Dicobalt II-II, II-III, and III-III complexes as spectroscopic models for dicobalt enzyme active sites

Article abstract of DOI:10.1021/ic7020534

A matched set of dinuclear cobalt complexes with II-II, II-III, and III-III oxidation states have been prepared and structurally characterized. In [(bpbp)Co₂(O₂P(OPh)₂)₂] ⁿ⁺ (n = 1, 2, or 3; bpbp- = 2,6-bis((N,/V-bis-(2-picolyl)amino)- methyl)-4-tertbutylphenolato), the nonbonded Co...Co separations are within the range 3.5906(17) to 3.7081(11) A, and the metal ions are triply bridged by the phenolate oxygen atom of the heptadentate dinucleating ligand and by two diphenylphosphate groups. The overall structures and geometries of the complexes are very similar, with minor variations in metal-ligand bond distances consistent with oxidation state assignments. The Co^IICo ^III compound is a valence-trapped Robin-Day class II complex. Solid state ³¹P NMR spectra of the diamagnetic Co^IIICo ^III (3) and paramagnetic Co^IICo^III (2) and Co^IICo^II (1) complexes show that ³¹P isotropic shifts broaden and move downfield by about 3000 ppm for each increment in oxidation state. Cyclic voltammetry corroborates the existence of the Co ^IICo^II, Co^IICo^III, and Co ^IIICo^III species in solution. The redox changes are not reversible in the applied scanning timescales, indicating that chemical changes are associated with oxidation and reduction of the cobalt centers. An investigation of the spectroscopic properties of this series has been carried out for its potential usefulness in analyses of the related spectroscopic properties of the dicobalt metallohydrolases. Principally, magnetic circular dichroism (MCD) has been used to determine the strength of the magnetic exchange coupling in the Co^IICo^II complex by analysis of the variable-temperature variable-field (VTVH) intensity behavior of the MCD signal. The series is ideal for the spectroscopic determination of magnetic coupling since it can occur only in the Co^IICo^II complex. The Co^IICo^III complex contains a nearly isostructural Co ^II ion, but since Co^III is diamagnetic, the magnetic coupling is switched off, while the spectral features of the Co^II ion remain. Analysis of the MCD data from the Co^IICo^III complex has been undertaken in the theoretical context of a ⁴T _1g ground-state of the Co^II ion, initially in an octahedral ligand field that is split by both geometric distortion and zero-field splitting to form an isolated doublet ground state. The MCD data for the Co^IICo^II pair in the [(bpbp)Co₂(O ₂P(OPh)₂)₂]⁺ complex were fitted to a model based on weak antiferromagnetic coupling with J = -1.6 cm^-1. The interpretation is confirmed by solid state magnetic susceptibility measurements.

Full text of DOI:10.1021/ic7020534

Inorg. Chem. 2008, 47, 5079-5092
Dicobalt II-II, II-III, and III-III Complexes as Spectroscopic Models for
Dicobalt Enzyme Active Sites
Frank B. Johansson,† Andrew D. Bond,† Ulla Gro Nielsen,† Boujemaa Moubaraki,‡ Keith S. Murray,‡
Kevin J. Berry,§ James A. Larrabee,*,| and Christine J. McKenzie*,†
Department of Physics and Chemistry, UniVersity of Southern Denmark, CampusVej 55, 5230
Odense M, Denmark, School of Chemistry, Building 23, Monash UniVersity, Clayton, Victoria
Australia, Department of Chemistry and Biochemistry, Middlebury College, Middlebury, Vermont,
U.S.A., and Westernport Secondary College, Hastings, Victoria, 3915, Australia
Received October 17, 2007
A matched set of dinuclear cobalt complexes with II-II, II-III, and III-III oxidation states have been prepared and
structurally characterized. In [(bpbp)Co2(O2P(OPh)2)2]n+ (n ) 1, 2, or 3; bpbp- ) 2,6-bis((N,N′-bis-(2-picolyl)amino)-
methyl)-4-tertbutylphenolato), the nonbonded Co · · · Co separations are within the range 3.5906(17) to 3.7081(11)
Å, and the metal ions are triply bridged by the phenolate oxygen atom of the heptadentate dinucleating ligand and
by two diphenylphosphate groups. The overall structures and geometries of the complexes are very similar, with
minor variations in metal-ligand bond distances consistent with oxidation state assignments. The CoIICoIII compound
is a valence-trapped Robin-Day class II complex. Solid state 31P NMR spectra of the diamagnetic CoIIICoIII (3)
and paramagnetic CoIICoIII (2) and CoIICoII (1) complexes show that 31P isotropic shifts broaden and move downfield
by about 3000 ppm for each increment in oxidation state. Cyclic voltammetry corroborates the existence of the
CoIICoII, CoIICoIII, and CoIIICoIII species in solution. The redox changes are not reversible in the applied scanning
timescales, indicating that chemical changes are associated with oxidation and reduction of the cobalt centers. An
investigation of the spectroscopic properties of this series has been carried out for its potential usefulness in
analyses of the related spectroscopic properties of the dicobalt metallohydrolases. Principally, magnetic circular
dichroism (MCD) has been used to determine the strength of the magnetic exchange coupling in the CoIICoII
complex by analysis of the variable-temperature variable-field (VTVH) intensity behavior of the MCD signal. The
series is ideal for the spectroscopic determination of magnetic coupling since it can occur only in the CoIICoII
complex. The CoIICoIII complex contains a nearly isostructural CoII ion, but since CoIII is diamagnetic, the magnetic
coupling is switched off, while the spectral features of the CoII ion remain. Analysis of the MCD data from the
4
CoIICoIII complex has been undertaken in the theoretical context of a T1g ground-state of the CoII ion, initially in
an octahedral ligand field that is split by both geometric distortion and zero-field splitting to form an isolated doublet
ground state. The MCD data for the CoIICoII pair in the [(bpbp)Co2(O2P(OPh)2)2]+ complex were fitted to a model
based on weak antiferromagnetic coupling with J ) -1.6 cm-1. The interpretation is confirmed by solid state
magnetic susceptibility measurements.
Introduction
interactions and reactivity of dinuclear CoII active sites in
proteins. Methionine aminopeptidase (MetAP) is an ubiq-
uitous dinuclear metallohydrolase responsible for removal
of the N-terminal methionine in nascent proteins. Extensive
studies have been conducted on the dicobalt form of the
enzyme, and structural determinations of human,1 Pyrococcus
furiousus,2 and Escherichia coli3a MetAPs have revealed
An understanding of the electronic properties of dinuclear
CoII complexes will provide insight into the electronic
* To whom correspondence should be addressed. E-mail: larrabee@
middlebury.edu (J.A.L.), chk@ifk.sdu.dk (C.J.M.).
†
University of Southern Denmark.
Monash University.
Westernport Secondary College.
Middlebury College.
‡
§
|
(1) Roderick, S. L.; Matthews, B. W. Biochemistry 1993, 32, 3907–3912.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5079
10.1021/ic7020534 CCC: $40.75  2008 American Chemical Society
Published on Web 05/22/2008

Johansson et al.
Scheme 1. Schematic Diagrams of Four Dicobalt Active Sites (Distances Shown in Å)
structurally similar dicobalt centers in the active sites. They
consist of two metal ions bridged by a hydroxide ion (or a
water molecule), one aspartate, and one glutamate. A recent
structure determination of MetAP from E. coli found one of
the CoII ions to be hexacoordinated in a distorted octahedral
environment, having a coordinated water and bidentate
aspartate completing its coordination sphere. The second CoII
ion is pentacoordinated in a distorted trigonal-bipyramidal
arrangement, with an imidazole nitrogen from histidine and
a monodentate glutamate completing the pentacoordination.
The carboxylate-bridged dimetallic active site is a common
structural motif in proteins and enzymes (Scheme 1). Other
proteins in this category include aminopeptidase, the dioxy-
gen carrier protein, hemerythrin, several phosphatases, and
phosphoesterases, such as glycerophosphodiesterase from
Enterobacter aerogenes (GpdQ).3 Few of these enzymes
have cobalt as the in vivo metal, but most are fully active or
even hyperactive in the dicobalt form. For example, MetAP
is likely a MnII enzyme, but most of the structural and
biochemical studies to date have been conducted on the CoII
form.3e,f Often the cobalt forms of these proteins are studied
because CoII is an excellent spectroscopic probe.
This is usually reported as “J” when using the isotropic
Heisenberg-Dirac-van Vleck (HDVV) theoretical model.
J is defined throughout this paper and in Table 7 by use of
the Hamiltonian -2JS1.S2. Many of the papers cited herein
use -JS1.S2 and thus their quoted J is equal to our 2J. In
MetAP it has been demonstrated that some inhibitors and
substrate analogs displace or interact with the bridging
hydroxide. In such situations, perturbation of the magnetic
coupling is expected.2 It has been demonstrated that the
diiron active site in pig purple acid phosphatase is weakly
antiferromagnetically coupled at low pH while the mixed
iron/manganese active site in a sweet potato purple acid
phosphatase is strongly coupled at the same pH.3c This has
been interpreted in terms of the presence of a µ-hydroxo
bridge in the diiron enzyme and a µ-oxo bridge in the mixed-
metal enzyme. The study of model systems will aid
understanding of the structural parameters that influence the
magnitude and sign of magnetic coupling so that measure-
ments in enzyme active sites can be interpreted. However,
there is currently a paucity of reported magnetic coupling
strengths for dinuclear CoII complexes that are structurally
similar to MetAP and related enzymes, partly because of
the experimental and theoretical difficulty in the quantitative
determination of magnetic coupling strength in high-spin CoII
dimers, and partly for the lack of suitable complexes. Some
dinuclear phosphate-bridged CoIICoIII complexes have been
structurally characterized, but these have been used pre-
dominantly to investigate phosphate hydrolysis.4 Brown et
One key parameter for comparing dinuclear CoII com-
plexes is the strength of the magnetic exchange coupling.
(2) (a) Lowther, W. T.; Orville, A. M.; Madden, D. T.; Lim, S.; Rich,
D. H.; Matthews, B. W. Biochemistry 1999, 38, 7678–7688. (b)
Lowther, W. T.; Zhang, Y.; Sampson, P. B.; Honek, J. F.; Matthews,
B. W. Biochemistry 1999, 38, 14810–14819.
(3) (a) Lowther, W. T.; Matthews, B. W. Chem. ReV. 2002, 102, 4581–
4608. (b) Wilcox, D. E. Chem. ReV. 1996, 96, 2435–2458. (c) Mitic,
N.; Smith, S. J.; Neves, A.; Guddat, L. W.; Gahan, L. R.; Schenk, G.
Chem. ReV. 2006, 106, 3338–3363. (d) Jackson, C. J.; Carr, P. D.;
Liu, J-W.; Watt, S. J.; Beck, J. L.; Ollis, D. L. J. Mol. Biol. 2007,
367, 1047–1062. (e) Wang, J.; Sheppard, G. S.; Lou, P.; Kawai, M.;
Park, C.; Egan, D. A.; Schneider, A.; Bouska, J.; Lesniewski, R.;
Henkin, J. Biochemistry 2003, 42, 5035–5042. (f) Copik, A. J.; Nocek,
B. P.; Swierczek, S. I.; Ruebush, S.; Jang, S. B.; Meng, L.; D’sousa,
V. M.; Peters, J. W.; Bennett, B.; Holz, R. C. Biochemistry 2005, 44,
121–129.
(4) (a) Wahnon, D.; Lebuis, A.-M.; Chin, J. Angew. Chem., Int. Ed. Engl.
1995, 34, 2412–2414. (b) Seo, J. S.; Sung, N.-D.; Hynes, R. C.; Chin,
J. Inorg. Chem. 1996, 35, 7472–7473. (c) Seo, J. S.; Hynes, R. C.;
Williams, D.; Chin, J. J. Am. Chem. Soc. 1998, 120, 9943–9944. (d)
Williams, N. H.; Cheung, W.; Chin, J. J. Am. Chem. Soc. 1998, 120,
8079–8087.
5080 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
al.,5 Ostrovsky et al.,6a,b Ye et al.7 and Sakiyama et al.8d
have prepared and characterized several dinuclear CoII
complexes having acetate bridging ligands, and in some
cases, an additional bridging oxygen atom from another
ligand. The reported coupling constants for these complexes
are all in the range of weak antiferromagnetic coupling (J
) -0.2 to -6.9 cm-1), except for one complex reported to
exhibit moderate ferromagnetic coupling (J ) 18.0 cm-1)
resulting in a µeff versus temperature plot that had a broad
maximum at ∼100 K.6b In some of these reports the
Hamiltonian used in the HDVV model neglected zero-field
splitting, which can affect the magnitude of J. Orbital
contributions, which can be important for highly symmetric
CoII dimers, have been considered in a number of complexes
after it was shown that the HDVV coupling model was
inadequate because of strong spin-orbit coupling and ligand-
field octahedral distortion effects.6b,8a,c,d
We have used magnetic circular dichroism (MCD) to probe
the magnetic coupling in the CoIICoII complex through
analysis of the temperature dependence of the MCD signal.
The CoIICoIII and CoIICoII complexes described herein are
ideally suited to test this spectroscopic method. In such
systems, properties in the ground-state that are due to
coupling can be isolated from those due to the uncoupled
CoII “monomeric” ground state. Such systems for cobalt have
been known for some time. For example, Wieghardt et al.
have reported a series of complexes having a Co2(µ-X)(µ-
OAc)2n+ (X ) OH, Cl, or Br; n ) 1-3) core, for which the
CoIICoII, CoIICoIII, and CoIIICoIII combinations were pre-
pared.9 However, quantitative magnetic properties were not
reported for that series. We report here an investigation of
the structural, spectroscopic, electrochemical, and electronic
properties of a matched set of related dinuclear CoIICoII,
CoIICoIII, and CoIIICoIII complexes in which the donor set to
each metal ion is identical and the metal ions are bridged
by two diphenyl phosphate bridging groups and a bridging
phenolate oxygen atom. With respect to the analogy to
dicobalt enzyme active sites, the phosphate bridging groups
resemble the bridging carboxylate groups in MetAP and a
proposed intermediate in the purple acid phosphatase mech-
anism.3 The series is ideal for the spectroscopic probing of
magnetic coupling since magnetic coupling can occur only
in the CoIICoII complex. The CoIICoIII complex contains a
nearly isostructural CoII ion, but since CoIII is diamagnetic,
the magnetic coupling is switched off, while the spectral
features of the CoII ion remain.
MCD has proved to be a generally useful probe of the
ground-state electronic properties of paramagnetic metal ions
in protein active sites,10 and the technique has been used
previously to examine some coupled diiron species.11 MCD
has distinct advantages over other techniques such as
temperature-dependent magnetic susceptibility and electron
paramagnetic resonance spectroscopy. In particular, MCD
is not a bulk property technique like magnetic susceptibility,
so it is not prone to errors introduced by paramagnetic
impurities that are likely in protein preparations. MCD is
particularly sensitive to transition metals having unpaired
electrons, so it is well-suited to study metalloenzymes that
are by nature magnetically dilute. In addition, MCD can be
used to examine a specific metal ion, even in mixed-metal
or mixed-valent systems. The present study serves as a first
demonstration of the application of MCD to measure weak
antiferromagnetic coupling in a dinuclear Co complex in
which the coupling can be turned “on” or “off”.
Results and Discussion
Synthesis. The CoIICoII complex [(bpbp)Co2(O2P(OPh)2)2]-
(ClO4)·H2O, 1-(ClO4)·H2O, is obtained from stoichiometric
reaction of the protonated ligand bpbpH, cobalt perchlorate,
and diphenylphosphonic acid in acetonitrile. We note that
under the same aerobic conditions with diphenylphosphonic
acid replaced by acetic acid, the product is a dicobalt(III)
acetate and peroxide-bridged complex, rather than the bis-
oxoacid bridged complex [(bpbp)Co2(CH3CO2)2]+.12 The
-
-
change of auxiliary bridge (RO2 ) from CH3CO2 to
-
(C6H5O)2PO2 clearly diminishes the propensity for the
[CoII (bpbp)(RO2)]2+ core to react with O2, consequently CoII
2
(5) (a) Brown, D. A.; Errington, W.; Glass, W. K.; Haase, W.; Kemp,
T. J.; Nimir, H.; Ostrovsky, S. M.; Werner, R. Inorg. Chem. 2001,
40, 5962–5971. (b) Brown, D. A.; Glass, W. K.; Fitzpatrick, N. J.;
Kemp, T. J.; Errington, W.; Clarkson, G. J.; Haase, W.; Karsten, F.;
Mahdy, A. H. Inorg. Chim. Acta 2004, 357, 1411–1436. (c) Brown,
D. A.; Glass, W. K.; Fitzpatrick, N. J.; Kemp, T. J.; Errington, W.;
Muller-Bunz, H.; Hussein, A. J.; Nimir, H. Inorg. Chim. Acta 2005,
358, 2755–2762.
complexes are relatively stabilized. The mixed-valence
CoIICoIII complex, [(bpbp)Co2(O2P(OPh)2)2](ClO4)2 · C3H6-
O · H2O, 2-(ClO4)2 · C3H6O · H2O, and the CoIIICoIII complex
[(bpbp)Co2(O2P(OPh)2)2](ClO4)3 ·3.5H2O, 3-(ClO4)3 ·3.5H2O,
were obtained, quite controllably, by reacting 1 with one
equivalent of CeIV and ten equivalents of CeIV, respec-
(6) (a) Ostrovsky, S. M.; Werner, R.; Brown, D. A.; Haase, W. Chem.
Phys. Lett. 2002, 353, 290–294. (b) Ostrovsky, S. M.; Falk, K.; Pelikan,
J.; Brown, D. A.; Tomkowicz, Z.; Haase, W. Inorg. Chem. 2006, 45,
688–694. (c) Turpeinen, U.; Ahlgren, M.; Hamalainen, R. Acta
Crystallogr. 1982, B38, 1580–1583.
1
tively. The 13C and H NMR of 3 give sharp signals, as
expected of a diamagnetic complex. Only five nonaromatic
(7) (a) Schultz, B. E.; Ye, B.-H.; Li, X.-Y.; Chan, S. I. Inorg. Chem. 1997,
36, 2617–2622. (b) Ye, B.-H.; Chen, X. M. Chin. J. Chem. 2003, 21,
531–536. (c) Ye, B.-H.; Williams, I. D.; Li, X.-Y. J. Inorg. Biochem.
2002, 92, 128–136.
(10) (a) Johnson, M. K. In Physical Methods in Bioinorganic Chemistry;
Que, L., Jr., Ed.; University Science Books: Sausilito, CA, 2000;
Chapter 5. (b) Pavel, E. G.; Solomon, E. I. In Spectroscopic Methods
in Bioinorganic Chemistry; Solomon, E. I., Hodgson, K. O., Eds.;
American Chemical Society: Washington, DC, 1998; Chapter 6. (c)
Wei, P.; Tomter, A. B.; Rohr, A. K.; Andersson, K. K.; Solomon,
E. I. Biochemistry 2006, 45, 14043–14051.
(8) (a) Fink, K.; Wang, C.; Staemmler, V. Inorg. Chem. 1999, 38, 3847–
3856. (b) Tsukerblat, B. S.; Palii, A. V.; Turta, K.; Jovmir, T.; Shova,
S.; Bartolome, J.; Evangelisti, M.; Filoti, G. J. Chem. Phys. 2001,
115, 9528–9535. (c) Hossain, M. J.; Yamasaki, M.; Mikuriya, M.;
Kuribayashi, A.; Sakiyama, H. Inorg. Chem. 2002, 41, 4058–4062.
(d) Sakiyama, H. Inorg. Chim. Acta 2006, 359, 2097–2100. (e) Aromi,
G.; Stoeckli-Evans, H.; Teat, S. J.; Cano, J.; Ribas, J. J. Mater. Chem.
2006, 16, 2635–2644.
(11) (a) Reem, R. C.; Solomon, E. I. J. Am. Chem. Soc. 1987, 109, 1216–
1226. (b) Pulver, S. C.; Tong, W. H.; Bollinger, J. M.; Stubbe, J.;
Solomon, E. I. J. Am. Chem. Soc. 1995, 117, 12664–12678. (c) Wei,
P.; Skulan, A. J.; Mitic, N.; Yang, Y.; Saleh, L.; Bollinger, J. M.;
Solomon, E. I. J. Am. Chem. Soc. 2004, 126, 3777–3788.
(9) Chaudhuri, P.; Querbach, J.; Wieghardt, K.; Nuber, B.; Weiss, J.
J. Chem. Soc., Dalton Trans. 1990, 271–278.
(12) Ghiladi, M.; Gomez, J. T.; Hazell, A.; Kofod, P.; Lumtscher, J.;
McKenzie, C. J. Dalton Trans. 2003, 1320–1325.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5081

Johansson et al.
bond lengths are as expected for CoII and CoIII with donor
O and N atoms,13 and the distinct difference in bond lengths
for 2 identifies it as a valence-trapped Robin-Day class II
complex.14
Comparison of the average deviation from 90° for the
twelve bite angles where the metal donor atoms are cis
coordinated (Table 3) shows that the low-spin CoIII sites are
closer to ideal octahedral geometry than the high-spin CoII
sites. The average deviation from 90° bond angles for the
two CoII sites in 1 are 5.19° and 5.06°, compared to 5.17/
5.15° for the CoII site in 2a/2b. The distortions for the CoIII
sites are 2.94/2.71 in 2a/2b, and 2.89/2.85° for the two sites
in 3, consistent with the preference for octahedral geometry
of the low-spin d6 configuration of CoIII. Of the three
complexes, the shortest metal-metal separation is 3.5906(17)
Åin3,despitethiscomplexexhibitingthelargestCo1-O1-Co2
angle, 132.2(3)°. Short metal-phenoxide distances (1.972(6)
and 1.956(6) Å) counterbalance this factor.
Solid-State NMR Spectroscopy. Paramagnetic shifts give
useful information about the local magnetic properties and
may be qualitatively predicted by the Goodenough-Kanamori
rules.15,16 It is relatively rare to have access to a series of
compounds in which chemically identical phosphate phos-
phorus atoms are located in three such different local
magnetic environments. Delineation of the phosphorus
chemical shifts of such a series would give data with
predictive capability for the assessment of electronic ground
states for other paramagnetic inorganic and organic phosphate
systems. Thus, we have measured the solid state 31P magic-
angle spinning (MAS) NMR spectra of the diamagnetic
CoIIICoIII (3) and the paramagnetic CoIICoIII (2) and CoIICoII
(1) complexes (Figure 2).
Figure 1. Cation of 1, showing the atom-labelling scheme (displacement
ellipsoids shown at 50% probability). For clarity, H atoms are omitted and
only the ipso carbon of the phenyl ring bound to O4 is shown. The cations
of 2 and 3 display closely comparable gross geometry.
carbon signals are seen, which indicates that 3 must be
symmetrical, rendering the C and H atoms in each half
of the molecule magnetically identical. The appearance
of proton signals in the nonaromatic region corresponding
to two protons with geminal coupling substantiates further
this assignment. The symmetry is not retained in the
crystal structure of 3.
X-ray Crystal Structures. The structure of the cation in
1 in the solid state is shown in Figure 1. The structures of 2
and 3 are geometrically similar. Two polymorphs of 2 have
been observed, both with stoichiometry 2-(ClO4)2 ·C3H6O·H2O.
The geometry of the CoIICoIII complex is comparable in the
two polymorphic structures. In each of the structures 1, 2,
and 3, the connectivities are equivalent and the gross
geometries of the complexes are closely comparable. The
heptadentate ligand provides four donor atoms to each metal
ion: one amine (N1 or N4), two pyridines (N2 and N3 or
N5 and N6), and one phenoxide (O1). The remaining two
coordination sites are occupied by an O atom from each of
the diphenylphosphate groups: O2 and O6 or O3 and O7.
The nitrogen donor set on each metal ion coordinates in a
facial geometry, and the dimetallic site is triply bridged by
the phenoxide O atom of the dinucleating ligand and the
two O,O′-bridging diphenylphosphate groups. Selected crys-
tallographic data and bond lengths and angles for the
complexes are given in Tables 1 and 2, respectively. The
coordination geometry of Co1 and Co2 for the homovalent
complexes 1 and 3 are comparable, while significant bond
length and angle variations are evident in the mixed-valence
complex 2. The coordination sphere contraction is apparent
for CoIII compared to CoII, with the average coordination
bond lengths being 1.93 and 2.11 Å for the two ions,
respectively. This conformational change can be ascribed to
both the change in charge on the metal ion and the spin
change between high-spin CoII and low-spin CoIII. In general,
The solution state 31P NMR spectra of the diamagnetic
complex 3 contains a single resonance with δiso(31P) ) 2.17
ppm, which is assigned to the two chemically equivalent
phosphorus atoms. In contrast the 31P MAS NMR spectra
of a polycrystalline sample of 3 shows two sites with δiso(31P)
) -0.7(1) and -15.5(1) ppm, respectively, the former signal
accounting for about 90% of the spectral intensity (Figure
2c). The two phosphorus atoms are chemically equivalent
but not crystallographically equivalent and two sites with a
1:1 intensity ratio might have been expected in 31P solid-
state NMR spectra. However, the signals are so different in
intensity this cannot be the explanation. Because only one
signal is observed by solution NMR spectroscopy, we suggest
the minor resonance with δiso(31P) ) -15.5(5) is due to the
presence of a trace of a second polymorph of 3. The two 31
P
sites in 3 have similar local coordination environments and
therefore observation of almost identical isotropic chemical
shifts resulting in overlapping resonances is not surprising.
In view of the structural characterization of two polymorphs
for 2, this does not seem unreasonable.
(13) Allen, F. H.; Orpen, A. G.; Taylor, R. J. Chem. Soc., Dalton Trans.
1989, 12, S1–S83.
(14) Robin, M. B.; Day, P. AdV. Inorg. Chem. Radiochem. 1967, 10, 247–
422.
(15) Goodenough, J. B. Magnetism and the Chemical Bonding; John Wiley
& Sons: New York, 1963.
(16) Grey, C. P.; Dupre, N. Chem. ReV. 2004, 104 (10), 4493–4512.
5082 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
Table 1. Crystallographic Data and Experimental Details
1-(B(Ph)4)
2-(ClO4)2 ·C3H6O·H2Oa
2-(ClO4)2 ·C3H6O·H2Ob
3-(ClO4)3 ·3.5H2O
empirical formula
Mr
C84H79BCo2N6O9P2
1507.14
P21/c
17.3667(12)
20.5374(15)
22.5306(15)
90
112.173(2)
90
7441.6(9)
4
C63H67Cl2Co2N6O19P2
1462.93
P1
C63H67Cl2Co2N6O19P2
1462.93
P1
C60H66Cl3Co2N6O24.5P2
1549.34
C2/c
45.883(7)
15.370(2)
19.544(2)
90
91.241(5)
90
13780(3)
8
j
j
space group
a [Å]
b [Å]
c [Å]
R [°]
ꢀ [°]
10.1939(4)
16.6030(7)
20.0035(8)
84.000(1)
77.223(2)
83.966(1)
3271.9(2)
2
10.341(2)
13.199(2)
24.018(3)
100.199(5)
90.062(5)
96.831(5)
3202.7(9)
2
γ [°]
V [Å3]
Z
T [K]
180(2)
1.345
0.552
44.0
30733
180(2)
1.485
0.715
50.0
46199
180(2)
1.517
0.730
50.0
39759
180(2)
1.494
0.727
44.9
52513
Fcalcd [g cm-3
]
µ(Mo KR) [mm-1
2θmax [°]
reflns measured
]
independent reflns
Rint
9058
0.086
10947
0.072
11097
0.113
8727
0.128
observed reflns [I > 2σ(I)]
parameters, restraints
R1 [I > 2σ(I)]
5075
937, 0
0.035
0.044
0.70
612974
6627
809, 0
0.059
0.168
1.04
6024
847, 0
0.062
0.155
0.94
5264
865, 40
0.085
0.263
1.04
wR2 [all data]
goodness of fit on F2
CCDC No.
a Polymorph 1 (2a). b Polymorph 2 (2b).
612975
612976
612977
Table 2. Selected Bond Lengths [Å] and Angles [°] for 1, 2, and 3
Table 3. Bite Angles [°] for Metal Donors with cis Geometry
2a 2b
1
2a
2b
3
1
3
Co1-O1
Co1-O2
Co1-O6
Co1-N1
Co1-N2
Co1-N3
2.058(2)
2.136(2)
2.076(2)
2.190(2)
2.131(2)
2.135(2)
2.154(3)
2.072(3)
2.068(3)
2.158(4)
2.110(4)
2.115(4)
2.158(3)
2.079(3)
2.031(3)
2.163(4)
2.110(4)
2.127(4)
1.972(6)
1.901(6)
1.922(6)
1.944(7)
1.915(7)
1.903(8)
O1-Co1-O2
O1-Co1-O6
O1-Co1-N1
O1-Co1-N3
O2-Co1-O6
O2-Co1-N1
O2-Co1-N2
O6-Co1-N2
O6-Co1-N3
N1-Co1-N2
N1-Co1-N3
N2-Co1-N3
89.25(7) 85.90(12) 86.84(12) 90.3(2)
98.12(7) 94.96(12) 91.77(12) 93.4(3)
89.36(8) 90.28(13) 90.50(12) 93.4(3)
91.54(8) 89.62(13) 91.13(13) 90.7(3)
90.43(7) 93.61(13) 92.73(12) 89.4(3)
97.39(8) 93.56(14) 93.78(13) 91.4(3)
85.11(8) 84.37(14) 84.09(14) 84.7(3)
96.58(9) 97.27(15) 100.23(14) 91.3(3)
93.04(9) 92.70(15) 94.09(14) 92.7(3)
76.88(10) 78.73(15) 78.56(15) 82.0(3)
79.00(10) 80.49(15) 79.46(15) 86.4(3)
93.19(9) 98.75(15) 96.47(15) 94.0(3)
Co2-O1
Co2-O3
Co2-O7
Co2-N4
Co2-N5
Co2-N6
2.054(2)
2.040(2)
2.116(2)
2.165(2)
2.133(3)
2.108(2)
1.924(3)
1.936(3)
1.935(3)
1.955(4)
1.926(4)
1.940(4)
1.939(3)
1.920(3)
1.916(3)
1.951(4)
1.905(4)
1.941(4)
1.956(6)
1.921(6)
1.923(6)
1.952(7)
1.899(8)
1.898(8)
Average deviation from 90°a 5.06
5.17
5.15
2.89
Co1· · ·Co2
3.6648(6)
126.1(1)
3.6889(9)
129.5(2)
3.7081(11)
129.6(2)
3.5906(17)
132.2(3)
O1-Co2-O3
O1-Co2-O7
O1-Co2-N4
O1-Co2-N5
O3-Co2-O7
O3-Co2-N5
O3-Co2-N6
O7-Co2-N4
O7-Co2-N6
N4-Co2-N6
N5-Co2-N4
N5-Co2-N6
98.02(8) 92.42(13) 92.44(13) 91.8(2)
88.08(7) 89.90(14) 90.61(13) 88.5(3)
91.52(9) 94.75(15) 94.14(15) 94.2(3)
90.59(9) 88.68(15) 87.77(14) 93.1(3)
95.43(8) 91.11(14) 91.31(13) 92.0(3)
94.17(10) 92.71(15) 91.91(15) 91.9(3)
94.30(10) 91.45(15) 91.07(15) 92.2(3)
90.33(8) 90.44(15) 90.48(14) 90.2(3)
83.04(8) 86.36(15) 87.10(14) 86.5(3)
77.07(10) 81.50(17) 82.43(16) 82.0(3)
80.20(10) 85.93(17) 86.50(16) 85.9(3)
96.25(9) 94.81(17) 94.32(16) 91.7(3)
Co1-O1-Co2
Four isotropic resonances with δ(31P) ) 2570(40), 2870(40),
2990(20), and 3170(40) ppm are identified for 2 by compar-
ing 31P MAS NMR spectra recorded with different spinning
speeds (Figure 2b). The observation of four sites confirms
the coexistence of 2a and 2b in comparable amounts in the
solid state. Each phosphorus atom has a single P-O-CoII
connectivity. In 2a these show bond angles of 128.9° and
133.7° and in 2b the corresponding angles are 132.5° and
131.4°. The size and sign of Fermi-contact shifts are very
sensitive to the P-O-CoII bond angles, as, for example,
observed for 6,7Li in lithium manganese oxides.16 Thus, the
about 600 ppm variation in 31P isotropic shifts for the four
sites is ascribed to these fairly small variations in P-O-CoII
bond angles making 31P NMR spectroscopy a strong tool
for detecting subtle structural changes.
In 1 each P atom has two P-O-CoII connectivities with
bond angles ranging from 127.5 to 137.8°. Thus, because
the two CoII are only weakly antiferromagnetically coupled,
isotropic shifts twice as large as for 2 are expected. A broad
resonance centered at ∼6000 (100) ppm (Figure 2a) can be
discriminated, in excellent agreement with the predicted value
Average deviation from 90°a 5.19
a Defined as the mean of the absolute differences between the angle and
90°.
2.94
2.71
2.85
for the isotropic shifts. Individual peaks for the one or more
isotropic resonances cannot be identified because of extremely
broad lines caused by the paramagnetism of the sample.
The 31P isotropic shifts (δiso(31P)) in the series of com-
pounds broaden and shift downfield by about 3000 ppm for
each increment in oxidation state. This corroborates the
changes in the local magnetic field. Furthermore, 31P solid-
state NMR confirms that the bulk samples of each of the
compounds are oxidation-state pure. For example, complex
1 contains no signal that would indicate contamination by 2
or 3.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5083

Johansson et al.
Figure 2. 31P MAS NMR spectra of 1, 2, and 3 recorded at 81.29 MHz
(υr ) 33 kHz), 81.06 MHz (υr ) 25 kHz), and 145.74 MHz (υr ) 6 kHz),
respectively. (a) The spectrum of the weakly antiferromagnetically coupled
1 shows an isotropic resonance for one or more P environments centered
around 6000 ppm, as indicated by the asterisk. (b) Paramagnetic 2: 1, 2, 3,
and 4 mark the position of the four isotropic resonances showing that the
two polymorphs, 2a and 2b, coexist in the solid phase in comparable
amounts. (c) The spectrum of diamagnetic 3. The isotropic resonance is
indicated by an asterisk. The isotropic resonance for a second site, tentatively
assigned to a second polymorph of 3, is marked by a triangle (see text).
Figure 3. Cyclic voltammograms of 1, 2, and 3, obtained with scan rate
0.20 V s-1. OCP used as start potential, switching potential -0.69 and
1.63 V. The dashed lines indicate the second scan.
Table 4. Peak Potentials (a ) anodic, c ) cathodic) for 1, 2, and 3
Measured by Cyclic Voltammetry (CV) and Differential Pulse
Voltammetry (DPV) in Acetonitrile at 25 °C
Electrochemistry. Cyclic voltammetry (CV) and dif-
ferential pulse voltammetry (DPV) of the three complexes
in acetonitrile under dry anaerobic conditions in the potential
range -0.69 to 1.63 V (vs Fc0/+) are shown in Figure 3, and
in the Supporting Information, Figures S1, S2 and S3.
Variations in the open circuit potential (OCP) for the three
complexes corroborate the assigned oxidation states: 1
[CoIICoII] has the lowest OCP of 0.00 V and 3 [CoIIICoIII]
has the highest OCP of 0.56 V, while the mixed-valence
complex 2 [CoIICoIII] has an OCP of 0.26 V. Peak potentials
are given in Table 4. The cyclic voltammograms are very
similar, each showing two anodic (Epa,1 and Epa,2) and two
cathodic (Epc,3 and Epc,4) peaks, widely separated. Because
the bpbp- and diphenylphosphate ligands are not redox active
in the applied potential window, the electrochemical pro-
cesses must arise from metal-based redox reactions. The
anodic peaks are assigned to the sequential one-electron
oxidations CoIICoIIfCoIICoIII and CoIICoIIIfCoIIICoIII, while
the cathodic peaks are counterpart one-electron reductions.
The redox waves are irreversible, suggesting that each
oxidation/reduction is followed by a rearrangement, which
is both electronic and structural. The CoIII ions are low spin
while the CoII ions are high spin. Thus, reduction of CoIII
CVa (DPVb) [V]
complex OCP
Epa,1
Epa,2
Epc,3
Epc,4
1
2
3
0.00 0.42 (0.50) 1.18 (1.24) 0.14 (0.05) -0.19 (-0.36)
0.26 0.36 (0.46) 1.13 (1.18) 0.17 (0.02) -0.15 (-0.32)
0.56 0.54 (0.32) 1.24 (1.05) 0.06 (0.13) -0.29 (-0.19)
a Peak potentials at scan rate 0.2 V s-1
.
b Peak potentials at scan rate
0.05 V s-1
may initially produce a low-spin CoII species, which presum-
ably undergoes a rapid spin change. These processes will
be accompanied by optimization of the Co-ligand bond
distances, in accordance with the differences observed in the
crystal structures of 1, 2, and 3. Because these geometrical
changes do not require any bonds to be broken, the rate of
conversion from destabilized to stabilized geometry might
well be fast, consistent with the fact that no reversible waves
are observed at high scan rates (2 V s-1). The reverse process
on the other hand must be slow on the electrochemical time
scale because reversible electrochemical behavior is not seen
at slow scan rates (0.02 V s-1). An overlay of the measured
voltammograms at scan rates from 0.02 to 0.50 V s-1 for 1,
2, and 3 is shown in the Supporting Information, Figure S1.
5084 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
Scheme 2. Assignment of the Four Electrochemical Peaks Detected by
Voltammetrya
Figure 4. UV-vis spectrum of 1 (solid line), 2 (dotted line), and 3 (dashed
line) in acetonitrile at 25 °C. Inset: expanded spectrum of 1 in the 400-600
nm region.
a Metal sites with a proposed or verified low-spin electronic configuration
are marked with an asterisk.
to oxidation or reduction of complexes that have not
undergone rearrangement are also seen, especially in the DPV
of 2 and 3. The appearance of these small peaks indicates
that the geometry rearrangement is not instantaneous.
Electronic and Magnetic Properties. UV-visible-NIR
data for 1, 2, and 3 in CH3CN are shown in Figure 4, and
the data are summarized in Table 5. For 1, a strong
absorption at 303 nm (4300 L mol-1 cm-1) is tentatively
assigned to a ligand-to-metal charge transfer (LMCT) transi-
tion, while the three weak overlapping absorptions at ∼500
The electrochemical behavior of 1, 2, and 3 is different
from that of related iron and manganese complexes, which
show reversible or quasi-reversible electrochemistry.17 We
believe this to be a consequence of the induced spin change
that occurs when the metal site is oxidized or reduced (i.e.,
high-spin CoII T low-spin CoIII) and the stereochemical
inertness of the generated low-spin CoIII ion. The first
oxidation wave (Epa,1) is the one-electron oxidation of 1,
generating a mixed-valence high-spin CoII-high-spin CoIII
complex which quickly rearranges into the high-spin CoII-
low-spin CoIII complex 2. The second oxidation wave then
corresponds to one-electron oxidation of 2 to generate a high-
spin CoIII-low-spin CoIII complex which then rearranges into
3. Similar to the oxidation waves, the reduction wave at Epc,3
can be assigned to the one-electron reduction of 3, which
rearranges to 2 and then is reduced at Epc,4 to regenerate 1.
These processes are summarized in Scheme 2. The assign-
ments are corroborated by two partial scans on 2 (Supporting
Information, Figure S2). Scans between -0.62 and 0.61 V
(solid line) show only oxidation of 1 (Epa,1) and reduction
of 2 (Epc,4), while scans between 0.00 and 1.52 V show
oxidation of 2 (Epa,2) and reduction of 3 (Epc,3).
The DPV of 1 and 2 (Supporting Information, Figure S3)
gives peak potentials that are almost identical to the potentials
measured by CV, while the potentials measured for 3 deviate.
This might be due to the high scan-rate dependence of the
peak potentials for 3 in the CV measurements (see Support-
ing Information). The shapes of the DPV diagrams are
interesting: the major peaks for 1 are from the two-oxidation
processes, while for 3 it is the reduction peaks that give the
largest current. The DPV diagram of 2 is almost intermediate
between 1 and 3. Small broad peaks which could correspond
4
nm (insert in Figure 4), are assigned to the T1gf4T1g(4P)
d-d transition in Oh symmetry, which is split because of
the lower symmetry.18 A broad, weak band at 1094 nm is
observed in both 1 and 2 (not shown in Figure 4) which is
assignable to the 4T1gf4T2g d-d transition in Oh symmetry
arising from the CoII ion. Complexes 2 and 3 both show
three strong absorption bands in the visible region. The
highest energy absorption at 390-400 nm is assigned to a
LMCT band from the phenolate to CoIII. The two remaining
1
1
peaks are assigned to the A1gf1T2g and A1gf1T1g d-d
transitions for octahedral low-spin CoIII, the latter having the
lowest energy. The transitions are separated by ∼5000-6000
cm-1 in both complexes, but the peaks for 3 are red-shifted
by ∼1000 cm-1 and weaker compared to 2.
MCD spectra of 1 in a poly(dimethylsiloxane) mull and 2
in ethanol solution are shown in Figure 5, and the data are
summarized in Table 5. (While acetonitrile is an excellent
solvent for 1, 2 and 3, it cannot be used for low temperature
MCD because it does not form an optical glass when frozen
as does ethanol. Unfortunately, ethanol is a poor solvent for
1, so the MCD spectra were collected on the mulled solid,
which produces higher noise than does an ethanol glass.)
All of the peaks are temperature-dependent C-terms, indicat-
ing that they arise from the high-spin CoII ground state. The
spectra show similar peaks in the d-d region, typical of
6-coordinate high-spin CoII species.19 It is clear from the
d-d region that 1 is an all high-spin CoII complex and 2 is
a mixed high-spin CoII and low-spin CoIII complex. The d-d
transitions from CoIII that are observed in the absorption
(17) (a) Borovik, A. S.; Que, L, Jr J. Am. Chem. Soc. 1988, 110, 1986–
1988. (b) Borovik, A. S.; Papaefthymiou, V.; Taylar, L. F.; Anderson,
O. P.; Que, L, Jr J. Am. Chem. Soc. 1989, 111, 6183–6195. (c)
Albedyhl, S.; Averbuch-Pouchot, M. T.; Belle, C.; Krebs, B.; Pierre,
J. L.; Saint-Aman, E.; Torelli, S. Eur. J. Inorg. Chem. 2001, 1457–
1464. (d) Ghiladi, M; Jensen, K. B.; Jiang, J.; McKenzie, C. J.; Mø
rup, S.; Søtofte, I.; Ulstrup, J. J. Chem. Soc., Dalton Trans. 1999,
2675–2681. (e) Diril, H.; Chang, H.-R.; Nilges, M. J.; Zhang, X.;
Potenza, J. A.; Schugar, H. J.; Isied, S. S.; Hendrickson, D. N. J. Am.
Chem. Soc. 1989, 111, 5102–5114.
(18) Das, R.; Nanda, K. K.; Paul, I.; Baitalik, S.; Nag, K. Polyhedron 1994,
13, 2639–2645.
(19) Kaden, T. A.; Holmquist, B.; Vallee, B. L. Inorg. Chem. 1974, 13,
2585–2590.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5085

Johansson et al.
Table 5. UV/Vis/NIR Absorption, MCD, and AOMX Calculations for the Complexes
complex
UV/vis/NIRa λmax [nm], ε [M-1cm-1
]
MCD λmax [nm]
calculatedb
assignment in Oh symmetry
1c
303, 4300
LMCT CoII
311
348
396
455
503
529
trace 2 impurity
phenolateO- fCoIIt2g
trace 2 impurity
4T1gf4T1g(P) CoII
4T1gf4T1g(P) CoII
4T1gf4T1g(P) CoII
4T1gf4T2g CoII
478, 5
510, 50
522, 45 (sh)
1094, 18
479
494
530
1110
2d
291, 7200 (sh)
390, 1900 (sh)
456, 2400
LMCT CoII + CoIII
LMCT CoII + CoIII
1A1gf1T2g CoIII
615, 500 (sh)
1A1gf1T1g CoIII
311
396
455
498
535
phenolateO- fCoII eg
phenolateO- fCoII t2g
4T1gf4T1g(P) CoII
4T1gf4T1g(P) CoII
4T1gf4T1g(P) CoII
4T1gf4T2g CoII
469
495
532
986
1094, 16
3
329, 4200 (sh)
397, 2950
480, 2200
LMCT CoIII
LMCT CoIII
1A1gf1T2g CoIII
634, 350 (sh)
1A1gf1T1g CoIII
a In acetonitrile at 25 °C b AOMX parameters (cm-1): c 1: εσ(N) ) 4500, εσ(O) ) 3170, εσ(phenolateO) ) 1790, B ) 770, ZFS ) 122. d 2: εσ(N) ) 4230,
εσ(O) ) 3980, εσ(phenolateO) ) 1790, B ) 681.
independent A-terms which are not observed. Differences
in the MCD spectra of 1 and 2 are seen in the 300-450 nm
region. These are tentatively attributed to charge transfer from
phenolate oxygen to CoII. Lever has presented a simplified
model for LMCT based on one-electron orbital energies in
which octahedral high-spin CoII will have vacancies in both
the t2g and eg levels.20 Thus, there are four possible LMCT
transitions from the phenolate O, namely πft2g, πfeg,
σft2g, and σfeg in order of increasing energy. The two
bands at 396 and 311 nm in 2 are assigned to the πft2g and
πfeg LMCT transitions, respectively, and should be sepa-
rated by 10Dq. These two bands are separated by ∼7000
cm-1 which is less than 10Dq for CoII with moderate field
ligands. This is an approximation because the actual sym-
metry is lower than Oh and the multiplicity that would be
introduced by seven electrons has been ignored. In 1, only
one LMCT is observed at 348 nm, which is presumably the
phenolate O πft2g LMCT transition. The πfeg would be
at shorter wavelength than 300 nm and could not be observed
since it is hidden beneath more intense electronic transitions
from the ligands. It is reasonable that the LMCT transitions
to CoII are at higher energies in 1 than in 2 because the
CoIICoII dinuclear site should be harder to reduce than the
CoIICoIII site. A fit of the d-d transitions arising from CoII
in 1 and 2 using the angular overlap model (Table 5)
demonstrated that the local environment around the metal
ions is substantially distorted from Oh symmetry. The
bridging phosphate oxygens exert a substantially higher
ligand field (εσ ) 3170 or 3980 cm-1) than the phenolate
oxygen (εσ ) 1790 cm-1). Thus the ligand field symmetry
immediately surrounding each CoII is no higher than C2V,
Figure 5. (a) MCD of 1 as a mull in poly(dimethylsiloxane), 10 K, 3.5 T.
(b) MCD of 2 in a saturated ethanol solution, 1.6 K, 3.5 T. Spectra were
fitted to a minimum number of Gaussian peaks using GRAMS/AI.
4
which gives a A ground state. The zero-field splitting
calculated by the AOM is 122 cm-1.
spectra are absent from the MCD spectra because a diamag-
netic ground-state can only have weak, temperature-
(20) Lever, A. B. P. Inorganic Electronic Spectroscopy; Elsevier: Amster-
dam, 1984; p 204.
5086 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
Figure 7. MCD intensity at 348 nm from 1 plotted versus 1/T at magnetic
fields ranging from 1.0 to 3.5 T. The lines through the data are there as
visual guides.
Table 6. Summary of Best Fit Parameters of the VTVH MCD of 1 and 2
complex 1 (eq 3)
E1 [cm-1
348 nm
503 nm
]
0
0
B1 [arbitrary]
0.16
7.4
15
-1.2
<0.1
4.3
0
0.09
7.4
73
-10
<0.1
4.3
0
E2 [cm-1
]
Isatlim2 [arbitrary]
B2 [arbitrary]
δ2 [cm-1
g|2
]
g⊥2
Mz/Mxy2
0
12.8
4.7
0
12.8
32
E3 [cm-1
]
B3 [arbitrary]
complex 2 (eq 1)
311 nm
396 nm
498 nm
Figure 6. (a) Magnetization plots of the 348 nm MCD band in 1. The
inset shows the magnetization plots for the 503 nm band. The lines through
the data represent the best fit to eq 3 with parameters given in Table 6. (b)
Magnetization plot of the 311 nm band in 2. The inset shows the
magnetization plots for the 396 and 498 nm bands. The lines through the
data represent the best fit to eq 1 with parameters given in Table 6. Only
the fits to the 1.6 K data are shown as all temperatures nearly overlay.
Isatlim [arbitrary]
B [arbitrary]
g|
g⊥
2.3
0.01
4.8
4.3
-0.06
0.8
-0.03
4.8
4.3
-0.1
0.3
0.01
4.8
4.3
-0.1
Mz/Mxy
intensity for an axial doublet at low temperatures and all
magnetic fields should be given by Equation 1.10
Figure 6 shows the magnetization plots for the 348 nm
MCD band in 1 and the 311 nm MCD band in 2. Similar
magnetization curves for the 503 nm band are shown in the
inset of 1, and curves for the 396 and 498 nm bands are
shown in the inset of 2. Despite having nearly identical
coordination environments, the shapes of the magnetizations
plots indicate that the CoII atoms in 1 and 2 have very
different ground states. The invariance with wavelength of
the shapes of the magnetization curves suggests that all of
the MCD transitions in both 1 and 2 are xy-polarized.10a The
temperature-dependent variations in the MCD signal at 348
and 503 nm for 1 indicate weak antiferromagnetic coupling
between the two CoII atoms, presumably mediated via the
bridging phenolate group. The antiferromagnetic behavior
in 1 is made clearer if the MCD intensity is plotted as a
function of 1/T at a fixed magnetic field, as shown in Figure
7. By contrast, the MCD signals collected at different
temperatures for the 311, 396, and 498 nm bands of 2 are
nearly superimposed, demonstrating temperature and mag-
netic field behavior qualitatively similar to a Kramer’s
doublet with an isotropic g-factor for an uncoupled CoII S
) 3/2 center. Accordingly the orientation-averaged MCD
2
π⁄2
cos θ sin θ
ΓꢀH
2kT
√
dθ - 2
∆ε ) Isatlim
g|| tanh
∫
(
)
0
[
Γ
3
Mz
π⁄2
sin θ
Γ
ΓꢀH
2kT
g⊥ tanh
dθ + BH (1)
∫
(
)
(
)
0
]
Mxy
Isatlim is the C-term intensity at saturation, Γ ) (g2|| cos2 θ
+ g2⊥ sin2 θ)1/2, ꢀ is the Bohr magneton, k is Boltzmann’s
constant, H is the magnetic field strength, Mz/Mxy is the ratio
of transition dipole moments of the z- and xy-polarized
transitions, and B is the linear B-term intensity. The
magnetization plots for 2 (Figure 6b) were fitted to eq 1.
The best fit parameters for the 311, 396, and 498 nm bands
are given in Table 6, and the solid lines in Figure 6b are the
calculated fits of the 1.6 K data using those parameters. The
C- and B-term intensity as well as the polarization should
be wavelength dependent since these are determined by the
properties of both the excited-state and the ground state. The
values of g| and g⊥ depend only on the ground-state and
should therefore be independent of wavelength. The small
values of Mz/Mxy indicate that all bands are xy-polarized,
which is expected for magnetization that is wavelength
independent. The fit to eq 1 is excellent at all wavelengths,
Inorganic Chemistry, Vol. 47, No. 12, 2008 5087

Johansson et al.
lending assurance that the model used for uncoupled high-
spin CoII is appropriate. The MCD magnetization data at
different temperatures nearly overlay, and g| and g⊥ are
nearly equal, indicating that 2 is practically isotropic, despite
the low symmetry.
and another singlet at ∼13 cm-1. All the remaining levels
are above 92 cm-1.
The low temperature VTVH MCD data can now be fit to
the MCD intensity expression for a series of singlets and
doublets for a non-Kramers system (integral spin) to
determine the ground-state parameters. This expression is
given in eq 3.10c
In complex 1, both metals are high-spin CoII and the
unpaired spins on each ion can couple through the phenolate
oxygen, with the phosphate pathways providing weaker
coupling. The magnitude of the exchange coupling can be
determined by analysis of the variable-temperature variable-
field (VTVH) MCD intensity (Figure 6a). The theory,
equations, and protocols for VTVH MCD data analysis and
interpretation in coupled dimers have been worked out and
reported over the past twenty years by Solomon and
co-workers.10c,11 The electronic structure of the ground-state
of the coupled CoII dimer can be described by the spin
Hamiltonian given in eq 2.10c This assumes that the ground-
state is an effective orbital singlet; an assumption that is
validated by the local symmetry around each metal (no higher
than C2V), the AOM calculations, and the excellent fit of the
magnetic susceptibility data using this spin Hamiltonian (vida
infra).
2
π⁄2
cos θ sinθ
∆ε )
(Isatlim i
)
g||iꢀHRi dθ -
∫
∑
(
0
[
Γi
i
3
Mz
π⁄2
sin θ
Γi
√
2
g⊥iꢀHRi dθ + BiHγi (3)
∫
)
0
]
Mxy
where
2
2
2
Γ ) δ + g ꢀH cos θ + g ꢀH sin θ
(
)
(
⊥i
)
√
i
i
||i
i
i
i
i
e
-(E -Γ ⁄2)⁄kT - e-(E +Γ ⁄2)⁄kT
Ri )
j
j
j
j
e
-(E -Γ ⁄2)⁄kT - e-(E +Γ ⁄2)⁄kT
∑
j
e
-(E -δ ⁄2)⁄kT - e-(E +δ ⁄2)⁄kT
i
i
i
i
γi )
-(E -δ ⁄2)⁄kT - e-(E +δ ⁄2)⁄kT
j
j
j
j
e
∑
j
S(S + 1)
2
H ) -2JS1·S2 + D1 Sz1
-
+ R1(Sx12 - Sy12) + D2
Equation 3 allows for a linear B-term from field-induced
mixing between states. (Isatlim)i, Bi, δi, g||i, and g⊥i are the
C-term and B-term MCD intensity, the ZFS of the doublet
levels, and the dimer g values of the ith doublet, respectively.
Ei is the energy of the ith sublevel, and the energy of the
ground-state is defined as zero. Mz/Mxy is the ratio of
transition dipole moments of the z- and xy-polarized transi-
tions. Because the shape of the magnetization curves for 1
are independent of wavelength, the transitions are assumed
to be xy-polarized; therefore, initial fits to eq 3 set Mz/Mxy
and g⊥ equal to zero. Furthermore, δ was initially fixed at
zero. The VTVH MCD data for 1 was fitted to eq 3 using
three energy levels. The best fit at both wavelengths resulted
in a ground-state singlet, a doublet at 7.4 cm-1 with g| )
4.3, and a second singlet at 12.8 cm-1. These energy spacings
are consistent with a coupling constant of J ) -1.6 cm-1.
No significant improvements in the fits were obtained when
Mz/Mxy and g⊥ were allowed to float. A small improvement
to the fits was observed when δ was allowed to float, but
the value for δ was less than 0.1 cm-1. Finally, no significant
improvements of the fits of the MCD or magnetic suscep-
tibility data (vida infra) were obtained if a nonzero rhombic
ZFS parameter (R in eq 2) was included.
(
)
3
S(S + 1)
2
Sz2
-
+ R2(Sx22 - Sy22) +
(
)
3
gx1ꢀHx·Sx1+ gy1ꢀHy·Sy1+ gz1ꢀHz·Sz1 + gx2ꢀHx·Sx2 +
gy2ꢀHy·Sy2 + gz2ꢀHz·Sz2 (2)
This Hamiltonian takes into account exchange coupling
(J) between the cobalt ions, single-ion axial (D1,2) and
rhombic (R1,2) zero-field splitting parameters and the Zeeman
terms (gx1ꢀHx ·Sx1, etc.) for each cobalt ion. The Hamiltonian
operates on the uncoupled basis set of wave functions (|S1,
Ms1, S2, Ms2| where Ms ) 3/2, 1/2, -1/2, -3/2) that yields a
16 × 16 matrix. Diagonalization of this matrix in zero
magnetic field yields the energies of the dinuclear ground
state. The total zero-field splitting, ZFS, is defined here as
ZFS ) |2D[1 + 3(R/D)2]1/2|. Equation 2 can be simplified
for 1 since the metal ions are symmetrically equivalent (S1
) S2, D1 ) D2, R1 ) R2, etc.). VTVH MCD data can be
directly fit to eq 2, as was recently demonstrated for the
coupled ferrous dimer in ribonucleotide reductase by Wei
et al.10c However, for the analysis of 1 it was more useful
to use eq 2 to simulate the 16 energy levels that are required
by the fit of the VTVH MCD data to the MCD intensity
expression for a series of singlets and doublets given by eq
3 (vida infra). Furthermore, experimentally derived param-
eters for eq 2 are available from the fit of the magnetic
susceptibility data (vida infra). The energy and corresponding
Ms for each level using J ) -1.6 and D ) 44 cm-1 are
listed in the Supporting Information, Table S1. It is seen
that the manifold of 16 energy levels consists solely of
singlets and doublets. This will always be the case as long
as the ZFS is nonzero. Furthermore, the only three levels
that will be populated at the low temperatures of the MCD
experiment are a ground-state singlet, a doublet at ∼7 cm-1,
Coupling constants for weak antiferromagnetically coupled
CoII dimers have exclusively been determined by magnetic
susceptibility (Table 7). Thus confirmation of the VTVH
MCD results was in order and accomplished by measurement
of magnetic susceptibility on a powder sample of 1 dispersed
in Vaseline to prevent a crystallite orientation that can yield
4
anomalous susceptibilities in anisotropic T1g (parent state)
systems. The maximum at 10 K observed in the 1/T plots of
the MCD intensity (Figure 7) was confirmed. The maximum
in ꢁM, per CoII , occurs at 9.4 K as shown in Figure 8 and
2
inset. The corresponding ꢁMT values decrease gradually from
5088 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
Table 7. Summary of Magnetic Coupling in Related Complexes
bridging core
Co · · ·Co [Å]
Co-O-Co [°]
J [cm-1
]
reference
CSD refcode
1, Co2(µ-OPhR)(µ-1,3-O2P(OPh)2)2
Co2(µ-OPhR′)(µ-1,3-O2CCH3)a
3.665
3.505
3.356
3.546
3.687
3.092
3.631
3.597
3.432
3.494
3.2
126
126
112
113
117
97
127
115
111
108
98
-1.6
-2.7
this work
31
8c
32
7a, c
24
HIMNEW
MOYYUU
ZALXAL
HUQQOZ
GONYOX
FIPTEE
BERWUQ
UDANIW
UDANES
b
Co2(µ-OPhR′′)(µ-1,3-O2CCH3)2
-0.22e,f
-0.2f
-1.6f
-6.9f
-1
Co2(µ-OH2)(µ-1,3-O2CCH3)2
Co2(µ-OH2)(µ-1,3-O2CCH3)2
Co2(µ-OPhR)2
Co2(µ-OG)(µ-1,3-O2CCH3)c
Co2(µ-OH2)(µ-1,3-O2CCH3)2
5b
-0.7
-3.6
18.0e
∼0
5a, 6a, 6c
5a, 6a
5a, 6b
2b, 20
d
Co2(µ-AA)(µ-1,3-O2CCH3)2
Co2(µ-1,3-O2CCH3)2(µ-1,1-O2CCH3)
MetAP, Co2(µ-OH(H))(µ-1,3-O2CR)2
a OPhR′ ) 2,6-cresolate-N,N,N′,N′-tetramethylene carboxylate. b OPhR′′ ) 2,6-bis[bis(2-hydroxyethyl)aminomethyl]-4-methylphenolate (bhmp-). c OG
d
) N-oxyglutarimide, C5H6NO3. AA ) acetohydroxamate, CH3(CdO)N(H)-O-. e No zero-field-splitting was included in the HDVV model. f This model
includes an orbital contribution.
J ) -0.9 cm-1 and D ) -53 cm-1. However, this results
in a doublet at 8.1 cm-1 having Ms ) ( 3, which is
inconsistent with the effective g ∼ 4 required to fit the MCD
data. The VTVH MCD data are very sensitive to g because
it governs the steepness of the magnetization curves. So even
though the susceptibility data are fitted better with a negative
D, a negative D does not produce the correct spin for the
lowest doublet. The best fits of the MCD and magnetic
susceptibility data are obtained using J ) -1.6 cm-1 and D
) 44 cm-1. This coupling constant is comparable in sign
and magnitude to those reported for related complexes (Table
7). Similar weak antiferromagnetic coupling constants have
been determined by temperature-dependent magnetic sus-
ceptibility measurements for a number of dinuclear bpbp-
complexes with other metals.17d The J’s for all but two of
Figure 8. Plot of molar susceptibility, per Co2, vs temperature (K) for
the complexes listed in Table 7 were obtained by fitting
magnetic susceptibility data to the HDVV model. The good
agreement between MCD and magnetic susceptibility in
complex 1 strongly suggests that the spin Hamiltonian in eq
2 (the HDVV model) is also appropriate. The only complexes
in Table 7 that required a model that included an orbital
contribution were [Co2(µ-OPhR′′)(µ-1,3-O2CCH3)2](BPh4),
with J very small and negative,8c and Co2(µ-1,3-
O2CCH3)2(µ-1,1-O2CCH3), and that complex has a relatively
large and positive coupling constant of J ) 18 cm-1.5a,6b
The HDVV model assumes that there is no orbital
contribution to the exchange interaction because the low-
symmetry ligand field removes all orbital degeneracy from
the ground state. The model further assumes isotropic
exchange between the CoII ions, although single-ion ZFS
splitting factors (D and R) and doublet ZFS (δ) were
included. This demonstrates that VTVH MCD can be used
to measure coupling in weak antiferromagnetic dimers and
that the shape of the magnetization curve is very sensitive
even to weak coupling. Ostrovsky et al. have demonstrated
previously that MCD magnetization is also sensitive to
ferromagnetic coupling.6b In the ferromagnetic case, the
MCD magnetization curves are qualitatively similar to those
observed for 2, but the curves taken at different temperatures
are nested rather than overlaid. The MCD of CoII in MetAP
has been reported and used to argue that the metals are not
antiferromagnetically coupled because the magnetization data
from the 6-coordinate CoII can be fitted to a simple doublet
complex 1. The expanded region of the maximum in ꢁM is shown in the
inset. The solid line is that calculated using the simplified eq 2 and the
parameter values g ) 2.4, J ) -1.6 cm-1, and D ) 44 cm-1
.
5.44 cm3 mol-1 K (µeff ) 6.59 µB), per Co2, to 4.2 cm3 mol-1
K at ∼50 K, then more rapidly to reach 0.22 cm3 mol-1
according to Bossek et al. to fit susceptibility data.21
K
at 2 K. The spin Hamiltonian in eq 2 was simplified
H ) -2JS ·S +
D S 2 - S S + 1 ⁄ 3 + µ g S ·H
( )
i iz i i B i i
{
[
]
}
∑
1
2
i)1,2
(simplified eq 2)
A very good fit over the whole 300-2 K temperature range
using the simplified Hamiltonian resulted for the parameter
set g ) 2.4, J ) -1.6 cm-1, D ) 44 cm-1. The shape of the
present ꢁMT plot is similar to those for two triply bridged
CoII(tacn) dinuclear species described by Bossek et al.,21 with
J ) -13 cm-1, D ) 40 cm-1, but contrasts with the plot
found by Ostrovsky et al.6b for a triply acetate bridged
derivative that showed a maximum in ꢁMT indicative of
ferromagnetic coupling (J ) 18 cm-1). A more complete
ligand field/orbital angular momentum model was used in
the latter case.6b Convergence on coupling parameters that
gave good fits to both the magnetic susceptibility data and
the MCD data for 1 was an iterative process. The low
temperature susceptibility data could also be well fitted using
(21) Bossek, U.; Nuhlen, D.; Bill, E.; Glaser, T.; Krebs, C.; Weyhermuller,
T.; Wieghardt, K.; Lengen, M.; Trautwein, A. X. Inorg. Chem. 1997,
36, 2834–2843.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5089

Johansson et al.
ground-state model, as in 2, with an isotropic g ) 4.45.22
The excellent fits of the MCD and susceptibility data to the
HDVV model suggest that orbital contributions are negligible
in 1.8c,d This is not always the case for weakly antiferro-
magnetically coupled dicobalt(II) as has been shown by
several groups.8
ions. Together with a recent study of a ferromagnetic
complex,6b this study demonstrates that MCD can be used
to probe the magnetic properties of dinuclear CoII sites in
hydrolase enzymes, providing a valuable alternative for
systems which may not lend themselves to magnetic
susceptibility studies.
There are too few data on dinuclear CoII complexes to
make general statements about the trends in the magnitude
of coupling with any certainty, but some speculation is now
warranted. First, it is generally assumed that ligands that
bridge with a single atom dominate the mediation of
magnetic coupling. Second, the nature of the bridging ligand
will be important since it affects the strength of the bond
between the metal and bridging ligand. Third, in complexes
with bridging water, hydroxide, oxide, or phenolate oxygen
atoms, in which the principal interaction will be through
p-orbitals, the CoII-O-CoII angle is very important because
the value of J is the sum of ferromagnetic and antiferro-
magnetic contributions, which will change with this
angle.23,24 The latter statement may serve to explain why 1
has J ) -1.6 cm-1 with a CoII · · ·CoII separation of 3.6648(6)
Å and CoII-O-CoII angle of 126.1(1)°, while no coupling
was detected in MetAP22 which has a CoII · · ·CoII separation
of only 3.2 Å, but the bridging hydroxide or water has a
CoII-O-CoII angle of 98°.2b
Experimental Section
Elemental analyses were performed at the Chemistry Department
II at Copenhagen University, Denmark and Atlantic Microlab, Inc.,
Norcross, Georgia 30091, U.S.A. UV/vis spectra were recorded on
a Shimadzu UV-3100 spectrophotometer. Electrospray Ionization
Mass Spectra (ESI MS) were obtained using a Finnigan TSQ 700
triple quadrupole instrument equipped with a Finnigan API source
in the nanoelectrospray mode. 13C{1H}- and 1H NMR were
measured on a Varian Gemini 300 spectrometer at 298 K. Chemical
shifts were referenced to the residual signals of the deuterated
solvent CD3CN and are reported versus SiMe4. 2,6-Bis[(bis-(2-
pyridylmethyl)amino)methyl]-4-tertbutylphenol (bpbpH) was pre-
pared as described previously.26
Synthetic procedure. Caution! Although no problems were
encountered in the preparation of the perchlorate salts, suitable
care should be taken when handling such potentially hazardous
compounds.
[(bpbp)Co2(O2P(OPh)2)2](ClO4)·H2O, 1-(ClO4)·H2O. Hbpbp
(0.8786 g, 1.53 mmol) and Co(ClO4)2 ·6H2O (1.1225 g, 3.08 mmol)
were dissolved in 15 mL of hot CH3CN to give a maroon solution,
to which HPO2(OPh)2 (0.7692 g, 3.08 mmol) dissolved in 6 mL of
CH3CN was added. Pink needles formed as the solution cooled.
The solution was left in the refrigerator overnight and the product
was then isolated and washed with a small amount of cold CH3CN.
Yield first crop, 1.1050 g. The mother liquor was concentrated to
∼10 mL and 1/2 mL of 2.5 M NaOH was added to give a second
crop. Yield second crop: 0.5631 g. The combined products were
recrystallized by slow evaporation of an acetonitrile solution (∼100
mL) containing 0.60 g of NaClO4 ·H2O. Yield 1.3900 g (69%).
Elemental analysis calcd (%) for 1-(ClO4) · H2O, C60H61ClCo2-
N6O14P2: C 55.20, H 4.71, N 6.44; found: C 54.85, H 4.29, N,
6.12. ESI MS (CH3CN): m/z 1187.9 ([(bpbp)Co2(O2P(OPh)2)2]+,
100%). Block-shaped pink crystals of 1-(B(Ph)4) suitable for single
crystal X-ray analysis were obtained by exchange of the perchlorate
counteranion for the tetraphenylborate counteranion, by slow
diffusion of toluene into a mixture of a CH2Cl2 solution of 1-(ClO4)
with 1.5 equiv of NaBPh4 dissolved in acetone.
Conclusion
This study represents the first analysis of a weakly
antiferromagnetically coupled CoIICoII system using MCD.
The J value so obtained compared well with that obtained
from magnetic susceptibility data. A key feature of the study
is the synthetic accessibility of the series of complexes. The
bis-diphenylphosphate bridged Co2II system is air stable (in
contrast to its carboxylate-bridged counterparts12,25) and
permits clean generation of the related CoIICoIII and Co2
III
systems by addition of one equivalent or an excess of a one-
electron oxidant. All three complexes can be isolated in the
solid state and they retain their oxidation states in solution.
This has enabled a structurally analogous CoIICoIII system
II
to be used to calibrate the Co2 result, since coupling is
removed in the CoIICoIII system, while the spectroscopic
properties of the CoII ion remain. A rudimentary correlation
of the system described here with the few other dicobalt
systems for which data are available suggests that the
antiferromagnetic coupling, albeit weak, may be dependent
more on the Co-O-Co bridging angle than on the spatial
proximity of the two CoII ions. This being the case, the lack
of observed antiferromagnetic coupling in MetAP can be
explained by its relatively small Co-O-Co bridging angle,
despite the relatively short distance between the two CoII
[(bpbp)Co2(O2P(OPh)2)2](ClO4)2 · C3H6O · H2O, 2-(ClO4)2 ·
C3H6O·H2O. (NH4)2[Ce(NO3)6] (0.2206 g, 0.40 mmol) was added
to a solution of 1-(ClO4) (0.4728 g, 0.37 mmol) dissolved in 10
mL of DMF to give a dark brown solution. NaClO4 ·H2O (0.2700
g, 1.90 mmol) dissolved in 2 mL of DMF was added and the
product was precipitated as a dark brown powder by addition of
40 mL of H2O. The powder was recrystallized from acetone/water,
providing brown crystals of 2a and 2b suitable for single crystal
X-ray analysis. Yield 0.3795 g (70%). Elemental analysis calcd
(%) for 2-(ClO4)2 ·C3H6O·H2O, C63H67Cl2Co2N6O19P2: C 51.72, H
4.62, N 5.74; found: C 51.82, H 3.89, N 5.93. ESI MS (CH3CN):
m/z 593.9 ([(bpbp)Co2(O2P(OPh)2)2]2+, 100%)
(22) Larrabee, J. A.; Leung, C-H.; Moore, R. L.; Thamrong-nawasawat,
T.; Wessler, B. S. H. J. Am. Chem. Soc. 2004, 126, 12316–12324.
(23) Hotzelmann, R.; Wieghardt, K.; Florke, U.; Haupt, H-J.; Weatherburn,
D. C.; Bonvoisin, J.; Blondin, G.; Girerd, J.-J. J. Am. Chem. Soc. 1992,
114, 1681–1696.
[(bpbp)Co2(O2P(OPh)2)2](ClO4)3 ·3.5 H2O, 3-(ClO4)3 ·3.5H2O.
(NH4)2[Ce(NO3)6] (2.2206 g, 4.02 mmol) was added to 1-(ClO4)
(0.4990 g, 0.39 mmol) dissolved in 10 mL of DMF to give a very
(24) Black, D.; Blake, A. J.; Dancey, K. P.; Harrison, A.; McPartlin, M.;
Parsons, S.; Tasker, P. A.; Whittaker, G.; Schroder, M. J. Chem. Soc.,
Dalton Trans. 1998, 3953–3960.
(25) Wackerbarth, H.; Larsen, F. B.; Glaargaard Hansen, A.; McKenzie,
C. J.; Ulstrup, J. Dalton Trans. 2006, 3438–3444.
(26) Ghiladi, M.; McKenzie, C. J.; Meier, A.; Powell, A. K.; Ulstrup, J.;
Wocaldo, S. J. Chem. Soc., Dalton Trans. 1997, 4011–4018.
5090 Inorganic Chemistry, Vol. 47, No. 12, 2008

Dicobalt II-II, II-III, and III-III Complexes
dark brown solution. NaClO4 ·H2O (0.4670 g, 3.34 mmol) dissolved
in 3 mL of DMF was added and the product was precipitated by
addition of 70 mL of H2O. The powder was recrystallized from
acetone/water to yield red needles suitable for single crystal X-ray
analysis. Yield 0.3211 g (53%). Elemental analysis calcd (%) for
3-(ClO4)3 ·3.5H2O, C60H61Cl3Co2N6O22P2: C 47.90, H 4.09, N 5.59;
found: C 47.81, H 3.78, N 5.32. ESI MS (CH3CN): m/z 643.4
([(bpbp)Co2(O2P(OPh)2)2(ClO4)]2+, 100%). NMR δC{1H} (75 MHz;
CD3CN) 31.19 (3 C, s, C(CH3)3), 35.33 (1 C, s, C(CH3)3), 40.15
(2 C, s), 68.31 (2 C, s), 68.56 (2 C, s), 120.25 (4 C, d, J 5.5,
P-O-C-C), 120.40 (4 C, d, J 5.8, P-O-C-C), 122.21 (2 C, s),
125.76 (2 C, s), 126.51 (2 C, s), 127.12 (4 C, s), 127.17 (4 C, s),
127.50 (2 C, s), 127.66 (2 C, s), 130.82 (2 C, s), 131.10 (2 C, s),
131.83 (2 C, s), 142.97 (2 C, s), 143.78 (2 C, s), 149.26 (1 C, s),
150.75 (2 C, s), 151.23 (2 C, d, J 7.5, P-O-C), 151.71 (2 C, d, J
7.4, P-O-C), 153.52 (2 C, s), 154.59 (1 C, s), 163.59 (2 C, s),
165.44 (2 C, s). δH (300 MHz; CD3CN) 1.19 (9 H, s, C(CH3)3),
3.18 (2 H, d, J 13), 3.95 (2 H, d, J 16), 4.03 (2 H, s), 4.06 (2 H,
s), 4.10 (2 H, d, J 13), 4.89 (2 H, d, J 16), 6.72 (2 H, d, J 7.3),
6.79-6.83 (4 H, m), 7.05-7.07 (6 H, m), 7.14 (2 H, s), 7.35-7.44
(8 H, m), 7.48-7.58 (6 H, m), 7.62-7.68 (4 H, m), 8.12 (2 H, dd,
J 7.8 and 1.2), 8.16 (2 H, d, J 5.4), 8.55 (2 H, d, J 5.5).
X-ray Crystallography. Analyses of 1-(B(Ph)4), 2-(ClO4)2 ·
C3H6O·H2O (two polymorphs), and 3-(ClO4)3 ·3.5 H2O were
performed at 180(2) K using a Bruker Nonius X8-APEXII
instrument. Crystallographic data and refinement details are given
in Table 1. All structures were solved by direct methods and refined
against all F2 data using SHELXTL ver. 6.10.27 All H atoms bound
to C atoms were placed in calculated positions and refined using a
riding model. Unless otherwise stated below, all non-H atoms were
refined with anisotropic displacement parameters. Crystals of
1-(B(Ph)4) were weakly diffracting, and data were truncated to 0.95
Å resolution, with about 50% of data observed at the 2σ(I) level to
that resolution. For one polymorph of 2-(ClO4)2 ·C3H6O·H2O (2a),
one phenyl ring belonging to one diphenyl phosphate group was
modeled as disordered over two orientations (site occupancy factors
0.75:0.25), related by rotation about the P-O bond. The geometries
of these two rings were constrained to be regular hexagons, and
the atoms were refined with isotropic displacement parameters. The
other phenyl ring of the same diphenyl phosphate group has
significantly distorted anisotropic displacement parameters, sug-
gesting some degree of disorder that corresponds to dynamic/static
disorder in the plane of the phenyl ring, which was not modeled
explicitly. For 3-(ClO4)3 · 3.5H2O, one perchlorate anion and all
solvent molecules were poorly resolved and some voids remain in
the structure, indicating the likely presence of more diffuse solvent.
Application of a continuous solvent-area model28 with the disor-
dered perchlorate and all water molecules omitted gave significant
improvements to the R values (R1 ) 0.068, wR2) 0.181) and
suggested up to 8 molecules of water per formula unit. The
empirical formula is therefore an approximation. Elemental analysis
of 3 gave the best agreement by including only one water solvent
molecule in the formula, suggesting that the lattice solvent
evaporates during vacuum drying of the product.
ringing and 25-35 kHz spinning. Spectra were recorded with carrier
frequencies ranging from 80.86 to 81.4 MHz to cover the entire
chemical shift range for the two compounds. For 3, which is
diamagnetic, a field of 8.1 T (145.7 MHz) and 5-8 kHz spinning
was employed. All spectra are referenced to neat 85% H3PO4 (δiso
) 0 ppm).
Electrochemistry. Cyclic Voltammetry (CV) and Differential
Pulse Voltammetry (DPV) were recorded in acetonitrile solution
under dry anaerobic conditions using an Autolab system (Eco
Chemie, The Netherlands), controlled by the GPES software. The
working electrode was a platinum disk, auxiliary electrode a
platinum wire, and reference electrode Ag/Ag+ (0.01 M AgNO3).
0.1 M TBAClO4 (TBA ) tertbutylammonium) was used as
electrolyte, and all potentials are given versus the ferrocene/
ferrocenium (Fc0/+) redox couple (E1/2 ) 88 mV vs Ag/Ag+, ∆E
) 75-80 mV). Concentrations of the sample solutions were 1.3
mM for 1, 2.3 mM for 2, and 2.2 mM for 3.
Magnetic Susceptibilities. Variable temperature magnetic sus-
ceptibilities on complex 1 were obtained using a Quantum Design
MPMS5 Squid magnetometer in a DC field of 1 T. To eliminate
any crystallite “torquing” at low temperatures, because of anisotropy
in ꢁM for high-spin Co(II) (4T1g) species, the ∼20 mg sample was
dispersed in Vaseline and the ꢁM value at 300 K was set at the
value using an accurately weighed powder sample.
Spectroscopy. Absorption spectra were recorded on a Cary 6000i
UV/visible/near-infrared absorption spectrometer. MCD spectra
were recorded at 2 nm bandwidth on a JASCO J-600 spectropo-
larimeter equipped with an Oxford SM-4 magnetic/cryostat with
an Oxford ITC-4 temperature controller. Temperatures were
measured with a Lake Shore Cryogenics carbon/glass resistor
calibrated between 1.45 and 99.99 K and located 2 mm from the
sample. Complex 2 was dissolved in ethanol, which forms an
optically clear glass when frozen, and placed in a 0.62 cm path-
length brass cell with quartz windows. Complex 1 was run as a
mull in poly(dimethysiloxane) because it was not sufficiently soluble
in ethanol. The zero-field spectrum was subtracted as a blank.
Details of the MCD data collection have been previously pub-
lished.29 MCD spectra were fitted to the minimum number of
Gaussian peaks using the GRAMS/AI software (Thermo Electron
Corporation).
Angular overlap model (AOM) calculations were made using
AOMX, a program based on routines developed by Hoggard for
d3 transition metal ions and extended to dn systems by Adamsky.30
AOMX determines the optimum ligand-field parameters needed to
fit an observed set of d-d transitions based on a given structure.
The two cobalt ions were treated separately. Input to AOMX
requires the ligand coordinates (polar r, θ, ꢂ or Cartesian x, y, z)
referred to the metal ion at the origin. The cobalt ligand set was
assumed to be 3 equivalent nitrogens, 2 equivalent oxygens from
the bridging phosphate groups, and the unique phenylate bridging
oxygen. AOMX does not use the bond distances to fit the spectrum,
(29) Larrabee, J. A.; Alessi, C. M.; Asiedu, E. T.; Cook, J. O.; Hoerning,
K. R.; Klingler, L. J.; Okin, G. S.; Santee, S. S.; Volkert, T. L. J. Am.
Chem. Soc. 1997, 119, 4182–4196.
(30) Hoggard, P. E. In Topics in Current Chemistry; Springer-Verlag:
Berlin, 1994; Vol. 171, pp 113-141. The AOMX program is
maintained by H. Adamsky, Institut fur Theoretische Chemie, Heinrich
HeineUniversitat,Duesseldorf,Germany;E-mail:adamsky@theochem.uni-
duesseldorfde. The program can also be run on the web using the
AOMX server at http://www.theochem.uni-duesseldorf.de/users/heri-
bert/aomx/start.main.html.
(31) Cai, L.; Xie, W.; Mahmoud, H.; Han, Y.; Wink, D. J.; Li, S.; O’Connor,
C. J. Inorg. Chim. Acta 1997, 263, 231–245.
(32) Coucouvanis, D.; Reynolds, R. A.; Dunham, W. R. J. Am. Chem. Soc.
1995, 117, 7570–7571.
Solid-State NMR Spectroscopy. 31P MAS NMR spectra of
polycrystalline samples were recorded at 4.7 T (80.86-81.3 MHz)
for 1 and 2 using a 2 mm MAS probe. These were recorded using
a Hahn echo sequence to alleviate spectrometer dead time and probe
(27) Sheldrick, G. M. SHELXTL; Bruker AXS: Madison, WI, 2000.
(28) (a) van der Sluis, P.; Spek, A. L. Acta Crystallogr. 1990, A46, 194–
201. (b) Spek A. L. SQUEEZE, implemented in PLATON, A
Multipurpose Crystallographic Tool; Utrecht University: Utrecht, The
Netherlands, 1998.
Inorganic Chemistry, Vol. 47, No. 12, 2008 5091

Johansson et al.
only the bond angles are important, and these are fixed by the crystal
structure. However, the bond distances are reflected by the
magnitude of the resulting ligand field parameters, εσ and επ.30
Stony Brook University, for generous access to NMR
instruments. K.S.M. acknowledges support from the Aus-
tralian Research Council for the magnetic work and the
Villum Kann Rasmussen (Velux) Foundation, Denmark, and
the Department of Physics and Chemistry, University of
Southern Denmark, for a Fellowship (May-June 2007) that
enabled this work to be finalized and discussed.
Acknowledgment. C.J.M. is grateful to the Danish Natural
Science Research Council. J.A.L. acknowledges the National
Science Foundation of the U.S.A. (Grant CHE-0554083) for
financial support, Professor Edward I. Solomon, Stanford
University, for providing the MCD fitting programs, and
Professor Natasa Mitic, University of Queensland for as-
sistance in setting up the MCD fitting programs. U.G.N.
acknowledges the Danish Science Foundation for a Steno
Fellowship (Grant 272-06-0058) and Professor Clare P. Grey,
Supporting Information Available: Figures S1-S3, cyclic
voltammetry; Figure S4, energy level diagrams; Table S1, energy
levels (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
IC7020534
5092 Inorganic Chemistry, Vol. 47, No. 12, 2008