Temperature-induced spin-transition in a low-spin cobalt(II) semiquinonate complex

Article abstract of DOI:10.1002/anie.200903789

(Chemical equation presented) A new aspect of cobalt dioxolene chemistry: The preparation and characterization of the first low-spin cobalt(II) semiquinonate complex is described. The temperature-induced change of the spin state in this cobalt dioxolene complex (see scheme) is linked to a spin-crossover process rather than to valence tautomerism.

Full text of DOI:10.1002/anie.200903789

Communications
DOI: 10.1002/anie.200903789
Spin-Crossover Complexes
Temperature-Induced Spin-Transition in a Low-Spin Cobalt(II)
Semiquinonate Complex**
Michꢀle Graf, Gotthelf Wolmershꢁuser, Harald Kelm, Serhiy Demeschko, Franc Meyer, and
Hans-Jꢂrg Krꢃger*
Dedicated to Professor Christoph Elschenbroich on the occasion of his 70th birthday
Spin crossover and valence tautomerism are examples of
processes that can be utilized as a basis for achieving
molecular switches.[1] Whereas the spin-crossover process is
characterized by a temperature-, pressure-, or light-induced
change of the electronic state of the metal ion to one with a
different spin multiplicity,[2] valence tautomerism entails an
intramolecular redox reaction between a metal ion and a
coordinated ligand, which, in a few instances, is accompanied
by a change in the spin state of the metal ion.[3] Various
reported low-spin cobalt(III) catecholate complexes, which
can be transformed into high-spin cobalt(II) semiquinonate
complexes by raising the temperature, provide excellent
examples of the latter process. In contrast, spin-crossover
chemistry is dominated by octahedral iron(II) complexes with
a FeN6 coordination sphere;[2] however, there are only very
few known octahedral cobalt(II)-containing spin-crossover
complexes.[4] Herein we describe the first cobalt(II) semi-
quinonate complex that displays spin-crossover properties
rather than valence tautomerism.
tetraphenylborate (Scheme 1). In accordance with the
description of 1 as a cobalt(III) catecholate complex,
solutions and solids of this substance are diamagnetic. X-ray
The starting point of our investigation was the olive-green
cobalt(III) 3,5-di-tert-butylcatecholate (dbc2À) complex [Co-
(L-N4Me2)(dbc)](BPh4)·0.8MeCN·0.2Et2O (1) containing the
dimethyl derivative of the tetraazamacrocyclic ligand 2,11-
diaza[3.3](2,6)pyridinophane (L-N4Me2) as coligand. This
complex was obtained in 42% yield by oxidation of the red
cobalt(II) catecholate complex [Co(L-N4Me2)(dbc)] (pre-
pared in situ from equimolar solutions of cobalt(II) perchlo-
rate, L-N4Me2, and 3,5-di-tert-butylcatecholate) with ferroce-
nium tetrafluoroborate ([Fe(Cp)2](BF4); Cp = cyclopenta-
dienyl), followed by a metathesis reaction with sodium
Scheme 1. Preparation of compounds 1 and 2.
structure analysis of 1 also supports this assignment.[6]
Figure 1 shows a perspective view of the complex cation in
1. Because of the small size of the macrocyclic ring, the
coordinated ligand L-N4Me2 is folded along the Namine–Namine
axis, thereby rendering a distorted cis-octahedral coordina-
[*] Dr. M. Graf, Dr. G. Wolmershꢁuser, Dr. H. Kelm, Prof. H.-J. Krꢂger
Fachbereich Chemie, Technische Universitꢁt Kaiserslautern
Erwin-Schrꢃdinger Strasse 54, 67663 Kaiserslautern (Germany)
Fax: (+49)631-205-4676
E-mail: krueger@chemie.uni-kl.de
Homepage: http://www.chemie.uni-kl.de/fachrichtungen/ac/
krueger/
Dr. S. Demeschko, Prof. Dr. F. Meyer
Institut fꢂr Anorganische Chemie
Georg-August-Universitꢁt Gꢃttingen (Germany)
[**] We gratefully acknowledge financial support from the Deutsche
Forschungsgemeinschaft (Priority Program “Molecular Magnet-
ism”, SPP 1137), the Technische Universitꢁt Kaiserslautern, and the
research center OPTIMAS.
Figure 1. Perspective view of the complex cation in 1 showing 50%
thermal ellipsoids; selected bond lengths [ꢀ]: Co(1)–N(1) 1.978(3),
Co(1)–N(2) 1.853(2), Co(1)–N(3) 1.980(2), Co(1)–N(4) 1.843(2),
Co(1)–O(1) 1.856(2), Co(1)–O(2) 1.865(2), C(17)–O(1) 1.362(3),
C(18)–O(2) 1.359(3).
Supporting information (synthetic procedures and characterization
data of complexes 1 and 2) for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200903789.
950
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 950 –953

Angewandte
Chemie
tion environment. Thus, the cobalt ion is coordinated to the
two oxygen donor atoms of the catecholate moiety and the
two pyridine nitrogen donor atoms of the tetraazamacrocyclic
ligand in the equatorial plane, and to the two amine nitrogen
donor atoms in the axial positions. In all cis-octahedral
transition-metal complexes containing L-N4Me2 as ligand, the
axial metal–Namine bonds are longer than the equatorial
metal–Npy bonds by at least 0.1 ꢀ.[7] In addition, the lengths
of the bonds between the metal ion and the nitrogen atoms of
the tetraazamacrocyclic ligand are quite characteristic of the
respective spin state and oxidation state of the coordinated
metal ion; they can, therefore, be used as diagnostic tools for
determining the electronic state of the metal ion. Thus, the
Figure 2. Electronic absorption spectra of 1 (dashed line) and 2 (solid
line) in acetonitrile.
À
À
observation of rather short Co Npy and Co Namine bond
lengths of (1.848 Æ 0.005) ꢀ[8] and (1.979 Æ 0.001) ꢀ, respec-
tively, is consistent only with the presence of a low-spin
À
cobalt(III) ion. Further, the average of the C O bond lengths
À
of (1.361 Æ 0.002) ꢀ as well as the more or less equal C C
bond lengths within the aromatic ring clearly identify the
coordinated dioxolene ligand as a catecholate unit.[3c,9] This
electronic ground state prevails even at higher temperatures.
To be able to reach the cobalt(II) electronic state by
raising the temperature, the capability of the ligand to donate
electron density to the metal ion has to be reduced. By
introducing bulky amine substituents such as tert-butyl groups
into the macrocyclic ligand, this objective can be accom-
plished because the amine substituents of the coordinated
tetraazamacrocycle are located above and below the two cis-
oriented coordination sites of the dioxolene ligand, and,
therefore, the larger steric interactions between the tert-butyl
À
groups and the dioxolene ligand will increase the Co Namine
bond length. Thereby, the capability of the ligand to donate
electron density to the coordinated metal ion is weakened and
at the same time the ligand-field strength of the di-tert-butyl-
substituted macrocycle L-N4tBu2 is reduced relative to that of
the ligand L-N4Me2.
Figure 3. Perspective view of the complex cation in 2 showing 50%
thermal ellipsoids; selected bond lengths [ꢀ] at 100 K [at 400 K]:
Co(1)–N(1) 2.308(2) [2.320(3)], Co(1)–N(2) 1.932(2) [1.994(3)],
Co(1)–N(3) 2.368(2) [2.371(3)], Co(1)–N(4) 1.931(2) [1.987(3)],
Co(1)–O(1) 1.888(1) [1.974(2)], Co(1)–O(2) 1.892(1) [1.964(2)],
C(23)–O(1) 1.311(2) [1.307(4)], C(24)–O(2) 1.315(3) [1.274(4)].
Employing an analogous synthetic procedure as before,
but starting with L-N4tBu2 instead of L-N4Me2 and using
potassium tetra(p-chlorophenyl)borate in the metathesis
reaction afforded the analytically pure, dark green compound
[Co(L-N4tBu2)(dbsq)](B(p-C6H4Cl)4) (2) in 61% yield
(dbsqÀ = 3,5-di-tert-butylsemiquinonate; Scheme 1) The big-
gest difference between the UV/Vis spectra of 1 and 2 is the
presence of an intense CT absorption band with a maximum
at 1075 nm (eM = 7640 LmolÀ1 cmÀ1) and a shoulder at
1042 nm (eM = 6890 LmolÀ1 cmÀ1) for 2 (Figure 2). This obser-
vation as well as the paramagnetism of 2 in solution (as
evidenced by its NMR spectrum) point towards a different
electronic ground state of 2 compared to that of 1. The
structure of 2 was determined at 100 K (Figure 3).[10] As
expected of a low-spin d7 metal ion, the Jahn–Teller effect
places the unpaired electron in the s-antibonding dz2 orbital,
whereas in the high-spin state both s-antibonding d orbitals
are singly occupied. Therefore, in cobalt(II) complexes
2.096 ꢀ).[11] In the structure of 2, the averages of the Co Npy
À
À
and the Co Namine bond lengths of (1.932 Æ 0.001) ꢀ and
(2.338 Æ 0.030) ꢀ,[12] respectively, indicate a low-spin cobal-
À
t(II) ion. In addition, the average of the C O bond lengths of
(1.313 Æ 0.002) ꢀ falls into the range commonly observed for
coordinated semiquinonate radicals. In contrast to 1, inspec-
À
tion of the C C bonds in 2 reveals a more localized p-bonding
pattern of an ortho-semiquinonate radical with short bonds
between the carbon atoms C(25) and C(26) as well as between
C(27) and C(28), and a rather long bond between the carbon
atoms C(23) and C(24).[3a,9] On the basis of these results, we
conclude that, at 100 K, complex 2 can be adequately
described as a low-spin cobalt(II) semiquinonate complex.
The magnetic properties of compound 2 were investigated
over a temperature range from 2 to 400 K with a SQUID
magnetometer. The dependence of cM T versus T is depicted
in Figure 4. The graph can be divided into three parts:
Between 40 and 200 K, the curve reaches a plateau with a
À
containing the macrocyclic ligand L-N4tBu2, the Co Namine
bonds are quite elongated in both the high-spin (2.351–
2.460 ꢀ) and the low-spin complexes (2.368–2.404 ꢀ),
À
whereas the Co Npy bond lengths differ considerably between
low-spin (1.902–1.940 ꢀ) and high-spin complexes (2.082–
Angew. Chem. Int. Ed. 2010, 49, 950 –953
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
951

Communications
To prove that a spin transition is indeed responsible for
the increase of the cM T value above 200 K, data collection for
a structure analysis was carried out at 400 K. The results of
À
this structural investigation demonstrate that the axial Co
Namine bond lengths remain nearly the same, whereas the
À
À
equatorial Co Npy and Co O bond lengths significantly
À
increase with temperature. Thus, the averages of the Co
À
Npy and of the Co O bond lengths change from (1.931 Æ
0.001) ꢀ to (1.991 Æ 0.004) ꢀ and from (1.890 Æ 0.002) ꢀ to
À
(1.969 Æ 0.005) ꢀ, respectively. The observed Co Npy bond
lengths at 400 K indicate, too, that the spin transition is not
complete at this temperature.[11] The slightly decreased C O
À
bond lengths at 400 K are still in the range of those in
Figure 4. Variation of the product cM T with temperature for solid 2 at
an applied magnetic field of 0.5 T.
coordinated semiquinonate radicals, but an increasing asym-
[3c,9]
À
metry between both C O bonds is discernible.
In summary, magnetic as well as structural data unequiv-
ocally demonstrate that a temperature-induced spin-cross-
over process takes place in complex 2. Spin-crossover
complexes with cobalt(II) ions are still considered rare and
are generally only observed with strong-field ligands. Further,
to best of our knowledge, until now temperature-induced
changes of the spin state in cobalt dioxolene complexes have
all been linked to valence tautomerism. Therefore, complex 2
is the first known low-spin cobalt(II) semiquinonate complex
and also the first known cobalt dioxolene complex that
undergoes temperature-induced spin changes as a result of
spin crossover rather than valence tautomerism.
We attribute this finding to the special structural features
of the coordinated ligand L-N4tBu2. In combination with a
semiquinonate ligand, this macrocyclic ligand still exerts
sufficient ligand-field strength to enforce a low-spin state as
the ground state upon the cobalt(II) ion. Because of the steric
interactions between the tert-butyl substituents and the ligand
value for cM T of about 1.16 cm3 KmolÀ1. This value corre-
sponds to a species with a spin state of S = 1. Taking the
assignment of 2 as a low-spin cobalt(II) semiquinonate
complex as the basis for describing the electronic ground
state, one unpaired electron resides in the s-antibonding
dz2 orbital, the other unpaired electron occupies a p-molec-
ular orbital of the semiquinonate radical. Considering that
both orbitals are orthogonal to each other and based on
related complexes with copper(II) ions,[13] a strong ferromag-
netic exchange coupling between both spin carriers is
expected. The observance of a ground state of S = 1 is,
therefore, consistent with the interpretation of the structural
data at 100 K. Below 40 K, the cM T curve falls to a value of
0.15 cm3 KmolÀ1. This behavior can be attributed to zero-field
splitting and/or to very weak intermolecular antiferromag-
netic coupling within the crystal lattice. A similar decrease
was found in copper(II) semiquinonate complexes with the
same spin ground state.[13b] However, the most remarkable
finding is that, above 200 K, there is a steady increase of the
cM T curve to about 1.98 cm3 KmolÀ1 at 400 K. This finding
can only be attributed to a gradually occurring spin transition
from a low-spin to a high-spin cobalt(II) state. This spin-
crossover process is, however, not complete at 400 K. In
reported high-spin cobalt(II) semiquinonate complexes, the
exchange coupling between the S = 3/2 spin of the high-spin
cobalt(II) ion and the S = 1/2 spin of the semiquinonate
radical is generally described as antiferromagnetic, leading to
a spin ground state of S = 1 with a moderate coupling
constant.[3,14] Any fitting of the magnetic data to a theoretical
model is problematic because both the spin transition process
and the exchange coupling contribute to the values of the cM T
curve above 200 K, no data are available to us above 400 K
(the high-temperature limit of the SQUID magnetometer),
and the removal of degeneracy of the 4T1g state in a high-spin
octahedral cobalt(II) ion as a result of distortion of the ligand
field and spin–orbit coupling can complicate a magnetic
analysis even without the occurrence of spin crossover.[14] We
felt that any attempt to simulate the powder susceptibility
data including all those effects would suffer from severe
overparametrization. Hence, no unambiguous and reliable
values for the exchange coupling constant and the critical
spin-crossover temperature can be derived at this time.
À
at the two cis-oriented coordination sites, the axial Co Namine
bonds are considerably elongated. In a low-spin cobalt(III)
À
complex containing L-N4tBu2, the axial Co Namine bond
lengths were found to be approximately 2.09 ꢀ.[15] Thus,
CoIII Namine bonds are about 0.1 ꢀ longer in low-spin cobalt-
À
(III) complexes containing L-N4tBu2 than in those containing
L-N4Me2. In the low-spin cobalt(II) complex, the prevailing
À
Jahn–Teller effect favors the long axial Co Namine bonds,
which are also enforced by the macrocyclic ligand L-N4tBu2.
Upon oxidizing the cobalt(II) ion to a cobalt(III) ion, the
À
anticipated decrease of the Co Namine bond lengths results in
stronger steric interactions between the tert-butyl substituents
and the dioxolene ligand. All of these factors destabilize the
cobalt(III) state relative to the cobalt(II) state in complexes
that contain the macrocycle L-N4tBu2. Consequently, this
increase in the redox potential of the [Co(L-N4tBu2)]2+/3+
fragment relative to that of the coordinated dioxolene
ligand is responsible for the presence of the low-spin
cobalt(II) semiquinonate ground state in 2.
Received: July 10, 2009
Published online: December 22, 2009
Publication delayed at the authors request
Keywords: cobalt · magnetic properties · N ligands · O ligands ·
.
spin crossover
952
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 950 –953

Angewandte
Chemie
can be obtained free of charge from The Cambridge Crystallo-
graphic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[7] H.-J. Krꢂger, Chem. Ber. 1995, 128, 531.
[1] a) O. Kahn, J. P. Launey, Chemtronics 1988, 3, 140; b) J.-F.
Lꢁtard, P. Guionneau, L. Goux-Capes, Top. Curr. Chem. 2004,
235, 221.
[8] The deviation in the arithmetic average of bond lengths, for
[2] a) P. Gꢂtlich, A. Hauser, H. Spiering, Angew. Chem. 1994, 106,
2109; Angew. Chem. Int. Ed. Engl. 1994, 33, 2024; b) Spin
Crossover in Transition Metal Compounds I–III (Eds.: P. Gꢂtlich,
H. A. Goodwin), Springer, Berlin, 2004 (Top. Curr. Chem. 2004,
233–235).
[3] a) D. A. Shultz in Magnetism: Molecules to Materials, Vol. II
(Eds.: J. S. Miller, M. Drillon), Wiley-VCH, Weinheim, 2001,
pp. 281 – 306; b) D. N. Hendrickson, C. G. Pierpont, Top. Curr.
Chem. 2004, 234, 63; c) C. G. Pierpont, Coord. Chem. Rev. 2001,
216–217, 99; d) A. Beni, C. Carbonera, A. Dei, J.-F. Lꢁtard, R.
Righini, C. Sangregorio, L. Sorace, J. Braz. Chem. Soc. 2006, 17,
1522.
[4] a) H. A. Goodwin, Top. Curr. Chem. 2004, 234, 23; b) J.
Zarembowitch, New J. Chem. 1992, 16, 255; c) I. Krivokapic,
M. Zerara, M. Lawson Daku, A. Vargas, C. Enachescu, C.
Ambrus, P. Tregenna-Piggott, N. Amstutz, E. Krausz, A. Hauser,
Coord. Chem. Rev. 2007, 251, 364; d) U. Beckmann, S. Brooker,
Coord. Chem. Rev. 2003, 245, 17.
À
example, Æ 0.003 ꢀ, indicates that both Co N bond lengths are
identical within 0.006 ꢀ, and should not be mistaken for a
statistical standard deviation.
[9] a) C. G. Pierpont, C. W. Lange, Prog. Inorg. Chem. 1994, 41, 331;
b) A. S. Attia, B. J. Conklin, C. W. Lange, C. G. Pierpont, Inorg.
Chem. 1996, 35, 1033; c) S. Bhattacharya, P. Gupta, F. Basuli,
C. G. Pierpont, Inorg. Chem. 2002, 41, 5810.
[10] Crystal data for 2: a) at 100 K: C60H68BCl4CoN4O2, Mr = 1088.72,
monoclinic, space group P21/n, a = 10.3507(1), b = 26.2575(2),
c = 20.5763(2) ꢀ, b = 101.778(1)8, V= 5474.6(1) ꢀ3, Z = 4,
1calcd = 1.321 gcmÀ3
;
T= 100 K, m(CuKa) = 4.613 mmÀ1 (l =
1.5418 ꢀ); 30291 reflections measured, 8658 unique, refinement
converged to R = 0.0340, wR2 = 0.0821 (I > 2s(I)), 661 parame-
ters and 0 restraints; min./max. residual electron density =
+ 0.284/À0.357 eꢀÀ3; b) at 400 K: a = 10.478(1), b = 26.873(1),
c = 20.977(1) ꢀ, b = 102.93(1)8, V= 5757(1) ꢀ3, Z = 4, 1calcd
=
1.256 gcmÀ3; T= 400 K, m(MoKa) 0.528 mmÀ1 (l = 0.71073 ꢀ);
35029 reflections measured, 11621 unique, refinement con-
verged to R = 0.0530, wR2 = 0.0689 (I > 2s(I)), 661 parameters
and 0 restraints; min./max. residual electron density = + 0.236/
[5] The presence of nonstoichiometric amounts of MeCN and Et2O
in compound 1 is based on the X-ray structure determination,
NMR spectroscopic results, and chemical analysis. The content
can vary depending of the method used for quantification,
rendering a general formulation of 1 as [Co(L-N4Me2)(dbc)]-
(BPh4)·(MeCN)x·(Et2O)1Àx (x = 0.7–0.9). We assumed an aver-
aged composition of the formulation as [Co(L-N4Me2)(dbc)]-
(BPh4)·0.8MeCN·0.2Et2O, which imposes a maximum error of
less than 0.5% on calculations concerning molar amounts.
[6] Crystal data for 1: C54H60BCoN4O2, Mr = 866.80, monoclinic,
space group P21/n, a = 19.020(2), b = 13.889(1), c = 21.177(2) ꢀ,
b = 110.98(1)8, V= 5223(1) ꢀ3, Z = 4, 1calcd = 1.102 gcmÀ3; T=
193 K, m(MoKa) 0.369 mmÀ1 (l = 0.71073 ꢀ); 55758 reflections
measured, 10900 unique, refinement converged to R = 0.0455,
À0.208 eꢀÀ3
.
[11] a) S. P. Meneghetti, P. J. Lutz, J. Fischer, J. Kress, Polyhedron
2001, 20, 2705; b) S. Reh, H. Kelm, H.-J. Krꢂger, unpublished
results.
[12] In many octahedral metal complexes containing L-N4tBu2 as
À
coligand, the two axial M Namine bonds differ substantially in
length. This result also explains the rather large range of values
found for Co Namine bond lengths in complexes of this type.
À
[13] a) O. Kahn, Molecular Magnetism, Wiley-VCH, Weinheim,
1993; b) O. Kahn, R. Prins, J. Reedijk, J. S. Thompson, Inorg.
Chem. 1987, 26, 3557; c) M. Ruf, B. C. Noll, M. D. Groner, G. T.
Yee, C. G. Pierpont, Inorg. Chem. 1997, 36, 4860; d) C. Benelli,
A. Dei, D. Gatteschi, L. Pardi, Inorg. Chem. 1990, 29, 3409.
[14] A. Caneschi, A. Dei, D. Gatteschi, V. Tangoulis, Inorg. Chem.
2002, 41, 3508.
wR2 = 0.0779 (I > 2s(I)), 567 parameters and
0 restraints.
Disordered solvent molecules (MeCN and Et2O) were found
within an area of a packing gap (x = 0.56, y = 0.60, z = 0.61). A
definite determination of the only partially occupied positions
was not possible. Therefore, a modified set of reflection data was
calculated with the option SQUEEZE of the program PLATON
so that the contribution to the reflection data from this area of
the packing gap is not included in the refinements. Min./max.
residual electron density = + 0.266/À0.279 eꢀÀ3. CCDC 748989
(1), 748991 (2, 100 K),[10] and 748990 (2, 400 K)[10] contain the
supplementary crystallographic data for this paper. These data
[15] The cobalt(III) complex [Co(L-N4tBu2)(NSC)2](BF4)·EtCN was
prepared and characterized by X-ray crystallography. The
À
À
averages of the Co Npy and of the Co Namine bond lengths in
this complex are 1.876 and 2.093 ꢀ, respectively. Interestingly, in
contrast to the cobalt(II) complexes, the lengths of both
À
Co Namine bonds differ only by 0.003 ꢀ. S. Reh, H. Kelm, H.-J.
Krꢂger, unpublished results.
Angew. Chem. Int. Ed. 2010, 49, 950 –953
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
953