Probing the influence of the ligands on the magnetism of dinuclear manganese, iron, and chromium complexes supported by aroylhydrazone

Article abstract of DOI:10.1002/ejic.201402084

The influence of aroylhydrazone and bridging ligands on the exchange coupling constant (J) of five dinuclear metal complexes is analyzed through ab initio calculations. The complete active space second-order perturbation theory with localized orbitals off

Full text of DOI:10.1002/ejic.201402084

FULL PAPER
DOI:10.1002/ejic.201402084
Probing the Influence of the Ligands on the Magnetism of
Dinuclear Manganese, Iron, and Chromium Complexes
Supported by Aroylhydrazone
Alex Domingo,*[a] David Specklin,[b] Vitor Rosa,[b,c]
Samir Mameri,[b] Vincent Robert,[a] and Richard Welter[b]
Keywords: Magnetic properties / Transition metals / Metal–metal interactions / Ab initio calculations
The influence of aroylhydrazone and bridging ligands on the
exchange coupling constant (J) of five dinuclear metal com-
culations are extended to three new dinuclear complexes of
MnIV, FeIII and CrIII constructed with a chemically simpler
plexes is analyzed through ab initio calculations. The com- ligand (HL). We provide the synthesis and structural charac-
plete active space second-order perturbation theory with lo-
calized orbitals offers not only estimates of J but also detailed
information about the contributions of the molecular regions
to the overall magnetism. This computational strategy is first
validated for two previously reported ferromagnetic MnIII/
MnIII and antiferromagnetic FeIII/FeIII complexes. Such cal-
terization (X-ray diffraction crystallography) of these three
new compounds. The HL ligand exhibits little electronic in-
fluence over the J value, but it directs the magnetic proper-
ties of the compound by conditioning the relative arrange-
ment of the metal centres.
pounds.[3] The intermetallic bridges of this class of com-
pounds create a superexchange pathway that often leads to
isotropic couplings and, in some cases, slow relaxation of
the magnetization to make a so-called single-molecule mag-
net (SMM).[4] Such magnets are promising candidates for
applications in molecular spintronics, high-density data
storage and quantum information processing.[5] The discov-
ery of SMMs of low nuclearity, down to four centres, stimu-
lated the research effort on dinuclear complexes, the sim-
plicity of which favours studies of structure–coupling-con-
stant relationships and time-consuming theoretical calcula-
tions.
In recent years, an ongoing effort has been made to pre-
pare a variety of transition metal compounds coordinated
to aroylhydrazone ligands and analyze the effect of system-
atic structural alterations on the magnetic properties
(Scheme 1).[6] As a result, it was shown that dinuclear MnIV,
FeIII and CrIII complexes bearing this type of ligand are
antiferromagnetic.[7] Specifically, the Fe-based species
proved to be of interest in solar energy conversion.[8]
Furthermore, a dinuclear MnIII/MnIII complex displays the
strongest ferromagnetic interaction reported to date be-
tween two bridged MnIII ions.[9]
Introduction
The design of molecular materials with specific electro-
magnetic properties requires an understanding of the effect
of the structure on the phenomenon of interest. In transi-
tion metal complexes, this knowledge has come a long way
and, nowadays, powerful predictions based on the ligand
field theory are at hand.[1] Nonetheless, small changes to
the molecular structure can still introduce, exacerbate or
even quench the macroscopic properties of such complexes.
One of the fields in which this kind of relationship is of the
most interest is molecular magnetism, for which the control
of exchange magnetic couplings at the molecular scale is of
unique significance. In recent decades, a class of complexes
characterized by a small number of paramagnetic transition
metal ions (Ͻ30) connected to each other by simple bridg-
ing anions, typically O2–, HO– or CH3O–, has provided
valuable insights as simple models of more complex sys-
tems.[2] Prime examples are found in catalysis and biological
processes, particularly in the case of Mn and Fe com-
[a] Laboratoire de Chimie Quantique, UMR 7177-CNRS/Unistra,
Université de Strasbourg,
1 rue Blaise Pascal, 67000 Strasbourg, France
E-mail: domingo@unistra.fr
The aim of the present work is to gain further insight
into the structural features that control the exchange cou-
pling constant (J) of dinuclear complexes based on aroyl-
hydrazone ligands. We have developed a computational
strategy based on ab initio multiconfigurational wave func-
tions that is capable of unravelling the influence of particu-
lar regions of the molecule on the overall magnetism of the
http://quantique.u-strasbg.fr
[b] Laboratoire DECOMET, UMR 7177-CNRS/Unistra,
Université de Strasbourg,
4 rue Blaise Pascal, 67000 Strasbourg, France
[c] REQUIMTE, Departamento de Química, Faculdade de
Ciâncias e Tecnologia, Universidade Nova de Lisboa,
Caparica 2829-516, Portugal
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejic.201402084.
Eur. J. Inorg. Chem. 2014, 2552–2560
2552
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
Scheme 2. Molecular structures of the HL and H2LS ligands. The
labels P, C and S refer to specific molecular regions of these ligands.
Scheme 1. Preparation of dinuclear Mn, Fe and Cr complexes (1–
3) with the HL ligand.
compound. The combination of the complete active space
second-order perturbation theory (CASPT2) method,[10]
which has proven to be very efficient for the treatment of
electron correlation effects on transition metal com-
pounds,[11] with localized molecular orbitals (MOs) allows
us to enable or disable arbitrarily defined regions of the
molecule.
Scheme 3. Molecular structures of the bridges found in the com-
plexes with HL and H2LS ligands. The label B refers to particular
molecular regions of these bridges.
We validated the proposed theoretical method through
its application to two previously reported and fully charac-
terized dinuclear complexes with the NЈ-(1,3-dithiolan-2-
ylidene)-2-hydroxybenzohydrazide (H2LS) ligand, namely,
[MnIII2(μ-OMe)2(HLS)4] with a strong ferromagnetic char-
acter[9] and its analogue [FeIII2(μ-OMe)2(HLS)4] with an
antiferromagnetic coupling.[7a] These two compounds will
Results and Discussion
Synthesis of MnIVL, FeIIIL and CrIIIL (1–3)
Ligand HL was readily and quantitatively prepared by
be referred to as MnIIIHLS and FeIIIHLS, respectively. Sub- using a previously reported procedure under classical con-
sequently, the theoretical strategy was applied to a new ditions or by applying microwave irradiation. It was not
series of three dinuclear complexes based on the ligand NЈ- straightforward to establish the appropriate conditions for
cyclopentylidenebenzohydrazide (HL). The simpler chemi- the synthesis of the corresponding dinuclear complexes.
cal structure of the HL ligand in comparison to that of Therefore, various conditions such as the reaction composi-
H2LS will help us to assess the effect of some functional tion, stoichiometry, concentration, time, pH and tempera-
groups on the ligand and the role of the basic aroyl- ture were investigated, and the optimal conditions are re-
hydrazone structure. The three HL complexes are [MnIV2(μ- ported in the Experimental Section. In general, the com-
O)2(L)4], [FeIII2(μ-OMe)2(L)4] and [CrIII2(μ-OMe)2(L)4], the plexes described in this work were obtained upon reaction
synthesis and structural characterization of which are de- of the ligand in the presence or absence of a base (KOAc
scribed herein. These three compounds will be referred to or pyridine) and Mn(OAc)2, Mn(acac)3, FeCl3 or CrCl3 in
as MnIVL, FeIIIL and CrIIIL, respectively.
organic solvents and crystallized either by slow evaporation
The present evaluation of the magnetic properties of or by the slow diffusion of a countersolvent into the com-
these five dinuclear compounds offers not only estimates plex solution at room temperature. The reactions of two
of their J values but also detailed information about the equivalents of HL with MnIII, CrIII or FeIII salts resulted
contributions of different parts of the molecule to the over- in the corresponding μ2-O-bridged dinuclear neutral spe-
all magnetism. The partitioning scheme for the HL/H2LS cies, in which each metal ion is coordinated to two ligands
ligands and the bridges, which was used to evaluate their in its deprotonated form (L–). The anionic benzo-
influence on the molecular magnetism, is illustrated in hydrazonate form L– results from the deprotonation of the
Schemes 2 and 3. The inspections along the series of com- benzohydrazonic acid tautomer of HL as testified by the
pounds with the HL ligand in combination with the data bond length analysis of the solid-state structure.
obtained from the two complexes with H2LS reveal clear
common traits between them. The role of the dimetallic in THF followed by diffusion of Et2O allowed the crystalli-
The reaction of manganese(III) acetylacetonate with HL
core and the bridges dominates the intermetallic exchange zation of [Mn2(μ-O)2(L)4]·THF (1, THF
=
tetra-
coupling constant. On the other hand, the HL and H2LS hydrofuran), which is composed of two MnIV ions bridged
ligands have a minor direct effect over J, but they have a by oxo anions that originate from residual water molecules
structural role that is determinant in the control of the in the THF. As the reaction occurs under anaerobic condi-
geometry of the dimetallic core and, thus, the magnetic in- tions, the formation of MnIV from MnIII seems to arise
teractions.
from a dismutation reaction. Notably, this phenomenon, al-
Eur. J. Inorg. Chem. 2014, 2552–2560
2553
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
though uncommon, have been observed previously.[12] De-
spite numerous efforts, the MnII derivative(s) that should
result from the dismutation reaction have not been isolated.
Complexes [Fe2(μ-OMe)2(L)4] (2) and [Cr2(μ-OMe)2
(L)4]·CHCl3·CH3OH (3) feature metal ions in the +3 oxi-
dation state bridged by methoxo anions. The reaction of
iron(III) chloride with HL in the presence of the weak base
KOAc in alcoholic media led to the precipitation of 2; one
should note that high concentrations were essential for the
formation of 2. Complex 3 was prepared by the reaction of
chromium(III) chloride with HL in methanol under reflux
containing zinc dust (Zn) in suspension. The prior re-
duction of CrIII to CrII with metallic Zn followed by air
oxidation to CrIII has proved to be necessary for the forma-
tion of the dimer complex over the monomeric complex (in
which the metallic centre is coordinated to three ligands).
Notably, 1–3 were isolated as single crystals in good yields,
and their solid-state structures are described hereafter.
Crystal Structures of the Dinuclear HL Complexes
Selected bond lengths and angles of 1–3 are collated in
Table 1. Complex 1 is composed of two six-coordinate
MnIV ions that adopt a distorted octahedral geometry (Fig-
ure 1). This dimer crystallizes in the monoclinic space group
C2/c with Z = 4. The core shows one THF solvent molecule
per complex. Each ion is surrounded by two μ2-oxo anions
and two chelating L– ligands. The structure is highly sym-
metrical with a C2 axis orthogonal to the plane formed by
Mn, MnЈ, O3 and O3Ј that passes through the centre of it.
The absence of Jahn–Teller deformation strongly supports
the oxidation state +4 (vs. +3). Moreover, as both metal
ions have two nitrogen atoms in axial positions, the re-
sulting disposition of the ligands is symmetrical. According
to PLATON calculations, the crystal structure is stabilized
by four weak intramolecular CH···N interactions that in-
volve the cyclopentane moieties and face imine nitrogen
atoms, C24H···N3Ј, C12H···N1Ј and their symmetrical
counterparts.
Figure 1. Top: ORTEP view of MnIVL with partial labelling
scheme; solvent molecules are omitted for clarity. Thermal ellip-
soids enclose 50% of the electron density. Bottom: packing of
MnIVL in the ac plane; the dashed lines and the dotted lines (red)
represent π–π interactions and CH–π interactions.
Complexes 2 and 3 are dinuclear species of six-coordi-
nate FeIII and CrIII ions that crystallize in the monoclinic
P21/a and P21/c space groups, respectively (Figures 2 and
3). Each ion is surrounded by two μ2-methoxo anions and
two chelating L– ligands. In comparison with 1, the differ-
ences arise mainly from the different oxidation states of the
metal ions and the type of bridging ligands. Complexes 2
and 3 show dissymmetrical dispositions of the ligands, as
one metal ion is coordinated to two nitrogen atoms in axial
positions, whereas the second metal ion is coordinated to
two oxygen atoms in axial positions. The coordination
spheres adopt distorted octahedral geometries in which the
longer M–N bonds show more deviation from the ideal oc-
tahedral geometry than the M–O bonds.
Table 1. Selected bond lengths [Å] and angles [°] of 1–3.
Bond/angle
MnIVL (1)
FeIIIL (2)
CrIIIL (3)
The structures of 2 and 3 slightly differ as a result of the
inclusion of solvent molecules in the lattice of 3. According
to PLATON calculations, 2 is stabilized by two CH–π inter-
actions that involve the cyclopentane moieties and face
phenyl moieties C34H–Ph(C13–18) and C46H–Ph(C1–6),
whereas 3 is stabilized by two weak CH–O nonclassical hy-
drogen bonds, C24H–O4 and C12H–O3. Finally, the struc-
ture of 2 shows that one of the CH–π interactions with the
cyclopentane moiety is well defined in an envelope confor-
mation on C46, whereas the other methylene group (CH2)
involved has a very broad thermal ellipsoid. Such structural
disorder could be related to the equidistant position of the
CH2 group between two phenyl moieties, which allows it to
partake either in the CH–π interaction discussed above or
M–O1
M–O2
M–N2
M–N4
1.948
1.950
2.009
2.014
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
M1–O1
M1–O2
M1–N2
M1–N4
M2–O3
M2–O4
M2–N6
M2–N8
M1–μO
M2–μO
M1–M2
M1–μO–M2
1.984
1.984
2.171
2.171
1.978
1.970
2.140
2.128
1.963
2.025
3.107
102.33
1.957
1.955
2.080
2.065
1.959
1.940
2.066
2.088
1.980
1.965
3.058
101.62
–
1.809
1.809
2.762
99.47
Eur. J. Inorg. Chem. 2014, 2552–2560
2554
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
in the intermolecular C34H–Ph(C25–30) interaction (see
Solid-State Packing section).
Solid-State Packing
As the complexes lack classical hydrogen-bond donors,
their packings are driven by weaker interactions such as
CH–X, π–π and CH–π interactions. Complex 1 crystallizes
with one THF solvent molecule and, according to
PLATON calculations, shows two major intermolecular in-
teractions that result in two interlocked π–π and CH–π net-
works in the ac plane. The π–π network arises from inter-
molecular interactions in which two phenyl rings stack on
top of each other at 3.64 Å with a slight slippage of 0.6 Å,
whereas the CH–π network consists of THF molecules in-
tercalated between two phenyl rings with a C(H)–π distance
of 2.76 Å (see Figure 1). In addition, some weak CH–X in-
teractions participate in the packing interactions.
A crossed network of intermolecular CH–π interactions
along ab stabilizes the packing of 2. The complex partici-
pates in the network with two phenyl CH donors and two
phenyl π-ring acceptors with C(H)–π distances of 3.703 and
3.814 Å, which results in closely packed layers along ab. The
packing along c is derived from weak interlayer interac-
tions, such as CH–X bonds and van der Waals interactions.
In the case of 3, the main packing interactions are the CH–
π interactions C9H–Ph(C2–7) and C45H–Ph(C38–43) with
C(H)–π distances of 3.804 and 3.749 Å, which result in two
collinear CH–π networks along b. The structure is further
stabilized through the formation of hydrogen bonds be-
tween MeOH solvent molecules and the N3 atom of the
complex.
Figure 2. Top: ORTEP view of FeIIIL with partial labelling scheme;
solvent molecules are omitted for clarity. Thermal ellipsoids enclose
50% of the electron density. Bottom: packing of FeIIIL in the ac
plane; the dotted lines (red) represent CH–π interactions.
Detailed bond lengths and angles and ORTEP views of
1, 2 and 3 together with the powder X-ray diffraction
pattern of 2 are provided in the Supporting Information.
Computational Analysis and Validation for MnIIIHLS and
FeIIIHLS
The exchange coupling constant values at the CASPT2
level for the two compounds with the H2LS ligand are listed
in Table 2. The reference J values for MnIIIHLS and
FeIIIHLS are presented in the first row of Table 2 and are
the result of the treatment of the electrons in the metal 3d
MOs excluding any correlation effects originating in their
environment. Each other row of Table 2 contains the J val-
ues obtained by activating other regions in addition to the
dimetallic core. The various regions defined in the compu-
tational model incorporate the π system and lone pairs lo-
calized in the zones depicted in Schemes 2 and 3 (see Com-
putational Details).
The J value obtained for FeIIIHLS with the interactions
included for all the regions is –22 cm–1. This result is in very
good agreement with the reported experimental value for
this complex of –27 cm–1[7a] and confirms the smaller im-
pact of the ligand σ orbital and the diffuse MOs on the
J value, as previously reported.[13] The exchange coupling
Figure 3. Top: ORTEP view of CrIIIL with partial labelling scheme;
solvent molecules are omitted for clarity. Thermal ellipsoids enclose
50% of the electron density. Bottom: Packing of CrIIIL in the ac
plane; H atoms are omitted for clarity, except those partaking in
the represented interactions or those in solvent molecules. The
dashed lines (red) represent CH–π interactions and the dotted lines
(blue) represent H bonds.
Eur. J. Inorg. Chem. 2014, 2552–2560
2555
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
Table 2. Exchange coupling constants of the MnIIIHLS and lations as a tool to screen and clarify the changes in the
FeIIIHLS compounds obtained at the CASPT2 level with the speci-
fied molecular regions enabled. The reference J values correspond
constant values from the CASPT2 calculations performed
to the results with all regions disabled.
magnetic behaviour of analogues. The exchange coupling
on the three compounds with the HL ligand are shown in
MnIIIHLS
J
FeIIIHLS J [cm–1]
Table 3. The reference J values for MnIVL, FeIIIL and
CrIIIL are presented in the first row of Table 3, and only
the electrons in the metal 3d MOs were treated excluding
correlation effects originating in their environment. The
other regions are added to the dimetallic core following the
partitioning scheme shown in Schemes 2 and 3 (see Compu-
tational Details).
Active regions[a]
[cm–1]
Reference
B
C
CP
CS
CPS
M
BCPSM
BCPSM+
6
2
5
5
5
4
4
4
4
–4
–9
–5
–5
0
–5
–8
–18
–22
Table 3. Exchange coupling constant of MnIVL, FeIIIL and CrIII
L
at the CASPT2 level with the specified molecular regions enabled.
The reference J values correspond to the result with all regions
disabled.
[a] See Schemes 2 and 3.
constant of the dimetallic core is relatively weak (–4 cm–1),
but it becomes almost eight times stronger owing to the
antiferromagnetic contributions from almost all molecular
regions. The only exception is the dithiolane group (S),
which significantly favours the ferromagnetic coupling. The
other two parts of the H2LS ligand (C and P) do not seem
to have any direct effect on the magnetism of this com-
pound. The largest antiferromagnetic contributions come
from the methoxo bridges (B) and the more diffuse virtual
orbitals located at the metal centres (M) and the first coor-
dination sphere (M+). This phenomenon is not uncommon
as the bridges open an indirect channel between the two
metal centres that favours the antiparallel arrangement of
their unpaired electrons (i.e., superexchange). Likewise, the
virtual orbitals allow direct through-space interactions be-
tween the unpaired electrons of each metal centre.
Active regions[a] MnIVL (1)
J [cm–1]
FeIIIL (2)
J [cm–1]
CrIIIL (3)
J [cm–1]
Reference
B
C
–15
–119
–32
0
–21
–1
–1
–4
–13
–17
5
0
3
3
3
–1
–3
CP
M
–32
–43
BCPM
BCPM+
–164
–201
[a] See Schemes 2 and 3.
The reference magnetic character within this series of
compounds with HL varies drastically from the antiferro-
magnetic MnIVL (–15 cm–1) to the ferromagnetic CrIII
L
(5 cm–1). However, the effect that each molecular region in-
duces on the overall J value is consistent among the three
dinuclear systems 1–3. The resulting alterations of J for the
dimetallic MnIV, FeIII and CrIII cores is quantitatively dif-
ferent in absolute terms, but they show a proportional trend
for all the explored regions that depends on the reference J
value (see Table 3). This behaviour is equally observed for
the two complexes with the H2LS ligand (MnIIIHLS and
FeIIIHLS), the sensitivity of which to the environment
seems to be reduced by the strength of the ferromagnetic
coupling between the metal centres.
On the other hand, the behaviour of MnIIIHLS is com-
pletely insensitive to its environment. The overall J value is
stable and remains ferromagnetic with a final value of
4 cm–1, in good qualitative agreement with the reported fer-
romagnetic character of this compound (20 cm–1).[9] Reach-
ing quantitative agreement for ferromagnetic systems is
usually complex because of specific coupling mecha-
nisms[13] or spin polarization effects,[14] which are not in-
cluded in the current CASPT2 treatment. The methoxo
bridges (B) still seem to have a significant antiferromagnetic
contribution, similar to that for FeIIIHLS. However, neither
the ferromagnetic effect of the dithiolane (S) nor the anti-
ferromagnetic contribution of the virtual orbitals (M+) ap-
pear in MnIIIHLS. These results indicate that the ferromag-
The MnIVL compound is expected to have an antiferro-
magnetic behaviour (–201 cm–1), one order of magnitude
stronger than that of its FeIII counterpart (–17 cm–1), and
the latter is also expected to be one order of magnitude
more antiferromagnetic than its CrIII analogue (–3 cm–1).
This ratio is maintained along the series of calculations with
some moderate deviations. Therefore, it is possible to ex-
tract some common effects that these ligands produce on
the three compounds. The bridging ligand (B), which is
either an oxo or methoxy group, significantly increases the
antiferromagnetic character of all complexes. This phenom-
enon is equally present for the two complexes with the H2LS
ligand, but now the proportionality of this effect with the
coupling of the dimetallic core (reference J value) is evi-
denced in the series of HL-based complexes. The diffuse
netism of MnIIIHLS is inherent to the dimetallic MnIII
/
MnIII core, and the factors that determine the J value are
the relative orientation between the metal centres and the
intermetallic distance. Once strong ferromagnetic coupling
is achieved between the metal centres, the influence of the
ligands becomes insignificant.
Computational Results for MnIVL, FeIIIL and CrIII
L
Prompted by the robustness of our theoretical approach, virtual orbitals located at the metal centres (M) and the
we inspected the new series of compounds to anticipate first coordination sphere (M+) also strengthen the anti-
their magnetic properties. Our goal is to use ab initio calcu- ferromagnetism of the three compounds. As observed for
Eur. J. Inorg. Chem. 2014, 2552–2560
2556
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
the bridges, their impact on the J value is also proportional fects induced by its environment. The strongest contri-
to the reference J value (see Table 3) but to a lesser extent. butions are antiferromagnetic and come from the bridges
The HL ligand causes a weak increase of the antiferro- (superexchange) and the virtual orbitals localized around
magnetic component that exclusively originates from the the metal centres (direct exchange). Notably, the dithiolane
central ring (C) directly coordinated to the metal cation. group significantly increases the ferromagnetic component
The aromatic system of the phenyl group (P) is inert with of the system. Nonetheless, the H2LS ligand has minor elec-
respect to the exchange coupling of the molecule. As HL tronic interactions with the dimetallic core and, thus, it di-
does not have the dithiolane group (S), which induces a rects the magnetic properties of the compound by condi-
ferromagnetic contribution for the H2LS complexes, it tioning the relative arrangement of the metal centres (the
could be expected that HL would exhibit a stronger antifer- M–M distance and M–μOMe–M angles). Usually, ex-
romagnetic contribution in comparison. The clearest exam- change coupling constants involving metal centres are
ple is the MnIVL compound, the J value of which becomes strongly sensitive to small distortions of the metallic coordi-
twice as antiferromagnetic with the CP regions activated. nation sphere.[15] One example is the different J value ob-
Unfortunately, the J values of the other compounds are too tained when the B regions are activated for FeIIIHLS
small to allow any conclusion about this matter.
Furthermore, the alterations of J induced by the different equivalent dimetallic core. Unfortunately, the analysis of
regions are additive for MnIVL complex but not for FeIII
this behaviour involves too many structural parameters to
(–9 cm–1) and FeIIIL (–21 cm–1), which have a chemically
L
and CrIIIL. The J value of the MnIVL system with the be addressed in the present work.
BCPM region active is –164 cm–1, and the sum of the inde-
Three dinuclear complexes with the HL ligand (1–3) have
pendent contributions of the B, CP and M regions to the been synthesized and fully characterized, and their solid-
reference value is exactly –164 cm–1. The contribution of a state structures have been elucidated by X-ray diffraction
certain region can be obtained by subtracting the reference crystallography. These systems extend the series of com-
J value from the J of the region. However, this is not the pounds with the HL ligand, which is chemically simpler
case for FeIIIL, which experiences an antiferromagnetic than H2LS. All three complexes are expected to be antifer-
contribution of –13 cm–1 with the BCPM region enabled, romagnetic with J = –201 cm–1 for MnIVL, J = –17 cm–1
whereas the sum of the individual components accounts for for FeIIIL and J = –3 cm–1 for CrIIIL (Table 3). The behav-
double that (–26 cm–1). A similar behaviour is observed for iour of these compounds is very similar to that of
the Cr complex, which has a J value of –1 cm–1 with BCPM MnIIIHLS and FeIIIHLS. Their strongest contributions are
and –6 cm–1 for the sum of the parts. The combined interac- equally antiferromagnetic and arise from the bridges and
tions between the individual regions are definitely not the virtual orbitals localized around the metal centres.
equivalent between the MnIV system and the FeIII and CrIII Nonetheless, the series of compounds based on HL show a
compounds. This distinct behaviour could be related to clearer dependency of these antiferromagnetic contri-
their bridge ligands, which are oxo for MnIVL and methoxo butions to the inherent intermetallic coupling. The larger
for FeIIIL and CrIIIL. Nevertheless, it is difficult to affirm the ferromagnetic coupling between the metal centres (ref-
whether this change is due solely to the different interaction erence J value), the less affected is the overall magnetism
of the bridges or also to the character of the involved metal by the superexchange and direct exchange. Therefore, the
centres.
control of the overall magnetism that the HL ligand im-
poses by determining the structure of the dimetallic core is
more effective than any external electronic interaction. The
HL ligand has the capacity to control the reference J and,
hence, the extent of the antiferromagnetic superexchange
Conclusions
We have developed a computational strategy based on ab and direct exchange effects. However, if the inherent ferro-
initio CASPT2 calculations that uses localized orbitals and magnetism of the dimetallic core is not sufficiently large,
is capable of selectively enabling or disabling different re- these effects could still switch the complex to (slightly) anti-
gions of the molecule. This technique has been applied to ferromagnetic behaviour.
analyze the specific influence on the exchange coupling con-
stant of various dinuclear transition metal compounds with magnetic exchange interaction in favourable circumstances,
aroylhydrazone ligands. our current investigation aims to include other bridging li-
As the dinuclear MnIII complexes can present a ferro-
The calculated J values for the two complexes with the gands as well as find a synthetic strategy that could prevent
H2LS ligand are in good agreement with the experimental the in situ oxidation of MnIII to MnIV.
ones, J = 4 cm–1 for MnIIIHLS and J = –22 cm–1 for
FeIIIHLS. The influence of the various molecular regions
(Scheme 2) over J shows that the different contributions to
Experimental Section
the overall magnetism largely depend on the inherent inter-
metallic coupling (Table 2). The ferromagnetic complex
MnIIIHLS is not significantly affected by the interaction
with the H2LS ligands and the bridges, whereas the antifer-
Syntheses of the Organic and Metallorganic Compounds: Unless
otherwise stated, manipulations were performed under anaerobic
conditions (Ar) by using standard Schlenk techniques. Commercial
romagnetic FeIIIHLS complex exhibits a wider range of ef- reagents were used as purchased, and solvents were degassed prior
Eur. J. Inorg. Chem. 2014, 2552–2560
2557
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
to use. IR spectra were recorded with a Nicolet 380 FTIR spec-
C50H58Cr2N8O6·CH3OH·CHCl3: calcd. C 55.64, H 5.66, N 9.98;
trometer. The yields of 1–3 are those of isolated crystals.
found C 55.23, H 5.74, N 9.83.
Ligand HL: Benzohydrazide (5 g, 36.7 mmol) and cyclopentanone
(3.09 g, 36.7 mmol, 1 equiv.) were placed in a quartz tube, which
was placed in a microwave oven and heated at 200 W for 3 min.
Then, the resulting mixture was cooled to room temperature. After
the addition of Et2O (20 mL) and further washing with Et2O (4ϫ
X-ray Crystallography: Single crystal X-ray diffraction measure-
ments were performed at 173 K by using a Nonius Kappa CCD or
a Bruker APEX-II CCD diffractometer with Mo-Kα radiation. The
crystal structures were solved by direct methods and refined by
employing full-matrix least-squares on F2 (SHELXL-97). All non-
hydrogen atoms were refined anisotropically, and the hydrogen
atoms of organic ligands were introduced by using a riding model
(SHELXL-97). Crystal data and refinement details of the structure
determination for 1–3 are listed in Table 4. CCDC-941599 (for 1),
-941600 (for 2) and -941601 contain the supplementary crystallo-
graphic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
1
15 mL), HL was isolated as a white solid (quantitative). H NMR
(400 MHz, [D6]DMSO): δ = 1.70–1.78 (m, 4 H), 2.30–2.48 (m, 4
H), 7.48 (d, 2 H, J = 6.5 Hz), 7.54 (d, 1 H, J = 6.1 Hz), 7.81 (d,
2H J = 5.6 Hz), 10.29 (s, 1 H) ppm. 13C NMR (100 MHz, [D6]-
DMSO): δ = 24.2, 24.3, 28.5, 33.0, 127.6, 128.2, 131.2, 134.1, 163.2,
170.2 ppm. All analyses are in agreement with those previously
reported.[16]
[MnIV2(μ-O)2(L)4]·THF (1): Mn(acac)3 (174 mg, 0.5 mmol) was
added to a stirred solution of HL (200 mg, 1 mmol, 2 equiv.) in
THF (20 mL). The mixture was further stirred for 24 h at room
temp. and then concentrated under reduced pressure. Slow dif-
fusion of Et2O into the THF solution afforded black crystals suit-
able for X-ray diffraction analysis in 5 d, yield 30% (based on Mn).
Computational Details: The exchange coupling constant (J) has
been calculated for the two dinuclear complexes with the H2LS li-
gand (i.e., MnIIIHLS, FeIIIHLS) and the three complexes with HL
(i.e., MnIVL, FeIIIL, CrIIIL). We used the molecular structures de-
rived from the X-ray data without any further computational geo-
metrical refinement or chemical simplification. The crystallo-
graphic structures of MnIVL, FeIIIL and CrIIIL are provided in the
present work, whereas the solid-state structures of MnIIIHLS and
FeIIIHLS are available in ref.[9] and ref.[7a], respectively. The elec-
tronic structure of a single molecule in vacuo was obtained from
ab initio calculations by using the CASPT2 method.[10] We calcu-
lated the low-spin (LS) state and the high-spin (HS) state for each
complex. As all systems have a dimetallic magnetic centre with a
maximum spin multiplicity larger than S = 1/2, for the sake of
simplicity, we chose to work with the highest spin state available
instead of the triplet state. For the diiron compounds, there are two
FeIII(3d5) centres that generate an undecuplet HS state (S = 5). The
ferromagnetic MnIIIHLS has two MnIV(3d4) centres that give rise
to a nonuplet HS state (S = 4). On the other hand, the CrIIIL and
MnIVL complexes have two CrIII(3d3) and two MnIV(3d3) centres,
respectively, and their HS state is a septuplet (S = 3). In all cases,
the LS state is a singlet state. Therefore, we can extract J from the
difference in energy between each pair of HS/LS states. From the
IR: ν = 1638, 1588, 1517, 1491, 1443, 1405, 1361, 1172, 1136, 1068,
˜
1026, 995, 940, 791, 707, 689, 667 cm–1. C48H56Mn2N8O6·THF:
calcd. C 61.29, H 5.93, N 10.99; found C 60.93, H 5.89, N 10.90.
[FeIII2(μ-OMe)2(L)4] (2):
A solution of FeCl3·6H2O (16 mg,
0.1 mmol) in MeOH (2 mL) was added to a stirred solution of HL
(40 mg, 0.2 mmol, 2 equiv.) in MeOH (2 mL). Then, a solution of
KOAc (20 mg, 0.2 mmol, 2 equiv.) in MeOH (2 mL) was added
dropwise to the resulting mixture. The mixture was stirred for 1 h
at room temp. A red precipitate formed and was separated by fil-
tration and dissolved in CHCl3 (5 mL). Slow diffusion of MeOH
into the CHCl3 solution gave red crystals suitable for X-ray diffrac-
tion analysis in 4 d, yield 18% (based on Fe). IR: ν = 1634, 1587,
˜
1519, 1491, 1439, 1410, 1369, 1358, 1212, 1171, 1137, 1068, 1038,
1028, 992, 936, 791, 759, 709, 679, 668 cm–1. C50H58Fe2N8O6
(978.75): calcd. C 61.36, H 5.97, N 11.45; found C 61.11, H 6.10,
N 11.34.
[CrIII2(μ-OMe)2(L)4]·CHCl3·MeOH (3): CrCl3 (32 mg, 0.2 mmol)
and HL (83 mg, 0.4 mmol, 2 equiv.) were sequentially added to a
stirred suspension of Zn dust (220 mg, 33 mmol) in MeOH
(20 mL). The resulting mixture was heated at reflux for 24 h, cooled
to room temp. and then evaporated to dryness. The resulting black
powder was dissolved in CH2Cl2 (10 mL), and the solution was
ˆ
ˆ ˆ
Heisenberg Hamiltonian model H = –JΣiSiSi+1, the HS/LS differ-
ence for the diiron compounds accounts for 15J = ELS – EHS
,
whereas for the MnIIIHLS system it is 10J, and for both CrIIIL and
MnIVL complexes it is 6J.
filtered through Celite. Slow evaporation of
a
1:1 MeOH/
The CASPT2 method performs a perturbational expansion on top
CHCl3 mixture (2 mL) gave black crystals suitable for X-ray dif-
fraction analysis in 2 d, yield 10% (based on Cr).
of
a zero-order complete active space self-consistent field
(CASSCF) Hamiltonian.[10] This CASSCF reference defines an
Table 4. Crystal data and refinement details of 1–3.
MnIVL (1)
FeIIIL (2)
CrIIIL (3)
Formula
Fw
Space group
C48H52Mn2N8O6·THF
1018.97
C2/c
C50H58Fe2N8O6
978.75
P21/a
C50H58Cr2N8O6·MeOH·CHCl3
1122.47
P21/c
a [Å]
b [Å]
c [Å]
β [°]
18.0900(10)
17.3950(11)
15.2690(7)
92.853(3)
4
18.9370(6)
11.3680(5)
22.5530(6)
93.555(2)
4
21.300(0)
10.796(8)
23.491(0)
97.114(0)
4
Z
T [K]
173
0.71073
1.410
5.88
173
0.71073
1.342
6.56
0.1086, 0.2439
173
0.71073
1.391
6.13
0.0727, 0.1921
λ (Mo-Kα) [Å]
Calculated density (gcm–3)
μ(Mo-Kα) [cm–1]
R1 and Rw [IϾ2σ(I)][a]
0.0717, 0.1744
2
2 2
2
[a] R1 = Σ(||Fo| – |Fc||)/Σ|Fo|. Rw = [Σw(Fo – Fc ) /Σw(Fo2)2]1/2. w = 1/[σ2(Fo)2 + (np)2 + 0.00p], p = max(Fo2) + (2Fc )/3.
Eur. J. Inorg. Chem. 2014, 2552–2560
2558
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
active space (CAS) of molecular orbitals (MO) and incorporates
all the allowed electronic excitations within the CAS into the wave
function. The dimetallic complexes under study have open 3d shells
on both metal centres. Therefore, to produce a balanced description
of the LS and HS states for all dinuclear complexes, we defined a
CAS formed by the ten 3d-like MOs. The active space for the diiron
compounds incorporates ten 3d electrons for a total CAS[10,10].
Analogously, the MnIIIHLS system employed a CAS[8,10] with
eight 3d electrons for both MnIII atoms. The active space employed
for the CrIIIL and MnIVL complexes is a CAS[6,10]. The LS and
HS calculations of the five systems were performed by solving, in
each case, one electronic state with the corresponding spin multi-
plicity. All the electronic structure calculations were performed
with the MOLCAS7 package.[17]
Table 6. Number and type of molecular orbitals constituting each
of the regions defined on the partitioning for MnIVL, FeIIIL and
CrIIIHL. A schematic representation of the spatial limits of each
region can be found in Schemes 2 and 3.
Region
Type of molecular orbital
n
π
π*
Diffuse
B
C
6[a]
20
0
8
0
8
0
0
CP
M
BCPM
BCPM+
20
20
0
20
20
20
0
20
20
0
0
82
82
212
26[a]
26[a]
[a] Two more MOs for the MnIVL complex.
We analyzed the influence of the coordination sphere and the dif-
ferent parts of the ligands on the magnetic properties of the com-
plexes. Once we achieved a good qualitative description of both
systems at the CASSCF level, we applied the CASPT2 step locally
to some regions of the molecule and incorporated selectively the
so-called dynamical correlation. This procedure is based on the lo-
calization of the delocalized MO set resulting from the CASSCF
calculation.[18] The orbitals are transformed by following a valence-
bond-like description of the electronic density, which is based on
one of the Kekulé structures of the molecule and allows us to spa-
tially separate the orbitals.[19] Once the partitioning is done, it be-
comes possible to enable/disable defined regions of the compound
(e.g., the bridges between the metal centres or some aromatic rings)
to analyze how they affect the magnetic properties of the molecule.
We used the DOLO set of tools, which are available in the CASDI
code,[20] to perform the localization.
and N atoms have a set of 4s3p1d contracted functions, the C
atoms have a 3s2p set of contracted functions, and the H atoms
have a 2s basis set. We took advantage of the Cholesky decomposi-
tion to reduce the computational cost of working with these rela-
tively large molecules.[23] We performed the CASPT2 calculations
with an IPEA value of 0.00 hartree. This original choice of the
zeroth-order Hamiltonian gives more accurate results for magnetic
coupling parameters than the modified Hamiltonian.[24] We also
had to apply an imaginary shift factor of 0.20 hartree for the
CASPT2 step,[25] which we previously tested to be the minimum
value to eliminate possible intruder states.
Supporting Information (see footnote on the first page of this arti-
cle): Detailed bond lengths and angles, ORTEP views of 1–3, pow-
der X-ray diffraction pattern of 2.
Acknowledgments
The schemes of the partition are shown in Schemes 2 and 3. It must
be noted that the regions established for the partitioning do not
contain all the electronic density localized on them. We always left
the core orbitals and the σ-type orbitals disabled, specifically, as
frozen for the occupied MOs or deleted for the virtual MOs in
the CASPT2 calculation, because this type of orbitals is largely
nonpolarizable and will not interact with the metal centres.[21]
Thus, the partition was performed upon the π-type MOs and the
nonbonding MOs. Tables 5 and 6 show the exact distribution of
orbitals over the different partition stages. Additionally, diffuse vir-
tual orbitals located on the metal centres (M) and its coordination
sphere (M+) are also enabled at a later stage to analyze their influ-
ence on the magnetic properties.
This work was supported by the Centre National de la Recherche
Scientifique (CNRS). A. D. and V. R. thank the Agence Nationale
de la Recherche (ANR) (project number ANR-2010-BLAN-703)
and the Excellence Network Chemistry of Complex Systems
(LabEx CSC, ANR-10-LABX-0026\_CSC) for financial support.
D. S., S. M. and R. W. thank the Ministère de la Recherche
(France) for PhD financial support. V. R. thanks the Fundaçao
para a Ciância e Tecnologia (FCT) for the funding of a fellowship
(SFRH/BPD/44262/ 2008).
[1] a) C. J. Ballhausen, Introduction to Ligand Field Theory,
McGraw-Hill, New York, 1962, p. 298; b) M. Atanasov, D.
Ganyushin, K. Sivalingam, F. Neese, in: Structure and Bonding
vol. 142: Molecular Electronic Structures of Transition Metal
Complexes II (Eds.: D. M. P. Mingos, P. Day, J. P. Dahl),
Springer, Berlin, 2012, p. 149–220.
Table 5. Number and type of molecular orbitals constituting each
of the regions defined on the partitioning for MnIIIHLS and
FeIIIHLS. A schematic representation of the spatial limits of each
region can be found in Schemes 2 and 3.
[2] O. Kahn, Molecular Magnetism, Wiley-VCH, New York, 1993,
p. 396.
[3] a) K. R. Reddy, M. V. Rajasekharan, S. Padhye, F. Dahan, J. P.
Tuchagues, Inorg. Chem. 1994, 33, 428; b) N. K. Thallaj, J.
Przybilla, R. Welter, D. Mandon, J. Am. Chem. Soc. 2008, 130,
2414.
[4] a) R. E. P. Winpenny, Structure and Bonding, vol. 122: Single-
Molecule Magnets and Related Phenomena, Springer, Berlin,
2006, p. 262; b) A. Caneschi, D. Gatteschi, R. Sessoli, A. L.
Barra, L. C. Brunel, M. Guillot, J. Am. Chem. Soc. 1991, 113,
5873; c) R. Sessoli, Mol. Cryst. Liq. Cryst. Sci. Technol., Sect.
A 1995, 274, 145; d) D. Gatteschi, R. Sessoli, Angew. Chem.
Int. Ed. 2003, 42, 268; Angew. Chem. 2003, 115, 278.
[5] a) M. N. Leuenberger, D. Loss, Nature 2001, 410, 789; b) A. R.
Rocha, V. M. García-suárez, S. W. Bailey, C. J. Lambert, J.
Ferrer, S. Sanvito, Nat. Mater. 2005, 4, 335; c) F. Troiani, A.
Ghirri, M. Affronte, S. Carretta, P. Santini, G. Amoretti, S.
Region
Type of molecular orbital
n
π
π*
Diffuse
B
C
6
0
8
0
8
0
0
0
0
20
28
36
44
0
CP
CS
CPS
M
BCPSM 50
BCPSM+ 50
20
8
20
0
20
8
20
0
0
82
82
270
20
20
20
20
We used the molecular structures obtained from the X-ray data of
the respective crystalline systems. All the atoms of the molecule are
described with ANO-RCC type basis sets.[22] The metal centres
have a basis set formed by 6s5p4d2f contracted functions, the O
Eur. J. Inorg. Chem. 2014, 2552–2560
2559
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
Piligkos, G. Timco, R. E. P. Winpenny, Phys. Rev. Lett. 2005,
94, 207208; d) L. Bogani, W. Wernsdorfer, Nat. Mater. 2008,
7, 179; e) G. A. Timco, S. Carretta, F. Troiani, F. Tuna, R. J.
Pritchard, C. A. Muryn, E. J. L. McInnes, A. Ghirri, A. Cand-
ini, P. Santini, G. Amoretti, M. Affronte, R. E. P. Winpenny,
Nat. Nanotechnol. 2009, 4, 173.
[13]
a) O. Oms, J.-B. Rota, L. Norel, C. J. Calzado, H. Rousselière,
C. Train, V. Robert, Eur. J. Inorg. Chem. 2010, 5373; b) J.-B.
Rota, C. J. Calzado, C. Train, V. Robert, J. Chem. Phys. 2010,
132, 154702.
A. Domingo, M. Vérot, F. Mota, C. de Graaf, J. J. Novoa, V.
Robert, Phys. Chem. Chem. Phys. 2013, 15, 6982.
A.-M. Pradipto, R. Maurice, N. Guihéry, C. de Graaf, R.
Broer, Phys. Rev. B 2012, 85, 014409.
M. M. Andrade, M. T. Barros, J. Comb. Chem. 2010, 12, 245.
F. Aquilante, L. De Vico, N. Ferré, G. Ghigo, P.-Å. Malmqvist,
P. Neogrády, T. B. Pedersen, M. Pitonˇák, M. Reiher, B. O.
Roos, L. Serrano-Andrés, M. Urban, V. Veryazov, R. Lindh, J.
Comput. Chem. 2010, 31, 224.
a) S. F. Boys, Rev. Mod. Phys. 1960, 32, 296; b) J. Pipek, P. G.
Mezey, J. Chem. Phys. 1989, 90, 4916; c) C. Angeli, G. Del Re,
M. Persico, Chem. Phys. Lett. 1995, 233, 102.
a) J.-P. Malrieu, N. Guihéry, C. J. Calzado, C. Angeli, J. Com-
put. Chem. 2007, 28, 35; b) C. J. Calzado, C. Angeli, D. Tarat-
iel, R. Caballol, J.-P. Malrieu, J. Chem. Phys. 2009, 131,
044327; c) J. Zapata-Rivera, R. Caballol, C. J. Calzado, Phys.
Chem. Chem. Phys. 2011, 13, 20241.
[14]
[15]
[6] a) K. Mogilaiah, M. Prashanthi, G. Randheer Reddy, Synth.
Commun. 2003, 33, 3741; b) G. I. Mustata, A. Brigo, J. M.
Briggs, Bioorg. Med. Chem. Lett. 2004, 14, 1447; c) L. Saíz-
Urra, M. P. González, Y. Fall, G. Gómez, Eur. J. Med. Chem.
2007, 42, 64; d) K. Mogilaiah, E. Anitha, K. S. Kumar, R. S.
Prasad, Ind. J. Chem. B 2011, 50, 126; e) V. F. Shul’gin, E. A.
Sarnit, O. V. Konnik, E. B. Rusanov, A. S. Bogomyakov, V. I.
Ovcharenko, V. V. Minin, Russ. J. Coord. Chem. 2012, 38, 44.
[7] a) N. Bouslimani, N. Clément, G. Rogez, P. Turek, M. Bernard,
S. Dagorne, D. Martel, H. N. Cong, R. Welter, Inorg. Chem.
2008, 47, 7623; b) N. Clément, C. Toussaint, G. Rogez, C.
Loose, J. Kortus, L. Brelot, S. Choua, S. Dagorne, P. Turek, R.
Welter, Dalton Trans. 2010, 39, 4579; c) D. Specklin, C. Tour-
billon, V. Rosa, M. Kurmoo, R. Welter, Inorg. Chem. Commun.
2012, 20, 172; d) W. Zuo, V. Rosa, C. Tourbillon, D. Specklin,
C. Khaled, M. Kurmoo, R. Welter, RSC Adv. 2012, 2, 2517.
[8] a) R. Welter, WO 2009/130562 A1, 2009; b) K. Cheaib, D. Mar-
tel, N. Clément, F. Eckes, S. Kouaho, G. Rogez, S. Dagorne,
M. Kurmoo, S. Choua, R. Welter, Dalton Trans. 2013, 42, 1406.
[9] C. Beghidja, G. Rogez, J. Kortus, M. Wesolek, R. Welter, J.
Am. Chem. Soc. 2006, 128, 3140.
[16]
[17]
[18]
[19]
[20] N. Ben Amor, D. Maynau, Chem. Phys. Lett. 1998, 286, 211.
[21] T. Heine, R. Islas, G. Merino, J. Comput. Chem. 2007, 28, 302.
[22] a) B. O. Roos, R. Lindh, P.-Å. Malmqvist, V. Veryazov, P.-O.
Widmark, J. Phys. Chem. A 2004, 108, 2851; b) B. O. Roos, R.
Lindh, P.-Å. Malmqvist, V. Veryazov, P.-O. Widmark, J. Phys.
Chem. A 2005, 109, 6575.
[23] F. Aquilante, P.-Å. Malmqvist, T. B. Pedersen, A. Ghosh, B. O.
Roos, J. Chem. Theory Comput. 2008, 4, 694.
[24] N. Queralt, D. Taratiel, C. de Graaf, R. Caballol, R. Cimirag-
lia, C. Angeli, J. Comput. Chem. 2008, 29, 994.
[25] N. Forsberg, P.-Å. Malmqvist, Chem. Phys. Lett. 1997, 274,
196.
[10] a) B. O. Roos, P. R. Taylor, P. E. M. Siegbahn, Chem. Phys.
1980, 48, 157; b) K. Andersson, P. A. Malmqvist, B. O. Roos,
A. J. Sadlej, K. Wolinski, J. Phys. Chem. 1990, 94, 5483; c) K.
Andersson, P.-Å. Malmqvist, B. O. Roos, J. Chem. Phys. 1992,
96, 1218.
[11] a) K. Pierloot, E. Van Praet, L. G. Vanquickenborne, B. O.
Roos, J. Phys. Chem. 1993, 97, 12220; b) C. de Graaf, R. Broer,
W. C. Nieuwpoort, Chem. Phys. 1996, 208, 35.
Received: February 19, 2014
Published Online: April 23, 2014
[12] K. Wieghardt, U. Bossek, B. Nuber, J. Weiss, J. Bonvoisin, M.
Corbella, S. E. Vitols, J. J. Girerd, J. Am. Chem. Soc. 1988, 110,
7398.
Eur. J. Inorg. Chem. 2014, 2552–2560
2560
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim