3]+PF6
and the appropriately
ing Information for more details of both structures). More-
over, the average Rh C bond lengths (2.182(11) and
deuterated mesitylene derivative[23] in dichloroethane.[15b] As
for the other dimers, 2a’2, 2a’’2, and a mixture containing
2a’2a’’ were obtained by reduction of the appropriate salts.
ꢀ
2.182(15) ꢃ for two crystallographically independent mole-
cules) in the structure obtained from reaction with 1b2 are
similar to those seen in the structures of other rhodocenium
ꢀ
Reaction products: As noted in Table 1, both mechanisms
proposed in Figure 2 are generally expected to give the
same final products: the 18-electron monomeric sandwich
salts, including that of 1b+PF6 (2.178(5) ꢃ),[10b] and some-
what shorter than that seen for the only structurally charac-
terized
19-electron
rhodocene,
[RhACHTUGNTERN(NUG C5Ph4H)2]
cations (M+) and the radical anion of the electron-transport
(2.262(15) ꢃ).[27] The average Ru C bond length
ꢀ
[11]
Cꢀ
molecule (A ). Several pieces of experimental evidence
(2.215(6) ꢃ) in the structure obtained using 2b2 is also typi-
do indeed suggest that the products are M+ and A . Firstly,
cal for a ruthenium mixed-cyclopentadienyl/arene cation
Cꢀ
ꢀ
1H NMR spectroscopy of solutions containing 3 and 1a2,
1b2, or 2b2 (ca. 1:1 to 1.5:1 molar ratios) in
[D5]chlorobenzene (1a2, 1b2) or [D8]THF (2c2) shows loss
of the dimer resonances and appearances of resonances at-
tributable to the corresponding cations, while the signals at-
tributed to 3 are considerably broadened, consistent with
(e.g., 2.219(5) ꢃ for [RuCp*
(C6Me6)]+PF6 ).[28]
Reactions in solution between excess dimer and 3, howev-
er, suggest mechanistic differences between the Group 9 and
Group 8 dopants. As discussed below, in the case of 1b2, the
Cꢀ
absorptions of 3 are replaced by those of 3 , which are in
turn replaced by features that we attribute to 32ꢀ, based on
the similarity to the spectrum obtained by prolonged expo-
sure of 3 in THF to Na–K alloy. Moreover, addition of neu-
tral 3 to a solution obtained by prolonged Na–K reduction
Cꢀ [24]
rapid electron exchange between 3 and 3 , with no other
significant diamagnetic species being observed. Secondly,
Vis/NIR spectroscopy of mixtures of dimeric dopants and 3
show the disappearance of the characteristic absorptions of
Cꢀ
leads to formation of 3 , consistent with the expected com-
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14763

S. R. Marder, S. Barlow et al.
proportionation reaction of 3 and 32ꢀ. In contrast, the ab-
sorptions attributed to the dianion are not seen when excess
2b2 or 2c2 is used as the dopant. In the case of mechanis-
Cꢀ
m II, ET from the dimers to 3 to form 32ꢀ is anticipated to
be considerably more endergonic than ET from dimer to 3,
meaning that the second reduction is likely to be many
orders of magnitude slower than the first, that is, that the
dianion is unlikely to be seen on a comparable timescale to
that over which the monoanion is generated. On the other
hand, if mechanism I is operative and the cleavage step is
rate-determining, and if the reaction of the 19-electron
Cꢀ
monoACHTUNGTRENNUNGmeric sandwich compound with 3 is sufficiently exer-
gonic to make the overall cleavage plus the ET reaction
Cꢀ
thermodynamically feasible (values of DG for ET to 3 are
C
C
estimated to be ꢀ0.13, ꢀ0.77, and ꢀ0.20 eV for 1b , 2b , and
2ꢀ
C
2c , respectively), 3 will, once all neutral 3 is consumed by
reduction to 3 , likely be formed at a similar rate to that at
which 3 is initially formed. Thus, these observations are
Cꢀ
Cꢀ
consistent with 1b2 reacting through mechanism I under
these conditions. Failure to observe 32ꢀ using 2b2 and 2c2 is
consistent with either reduction by these dimers proceeding
by mechanism I, but with the cleavage reactions being more
Cꢀ
endergonic than for 1b2, meaning the ET to 3 is insuffi-
ciently exergonic to render the overall reaction feasible, or
with these dimers reacting by mechanism II.[29]
Cꢀ
The formation of M+ and A (or A2ꢀ) can be contrasted
with the mechanisms proposed for other “air-stable precur-
sor” dopants. In the case of heating organic salts,[6a–d,f] for ex-
ample, the highly reducing organic radicals believed to be
the active dopant cannot be formed from the salt without
formation of side products; moreover, another species, be-
lieved to the corresponding leuco form of the dye, has been
detected in the sublimate obtained by heating pyronin B in
vacuum.[6f] In the case of thermally or photochemically acti-
vatable dopants such as leucocrystal violet[6e] and 1,3-di-
methyl-2-[4-(dimethylamino)phenyl]-2,3-dihydro-1H-benzo-
Figure 4. 1D NOE spectra of a) a mixture of 1b’2 and 1b’’2 in [D8]toluene
(* indicates residual 1H signal of the solvent) and b) of the product ob-
tained after sublimation (120–1258C, ca. 25–30 mTorr) of a mixture of
1b’2 and 1b’’2 in [D6]benzene, in both cases irradiating at the bridgehead
methyl group (i-CH3) of the h4-Cp* moiety. A spectrum similar to b) is
also observed for the product obtained by reducing a mixture of 1b’+
ACHTUNGTRENNUNG[d]ACHTUNGTRENNUNG
imidazole[6h,30] a hydrogen radical must also be lost from
the dopant precursor; indeed doping fullerenes with the
former compound has been shown to result in the formation
of partially hydrogenated fullerenes.[6e]
ꢀ
PF6 and 1b’’+PF6ꢀ. The only NOE seen in a) is to the a-CH3 group of
the h4-Cp* ring, whereas in b) an additional interring NOE is seen to the
i-H/a-H resonances of the h4-Cp moiety. The absence of the interring
NOEs in a) indicates no significant formation of 1b’1b’’ in solution on
the timescale of the experiment (on longer timescales H/D scrambling is
observed for the methyl groups), whereas its observation in b) indicates
that sublimation leads to the formation of isomer 2 (and presumably also
isomer 1) of 1b’1b’’. c) The structures of the relevant dimers and the pro-
tons of 1b’’2 and isomer 2 of 1b’1b’’ between which NOEs are observed
in a) and b).
Crossover experiments: The synthesis of partially deuterated
versions of 1b2 and 2a2 described above enabled us to per-
form crossover experiments to test whether the dimers are
in equilibrium with the corresponding monomers in solution
or in the vapor phase. Although mass spectrometry superfi-
cially appears an attractive means of detecting the mixed
species, the ready fragmentation of the dimer cations leads
to detection of only the monomeric 18e cations by using
either EI and ESI modes of ionization. Accordingly, we
used 1H NMR techniques. Nuclear Overhauser effect
(NOE) experiments[31] proved particularly useful; examples
are shown in Figure 4 and in the Supporting Information. In
the case of 1b2, NOEs between the bridgehead (i-H) and/or
neighboring (a-H) protons and the bridgehead Cp* methyl
group (i-CH3) are seen for the product of a mixed reduction
of 1b’2 and 1b’’2 and are attributable to isomer 2 of 1b’1b’’
(Scheme 2, Figure 4c), but are not seen for either pure 1b2’
or 1b’’2 in which the bridgehead Cp* and Cp positions, re-
spectively, are deuterated. The most useful signature of the
mixed ruthenium dimer 2a’2a’’ was found to be the NOE
between the o-CH3 group of a C6D CHTUNGTREN(NUNG CH3)3 ring and the m-H
3A
of a directly attached C6H3ACHTUGNRTEN(NUNG CD3)3 ring (see Supporting In-
formation).
14764
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

Organic Electronic Materials
FULL PAPER
Table 3. DFT enthalpies and free energies [kJmolꢀ1] of cleavage for or-
We have previously shown that there is no detectable for-
mation of 2a2b in crossover experiments between 2a2 and
2b2 in [D8]toluene, even over seven days at over 1008C,[9a]
strongly suggesting that mechanism I is not responsible for
the solution doping reactions observed using the ruthenium
dimers. Solution crossover experiments between 1b’2 and
1b’’2 are less straightforward: spectra recorded in
[D8]toluene immediately after mixing (Figure 4a) show no
evidence for the formation of 1b’1b’’, but spectra acquired
after 1 h at room temperature exhibit significant changes,
which can be attributed to H/D scrambling between the
methyl groups of 1b’2 and 1b’’2 and which complicate the re-
liable detection of any NOEs characteristic of 1b’1b’’.[32] Al-
though the mechanism for the deuterium scrambling is un-
clear, it is worth noting that it occurs for each of the four in-
equivalent Cp* CH3 resonances at the same rate, suggesting
that the different methyl groups are rendered equivalent by
some process on a timescale shorter than or comparable to
that of the scrambling reaction, which is at least consistent
with dissociation to the monomer.[33] Moreover, co-sublima-
tion of 1b’2 and 1b’’2 (120–1258C, ca. 25–30 mTorr) gave a
ganometallic dimers.[a]
+
C
C
C
M2!2M
M2 !M +M+
DH
DG
DH
DG
1a
1b
2b
2c
+98.2
+99.5
+193.3
+99.8
+12.3
+42.7
+110.1
+14.2
+21.8
+11.7
+43.2
+9.4
ꢀ34.5
ꢀ48.9
ꢀ26.8
ꢀ52.0
[a] These values refer to isolated molecules; see Supporting Information
for calculations performed for a dielectric continuum representing THF.
Free energies are calculated at 298.15 K.
are exergonic at 298 K (we note that these enthalpies and
free energies may be more positive/less negative than their
true values due to overstabilization of the dimer cation as a
result of the known tendency of DFT to give overdelocal-
ized structures for odd-electron species such as these).[36]
Moreover, if the dielectric effects of a solvent are simulated,
+
C
the cleavage reactions of the M2 species are calculated to
be markedly more exothermic and exergonic (see Support-
ing Information). The DFT calculations were also used to
obtain estimates of the ionization potentials of the neutral
dimers and monomers; as shown in Table 1, the trends are
broadly consistent with those seen in the corresponding
electrochemical potentials, although it is worth noting that
experimental and computational values for the dimers are
potentially affected by electrochemical irreversibility and
yellow solid with essentially identical H, H, 13C{1H}, and
1H NOE (Figure 4b) NMR spectra to those seen for the
product of the mixed reduction of 1b’+ and 1b’’+, in partic-
ular exhibiting strong NOEs characteristic of 1b’1b’’
(isomer 2), indicating that the dimers dissociate to mono-
mers in the gas phase, consistent with previous observations
for 1a2.[7a] In contrast, 2a’2a’’ was not formed to a detectable
extent when a mixture of 2a’2 and 2a’’2 was sublimed at
145–1508C (25–30 mTorr).[34] Thus, although the crossover
experiments do not provide unambiguous evidence for the
formation of monomeric 1b in solution, the vapor-phase
crossover experiments do indicate that the energetic barrier
to dissociation is much lower for the rhodium dimer 1b2
than for the ruthenium species 2a2. Moreover, the sublima-
tion crossover experiments indicate that 1b2 dissociates to
monomeric 1b, as previously reported for 1a2,[7a] and that
vapor-phase deposition of electron-transport materials ach-
ieved by co-deposition with the rhodium dimers may pro-
ceed by mechanism I. On the other hand, 2a2 dimers appa-
rently sublime intact, inconsistent with doping by vapor-
phase deposition doping with 2a2 proceeding by mechanis-
m I and, therefore, suggesting that, as in solution, doping
with the ruthenium species proceeds by mechanism II.
1
2
the effects of overdelocalization in M2 +, respectively.
C
Rate laws: As noted in the discussion of reaction products,
the Vis/NIR spectra of 3/dimer mixtures evolve to show the
Cꢀ
appearance of signals that we attribute to 3 ; this evolution
is relatively slow, with reactions at reagent concentrations of
10ꢀ2–10ꢀ4 m approaching completion on a timescale of hours
at room temperature, making determination of the rate law
and rate constants possible by monitoring the signals of 3
Cꢀ
and/or 3 over time. Figure 5a shows, as an example, the
temporal evolution of spectra of a chlorobenzene solution
of 3 and 2b2 with the dimer in a large excess, that is, under
“pseudo-single-reactant” conditions in which [2b2] is effec-
tively constant over time. Figure 5b also shows how the ab-
Cꢀ
sorbances at 644 nm (attributable to 3) and 750 nm (3 )
change over time; in both cases these evolutions can be fit
to exponential functions indicating that the reaction is first
order in the rate-limiting reagent, 3. Similar experiments
performed with excess 3 indicate that the reaction is also
first order in 2b2. These experiments therefore suggest that
the rate law can be described by Equation (1), in which k
has a value of 2.0ꢀ10ꢀ1 mꢀ1 sꢀ1 at room temperature.
DFT calculations: The differences in the sublimation cross-
over experiments for rhodium and ruthenium systems are
qualitatively consistent with quantum-mechanical estimates
of the enthalpies and free energies of cleavage obtained
using density functional theory (DFT) with the M06 func-
tional (Table 3); these indicate that the ruthenium dimer
2b2 is much less readily dissociated than the rhodocene
dimers 1a2 or 1b2, and also suggest that the iron dimer 2c2
should be cleaved much more readily than its ruthenium an-
alogue.[35] The calculations also show that the radical cations
of the dimers are all more readily cleaved than the corre-
sponding neutral species and that these cleavage reactions
d½2b2ꢁ 1 d½3ꢁ
1 d½3ꢂꢀꢁ
ð1Þ
¼
¼ ꢀ
2 dt
¼ ꢀk½2b2ꢁ½3ꢁ
dt
2 dt
The rate law is, therefore, fully consistent with mecha-
AHCTUNGTRENNUNG
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14765

S. R. Marder, S. Barlow et al.
Figure 5. Top: Vis/NIR spectra of a solution of 3 and excess 2b2 in
chlorobenzene at various times after mixing. The peaks decreasing in ab-
sorbance are attributed to 3, while those increasing are attributed to 3 .
Bottom: plots of the absorbances at 750 and 646 nm versus time for the
reaction shown in the top part, the solid lines being fits to a first-order
expression.
Figure 6. Top: Vis/NIR spectra of a solution of 3 and excess 1b2 in chlor-
ACHTUNGTRENNUNG
obenzene at various times after mixing, showing the rise and fall of the
Cꢀ
Cꢀ
peak at 740 nm attributed to 3 and the growth of a feature at 670 nm
that we attribute to 32ꢀ. Bottom: plots of the absorbances at 740 and
646 nm versus time.
The rhodium dimer 1b2 shows less straightforward behav-
Cꢀ
ior. When 1b2 is in excess, the absorptions attributed to 3
do not asymptopically approach a saturated value, but ap-
proach a similar maximum absorbance to that seen when
2b2 or 2c2 is the dopant, yet in an almost, but not exactly,
linear fashion (Figure 6); thus, this temporal evolution sug-
gests that neither mechanism I nor II alone can be used de-
scribe the process (when 3 is in excess, the temporal evolu-
tion is consistent with a reaction that is first order in 1b2, as
is expected for either mechanism I or II, or with both occur-
ring in parallel). Accordingly, initial rates were determined
from the initial slope of the plots of absorbance versus time
for different initial concentrations of 1b2 and 3 The varia-
tion in rate with initial concentrations is fully consistent
with a rate law of the form given in Equation (2), which is
also consistent with the unusual temporal evolution of the
spectra and suggesting that, for this reaction, both mecha-
nisms I and II are in competition at room temperature.
Figure 7. Plot showing the initial rate divided by the initial concentration
of dimer versus the initial concentration of acceptor for the reaction of
1b2 and 3 in chlorobenzene at room temperature. The linear fit with non-
zero intercept shown indicates that the rate can be expressed as a sum of
terms zero and first order in 3, as shown in Equation (2).
placed by bands that we attribute to 32ꢀ, the similar time-
scale suggesting that the dianion is formed by mechanism I.
Furthermore, fitting of the data for this second reduction for
several such experiments is consistent with the rate law
given by Equation (3).
d½1b2ꢁ 1 d½3ꢁ
1 d½3ꢂꢀꢁ
ð2Þ
[1b2]0 versus [3]0 (where the zero
¼
¼ ꢀ
2 dt
¼ ꢀkI½1b2ꢁ ꢀ kII½1b2ꢁ½3ꢁ
dt
2 dt
A plot of (d
[1b2]/dt) /
N
d½3ꢂꢀꢁ
subscript denotes the value at t=0, Figure 7) allows the rate
constants kI and kII to be determined as 1.1ꢀ10ꢀ4 sꢀ1 and
7.2ꢀ10ꢀ3 mꢀ1 sꢀ1, respectively. As noted above and as shown
in Figure 6, further complications are observed at longer re-
0:5
¼ ꢀk½1b2ꢁ ½3ꢂꢀꢁ
ð3Þ
dt
This rate law suggests that this reaction also proceeds by
C
Cꢀ
mechanism I, but with the second step, ET from 1b to 3 ,
Cꢀ
action times, at which the signals assigned to 3 are re-
14766
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

Organic Electronic Materials
FULL PAPER
being rate-determining,[14] in contrast to what is
seen for the reduction of 3 to 3 . This change of
Table 4. Arrhenius and Eyring parameters for reactions of dimers and 3 in chloroben-
zene.
Cꢀ [37]
Arrhenius
Eyring
rate-determining step implies that the barrier for
¼
¼
¼
Cꢀ
C
the reduction of 3 to 32ꢀ by 1b is larger than that
lnA
Ea
DH
[kJmolꢀ1
DS
DG300
[kJmolꢀ1
]
A
]
G
]
[Jmolꢀ1 Kꢀ1
]
U
Cꢀ
for 3 to 3 ; this is consistent with expectations
1b2 kI (32.0ꢃ0.7)[a]
N
(+105ꢃ1.8)[a] (+12.8ꢃ6.0)[a]
ACHTUNGTRENNUNG
based on Marcus theory, the estimated driving
forces (ꢀDG) for these two reductions being 0.13
and 0.61 eV, respectively.
1b2 kII (17.1ꢃ1.7)[a] (54.3ꢃ4.5)[a] (+51.7ꢃ4.5)[a]
(ꢀ112ꢃ14)[a]
(+85.3ꢃ8.8)[a]
(+77.0ꢃ1.2)
(+80.9ꢃ0.6)
2b2
2c2
(16.6ꢃ0.2)
(+42.4ꢃ0.6)
ACHTUNGTRENNUNG
(18.9ꢃ0.1)
(+52.0ꢃ0.3)
[a]Acquired from plots of initial rate data.
Temperature-dependent kinetic data: Kinetic data
were also acquired at elevated temperatures. In the
case of the Group 8 dimers, the rate constants at each tem-
perature can be easily extracted from the temporal evolu-
+27 to +64 Jmolꢀ1 Kꢀ1, depending on the solvent,[40] and a
value of +27 Jmolꢀ1 Kꢀ1 is found for the dissociation of cin-
namyl-4-nitrobenzenesulfenate to a radical pair in tol-
uene.[41] However, as noted in the Experimental Section, ad-
ditional complications associated with the use of initial rate
Cꢀ
tion of the 3 absorption, whereas for the rhodium dimer,
1b2, the first- and second-order rate constants, kI and kII,
were determined from plots of initial rate data analogous to
that shown in Figure 7 at each temperature. In all cases
linear Arrhenius and Eyring plots are obtained (Figure 8).
¼
data may mean that this values of DS is artificially low. In
any case, the Eyring plot of the kI data is qualitatively differ-
ent from that of the kII data or from those for the Group 8
compounds and is completely consistent with a dissociative
rate-determining step, as required for mechanism I. Second-
ly, the Arrhenius activation barriers, Ea, and the room-tem-
¼
perature free energies of activation, DG (Table 4), for the
reactions that we assign to mechanism II are all of compara-
ble magnitude to the electrochemical estimates for DGET for
electron transfer from the dimers to 3 (see Table 1). The
¼
values of DG are all larger than the corresponding values
of DGET, as expected from the expected contribution of the
reorganization energy, l, to the barrier. The values of l im-
¼
plied by comparison of DG and DGET are rather large, but
differences in solvation and ion pairing between the kinetic
and electrochemical experiments may also play a role. Im-
¼
portantly, the trends in the values of DG obtained from
the Eyring plots follow those in the electrochemically esti-
mated values of DGET.[42]
Discussion
~
*
^
ACHTUNGTRENN(GU kIIACHTUNTGNER(NGNU 1b2)); & lnACHTUGNRETNNUG IIAHCUTGNNERNG(U 2c2));
Figure 8. Arrhenius [top:
ln(kIIA(2b2))] and Eyring [bottom: ln(k
(kIIA(2c2)/T); ln
ln(k
G
ln
(1b2)/T); ln
ACHTUNGTRENNUNG(kIIACHTUNGTRENNUNG(2b2)/T)] plots of variable-temperature rate-constant
~
The reaction order, the negative entropy of activation, the
magnitude of the activation barriers, and the resistance of
2a2 to cleavage observed in crossover experiments and indi-
cated by DFT calculations, suggest that the organorutheni-
um compounds exclusively react with 3 by means of mecha-
nism II.[43] The kinetic studies reveal the reduction of 3 by
2b2 is fairly slow; for example, in the pseudo-first-order re-
action of 2b2 and 3 shown in Figure 5, the reaction takes
about 30 min to be half complete. This can be contrasted to
simple exergonic one-electron reductions, such as reduction
of 3 by the monomeric 19-electron decamethylcobaltocene,
which are typically rather fast. However, in the presence of
higher concentrations of 3, the half life for the dopant, given
by Equation (4), is considerably shorter.
G
E
IA
ACHUTNGTNRENUN(G kIIACHTUNGTRENNUNG
^
E
U
data for the reduction of 3 with organometallic dimers.
Eyring plots of the second-order rate constants for the
Group 8 dimers and for the 1b2 pathway dominant at higher
3·ꢀ concentrations (kII) yield negative entropies of activa-
¼
tion, DS , consistent with an associative reaction, which in
turn is consistent with mechanism II. Although the extent to
which the Eyring equation can be applied to ET reactions is
¼
unclear,[38] the values of DS are comparable in magnitude
with Eyring-based values reported for many other intermo-
lecular outer-sphere ET reactions in solution.[39] The value of
¼
DS obtained from an Eyring plot of kI data for 1b2 is posi-
tive, but small in magnitude. A somewhat larger value might
be anticipated for a dissociation reaction that, owing to its
endergonicity, has a relatively late transition state; for exam-
ln 2
k½3ꢁ0
t1=2
¼
ð4Þ
¼
ple, DS for the homolysis of tert-butyl peroxide varies from
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14767

S. R. Marder, S. Barlow et al.
For example, at a typical spin-coating concentration of 3
of 18 mgmLꢀ1 (ca. 0.028m), t1/2 is expected to be about
2 min, indicating that the doping reaction is likely to be
complete within the time taken to prepare and deposit a sol-
ution.[9b] Thus, such spin-coating should be carried out in
inert atmosphere to prevent aerial oxidation of the highly
Cꢀ
sensitive 3 . This rate is also expected to be highly depen-
ACHTUNGTRENNUNGdent on the reduction potential of the electron-transport
material, occuring even more rapidly for more easily re-
duced materials, as in the successful doping of a naptha-
lene–diimide-based polymer reported in references [9a,b],
and more slowly for more challenging substrates. Indeed, we
do not observe the formation of 32ꢀ in the presence of
excess Group 8 dopants, although from our current data it is
not clear whether formation of this species is thermodynam-
ically feasible using these dimers at room temperature (see
below).
On the other hand, 1b2 appears to react by means of both
mechanisms I and II, with the dominant mechanism being
dependent on the experimental conditions. In particular, the
ratio of the contributions to the rate from mechanism II to I
is given by Equation (5).
Figure 9. Free-energy diagram for the reaction of 1b2 and 3. Quantities
marked * are upper limits and may in fact be more endergonic/less exer-
gonic than indicated.
tively). It is worth noting that overall free energy for reac-
tion between dopants and acceptors depends, regardless of
mechanism, on the redox potentials for the monomeric 19-
electron sandwich compounds and of the acceptor and on
the free energy required to dissociate the neutral dimer to
monomers. Given our failure to observe crossover for the
ruthenium species and the DFT values of DHdiss and DGdiss
for Rh and Ru dimers, this free energy of dissociation is
likely much greater for 2a/b2 than the ꢄ25 kJmolꢀ1 estimat-
ed for 1b2.
ꢀ
ꢁ ꢂꢀ
ꢁ
k
d½1b2ꢁ
dt
d½1b2ꢁ
dt
II
¼ ꢀ ½3ꢁ
kI
I
ð5Þ
II
Thus, in spin-coating of doped films based on 3, that is, at
much higher [3] than used here, mechanism II is likely to
dominate. However, kII is expected to be strongly dependent
on the redox potential of the acceptor, such that mecha-
ACHTUNGTRENNUNGnism I may still be important in solution processing of
doped films based on more challenging acceptors (such as
those with redox potentials similar to the second reduction
potential of 3). The different temperature dependences of kI
and kII also mean that mechanism I may be more important
when the doping is carried out at elevated temperature.
The data obtained for 1b2 and 3 also enables us to partial-
ly construct a free-energy diagram for the potential reaction
pathways for this system (Figure 9). A key observation is
the formation of 32ꢀ in the presence of excess 1b2,[44] indicat-
Finally, we note some implications for the selection and/
or design of new dimers for use as n-dopants in organic elec-
tronics applications. It would be desirable if films consisting
of electron-transport materials and dopants could be pro-
AHCTUNGTREGcNNUN essed in air. In the case of acceptors forming air-sensitive
radical anions, this could be achieved if the solution reaction
with the dopant were slow at room temperature, the doping
being activated thermally or photochemically subsequent to
deposition in inert atmosphere (possibly subsequent to
device encapsulation), as has been demonstrated for the
doping of fullerenes with organic air-stable dopant precur-
sors, such as leucocrystal violet.[6e] In order to achieve this
with dimeric dopants, it will be necessary to effectively sup-
press both mechanisms I and II at room temperature. The
rate of the latter process can be limited (for a given accept-
or) by selecting a dimer with a more cathodic value of Eox
and, as noted above, will be slower for more challenging ac-
ceptors. While suppression of mechanism I can be achieved
Cꢀ
ing that DGꢄ0 for the reaction of 1b2 and 3 . According to
the electrochemical data, the contribution of ET from
Cꢀ
monoACHTUNGTRENNUNG
ing that the free energy associated with cleavage of the 1b2
dimer is ꢄ +25 kJmolꢀ1 (DFT affords values of +43.7 and
+1.8 kJmolꢀ1 in the gas phase and in THF respectively; see
Table 3 and Table S2 in the Supporting Information). This
C
value, in combination with the free energy of ET from 1b
to 3 allows us to estimate the overall free-energy change for
Cꢀ
the reaction of 1b2 and 3 to give 1b+ and 3 to be more ex-
by selecting a dimer with a stronger C C central bond, an
ꢀ
ergonic than ꢀ93 kJmolꢀ1. Since the products are the same
increased free energy of dissociation, as noted above, will
limit the free energy of the overall doping reaction and,
therefore, the range of materials that can be doped. Thus, in
order to achieve dopants that can both be processed in air
and that have the strongest possible reducing abilities, care-
ful control of both the redox potential and dissociation
energy of the dimer will be required.
for both pathways, the overall free-energy of reaction can
be used, in conjunction with electrochemical data, to deduce
+
C
that the free energy change for the cleavage of 1b2 is
highly exergonic ( ꢄꢀ84 kJmolꢀ1; cf. DFT estimates in
Table 3 and Table S2 in the Supporting Information of
ꢀ48.9 and ꢀ91.8 kJmolꢀ1 for the gas phase and THF respec-
14768
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

Organic Electronic Materials
FULL PAPER
[{RhACTHNUTRGNEGN(U [D15]Cp*)Cl2}2]: This compound (735 mg, 86%, deuteration of CH3
Conclusion
groups estimated as 97% using 1H and 2H NMR spectroscopy and
MALDI mass spectrometry) was prepared according to the literature
procedure.[17] 2H NMR (61.4 MHz, chloroform): d=1.58 ppm (brs, CD3,
Crossover experiments and kinetic data indicate that solu-
tion reduction of acceptors using the air-stable dimers of
rhodocenes and Group 8 mixed-ring sandwich compounds
as n-dopants can take place by at least two distinct mecha-
nisms. In the case of the pentamethylrhodocene dimer, 1b2,
both mechanisms are operative and their relative impor-
tance is expected to depend on the choice of acceptor and
of experimental conditions, whereas only one mechanism
seems to be feasible for the ruthenium derivatives 2a2 and
2b2, at least at the temperature ranges investigated. This
mechanistic information may be useful in the development
of new dopants/reductants with different properties.
[D15]C5Me5); 13C{1H} NMR (100 MHz, [D]chloroform): d=94.0 (d,
1
1JC,Rh =8.8 Hz, CCD3, [D15]C5Me5), 8.6 ppm (1:3:5:7:5:3:1 septet, JC,D
=
19.8 Hz, CD3, [D15]C5Me5).
ꢀ
[Rh
N
2.25 mmol) was added to a suspension of [{RhACTHUNRTGNEUNG([D15]Cp*)Cl2}2] (730 mg,
1.12 mmol) in aqueous DMSO (5%, 10 mL). The reaction was vigorously
stirred overnight at room temperature under nitrogen with exclusion of
light. Then the mixture was diluted with water (10 mL) and filtered
through Celite and the residue was washed with water until it become
ꢀ
colorless. An excess of NH4+PF6 was added to the brown aqueous fil-
trate to precipitate the salt. The solid was filtered, washed with water,
dried under vacuum, and recrystallized from CH2Cl2 (10 mL)/Et2O
(40 mL) to give an off-white solid (753 mg, 73%, deuteration of CH3
groups ca. 95%). 1H NMR (300 MHz, [D6]acetone): d=5.72 ppm (d,
2JH,Rh =0.9 Hz, C5H5); 2H NMR (61.4 MHz, acetone): d=2.15 ppm (brs,
CD3, [D15]C5Me5); 13C{1H} NMR (75.4 MHz, [D6]acetone): d=102.2 (d,
1
1JC,Rh =7.1 Hz, CCD3, [D15]C5Me5), 89.1 (d, JC,Rh =6.6 Hz, C5H5), 9.9 ppm
Experimental Section
(1:3:5:7:5:3:1 septet, 1JC,D =19.8 Hz, CD3, [D15]C5Me5). MALDI-MS:
m/z: 318.1 [M]+.
General details: All operations were performed under an atmosphere of
nitrogen, either using standard Schlenk techniques or in a glove box. Tol-
uene was dried using a solvent-purification system from MBraun, THF
was distilled from sodium benzophenone, while chlorobenzene and tol-
uene used for UV/Vis/NIR studies were dried over CaH2. All the sol-
vents were deoxygenated by three freeze-pump-thaw cycles and stored in
a glove box under an atmosphere of purified N2. [D6]Benzene and
[{RhACTHUNRTGNENG(U [D15]Cp*)Cp}2] (1b’2): This compound was obtained as a yellow
ꢀ
solid (110 mg, 53%) in an analogous fashion to 1b2, using 1b’+PF6
(300 mg, 0.647 mmol) and 1 wt% sodium amalgam (5.8 g, 26 mmol) in
THF (20 mL). 1H NMR (400 MHz, [D6]benzene): d=4.94 (s, 5H; C5H5),
4.41 (s, 2H; b-H, C5H5), 2.50 ppm (s, 3H; a-H, i-H, C5H5); 2H NMR
(61.4 MHz, benzene): d=1.98 (brs, 6D; b-CD3, [D15]C5Me5), 1.84 (brs,
15D; [D15]C5Me5), 1.34 (brs, 3D; i-CD3, [D15]C5Me5), 1.16 ppm (brs, 6D;
a-CD3, [D15]C5Me5); 13C{1H} NMR (100 MHz, [D6]benzene): d=93.9 (d,
1JC,Rh =6 Hz, CCD3, [D15]C5Me5), 89.5 (d, 1JC,Rh =10 Hz, b-CCD3
[D15]C5Me5), 83.8 (d, 1JC,Rh =5 Hz, C5H5), 73.4 (d, 1JC,Rh =9 Hz, b-C
[D8]toluene were dried over Na–K alloy and [D5]chlorobenzene was puri-
ꢀ [15a]
fied using CaH2. 1b+PF6
,
[{RhCp*Cl2}2],[16] TlCp,[18] [RuCp*-
ꢀ
[22]
ꢀ
[15b]
[8b]
A
,
2a+PF6
,
2a2,[8a,9a] 2b2,[9a] and 2c2 were synthe-
sized according to literature methods. 6,13-Bis
ACHTUNGTRENNUNGethynyl]pentacene (3) was purchased from Aldrich and used without fur-
ACHTUNGTREN[UNNG triACHTUNRTGEGUN(NN isopropyl)ACTHNUGTRENsNUGN ilyl-
1
C5H5), 66.8 (i-CCD3, [D15]C5Me5), 64.1 (i-C, C5H5), 60.7 (d, JC,Rh =12 Hz,
a-CCD3, [D15]C5Me5), 47.8 (d, 1JC,Rh =12 Hz, a-C, C5H5), 21.8
(1:3:5:7:5:3:1 septet, 1JC,D =19 Hz, i-CD3, [D15]C5Me5), 13.9 (1:3:5:7:5:3:1
ther purification. Sodium amalgam (1 wt%) was prepared from Na metal
and Hg (electronic grade, 99.99%) immediately before use. 1H, 2H,
13C{1H}, and NOESY NMR spectra were recorded either on a Bruker
AMX 400 or Varian Mercury 300 MHz spectrometer. Frequencies are
referenced to the residual resonances of 1H and 13C chemical shifts were
referenced to tetramethylsilane using the residual proton signal of the
solvent and the carbon resonances of the deuterated solvent, respectively.
Unless stated otherwise, carbon signals were observed as singlets. Mass
spectra were measured on an Applied Biosystems 4700 Proteomics Ana-
lyzer using MALDI mode. The electrochemical data were acquired using
cyclic voltammetry in 0.1m nBu4NPF6 in dry THF under nitrogen, using a
CH Instruments 620D potentiostat, a glassy carbon working electrode, a
platinum wire auxiliary electrode, and, as a pseudo-reference electrode, a
silver wire anodized in 1m aqueous potassium chloride solution, at a scan
rate of 50 mVsꢀ1. Ferrocene was used as an internal reference.
septet, 1JC,D =19 Hz, a-CD3, [D15]C5Me5), 12.0 (1:3:5:7:5:3:1 septet,
1
1JC,D =19 Hz, b-CD3, [D15]C5Me5), 10.4 ppm (1:3:5:7:5:3:1 septet, JC,D
=
19 Hz, [D15]C5Me5).
[D5]TlCp: A fresh solution of NaOD was prepared by gradually adding
freshly cut sodium (21 g, 0.91 mol) to deoxygenated D2O (180 mL, 9.0
mol) at 08C. Freshly cracked cyclopentadiene (35 mL, 0.41 mol) was
added by syringe to a solution of DMSO (35 mL) and a portion of the
NaOD/D2O solution (35 mL) under nitrogen. The mixture was stirred
vigorously for 1 h at room temperature. The organic layer was separated
using a cannula and then immediately added to a new flask containing
fresh DMSO (30 mL) and NaOD/D2O solution (35 mL) and the above
exchange step was repeated. In total, five such exchange reactions were
carried out; the organic fraction from the final exchange was then distil-
led at 408C to give [D6]cyclopentadiene (12 mL, 38%). Tl2SO4 (4 g,
7.92 mmol) was added to a portion of the [D6]cyclopentadiene (3 mL,
32 mmol) in a portion of the NaOD/D2O solution (30 mL). An off-white
precipitate was formed immediately. Stirring at room temperature was
continued for 4 h in the absence of light. The solid was filtered, washed
with D2O (3ꢀ5 mL), dried under high vacuum, and sublimed (65–808C,
20–30 mTorr) to give a crystalline off-white light-sensitive solid (3.5 g,
ACHTUNGTRENNUNG[{RhCp*Cp}2] (1b2): This compound was synthesized by minor modifica-
ꢀ
tion of a published procedure.[10a] 1b+PF6 (444 mg, 0.99 mmol) in THF
(25 mL) was added to 1 wt% sodium amalgam (8.8 g, 39.66 mmol). After
stirring for 2 h at room temperature, the volatiles were removed under
reduced pressure and the resulting black-brown solid was extracted with
toluene (15 mL). The resulting solution was decanted and filtered
through Celite. The solution was evaporated under reduced pressure to
afford a yellow solid which was further washed with deoxygenated cold
pentane (2ꢀ3 mL) then dried under vacuum to give the desired product
as a yellow solid (157 mg, 53%). 1H NMR (300 MHz, [D6]benzene): d=
4.90 (d, 2JH,Rh =0.9 Hz, 5H; C5H5), 4.37 (s, 2H; b-H, C5H5), 2.48 (s, 3H;
a-H, i-H, C5H5), 2.02 (s, 6H; b-CH3, C5Me5), 1.88 (s, 15H; C5Me5), 1.37
(s, 3H; i-CH3, C5Me5), 1.20 ppm (s, 6H; a-CH3, C5Me5); 13C{1H} NMR
(75.4 MHz, [D6]benzene): d=94.0 (d, 1JC,Rh =6 Hz, C5Me5), 89.6 (d,
1JC,Rh =10 Hz, b-CCH3 C5Me5), 83.8 (d, 1JC,Rh =5 Hz, C5H5), 73.4 (d,
1JC,Rh =9.4 Hz, b-C, C5H5), 67.1 (i-CCH3, C5Me5), 64.0 (i-C, C5H5), 60.8
(d, 1JC,Rh =12.6 Hz, a-CCH3, C5Me5), 47.8 (d, 1JC,Rh =12 Hz, a-C, C5H5),
22.8 (i-CCH3, C5Me5), 14.8 (a-CCH3, C5Me5), 12.9 (b-CCH3, C5Me5),
11.2 ppm (C5Me5).
81%, deuteration ca. 97%). EI-MS: m/z: 275 [M]+
203Tl).
AHCTUNGTRENNUNG
[RhCp*([D5]Cp)]+PF6 (1b’’+PF6ꢀ): This compound was obtained as an
(
205Tl), 273 [M]+
(
ꢀ
off-white solid (630 mg, 61%, deuteration of Cp rings ca. 93%) in an
ꢀ
analogous fashion to 1b’+PF6 using [{RhCp*Cl2}2] (708 mg, 1.15 mmol)
and freshly sublimed [D5]TlCp (630 mg, 2.3 mmol) in a 5% solution of
[D6]DMSO in D2O (8 mL). 1H NMR (300 MHz, [D6]acetone): d=
2.19 ppm (s, C5Me5); 2H NMR (61.4 MHz, acetone): d=5.74 ppm (brs,
1
C5D5); 13C{1H} NMR (75.4 MHz, [D6]acetone): d=101.4 (d, JC,Rh
=
7.6 Hz, CCH3, C5Me5), 88.9 (1:1:1 t of d, 1JC,Rh =7.3 Hz, 1JC,D =14 Hz,
C5D5) 9.7 ppm (C5Me5); MALDI-MS: m/z: 309.1 [M]+.
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14769

S. R. Marder, S. Barlow et al.
A
impurities. Evaporation of the solvent gave an off-white solid, which was
further purified by recrystallization from CH2Cl2 (12 mL) and diethyl
ether (60 mL) to give the desired product as an off-white crystalline solid
(1.52 g, 47%, deuteration of aromatic positions ca. 95%). 1H NMR
(400 MHz, [D6]acetone): d=2.20 (s, 9H; CH3, C6Me3D3), 1.92 ppm (s,
15H; C5Me5); 2H NMR (61.4 MHz, acetone): d=5.80 ppm (brs,
C6Me3D3); 13C{1H} NMR (100 MHz, [D6]acetone): d=101.0 (CCH3,
C6Me3D3), 95.4 (C5Me5), 89.2 (1:1:1 t, 1JC,D =27 Hz, CD, C6Me3D3), 18.1
(CH3, C6Me3D3), 9.9 ppm (C5Me5). MALDI-MS: m/z: 359.1 [M]+.
ꢀ
solid (105 mg, 51%) in an analogous fashion to 1b2, using 1b’’+PF6
(300 mg, 0.661 mmol), 1 wt% sodium amalgam (5.9 g, 26.5 mmol) in
THF (20 mL). 1H NMR (400 MHz, [D6]benzene): d=2.06 (s, 6H; b-CH3,
C5Me5), 1.92 (s, 15H; C5Me5), 1.20 (s, 3H; i-CH3, C5Me5), 1.24 ppm (s,
6H; a-CH3, C5Me5); 2H NMR (61.4 MHz, benzene): d=4.93 (brs, 5D;
C5D5), 4.45 (brs, 2D; b-D, C5D5), 2.50 (brs, 2D; a-D, C5D5), 2.01 ppm
(br, 1D; i-D, C5D5). 13C{1H} NMR (100 MHz, [D6]benzene): d=94.0 (d,
1JC,Rh =5.8 Hz, C5Me5), 89.6 (d, 1JC,Rh =10.3 Hz, b-CCH3 C5Me5), 83.4
(1:1:1 t of d, JC,Rh =4.8 Hz, JC,D =21 Hz, C5D5), 73.1 (1:1:1 t of d, JC,Rh
AHCTUNGERTN[GNUN {RuCp*(1,3,5-C6Me3D3)}2] (2a’’2): This compound was obtained (101 mg,
9.8 Hz, 1JC,D =22 Hz, b-C, C5D5), 67.0 (i-CCH3, C5Me5), 63.4 (1:1:1 t,
1JC,D =21 Hz, i-C, C5D5), 60.7 (d, 1JC,Rh =12.7 Hz, a-CCH3, C5Me5), 47.3
(br, a-C, C5D5), 22.7 (i-CCH3, C5Me5), 14.8 (a-CCH3, C5Me5), 12.8 (b-
CCH3, C5Me5), 11.2 ppm (C5Me5).
ꢀ
24%) in a similar fashion to 2a’2 using 2a’’+PF6 (600 mg, 1.2 mmol) and
1 wt% Na-Hg (10.5 g, 46.5 mmol). 1H NMR (100 MHz, [D6]benzene):
d=1.99 (s, 3H; p-CH3, C6Me3D3), 1.66 (s, 15H; C5Me5), 1.52 ppm (s, 6H;
o-CH3, C6Me3D3); 2H NMR (61.4 MHz, benzene): d=3.46 (brs, 2D; m-
CD, C6Me3D3), 1.65 ppm (brs, 1D; i-CD, C6Me3D3); 13C{1H} NMR
(100 MHz, [D6]benzene): d=87.5 (C5Me5), 87.1 (p-quat., C6Me3D3), 81.9
(1:1:1 t, 1JC,D =25 Hz, m-quat., C6Me3D3), 58.1 (1:1:1 t, 1JC,D =19 Hz, i-
quat., C6Me3D3), 41.6 (o-quat., C6Me3D3), 24.4 (o-CCH3, C6Me3D3), 19.4
(p-CCH3, C6Me3D3), 10.5 ppm (C5Me5).
Mixture of 1b’2, 1b’’2, and 1b’b’’: This mixture was obtained as a yellow
solid (60 mg) in an analogous fashion to 1b2, using 1b’+PF6 (100 mg,
0.215 mmol), 1b’’+PF6 (97.8 mg, 0.215 mmol) and 1 wt% sodium amal-
gam (4 g, 18 mmol) in THF (25 mL). Note that 1b’1b’’ is formed as a
mixture of two isomers differing in whether they contain a Cp–Cp* link-
age or ([D5]Cp)–([D15]Cp*) linkage.
Mixture of 2a’2, 2a’’2, and 2a’2a’’: Na–K alloy (3:1, ca. 500 mg, CAU-
ꢀ
ꢀ
[RuCp*
N
N
TION, HIGHLY PYROPHORIC) was added to a mixture of 2a’+PF6
tion of NaOD in D2O (ca. 2m, 10 mL) was added to a solution of 2a+
PF6 (625 mg, 1.25 mmol) in a minimum amount of [D6]acetone. The
(90 mg, 0.18 mmol) and 2a’’+PF6 (90.9 mg, 0.18 mmol) in THF (15 mL)
ꢀ
ꢀ
inside a N2-filled glove box. The reaction was stirred for 1 h at room tem-
perature, after which time the THF solution was decanted off the metal
and evaporated under reduced pressure. The resulting solid was extracted
in toluene (10 mL) and the extracts were filtered through Celite and
evaporated under reduced pressure to give a light yellow solid, which
was then washed with cold pentane (2ꢀ3 mL) and dried under vacuum
mixture was heated up to 1008C for 3 d, during which time the progress
of the deuteration was checked by 1H NMR spectroscopy. The volatiles
were removed under reduced pressure and the resulting solid was fil-
tered, washed with water, dried under high vacuum, and recrystallized
from CH2Cl2 (5 mL)/Et2O (20 mL) (514 mg, 81%, deuteration of mesity-
1
lene CH3 groups ca. 98%). H NMR (400 MHz, [D6]acetone): d=5.76 (s,
(53 mg).
3H; [D9]C6H3Me3), 1.92 ppm (s, 15H; C5Me5); 2H NMR (61.4 MHz, ace-
tone): d=1.90 ppm (brs, CD3, [D9]C6H3Me3); 13C{1H} NMR (100 MHz,
[D6]acetone): d=101.0 (CCD3, [D9]C6H3Me3), 95.4 (C5Me5), 89.5 (CH,
[D9]C6H3Me3), 17.3 (1:3:5:7:5:3:1 septet, 1JC,D =19.6 Hz, CD3,
+
Cꢀ
Crystal growth and structure determinations: Crystals of 1b 3 and 2b+
3
Cꢀ
were obtained by dissolving dimer 1b2 (6.6 mg, 11 mmol) or 2b2
(5.0 mg, 7.5 mmol) and 3 (7.0 mg, 11 mmol or 3.1 mg, 3.5 mmol in the cases
of reaction with 1b2 or 2b2, respectively) in benzene (2 mL) in a glove-
box, immediately filtering the solution using a syringe filter, and leaving
the solution to stand overnight at room temperature in a 3 mL vial. In
both cases the mixture turned from blue to green and green hair-like
[D9]C6H3Me3), 9.9 ppm (CH3, C5Me5); MALDI-MS: m/z 365.1 [M]+.
ꢀ
A
E
N
added to 1 wt% sodium amalgam (8 g, 32 mmol) in THF (25 mL). The
reaction mixture was stirred for 2 h at room temperature. The volatiles
were removed under reduced pressure and the crude solid was extracted
with toluene (12 mL). The toluene extracts were decanted and filtered
through Celite. Removal of the solvent yielded a light yellow solid, which
was washed with cold deoxygenated pentane (1ꢀ3 mL) and dried under
high vacuum (40 mg, 24%). 1H NMR (400 MHz, [D6]benzene): d=3.49
(s, 2H; m-CH, [D9]C6H3Me3), 1.72 (s, 1H; i-CH, [D9]C6H3Me3), 1.66 ppm
(s, 15H; C5Me5); 2H NMR (61.4 MHz, benzene): d=1.92 (brs, 3D; p-
CD3, [D9]C6H3Me3), 1.45 ppm (brs, 6D; o-CD3, [D9]C6H3Me3); 13C{1H}
NMR (100 MHz, [D6]benzene): d=87.5 (C5Me5), 87.2 (p-quat.,
[D9]C6H3Me3), 82.2 (m-quat., [D9]C6H3Me3), 58.7 (i-quat., [D9]C6H3Me3),
41.4 (o-quat., [D9]C6H3Me3,), 23.5 (1:3:5:7:5:3:1 septet, 1JC,D =19 Hz, o-
CD3, [D9]C6H3Me3), 18.7 (1:3:5:7:5:3:1 septet, 1JC,D =18 Hz, p-CD3,
[D9]C6H3Me3), 10.5 ppm (C5Me5).
crystals were obtained. Additional views and description of the structures
+
Cꢀ
are given in the Supporting Information. CCDC-892695 (1b 3 ) and
+
Cꢀ
CCDC-892696 (2b 3 ) contain the supplementary crystallographic data
for this paper. These data can be obtained free of charge from The Cam-
bridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_
request/cif.
Crossover experiments: Crossover experiments were carried out by dis-
solving the relevant compounds in deuterated solvents under nitrogen
and acquiring NMR spectra essentially immediately and again after vari-
ous times standing at various temperatures. The sublimation crossover
experiment between 1b’2 and 1b’’2 was carried out by placing equimolar
quantities of 1b’2 (25.0 mg, 39 mmol) and 1b’’2 (24.2 mg, 39 mmol) in a
small sublimation flask. The mixture was sublimed onto a cold-water-
cooled probe by slowly heating from 60–1258C under vacuum (20–
30 mTorr) for about 8 h. The color of the mixture in the flask turned
from yellow to slightly brown and a yellow solid was obtained on the
cooling finger. The solid was collected inside the glove-box and used for
NMR characterization without any further purification. 1H, 2H, and
13C NMR spectra were identical with the solid obtained from the mixed
1,3,5-C6Me3D3:[23,45] Mesitylene (10.4 g, 8.6 mmol) was added to a flask
containing 99% D2SO4 (15 g) and D2O (2 g) and the mixture was vigo-
rously stirred at room temperature for 24 h under nitrogen. After stirring
for 24 h, the organic layer was separated and transferred into another
flask containing fresh D2SO4/D2O. A total of three exchange reactions
were repeated. The organic layer from the final exchange was purified by
stirring over CaH2 overnight and then distilling to give the product (6.6 g,
62%, deuteration of aromatic positions ca. 97%). 1H NMR (400 MHz,
[D]chloroform): d=2.35 ppm (s, CH3, C6Me3D3); 2H NMR (61.4 MHz,
ꢀ
reduction of the mixture of 1b’+PF6 and 1b’’+PF6ꢀ. An analogous subli-
mation experiment was carried out using 2a’2 (15.0 mg, 21 mmol) and
2a’’2 (14.7 mg, 21 mmol), heating slowly from 80–1508C under vacuum
(20–30 mTorr) overnight. The sublimate, which was light yellow, was col-
lected inside the glove-box; 1H, 2H, and 13C NMR spectra were identical
to those seen for a solution of freshly mixed 2a’2 and 2a’’2.
chloroform): d=7.03 ppm (s, C6Me3D3); 13C{1H} NMR (100 MHz,
[D]chloroform): d=137.6 (CCH3, C6Me3D3), 126.63 (1:1:1 t, JC,D
1
=
23.4 Hz, CD, C6Me3D3), 21.12 ppm (CH3).
Vis/NIR kinetics studies: The solutions for Vis/NIR measurements were
prepared in a glove-box at room temperature. The solutions of the two
reactants were mixed in the desired ratio so that initial reactant concen-
trations were in the range ca. 10ꢀ5–10ꢀ3 m. Reactions carried out under
“pseudo-single-reactant” conditions with one reactant in a greater than
10-fold excess. The mixtures were immediately transferred into PTFE-
stopcock-sealed quartz cuvettes (175–2700 nm) with path lengths of
ꢀ
G
ꢀ
(NCMe)3]+PF6 (3.2 g, 6.3 mmol) was added to
a
solution of 1,3,5-
C6Me3D3 (6.6 g, 53 mmol) in dichloroethane (25 mL) deoxygenated with
nitrogen. The mixture was heated to reflux for 24 h; the solvent was then
evaporated to yield a brown oily residue, which was dissolved in acetone
(28 mL) and passed through a neutral alumina column to remove brown
14770
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772

Organic Electronic Materials
FULL PAPER
10 mm or 1 mm. The cuvettes were then sealed and taken out of the
glove-box to a Varian Cary 5E spectrometer, the acquisition of Vis/NIR
data being begun in less than 3 min from the initial mixing of the two re-
actant solutions. A Quantum Northwest TC 125 temperature controller
was used to control the temperature of the sample cuvette holder. In
order to determine rate constants and reaction order for the reactions of
Chan, F. Amy, Q. Zhang, S. Barlow, S. R. Marder, A. Kahn, Chem.
Phys. Lett. 2006, 431, 67.
[4] a) G. Parthasarathy, C. Shen, A. Kahn, S. R. Forrest, J. Appl. Phys.
2001, 89, 4986; b) J. Kido, T. Matsumoto, Appl. Phys. Lett. 1998, 73,
2866.
[5] D. Grozea, A. Turak, X. D. Feng, Z. H. Lu, D. Johnson, R. Wood,
Appl. Phys. Lett. 2002, 81, 3173.
Cꢀ
2b2 or 2c2 with 3, the temporal evolution of the 3 and 3 absorbances
was fitted, whereas, in the case of 1b2 initial rate data was used, due to
the complex temporal evolution resulting from two competing mecha-
nisms. One complication in the variable-temperature data for 1b2 is
likely to be that the reaction temperature is unlikely to have stabilized
on the timescale over which the initial rate is estimated; accordingly, at
increasingly higher temperatures, the nominal and actual temperatures
are likely to be increasingly different; thus, these data likely underesti-
mate the degree of temperatures dependence and may, therefore, lead to
[6] a) A. G. Werner, F. Li, K. Harada, M. Pfeiffer, T. Fritz, K. Leo,
Appl. Phys. Lett. 2003, 82, 4495; b) D. Gebeyehu, B. Maennig, J.
Dreschel, K. Leo, M. Pfeiffer, Sol. Energy Mater. Sol. Cells 2003, 79,
81; c) B. Maennig, J. Dreschel, D. Gebeyehu, P. Simon, F. Kozlowski,
A. Werner, A. Li, S. Grundmann, S. Sonntag, M. Koch, K. Leo, M.
Pfeiffer, H. Hoppe, D. Meissner, N. S. Sariciftci, I. Riedel, V. Dyako-
nov, J. Parisi, Appl. Phys. A 2004, 79, 1; d) A. G. Werner, F. Li, K.
Harada, M. Pfeiffer, T. Fritz, K. Leo, S. Machill, Adv. Funct. Mater.
2004, 14, 255; e) F. Li, A. Werner, M. Pfeiffer, K. Leo, J. Phys.
Chem. B 2004, 108, 17076; f) C. K. Chan, E.-G. Kim, J.-L. Brꢁdas,
A. Kahn, Adv. Funct. Mater. 2006, 16, 831; g) P. Wei, J. H. Oh, G.
Dong, Z. Bao, J. Am. Chem. Soc. 2010, 132, 8852; h) N. Cho, H.-L.
Yip, J. A. Davies, P. D. Kazarinoff, D. F. Zeigler, M. M. Durban, Y.
Segawa, K. M. OꢄMalley, C. K. Luscombe, A. K.-Y. Jen, Adv.
Energy Mater. 2011, 1, 1148; i) P. Wei, T. Menke, B. D. Naab, K.
Leo, M. Riede, Z. Bao, J. Am. Chem. Soc. 2012, 134, 3999.
[7] a) E. O. Fischer, H. Wawersik, J. Organomet. Chem. 1966, 5, 559;
b) N. El Murr, J. E. Sheats, W. E. Geiger, J. D. L. Holloway, Inorg.
Chem. 1979, 18, 1443.
¼
¼
artificially low values of Ea, DH , and DS . In an attempt to minimize
this problem, the initial rate (corresponding to the point at which absorb-
Cꢀ
ance of 3 is zero) was obtained by extrapolating the temperature-stabi-
Cꢀ
lized part of the 3 absorption trend over time.
Quantum-mechanical calculations: All calculations were performed with
the Gaussian 09 quantum-chemical package.[46] The geometry optimiza-
tions of the neutral and cationic forms of the monomeric and dimeric
sandwich molecules were carried out at the DFT level with the M06
functional,[47] using the 6–31G** and LANL2DZ basis sets for the first-
row atoms and transition-metal atoms respectively. Vibrational frequency
analyses were performed at the same level of theory to ensure that the
geometries represent minima on the potential energy surface. Calcula-
tions in the presence of an implicit solvent dielectric were performed
with the conductor polarizable continuum model (CPCM)[48] and a die-
lectric medium representing tetrahydrofuran (THF, e=7.43) for direct
[8] a) O. V. Gusev, M. A. Ievlev, M. G. Peterleitner, S. M. Peregudova,
L. I. Denisovich, P. V. Petrovskii, N. A. Ustynyuk, J. Organomet.
Chem. 1997, 534, 57; b) J.-R. Hamon, D. Astruc, P. Michaud, J. Am.
Chem. Soc. 1981, 103, 758.
comparison with experiment. The enthalpy changes (DH) for the cleav-
[9] a) S. Guo, S. B. Kim, S. K. Mohapatra, Y. Qi, T. Sajoto, A. Kahn,
S. R. Marder, S. Barlow, Adv. Mater. 2012, 24, 699; b) Y. Qi, S. K.
Mohapatra, S. B. Kim, S. Barlow, S. R. Marder, A. Kahn, Appl.
Phys. Lett. 2012, 100, 083305.
+
C
C
age of the neutral dimers (M2!2M ) and the radical-cation dimers (M2
!M+ +M ) at 0 K were calculated from Equations (6) and (7), respec-
tively. The free energy changes at 298.15 K were then calculated from
Equation (8).
C
[10] a) O. V. Gusev, L. I. Denisovich, M. G. Peterleitner, A. Z. Rubezhov,
N. A. Ustynyuk, P. M. Maitlis, J. Organomet. Chem. 1993, 452, 219;
b) S. K. Mohapatra, A. Romanov, G. Angles, T. V. Timofeeva, S.
Barlow, S. R. Marder, J. Organomet. Chem. 2012, 706–707, 140.
[11] For simplicity, hereafter we refer only to closed-shell acceptors, A,
C
C
DHðM2 ! 2 M Þ ¼ 2HðM ÞꢀHðM2Þ
ð6Þ
ð7Þ
ð8Þ
DHðM2 ! Mþ þ M Þ ¼ HðM Þ þ HðM ÞꢀHðM2Þ
DG ¼ DHꢀTDS
þ
Cþ
C
C
Cꢀ
that form radical anions, A , on one-electron reduction, since most
molecules of interest for n-doping, including 3, are indeed closed-
shell species. However, it is worth noting that some, such as CuPc,
are in fact odd-electron species that are reduced to even-electron
anions when n-doped, and that our discussion applies equally well in
these cases.
Acknowledgements
[12] a) O. V. Gusev, M. G. Peterleitner, M. A. Ievlev, A. M. Kalꢄsin, P. V.
Petrovskii, L. I. Denisovich, N. A. Ustynyuk, J. Organomet. Chem.
1997, 531, 95; b) a couple of examples have also been reported from
We thank Solvay S.A., the National Science Foundation (through the Sci-
ence and Technology Center Program, DMR-0120967, and the PREM
program, DMR-0934212), and the Office of Naval Research (N00014-11-
1-0313) for support. We also thank Sang Bok Kim and Tissa Sajoto for
the synthesis of 1a2 and 2c2, respectively, and Les Gelbaum for help with
NMR experiments.
Group 7: the dimers of [MACHTUNRGTNE(UGN C6Me6)2] (M=Tc, Re): E. O. Fischer,
M. W. Schmidt, Chem. Ber. 1969, 102, 1954; c) E. O. Fischer, M. W.
Schmidt, Chem. Ber. 1966, 99, 2206; J.-R. Hamon, D. Astruc, Orga-
nometallics 1989, 8, 2243; d) analogous dimers of [MnACHTUNTGRNENG(U arene)(CO)3]
piano stool compounds have also been reported: F.-W. Grevels, J.
Jacke, P. Betz, C. Krꢅger, Y.-H. Tsay, Organometallics 1989, 8, 293;
e) R. L. Thompson, S. J. Geib, N. J. Cooper, J. Am. Chem. Soc. 1991,
113, 8961; f) W. Dai, S. B. Kim, R. D. Pike, C. L. Cahill, D. A. Swei-
gart, Organometallics 2010, 29, 5173.
[1] K. Walzer, B. Maenning, M. Pfeiffer, K. Leo, Chem. Rev. 2007, 107,
1233.
[2] a) S. P. Tiwari, W. J. Potscavage, T. Sajoto, S. Barlow, S. R. Marder,
B. Kippelen, Org. Electron. 2010, 11, 860; b) W. Zhao, Y. Qi, T.
Sajoto, S. Barlow, S. R. Marder, A. Kahn, Appl. Phys. Lett. 2010, 97,
123305.
[3] a) S. Tanaka, E. Kawabe, K. Kanai, T. Iwahashi, T. Nishi, Y. Ouchi,
K. Sekia, J. Electron Spectrosc. Relat. Phenom. 2005, 144–147, 533;
b) K. Harada, A. G. Werner, M. Pfeiffer, C. J. Bloom, C. M. Elliott,
K. Leo, Phys. Rev. Lett. 2005, 94, 036601; c) L.-J. Li, A. N. Khlobys-
tov, J. G. Wiltshire, G. A. D. Briggs, R. J. Nicholas, Nat. Mater. 2005,
4, 481; d) C. K. Chan, A. Kahn, Appl. Phys. A 2009, 95, 7; e) C. K.
Chan, W. Zhao, A. Kahn, I. G. Hill, Appl. Phys. Lett. 2009, 94,
203306; f) C. K. Chan, W. Zhao, S. Barlow, S. R. Marder, A. Kahn,
Org. Electron. 2008, 9, 575; g) C. K. Chan, A. Kahn, Q. Zhang, S.
Barlow, S. R. Marder, J. Appl. Phys. 2007, 102, 014906; h) C. K.
[13] The electrochemical oxidation of 1a2 has previously been shown to
result in formation of 1a+ (reference [7b]) and the signatures of the
corresponding cations are also found in the cyclic voltammograms
of 1b2 and 2a2–2c2 subsequent to scanning the oxidation peak of the
dimer.
[14] For mechanism I the general form of the rate law is very complex,
but if the cleavage step can be regarded as a pre-equilibrium preced-
ing a rate-determining ET, the expression reduces to: d[M2]/dt=
ꢀ0.5k
(k1/k-1)0.5[M2]0.5[A], in which k1 and k2 are the forward rate
2ACHTUNGTRENNUNG
constants for the first and second steps repectively and kꢀ1 is the re-
verse rate constant for the first step. In the case of mechanism II,
the general rate law obtained, assuming a steady-state concentration
Chem. Eur. J. 2012, 18, 14760 – 14772
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
14771

S. R. Marder, S. Barlow et al.
+,
C
Cꢀ
of M2 is given by: d[M2]/dt=ꢀk1k2[M2][A]/
(k
R
dium dimers. While we have not performed crossover experiments
Cꢀ
Cꢀ
[A ], that
C
reduces to: d[M2]/dt=ꢀ
(k1k2/kꢀ1)[M2][A]/
[A ], if k2 !kꢀ1
to test for establishment of a 2c2 =2ꢀ2c equilibrium, we do not ob-
is, the first ET step can be regarded as a pre-equilibrium preceding
rate-determining cleavage of the dimer cation.
serve formation of 32ꢀ at long reaction times in kinetic studies using
excess 2c2, despite a more favorable DGET (see electrochemical data
¼
[15] a) K. Moseley, J. W. Kang, P. M. Maitlis, J. Chem. Soc. A 1970, 2875;
b) J. L. Schrenk, A. M. McNair, F. B. McCormick, K. R. Mann,
Inorg. Chem. 1986, 25, 3501.
[16] B. L. Booth, R. N. Haszeldine, M. Hill, J. Chem. Soc. A 1969, 1299.
[17] a) J. W. Kang, P. M. Maitlis, J. Organomet. Chem. 1971, 30, 127;
b) C. S. Wei, C. A. Jimꢁnez-Hoyos, M. F. Videa, J. F. Hartwig, M. B.
Hall, J. Am. Chem. Soc. 2010, 132, 3078.
in Table 1), suggesting that the barrier, DG , to cleavage in the iron
case is higher than that for the rhodium species.
[36] For an overview of the self-interaction error, see: W. Koch, M. C.
Holthausen, A Chemistꢀs Guide to Density Functional Theory, 2nd
ed., Wiley-VCH, Weinheim, 2002.
[37] This fit is of somewhat poorer quality than the fits obtained for the
Cꢀ
formation of 3 . Borderline thermodynamic feasibility, leading to
[18] A. F. Neto, A. D. L. Borges, J. Miller, V. A. Darin, Synth. React.
Inorg. Met.-Org. Chem. 1997, 27, 1299.
incomplete reaction and instability of the highly sensitive dianion
under the reaction conditions may play a role.
[19] a) H.-V. Wong, J. S. O. Evans, S. Barlow, S. J. Mason, D. OꢄHare,
Inorg. Chem. 1994, 33, 5515; b) J. S. O. Evans, D. OꢄHare, Chem.
Mater. 1995, 7, 1668.
[20] a) G. K. Anderson, R. J. Cross, I. G. Phillips, J. Chem. Soc. Chem.
Commun. 1978, 709; b) P. Mills, S. Korlann, M. E. Bussell, M. A.
Reynolds, M. V. Ovchinnikov, R. J. Angelici, C. Stinner, T. Weber,
R. Prins, J. Phys. Chem. A 2001, 105, 4418.
[21] J.-R. Hamon, P. Hamon, S. Sinbandhit, P. Guenot, D. Astruc, J. Or-
ganomet. Chem. 1991, 413, 243.
[22] B. Steinmetz, W. A. Schenk, Organometallics 1999, 18, 943.
[23] D. OꢄHare, S. J. Heyes, S. Barlow, S. Mason, J. Chem. Soc. Dalton
Trans. 1996, 2989.
[38] The Eyring model assumes that the rate at which reactants are
transformed into products in a chemical reaction is limited by a “tip-
ping frequency”, small relative to kBT/h, whereas, in the Marcus
model the rate of non-adiabatic electron transfer is limited by the
intermolecular electronic coupling, V, and reorganization energy, l,
according to the pre-exponential factor V2[4p3 h2lkBT)]0.5
.
[39] a) A. Loewenstein, G. Ron, Inorg. Chem. 1967, 6, 1604; b) R.
Stasiw, R. G. Wilkins, Inorg. Chem. 1969, 8, 156; c) J. L. Cramer,
T. J. Meyer, Inorg. Chem. 1974, 13, 1250; d) S. F. Nelsen, Y. Wang,
M. T. Ramm, M. A. Accola, J. R. Pladziewicz, J. Phys. Chem. 1992,
96, 10654; e) G. Novitchi, L. Helm, C. Anson, A. K. Powell, A. E.
Merbach, Inorg. Chem. 2011, 50, 10402 and references therein.
[40] C. Walling, D. Bristol, J. Org. Chem. 1971, 36, 733.
[41] J. Amaudrut, O. Wiest, J. Am. Chem. Soc. 2000, 122, 3367.
[42] We have also acquired some variable-temperature rate data in sol-
vents of varying polarity. However, only a limited range of polarities
were examined due to practicalities of reactant and product solubili-
ty and stability and the results are not helpful in identifying the
mechanisms operative. See Supporting Information for details.
[43] In the case of the Fe dopant, 2c2, the rate law clearly indicates that
reduction of 3 also proceeds by mechanism II, but we have not car-
ried out crossover experiments that would address the possibility of
mechanism I playing a role for other acceptors and/or under condi-
tions.
[24] Over time the NMR and Vis/NIR absorptions characteristic of 3 re-
Cꢀ
appeared, consistent with reoxidation of the highly air-sensitive 3
to the neutral parent molecule due to adventitious oxygen and/or
moisture leaking into the samples, again consistent with clean for-
Cꢀ
mation of 3 .
[25] T. Shida, S. Iwata, J. Am. Chem. Soc. 1973, 95, 3473.
[26] a) R. Mondal, C. Tçnshoff, D. Khon, D. C. Neckers, H. F. Bettinger,
J. Am. Chem. Soc. 2009, 131, 14281; b) T. M. Halasinski, D. M.
Hudgins, F. Salama, L. J. Allamandola, T. Bally, J. Phys. Chem. A
2000, 104, 7484.
[27] J. E. Collins, M. P. Castellani, A. L. Rheingold, E. J. Miller, W. E.
Geiger, A. L. Rieger, P. H. Rieger, Organometallics 1995, 14, 1232.
[28] P. J. Fagan, M. D. Ward, J. C. Calabrese, J. Am. Chem. Soc. 1989,
111, 1698.
[44] Although Figure 6 shows the formation of 32ꢀ in chlorobenzene,
qualitatively similar behavior is seen in THF, the same solvent used
for the electrochemical measurements.
[45] a) J. A. Dixon, R. W. Schiessler, J. Am. Chem. Soc. 1954, 76, 2197;
b) E. L. Mackor, P. J. Smit, J. H. van der Waals, Trans. Faraday Soc.
1957, 53, 1309; c) C. K. Ingold, C. G. Raisin, C. L. Wilson, J. Chem.
Soc. 1936, 915.
[46] Gaussian 09 (Revision A.2), M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V.
Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X.
Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Son-
nenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
J. A. Montgomery, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd,
E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Nor-
mand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J.
Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B.
Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Strat-
mann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochter-
ski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P.
Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, ꢆ. Farkas,
J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian, Inc.,
Wallingford CT, 2009.
[29] In fact the electrochemically estimated values for DG for reduction
Cꢀ
C
C
of 3 with 2b or 2c coupled with the failure to observe formation
of 32ꢀ suggests that if the corresponding dimers did operate through
mechanism I, DG for cleavage of 2b2 or 2c2 would have to be in
excess of +1.54 or +0.40 eV respectively. In the case of 2b2 this is
much larger than the +0.50 eV estimated for ET from the dimer to
neutral 3 (Table 1), suggesting that reduction of 3 with 2b2 is much
more likely to proceed by mechanism II.
[30] J. H. Oh, P. Wei, Z. Bao, Appl. Phys. Lett. 2010, 97, 243305.
[31] a) F. A. L. Anet, A. J. R. Bourn, J. Am. Chem. Soc. 1965, 87, 5250;
b) G. E. Bachers, T. Schaefer, Chem. Rev. 1971, 71, 617; c) H. Mo,
T. C. Pochapsky, Prog. Nucl. Magn. Reson. Spectrosc. 1997, 30, 1.
[32] Control experiments show that under the same conditions there is
neither measurable incorporation of deuterium from the solvent
into the methyl positions of 1b2 nor H/D deuterium scrambling be-
tween the methyl and Cp positions of 1b’2.
[33] One possibility is the existence of an additional hydrogen-atom
transfer equilibrium between pairs of monomers via species of the
type [RhCp(h4-C5Me5H)] and [RhCp(h4-C5Me4CH2)], as shown in
the Supporting Information (Figure S9).
[34] Cosublimation of 2a2 and 2b2 also fails to yield any of the crossover
product, but this result is somewhat more ambiguous in that the two
species may have sufficiently different vapor pressures that the two
species essentially sublime sequentially.
[47] Y. Zhao, D. G. Truhlar, Theor. Chem. Acc. 2008, 120, 215.
[48] M. Cossi, N. Rega, G. Scalmani, V. Barone, J. Comput. Chem. 2003,
24, 669.
[35] The calculations give relatively low values of DHdiss and DGdiss for
the iron dimer 2c2, which are comparable to the values for the rho-
Received: July 23, 2012
Published online: October 25, 2012
14772
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 14760 – 14772