Neutral bis(α-iminopyridine)metal complexes of the first-row transition ions (Cr, Mn, Fe, Co, Ni, Zn) and their monocationic analogues: Mixed valency involving a redox noninnocent ligand system

Article abstract of DOI:10.1021/ja710663n

A series of bis(α-iminopyridine)metal complexes featuring the first-row transition ions (Cr, Mn, Fe, Co, Ni, and Zn) is presented. It is shown that these ligands are redox noninnocent and their paramagnetic π radical monoanionic forms can exist in coordination complexes. Based on spectroscopic and structural characterizations, the neutral complexes are best described as possessing a divalent metal center and two monoanionic π radicals of the α-iminopyridine. The neutral M(L^?)₂ compounds undergo ligand-centered, one-electron oxidations generating a second series, [(L^x)₂M(THF)][B(Ar_F)₄] [where L ^x represents either the neutral α-iminopyridine (L)⁰ and/or its reduced π radical anion (L^?)^-]. The cationic series comprise mostly mixed-valent complexes, wherein the two ligands have formally different redox states, (L)⁰ and (L^?) ^-, and the two ligands may be electronically linked by the bridging metal atom. Experimentally, the cationic Fe and Co complexes exhibit Robin-Day Class III behavior (fully delocalized), whereas the cationic Zn, Cr, and Mn complexes belong to Class I (localized) as shown by X-ray crystallography and UV-vis spectroscopy. The derealization versus localization of the ligand radical is determined only by the nature of the metal linker. The cationic nickel complex is exceptional in this series in that it does not exhibit any ligand mixed valency. Instead, its electronic structure is consistent with two neutral ligands (L)⁰ and a monovalent metal center or [(L) ₂Ni(THF)][B(Ar_F)₄]. Finally, an unusual spin equilibrium for Fe(II), between high spin and intermediate spin (S_Fe = 2 ? S_Fe = 1), is described for the complex [(L ^?)(L)Fe(THF)][B(Ar_F)₄], which consequently is characterized by the overall spin equilibrium (S_tot = 3/2 ? S_tot = 1/2). The two different spin states for Fe(II) have been characterized using variable temperature X-ray crystallography, EPR spectroscopy, zero-field and applied-field Moessbauer spectroscopy, and magnetic susceptibility measurements. Complementary DFT studies of all the complexes have been performed, and the calculations support the proposed electronic structures.

Full text of DOI:10.1021/ja710663n

Published on Web 02/20/2008
Neutral Bis(r-iminopyridine)metal Complexes of the First-Row
Transition Ions (Cr, Mn, Fe, Co, Ni, Zn) and Their
Monocationic Analogues: Mixed Valency Involving a Redox
Noninnocent Ligand System
Connie C. Lu, Eckhard Bill, Thomas Weyhermu¨ller, Eberhard Bothe, and
Karl Wieghardt*
Max-Planck-Institut fu¨r Bioanorganische Chemie, Stiftstrasse 34-36,
D-45470 Mu¨lheim an der Ruhr, Germany
Received December 4, 2007; E-mail: wieghardt@mpi-muelheim.mpg.de
Abstract: A series of bis(R-iminopyridine)metal complexes featuring the first-row transition ions (Cr, Mn,
Fe, Co, Ni, and Zn) is presented. It is shown that these ligands are redox noninnocent and their paramagnetic
π radical monoanionic forms can exist in coordination complexes. Based on spectroscopic and structural
characterizations, the neutral complexes are best described as possessing a divalent metal center and
two monoanionic π radicals of the R-iminopyridine. The neutral M(L•)2 compounds undergo ligand-centered,
one-electron oxidations generating a second series, [(Lx)2M(THF)][B(ArF)4] [where Lx represents either the
neutral R-iminopyridine (L)0 and/or its reduced π radical anion (L•)-]. The cationic series comprise mostly
mixed-valent complexes, wherein the two ligands have formally different redox states, (L)0 and (L•)-, and
the two ligands may be electronically linked by the bridging metal atom. Experimentally, the cationic Fe
and Co complexes exhibit Robin-Day Class III behavior (fully delocalized), whereas the cationic Zn, Cr,
and Mn complexes belong to Class I (localized) as shown by X-ray crystallography and UV-vis
spectroscopy. The delocalization versus localization of the ligand radical is determined only by the nature
of the metal linker. The cationic nickel complex is exceptional in this series in that it does not exhibit any
ligand mixed valency. Instead, its electronic structure is consistent with two neutral ligands (L)0 and a
monovalent metal center or [(L)2Ni(THF)][B(ArF)4]. Finally, an unusual spin equilibrium for Fe(II), between
high spin and intermediate spin (SFe ) 2 T SFe ) 1), is described for the complex [(L•)(L)Fe(THF)][B(ArF)4],
which consequently is characterized by the overall spin equilibrium (Stot ) 3/2 T Stot ) 1/2). The two different
spin states for Fe(II) have been characterized using variable temperature X-ray crystallography, EPR
spectroscopy, zero-field and applied-field Mo¨ssbauer spectroscopy, and magnetic susceptibility measure-
ments. Complementary DFT studies of all the complexes have been performed, and the calculations support
the proposed electronic structures.
1. Introduction
dianions, but their electron-rich π-systems can be oxidized in
one-electron steps to produce first the benzosemiquinone radical
and then the benzoquinone. Recently, we have turned our atten-
tion to ligands with electron-accepting π-systems, such as R-di-
imines,15,16 R-iminoketones,17 and R-diketones.18 Since these
latter ligands resemble the benzoquinone structure, it is not sur-
prising that they can be reduced by one electron to generate
the monoanionic π-radical equivalent of the benzosemi-
quinone.15,19-21
Open-shell ligand radicals are found throughout the coordina-
tion chemistry of transition metal ions. The classical benzosemi-
quinone redox state is well-known for catechols,1,2 o-phenylene-
diamines,3-6 o-benzene-1,2-dithiolates,7-10 and their mixed
donor analogues.11-14 These ligands are typically closed-shell
(1) Hendrickson, D. N.; Pierpont, C. G. In Spin CrossoVer in Transition Metal
Compounds II; Springer: Berlin, London, 2004; Vol. 234, p 63.
(2) Pierpont, C. G.; Lange, C. W. In Progress in Inorganic Chemistry; Karlin,
K. D., Ed.; John Wiley & Sons, Inc.: 1994; Vol. 41, p 331.
(3) Herebian, D.; Wieghardt, K. E.; Neese, F. J. Am. Chem. Soc. 2003, 125,
10997.
(11) Bill, E.; Bothe, E.; Chaudhuri, P.; Chłopek, K.; Herebian, D.; Kokatam,
S.; Ray, K.; Weyhermu¨ller, T.; Neese, F.; Wieghardt, K. Chem.sEur. J.
2004, 11, 204.
(12) Chaudhuri, P.; Verani, C. N.; Bill, E.; Bothe, E.; Weyhermu¨ller, T.;
Wieghardt, K. J. Am. Chem. Soc. 2001, 123, 2213.
(13) Kokatam, S.; Weyhermu¨ller, T.; Bothe, E.; Chaudhuri, P.; Wieghardt, K.
Inorg. Chem. 2005, 44, 3709.
(14) Ghosh, P.; Bill, E.; Weyhermu¨ller, T.; Wieghardt, K. J. Am. Chem. Soc.
2003, 125, 3967.
(15) Muresan, N.; Chlopek, K.; Weyhermu¨ller, T.; Neese, F.; Wieghardt, K.
Inorg. Chem. 2007, 46, 5327.
(16) Muresan, N.; Weyhermu¨ller, T.; Wieghardt, K. Dalton Trans. 2007, 4390.
(17) Lu, C. C.; Bill, E.; Weyhermu¨ller, T.; Bothe, E.; Wieghardt, K. Inorg. Chem.
2007, 46, 7880.
(4) Herebian, D.; Bothe, E.; Neese, F.; Weyhermu¨ller, T.; Wieghardt, K. J.
Am. Chem. Soc. 2003, 125, 9116.
(5) Chłopek, K.; Bothe, E.; Neese, F.; Weyhermu¨ller, T.; Wieghardt, K. Inorg.
Chem. 2006, 45, 6298.
(6) Chłopek, K.; Bill, E.; Weyhermu¨ller, T.; Wieghardt, K. Inorg. Chem. 2005,
44, 7087.
(7) Kapre, R. R.; Bothe, E.; Weyhermu¨ller, T.; George, S. D.; Wieghardt, K.
Inorg. Chem. 2007, 46, 5642.
(8) Kapre, R.; Ray, K.; Sylvestre, I.; Weyhermu¨ller, T.; George, S. D.; Neese,
F.; Wieghardt, K. Inorg. Chem. 2006, 45, 3499.
(9) Ray, K.; George, S. D.; Solomon, E. I.; Wieghardt, K.; Neese, F. Chem.s
Eur. J. 2007, 13, 2783.
(10) Ray, K.; Petrenko, T.; Wieghardt, K.; Neese, F. Dalton Trans. 2007, 1552.
9
10.1021/ja710663n CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 3181-3197
3181

A R T I C L E S
Lu et al.
delocalization.26-29 We now show that this is the case for our
bis(ligand)metal systems, wherein the extent of ligand-ligand
electronic communication is dependent on the nature of the
“bridging” metal atom. The spectroscopic and physical char-
acterizations of these complexes reveal distinct trends that may
correlate with the position of the metal ion in the periodic table.
Scheme 1. Different Redox States of the R-Iminopyridine Ligand
with Their Characteristic Bond Distances (Å)
2. Results and Discussion
2.1. Synthesis and Characterization of 1-6. For the present
study, we use the neutral R-iminopyridine ligand, 2,6-bis(1-
methylethyl)-N-(2-pyridinylmethylene)phenylamine, henceforth
represented by L. The preparation of the neutral bis(ligand)-
metal series is depicted in Scheme 2. With the exception of
nickel, the M(L•)2 complexes (1-4, 6; where M ) Cr, 1; Mn,
2; Fe, 3; Co, 4; Zn, 6) were prepared by mixing the metal
dichloride precursors with 2 equiv of ligand and 2 equiv of
sodium metal in dimethoxyether (DME). To obtain the bis-
(ligand)nickel complex, Ni(L•)2 5, 2 equiv of ligand were added
to Ni(COD)2 in pentane (where COD ) 1,5-cyclooctadiene).
The synthetic methodologies have been adapted from those
reported by tom Dieck,30-34 Walter,35,36 and Holm.37
We and others22,23 have previously reported that the terdentate
bis(R-diimine)pyridine ligand is redox noninnocent. With its
extensive π-network, bis(R-diimine)pyridines have been shown
to exist beyond the neutral closed-shell form, as monoanionic
and dianionic π-radical(s) within coordination complexes. The
present endeavor focuses on the related R-iminopyridines. In
principle, R-iminopyridines can have three different redox states,
which are shown in Scheme 1 along with their characteristic
bond distances. Among the three, the neutral R-iminopyridine
(L)0 is the most commonly found in the Cambridge Structural
Database. The open-shell monoanion (L•)- and the doubly
Compounds 1-6 dissolve readily in benzene, THF, and Et2O
1
reduced dianion (L )
red 2- are rare. To our knowledge, the former
but are only sparingly soluble in pentane. Based on H NMR
spectroscopy, only the nickel complex, 5, is diamagnetic. The
1H NMR spectrum of 5 contains broad, fluxional peaks at room
temperature. Upon cooling to -60 °C, two distinct sets of ligand
resonances are clearly resolved in nearly equal proportions. (The
has not been structurally characterized, and the latter has been
observed in only two magnesium complexes.24
Using an R-iminopyridine, we have prepared here bis(ligand)-
metal complexes of the first-row transition metal ions, Cr, Mn,
Fe, Co, Ni, and Zn. Collectively, these neutral complexes com-
prise two ligand radicals and a divalent metal center. Moreover,
the unpaired spins on the ligands and at the metal center are
strongly antiferromagnetically coupled. This general electronic
structure is supported by physical measurements including X-ray
crystallography, magnetic susceptibility measurements, cyclic
voltammetry, UV-vis, NMR, EPR, and Mo¨ssbauer spectros-
copy, and density functional theoretical (DFT) calculations.
We were particularly interested in studying the one-electron
oxidized analogues because such systems would display mixed
valency that arises from two ligands in different formal redox
states if the oxidation process is ligand-centered and not metal-
centered. Indeed, we show here that the first electron oxidation
of the neutral series is ligand-centered and proceeds cleanly to
generate their cationic counterparts. Interestingly, the cationic
complexes exhibit vastly different electronic structures, spanning
both the Robin-Day Class I (localized) and Class III (fully
delocalized) mixed-valent systems.25 Previous studies on metal-
centered mixed-valent systems have shown that the nature and
electronic structure of the organic linker between two mixed-
valent metal ions are critical in deciding the degree of
1
exact ratio determined from the integrations of the H NMR
spectrum is 1.3 to 1.0. See SI Figure 1.) The two sets correspond
to two equilibrating geometric isomers, which can arise in four-
coordinate bis(ligand)metal systems when the ligands are
asymmetric and the angle between the two metal-ligand planes
is not 90°.15,16,38 For Ni 5, the dihedral angle is 70.9° as shown
below by X-ray crystallography, and thus, two geometric
isomers are possible and can interconvert by twisting of the
two metal-ligand planes relative to one another.
The cyclic voltammograms for 1-6 in THF at 22 °C are
shown in Figure 1. In general, the complexes have similar
electron-transfer properties. In all cases, an irreversible reduction
event occurs at -2.2 V which is ligand centered, (L•)- + e f
(L )
red 2- (shown only for Zn). Also, each complex exhibits two
fully or quasi-reversible peaks (Table 1). These features are
assigned to the two discrete one-electron oxidations of the ligand
radicals,
E2°+′ /+
E2°+′ /+
-e
[MII(L•)2]0 y -+ee z [MII(L•)(L)]1+ y z [MII(L)2]2+
+e
(18) Spikes, G.; Bill, E.; Weyhermu¨ller, T.; Wieghardt, K. Angew. Chem., Int.
Ed. 2008, in press.
(26) Creutz, C. In Progress in Inorganic Chemistry; Lippard, S. J., Ed.; Wiley
& Sons, Inc.: 1983; Vol. 30, p 1.
(19) Rijnberg, E.; Richter, B.; Thiele, K. H.; Boersma, J.; Veldman, N.; Spek,
A. L.; van Koten, G. Inorg. Chem. 1998, 37, 56.
(27) Cotton, F. A.; Liu, C. Y.; Murillo, C. A.; Zhao, Q. Inorg. Chem. 2007, 46,
2604.
(20) Khusniyarov, M. M.; Harms, K.; Burghaus, O.; Sundermeyer, J. Eur. J.
Inorg. Chem. 2006, 2985.
(28) de Biani, F. F.; Corsini, M.; Zanello, P.; Yao, H. J.; Bluhm, M. E.; Grimes,
R. N. J. Am. Chem. Soc. 2004, 126, 11360.
(21) Gardiner, M. G.; Hanson, G. R.; Henderson, M. J.; Lee, F. C.; Raston, C.
L. Inorg. Chem. 1994, 33, 2456. See also: Lange, C. W.; Conklin, B. J.;
Pierpont, C. G. Inorg. Chem. 1994, 33, 1276.
(29) Gamelin, D. R.; Bominaar, E. L.; Kirk, M. L.; Wieghardt, K.; Solomon,
E. I. J. Am. Chem. Soc. 1996, 118, 8085.
(30) Svoboda, M.; tom Dieck, H.; Kruger, C.; Tsay, Y. H. Z. Naturforsch., B:
Chem. Sci. 1981, 36, 814.
(22) de Bruin, B.; Bill, E.; Bothe, E.; Weyhermu¨ller, T.; Wieghardt, K. Inorg.
Chem. 2000, 39, 2936.
(31) tom Dieck, H.; Bruder, H. J. Chem. Soc., Chem. Commun. 1977, 24.
(32) tom Dieck, H.; Svoboda, M.; Kopf, J. Z. Naturforsch., B: Chem. Sci. 1978,
33, 1381.
(23) Bart, S. C.; Chłopek, K.; Bill, E.; Bouwkamp, M. W.; Lobkovsky, E.; Neese,
F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2006, 128, 13901.
(24) Westerhausen, M.; Bollwein, T.; Makropoulos, N.; Schneiderbauer, S.;
Suter, M.; No¨th, H.; Mayer, P.; Piotrowski, H.; Polborn, K.; Pfitzner, A.
Eur. J. Inorg. Chem. 2002, 389.
(33) tom Dieck, H.; Svoboda, M.; Greiser, T. Z. Naturforsch., B: Chem. Sci.
1981, 36, 823.
(34) tom Dieck, H.; Dietrich, J. Angew. Chem., Int. Ed. Engl. 1985, 24, 781.
(35) Kirmse, R.; Stach, J.; Walther, D.; Bottcher, R. Z. Chem. 1980, 20, 224.
(25) Robin, M. B.; Day, P. AdV. Inorg. Chem. Radiochem. 1967, 10, 247.
9
3182 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
Scheme 2
The separation between the potentials is the smallest for Zn 6
(∆E° ) 0.17 V). This difference steadily increases as one
traverses down the period from Cr 1 and Mn 2 (0.27 and 0.26
V, respectively) to Ni 5 (0.63 V). The increasing ∆E° may
reflect a growing electronic interaction between the two ligand
sites from Cr to Co.39 Another interpretation is that the trend in
∆E° suggests an increasing metal character in these predomi-
nantly ligand-centered oxidations. Indeed, we will show later
that the second oxidation event for Ni 5 is actually metal-
centered (NiII/NiI, E°′2+/+ ) -0.57 V, vs Fc+/Fc). These two
explanations are not contrary since the metal center links the
two ligand sites and, thus, is directly responsible for any
electronic interaction between them.
event: Co(II) + e f Co(I). An analogous feature is present,
albeit completely irreversible, for the earlier transition metals
Cr 1, Mn 2, and Fe 3, and this event is completely absent in
the cyclic voltammograms of the later metals Ni 5 and Zn 6.
The electronic absorption spectra of 1-6 recorded in a
benzene solution at 22 °C are shown in Figure 2. All these
complexes are intensely colored with a high molar absorptivity
between 300 and 600 nm (∼104 M-1 cm-1). The zinc species
6 is lime-green in color with several weak bands (ꢀ ) 1500
M-1 cm-1) from 600 to 900 nm. Because the Zn(II) center has
a filled d10 configuration, these bands are attributed to the ligand
π radicals and are assigned as intraligand (π f π*) charge-
transfer bands.40 The Cr 1,41 Mn 2, Fe 3, and Co 4 complexes
have similar transitions to those observed for 6 from 600 to
900 nm and are various shades of dark brown. Only the Ni 5
complex, which is dark purple, stands apart in that its absorption
bands are shifted much further into the near-infrared region (up
to 1350 nm). Preliminary TD-DFT studies of Ni 5 suggest that
this NIR transition is predominantly LMCT in character, which
is notably absent in the rest of the series.42
As a side note, the Co compound 4 is unique in this series
due to an additional quasi-reversible event at -1.96 V versus
Fc+/Fc. This wave is assigned to a metal-centered reduction
2.2. Solid-State Structures of 1-6. Single-crystal X-ray
diffraction studies of 1-6 were performed at 100(2) K (Table
2). The neutral complexes 2-6, which crystallized from a
benzene/pentane mixture, are distorted tetrahedrons in the solid
state. As an example, the molecular structure of the Zn
compound 6 is shown in Figure 3. The geometry around zinc
is approximately tetrahedral since the bite angle of the ligand
is only 85°. Further distortion from tetrahedral geometry is
evident in the dihedral angle between the two Zn-ligand planes
of 83.5°. The Mn, Fe, Co, and Ni complexes have similar
geometries with slightly varying dihedral angles: Mn 2, 83.5°;
Fe 3, 80.6°; Co 4, 79.7°; Ni 5, 70.9°. Their solid-state structures
are located in the Supporting Information (SI Figures 2-5).
Unlike the rest of the neutral series, the Cr compound 1 does
not readily crystallize from benzene. However, block crystals
were grown from Et2O at -30 °C that were suitable for X-ray
diffraction analysis. The structure reveals a five-coordinate
solvento complex, 1‚Et2O, which is best described as a distorted
trigonal bipyramid (Figure 4). The angle between the two Cr-
(36) Walther, D.; Kreisel, G.; Kirmse, R. Z. Anorg. Allg. Chem. 1982, 487,
149.
(37) Balch, A. L.; Holm, R. H. J. Am. Chem. Soc. 1966, 88, 5201.
(38) Blanchard, S.; Neese, F.; Bothe, E.; Bill, E.; Weyhermu¨ller, T.; Wieghardt,
K. Inorg. Chem. 2005, 44, 3636.
(39) Richardson, D. E.; Taube, H. Inorg. Chem. 1981, 20, 1278.
(40) Time-dependent DFT calculations on the neutral and cationic zinc
complexes 6 and 6ox support this assignment. V. Bachler and K. Wieghardt,
unpublished results.
Figure 1. Cyclic voltammograms of 1-6 (0.1 M [N(n-Bu)4](PF6) in THF,
22 °C, glassy carbon working electrode, 100 mV/s scan rate (except for 1,
50 mV/s), internal ferrocene (Fc) standard).
(41) Compound 1 has an additional transition at 495 nm when dissolved in the
coordinating solvent, Et2O.
(42) Bachler V. and Wieghardt, K. Unpublished results.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3183

A R T I C L E S
Lu et al.
Table 1. Potentials (V, vs Fc+/Fc) for 1-6
M
E
°′+/
E
°′2
∆E°
0
+/+
1
2
3
4
5
6
Cr
Mn
Fe
Co
Ni
Zn
-1.17
-1.40
-1.35
-1.26
-1.20
-1.33
-0.90
-1.14a
-1.00a
-0.76
-0.57
-1.16
0.27
0.26
0.35
0.50
0.63
0.17
a Quasi-reversible.
ligand planes is dramatically reduced to 41.4°. The preference
for penta-coordination may be caused by the instability of d4
ions in a tetrahedral environment. Jahn-Teller distortion from
ideal tetrahedral geometry could create an open site for solvent
binding.
An inspection of the bond lengths in the ligand backbones
reveals that all the ligands in 1-6 are essentially identical. This
can be readily seen in Figure 5, wherein the C-C and C-Nimine
bond distances are plotted in a histogram with the estimated
standard deviations ((3 esd).43 For 1-6, the averages of the
C-C and C-Nimine bond lengths are 1.409 ( 0.005 and 1.340
( 0.005 Å, respectively. These values are consistent with the
monoanionic π-radical state of the ligand. The C-Npyridine bond
distances were also evaluated, but this bond parameter is
relatively insensitive to the oxidation level of the ligand,
probably due to delocalization of electron density throughout
the pyridine ring. Thus, the C-Npyridine bond length will not be
considered further.
2.3. Magnetic Susceptibility Measurements of the Neutral
M(L•)2 Series with EPR and Mo1ssbauer Spectroscopic
Characterization. Excluding the S ) 0 nickel complex 5, the
magnetic susceptibilities of the neutral series were measured
on powder samples from 4 to 290 K using an applied field of
0.01 T. The temperature-dependent plots of the effective
magnetic moments, µeff, for complexes 1, 2, 3, 4, and 6 are
shown collectively in Figure 6. For all these compounds, the
magnetic moments are independent of temperature from 60 to
300 K with the following values: 1, 2.59; 2, 3.84; 3, 3.62; 4,
2.28; and 6, 2.52 µB.
For the Cr and Mn complexes, 1 and 2, the µeff values of
2.59 and 3.84 µB are near the spin-only values for an S ) 1
(2.83 µB) and an S ) 3/2 (3.87 µB) system, respectively. These
magnetic ground states are the result of antiferromagnetic
coupling between the two ligand radicals (combined ligand spin
vector SL ) 1) and a high-spin divalent metal (SCr ) 2; SMn
)
5/2). The temperature independent µeff suggests that this in-
tramolecular coupling is quite strong. Thus, assuming energeti-
cally well-isolated magnetic ground states with total spin Stot
3
) 1 and Stot ) /2 for 1 and 2, the plots can be well simulated
by adopting isotropic gtot values of 1.87 and 1.98, respectively.44
To isolate the metal contribution in gtot, the total spin vector is
divided into two spin components: Sˆmetal and SˆL, where SˆL is
the combined spin vector of the two ligand radicals with SL )
1.45 Projecting these spin components results in the equations
shown in Table 3, from which gCr ) 1.91 and gMn ) 1.99 are
Figure 2. Electronic absorption spectra for 1-6 in benzene (shown in red)
and 1ox‚THF-6ox‚THF in THF (shown in green) at 22 °C.
calculated. While the gmetal value is agreeable for Mn, the value
for Cr is a bit low.46 Nonetheless, g values as low as 1.90 have
been reported for high-spin Cr(II).47-49 Figure 7 shows the
(45) Since the ligands radicals are strongly coupled antiferromagnetically to the
metal, they remain parallel to one another. Consequently, the two ligand
radicals can be treated as one total ligand spin vector with SL ) 1.
(46) Casey, A. T.; Mitra, S. In Theory and Applications of Molecular
Paramagnetism; Boudreaux, E. A., Mulay, L. N., Eds.; Wiley: New York,
1976; p 135.
(47) Gill, N. S.; Nyholm, R. S.; Barclay, G. A.; Christie, T. I.; Pauling, P. J. J.
Inorg. Nucl. Chem. 1961, 18, 88.
(48) Earnshaw, A.; Larkworthy, L. F.; Patel, K. S. J. Chem. Soc. A 1969, 1339.
(43) For complexes 2, 3, 4, and 6, two independent molecules were present in
the asymmetric unit cell. Thus, the histogram contains a total of four
individual bond lengths for each bond type in 2, 3, 4, and 6.
(44) For 1, the drop of µeff below 50 K may be caused by a weak intermolecular
antiferromagnetic coupling, which was effectively modeled by introducing
a Weiss constant, θ, of -3.5 K. No other parameters were necessary to
create the fit shown. For 2, the small rise in µeff below 50 K was not
simulated but may reflect a weak intermolecular ferromagnetic coupling.
9
3184 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
Table 2. Crystallographic Data for 1-6
1‚
Et2O
2
3
4
5
6
formula
crystal size, mm3
fw
space group
a, Å
C40H54CrN4O
C36H44MnN4
C36H44FeN4
C36H44CoN4
C36H44NiN4
C36H44ZnN4
0.06 × 0.15 × 0.16 0.14 × 0.16 × 0.18 0.02 × 0.14 × 0.06 0.12 × 0.14 × 0.15 0.14 × 0.30 × 0.48 0.07 × 0.10 × 0.15
658.87
587.69
Pca21, no. 29
18.6955(4)
17.8857(4)
19.2902(4)
90
90
90
6450.3(2)
8
588.60
591.68
591.46
598.12
Pca21, no. 29
18.5855(4)
17.9390(4)
19.2006(4)
90
90
90
6401.6(2)
8
P21/n, no. 14
11.4200(2)
24.7034(5)
13.1336(2)
90
97.899(3)
90
3670.0(1)
4
P21, no. 4
9.4383(3)
18.6250(6)
18.2581(6)
90
100.527(3)
90
3155.5(2)
4
P21, no. 4
9.3884(3)
18.4602(6)
18.3906(6)
90
100.704(3)
90
3131.85(18)
4
P21/n, no. 14
10.0840(3)
18.8318(5)
17.5447(4)
90
102.104(3)
90
3257.67(15)
4
b, Å
c, Å
R, deg
â, deg
γ, deg
V, Å3
Z
T, K
100(2)
1.192
100 824/72.6
100(2)
1.210
141 838/65.0
100(2)
1.239
22 737/65.0
100(2)
1.255
70 888/66.4
100(2)
1.206
99 719/69.0
100(2)
1.241
153 279/65.0
F calcd, g cm-3
refl. collected/
2σmax
unique refl./
I > 2σ(I)
no. params./
restraints
17 717/14 090
425/0
23 276/20 104
755/1
22 737/65 093
763/11
23 746/20 802
753/1
13 601/11 296
378/0
23 091/18 287
755/1
λ, Å /µ(KR), cm-1 0.710 73/3.5
0.710 73/4.4
0.0362/1.012
0.0789
0.710 73/5.1
0.0457/1.030
0.0976
0.710 73/5.8
0.0349/1.013
0.0820
0.710 73/6.2
0.0362/1.028
0.0973
0.710 73/8.0
0.0397/1.034
0.0812
R1a/ GOFb
0.0383/1.038
0.0999
+0.53/-0.49
wR2c (I > 2σ(I))
residual density,
e Å -3
+0.26/-0.29
+0.94/-0.35
+0.78/-0.54
+0.86/-0.37
+0.43/-0.43
2
2
2
2
2
a Observation criterion: I > 2σ(I). R1 ) Σ||Fo| - |Fc||/Σ|Fo|. b GOF ) [Σ[w(Fo - Fc )2]/(n - p)]1/2
.
c wR2 ) [Σ[w(Fo - Fc )2]/Σ[w(Fo )2]]1/2 where
w ) 1/σ2(Fo ) + (aP)2 + bP, P ) (Fo + 2Fc )/3.
2
2
2
Figure 4. Thermal ellipsoid representation (50%) of the neutral Cr complex
4 with a bound Et2O. Hydrogen atoms have been omitted. Selected bond
distances (Å) and angles (deg) for 1‚(Et2O): Cr-O 1.362(1), Cr-N1
2.0485(7), Cr-N8 2.0214(7), Cr-N21 2.0428(7), Cr-N28 2.0151(7), N1-
Cr-N8 78.96(3), N21-Cr-N28 78.97(3), N1-Cr-N21 170.27(3), N8-
Cr-N28 142.02(3), O-Cr-N8 107.12(3), O-Cr-N28 110.81(3).
Figure 3. Thermal ellipsoid representation (50%) of the neutral zinc
complex 6. Hydrogen atoms have been omitted. Only one of two
independent molecules is shown.
Presuming that the magnetic susceptibilities for Fe 3 and Co
4 are also the result of strong antiferromagnetic coupling
between two ligand radicals and a high-spin divalent metal
X-band EPR spectrum for Mn 2, which is consistent with an S
) 3/2 system. Interestingly, transitions for both Kramers doublets
ms ) (1/2 and ms ) (3/2 are observed, and the spectrum could
be fitted by fixing all g-components to 2.00 and using hyperfine
1
center, we would expect an Stot ) 1 and an Stot ) /2 system,
respectively. For Fe 3 and Co 4, their µeff values of 3.68 and
2.28 µB are significantly higher than the spin-only values of
2.83 and 1.73 µB, respectively. Indeed, the simulations of the
µeff versus T plots required adopting large isotropic gtot values
of 2.55 for 3 and 2.65 for 4.51 Using the equations in Table 3,
we obtain gFe ) 2.37 and gCo ) 2.39. The obtained gCo value
5
coupling constants to the 55Mn nucleus (I ) /2) of (280, 110,
40) × 10-4 cm-1, a zero-field splitting D of 0.65 cm-1 and a
rhombicity E/D of 0.11.50
(49) Earnshaw, A.; Larkworthy, L. F.; Patel, K. S. J. Chem. Soc. A 1966, 363.
(50) Unfortunately, the transition from the ms ) (3/2 states was not so well
simulated given the shifted peaks between the experimental data and the
simulation at ∼100 mT. Because this g component also determines the
transition from the ms ) (1/2 states at 350 mT, it was not possible to
reproduce both parts perfectly using a first-order spin Hamiltonian
simulation.
(51) For 3, the plot was simulated using a zero-field splitting D ) -24.8 cm-1
(DFe ) -11.8 cm-1) and rhombicity E/D ) 0.04. For 4, the plot was
corrected for temperature-independent paramagnetism, øTIP, of 950 × 106
cm3 mol-1, and a Weiss constant θ ) -3.0 K was used in the simulation.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3185

A R T I C L E S
Lu et al.
Figure 7. X-band EPR spectrum (dX′′/dB) of Mn 2 in toluene glass shown
in bold (1.0 mM, 10.0 K, frequency ) 9.45 GHz, modulation ) 15.0 G,
power ) 252 µW). The spectrum was simulated (shown in gray) by adopting
the following values: g (fixed) ) 2.00; line widths, Γ ) (280, 110, 40) ×
10-4 cm-1; magnetic hyperfine coupling constants, A(55Mn, I ) 5/2) ) (280,
110, 40) × 10-4 cm-1; zero-field splitting, D ) 0.65 cm-1; rhombicity
E/D ) 0.11.
Figure 5. Histogram of all the individual C-C (white) and C-Nimine (gray)
bond distances (Å, (3 esd) of the ligand backbone in 1-6.
Figure 6. Temperature dependence of the magnetic moment, µeff, of 1, 2,
3, 4, and 6. The solid lines represent the spin-Hamiltonian simulations. See
text for fitting parameters.
Figure 8. Zero-field Mo¨ssbauer spectrum of 3 at 80 K. The fit is shown
as a solid line, with an isomer shift δ of 0.75 mm s-1 and a quadrupole
Table 3. Values for S and g of the Neutral Bis(ligand)metal
Complexes 1-4a
splitting |∆EQ| of 1.29 mm s-1
.
b
M
Stot
gtot
gtot
)
a(gmetal
)
−
b(gL)c
Smetal
gmetal
between the two ligand radicals need to be considered. The
ligand radicals can align in two possible ways: (1) antiparallel,
giving rise to a singlet S ) 0 ground state, or (2) parallel,
producing a triplet S ) 1 ground state. In general, organic
radicals have a g value of 2.0, and thus, their µeff values do not
deviate from the spin-only values. However, the µeff of 6 (2.52
µB) is much smaller than that expected for an S ) 1 system. It
is, however, very near 2.45µB, which is the spin-only value of
a ground state comprising isoenergetic S ) 0 and S ) 1 states.
To arrive at this value, we use the general formula for
determining µtot when two isolated spin ground states A and B
1
2
3
4
Cr
Mn
Fe
1
/
1.87
1.98
2.55
2.65
a ) 3/2, b ) 1/2
a ) 7/5, b ) 2/5
a ) 3/2, b ) 1/2
a ) 5/3, b ) 2/3
2
/
1.91
1.99
2.37
2.39
3
5
2
2
1
2
1
3
Co
/
/
2
2
a Based on magnetic susceptibility measurements (0.01 T, 60 to 290 K).
b Error ) (0.05; See Supporting Information, Figure 6. c gL ) 2 is assumed.
is in the range expected for tetrahedral high-spin Co(II) centers,
which due to spin-orbit coupling can have remarkably large g
values (2.37 to 2.46).52 Moreover, the X-band EPR spectrum
of Co 4 correlates with a gave of 2.60, which is in excellent
agreement with the value of 2.65 obtained from the magnetic
susceptibility measurement (SI Figure 7).
Tetrahedral high-spin Fe(II) centers have also been reported
to have deviant g values in the range 2.20 to 2.28.46 The obtained
Fe value 2.37 is slightly higher than these reported values. The
presence of a high-spin Fe(II) center, however, is irrefutably
based on Mo¨ssbauer studies. The zero-field Mo¨ssbauer spectrum
of solid 3 at 80 K shows one quadrupole doublet with an isomer
shift δ of 0.75 mm s-1 and a quadrupole splitting |∆EQ| of 1.29
mm s-1 (Figure 8). These parameters are fully consistent with
high-spin Fe(II).53
are degenerate, µtot
)
(µ )2+(µ )2. In this instance, we
x
Α Β
obtain µtot
)
0.25(0)2+0.75(2.828)2 ) 2.45µ upon consid-
x
Β
ering the spin-only values for the two spin states and factoring
in the degeneracy. The proximity of this value with the
experimental µeff indicates that the singlet and triplet states are
approximately degenerate in 6.
g
Due to the nonexisitent or the miniscule zero-field splitting
in Zn 6, the X-band EPR spectrum of the triplet state of 6 is
observable and shown in Figure 9. At g ) 2, features are
observed that result from the dipole-dipole coupling of the two
ligand radicals. A similar pattern has been reported by Raston
et al. for a similar bis(ligand)Zn complex featuring two
R-diimine radicals.21 Remarkably, the forbidden transition ∆ms
) 2 is clearly observable at half-field (g ) 4.0).
The Zn complex 6 has a temperature independent µeff of 2.52
µB. Because the Zn(II) center is redox inert, only the interactions
(52) Girerd, J.-J.; Journaux, Y. In Physical Methods in Bioinorganic Chemis-
try: Spectroscopy and Magnetism; Que, Jr., L., Ed.; University Science
Books: Sausalito, CA, 2000; p 321.
(53) Greenwood, N. N.; Gibb, T. C. Mo¨ssbauer Spectroscopy; Chapman and
Hall Ltd.: London, 1971.
2.4. Synthesis and Characterization of 1ox‚THF-6ox.
Spin-spin interactions between the metal-ligand and ligand-
ligand pairs have been explored in the neutral bis(ligand)metal
9
3186 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
ligand intervalence charge transfer (LLIVCT). Table 4 lists the
physical details of these bands.
Building on the Marcus seminal model of electron-transfer,
Hush has developed relationships to calculate electron-transfer
parameters in symmetric metal-centered systems based on the
intervalence charge-transfer band.56 The following Hush equa-
tion relates the theoretical bandwidth at half-intensity ∆ν1/2 to
1/2
the band maximum νmax of the IVCT, ∆ν1/2 ) [2310 νmax
]
(cm-1). When experimental values for ∆ν1/2 are smaller than
their estimated theoretical values, then full delocalization of the
electron typically is occurring.26 The ∆ν1/2 values for Fe 3ox‚
THF and Co 4ox‚THF are well within the theoretical estimates,
and hence both of these complexes are assigned to Class III
(Table 4). We note that this is a tentative assignment because
it can be difficult to experimentally differentiate between Class
II and Class III borderline systems.26
Figure 9. X-band EPR spectrum (dX′′/dB) of Zn 6 in toluene glass (1.0
mM, 15.0 K, frequency ) 9.45 GHz, modulation ) 6.0 G, power ) 500
µW). Inset corresponds to the forbidden ∆ms ) 2 transition.
The electronic spectrum for Ni 5ox‚THF preserves both the
intensity and the location of the far-infrared absorption of its
neutral counterpart 5. Currently, we do not wish to speculate
on the character of this transition but note that a similar complex
(R-diimine)2Ni had an analogous feature in its electronic
absorption spectrum.16
series (Vide supra). We were interested in extending the study
of spin-spin interactions to the mixed-valent systems featuring
the two ligands in different redox states. Indeed, oxidation of
just one ligand radical appears feasible based on the cyclic
voltammograms of 1-6.
The one-electron oxidations of 1-6 were effected by adding
1 equiv of [Cp2Fe][B(ArF)4] in THF (where ArF ) 3,5-
(CF3)2C6H3), generating the oxidized bis(ligand)metal com-
plexes, [(Lx)2M(THF)][B(ArF)4] (1ox‚THF-4ox‚THF) or [(Lx)2M]-
[B(ArF)4] (5ox, 6ox) (Scheme 3). All these compounds are soluble
in THF, and in this solvent, they exist as five-coordinate species
with a bound THF molecule [(Lx)2M(THF)]+ (Vide infra).54 Only
the cationic Ni complex has a visibly different color as a powder
(brown-purple) and as a THF solution (green-black), suggesting
that as a powder it is the four-coordinate bis(ligand)nickel com-
plex [(Lx)2Ni][B(ArF)4] 5ox. Combustion analysis results for all
the cationic compounds are consistent with four-coordinate spe-
cies and the absence of THF. However, a bound THF molecule
could be labilized under the conditions of the combustion
experiment. Indeed, mass analysis of well-dried crude powders
after thorough washings with hexane indicates that complexes
1ox‚THF-4ox‚THF remain solvated by THF (Cr, 1ox‚THF;
Mn, 2ox‚THF; Fe, 3ox‚THF; Co, 4ox‚THF), while THF is absent
in Ni 5ox and Zn 6ox. The solvated nature of Fe 3ox‚THF is
corroborated by Mo¨ssbauer experiments on both the dry powder
and wet crystalline samples of 3ox‚THF (Vide infra).
The electronic absorption spectra of 1ox‚THF-6ox‚THF were
shown previously in Figure 2 (in THF, 22 °C). Akin to the
neutral species 1-6, the cations 1ox‚THF-6ox‚THF are intensely
colored with high molar absorptivity between 300 and 500 nm
(∼104 M-1 cm-1). The intraligand charge-transfer bands
between 600 and 900 nm are still preserved in Cr (1ox‚THF),55
Mn (2ox‚THF), and Zn (6ox‚THF). Indeed, the electronic spectra
of the neutral and cationic species are similar for Cr, Mn, and
Zn. On the other hand, striking differences between the neutral
and cationic species are seen for Fe and Co. The electronic
spectra of Fe 3ox‚THF and Co 4ox‚THF are dominated by a
strong broad band in the near-infrared (900 to 1400 nm, ∼3 ×
103 M-1 cm-1), which is the result of intramolecular ligand-
2.5. Solid-State Structures of 1ox‚THF-6ox‚THF. Com-
pounds 1ox‚THF-6ox‚THF crystallized from THF/pentane solu-
tions as five-coordinate cations with B(ArF)4 counteranions
(Table 5). Figures 10 and 11 show the molecular structures of
the cationic Mn 2ox‚THF and Fe 3ox‚THF complexes, respec-
tively. The structures of the other five-coordinate cationic
complexes are provided in the Supporting Information (SI
Figures 8-11). The solid-state structures are all isostructural
with remarkably similar bond angles around the metal center
(SI Figure 12). These cationic complexes adopt a distorted
trigonal bipyramidal geometry, with the two pyridyl nitrogen
atoms (Npyr) occupying the axial positions (Npyr-M-Npyr, 168°
to 178°). The equatorial plane is occupied by the two iminyl
nitrogens (Nim) and the oxygen atom of THF. Though these
atoms are perfectly coplanar, they distort from tbp geometry
due to the large Nim-M-Nim angle (132° to 139°) and the
asymmetric positioning of the oxygen atom between the two
iminyl nitrogens (O-M-Nim ) 103° to 100°, 116° to 119°).
Close scrutiny of the solid-state structures reveals that the
bond lengths in the backbone of both ligands vary gradually
within the cationic complexes. The two R-iminopyridine ligands
are not always identical as in the neutral series. Figure 12 depicts
the range of C-C and C-Nim bond distances plotted in
descending order of the former with (3 esds. At the far left of
the histogram, the observed distances are representative of the
neutral ligand (L)0, and at the far right, the distances are
characteristic of the monoanionic π radical ligand (L•)-. These
divisions are denoted in Figure 12 with dashed red lines. The
Cr 1ox‚THF, Mn 2ox‚THF, and Zn 6ox‚THF compounds haVe
one ligand of each type and thus are formulated as [(L•)(L)M-
(THF)]+. The average C-C and C-Nim bond distances in 1ox‚
THF, 2ox‚THF, and 6ox‚THF are, for (L)0, 1.469(4) and 1.279(6)
Å, respectively, and, for (L•)-, 1.411(8) and 1.345(4) Å,
respectively.
In stark contrast, the C-C and C-Nim distances of both
(54) Dichloromethane would have been a more ideal solvent than THF because
of its reduced tendency to coordinate. However, some complexes showed
rapid decomposition.
ligands in Ni 5ox‚THF are nearly identical with average values
(55) The electronic spectrum of Cr 1ox has a band at 455 nm which is akin to
the 495 nm band in 1 when dissolved in Et2O. Both bands may be the
result of a coordinated solvent (LMCT).
(56) Hush, N. S. In Progress in Inorganic Chemistry; Cotton, F. A., Ed.;
Interscience: New York, 1967; Vol. 8, p 391.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3187

A R T I C L E S
Scheme 3
Lu et al.
series 1ox‚THF-6ox, which, despite being isostructural in THF,
exhibit vastly different electronic structures. The Cr 1ox‚THF,
Mn 2ox‚THF, and Zn 6ox‚THF compounds are classified as
Robin-Day Class I with a localized ligand radical, while the
Fe 3ox‚THF and Co 4ox‚THF complexes belong to Class III with
a fully delocalized radical. The Ni 5ox‚THF species stands apart
in that a ligand radical is absent. In the case of Ni, we suggest
that the energy incurred by the pairing of two metal d-electrons
(in addition to the loss of exchange interaction) is compensated
by the stabilization of the radical at the Ni center. This may be
reasonable since nickel has the greatest effective nuclear charge
of the series and, thus, the energetically lowest-lying d-orbitals.
Table 4. IVCT Band Details
νmaxa (cm-
)
ꢀmax
∆ν1/2 (th)c
1
b
3ox‚THF(Fe)
4ox‚THF(Co)
1175 (8510)
1077 (9285)
2.6 × 103
2120 (4430)
3590 (4630)
3.0 × 103
a In nm. b In M-1 cm-1
.
c In cm-1; th ) theoretical.
of 1.451(8) Å for the C-C bond and 1.30(1) Å for the C-Nim
bond. These values indicate that the ligands are neutral. Thus,
the Ni complex 5ox‚THF is best described as [(L)2Ni(THF)]+,
wherein one electron (originally ligand-based) has been trans-
ferred to the Ni center, reducing it to Ni(I). This oxidative
phenomenon has been observed in an analogous system with
nickel and two R-diimine ligands.15,16 The expected increase in
metal-ligand π-back-bonding for a monovalent versus divalent
metal explains the small perturbation in the bond distances
observed in Ni 5ox‚THF relative to the other (L)0-containing
complexes.
The ligand geometries in Fe 3ox‚THF and Co 4ox‚THF are
also nearly identical. The average distances for the C-C and
C-Nimine bond are 1.425(3) and 1.317(6) Å, respectively, which
are almost exactly intermediate between those expected for the
(L)0 and (L•)- oxidation states. These compounds are also
formulated as [(L•)(L)M(THF)]+. However, they are distinct
from the Cr 1ox‚THF, Mn 2ox‚THF, and Zn 6ox‚THF complexes
in that the unpaired electron of the ligand radical is delocalized
over both ligands. Consequently, each ligand is essentially
reduced by “half” an electron giving rise to the observed bond
distances. The delocalization of the ligand-based electron is
consistent with the LLIVCT band observed for Fe 3ox‚THF and
Co 4ox‚THF. Admittedly, a quarter of a bond order is too small
to be statistically differentiated using low-temperature X-ray
diffraction studies. We do, however, have the advantage of
possessing a whole series to track and interpret these subtle
structural changes.
2.6. Magnetic Susceptibility Measurements of the Cationic
Series [M(Lx)2(THF)]+ or [M(Lx)2]+ with EPR and Mo1ss-
bauer Spectroscopic Characterization. The magnetic suscep-
tibilities of the cationic compounds were measured on powder
samples from 4 to 300 K using an applied field of 0.01 T. In
the cases where the sample’s magnetism was too low to be
measured accurately, a higher field of 1.0 T was applied (for
5ox, 6ox). The temperature-dependent plots of the effective
magnetic moments, µeff, for 1ox‚THF-6ox are collectively shown
in Figure 13. With the exception of Fe 3ox‚THF and Co 4ox‚
THF, the rest of the series have magnetic moments that are
independent of temperature from 30 to 290 K.
The µeff of Cr 1ox‚THF is 3.73µB, which is near the spin-
3
only value of an S ) /2 system (3.87µB). The data were fit
using a gtot of 1.93 (øTIP ) 100 × 106 cm3 mol-1 and θ ) -0.7
K). As described previously, the metal contribution can be
isolated using the spin projection equations in Table 6 to yield
a reasonable gCr of 1.94. The µeff of Mn 2ox‚THF is 4.90µB,
which is the exact spin-only value of an S ) 2 system. From
the data simulation57 a very agreeable gMn of 2.00 was
calculated. The ground-state spin systems of the Cr 1ox‚THF
and Mn 2ox‚THF compounds are the result of a strong
The one-electron oxidation of the neutral complexes 1-6,
which share a common electronic structure, produces the cationic
(57) The data were fit using a gtot of 2.00 (with øTIP ) 200 × 106 cm3 mol-1
and θ ) -0.5 K).
9
3188 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
Table 5. Crystallographic Data for 1ox‚THF-6ox‚THF
1ox
‚
THF
2ox
‚
THF
3ox
‚
THF
4ox
‚
THF
5ox
‚THF
6ox
‚THF
formula
C72H64BCr-
F24N4O
C72H64BF24-
MnN4O
C72H64BF24-
FeN4O
C72H64BF24-
CoN4O
C72H64BF24-
NiN4O
C72H64BF24-
ZnN4O
crystal size, mm3
fw
space group
a, Å
b, Å
c, Å
0.10 × 0.12 × 0.18 0.16 × 0.16 × 0.20 0.08 × 0.09 × 0.26 0.10 × 0.12 × 0.20 0.05 × 0.25 × 0.32 0.17 × 0.18 × 0.24
1520.08
P21/n, no. 14
12.7789(4)
29.4954(9)
18.8736(6)
90
95.181(3)
90
7084.7(4)
4
1523.02
P21/n, no. 14
12.7681(3)
29.7624(9)
18.8399(5)
90
94.691(3)
90
7135.4(3)
4
1523.93
P21/n, no. 14
12.7624(2)
29.4794(4)
18.9134(3)
90
95.431(3)
90
7083.81(18)
4
1527.01
P21/n, no. 14
12.7059(4)
29.3218(11)
18.9079(5)
90
95.674(3)
90
7009.8(4)
4
1526.79
P21/n, no. 14
12.7517(5)
29.4819(11)
18.8847(8)
90
95.573(3)
90
7066.0(5)
4
1533.45
P21/n, no. 14
12.8399(4)
29.4188(7)
18.7498(4)
90
95.159(3)
90
7053.8(3)
4
R, deg
â, deg
γ, deg
V, Å3
Z
T, K
100(2)
1.425
154 718/60.0
100(2)
1.418
160 388/60.0
120(2)
1.429
1 419 135/55.0
100(2)
1.447
174 881/60.0
100(2)
1.435
150 361/60.0
100(2)
1.444
210 459/65.0
F calcd, g cm-3
refl. collected/
2σmax
unique refl./
I > 2σ(I)
no. params./
restraints
20 641/13 953
949/7
20 778/15 615
949/7
16 234/14 313
949/7
20 419/15 265
949/7
20 584/15 870
949/7
25 488/18 550
949/7
λ, Å /µ(KR), cm-1 0.710 73/2.7
0.710 73/2.9
0.0541/1.007
0.1457
0.710 73/3.23
0.0515/1.021
0.1309
0.710 73/3.6
0.0504/1.031
0.1231
0.710 73/3.8
0.0560/1.036
0.1379
0.710 73/4.6
0.0501/1.007
0.1231
R1a/GOFb
0.0603/1.009
0.1508
+0.66/-0.70
wR2c (I > 2σ(I))
residual density,
e Å -3
+0.82/-0.76
+1.09/-0.79
+1.00/-0.63
+1.04/-0.81
+1.08/-1.02
2
2
2
2
2
a Observation criterion: I > 2σ(I). R1 ) Σ||Fo| - |Fc||/Σ|Fo|. b GOF ) [Σ[w(Fo - Fc )2]/(n - p)]1/2
.
c wR2 ) [Σ[w(Fo - Fc )2]/Σ[w(Fo )2]]1/2 where
w ) 1/σ2(Fo ) + (aP)2 + bP, P ) (Fo + 2Fc )/3.
2
2
2
Figure 10. Thermal ellipsoid representation (50%) of the cationic Mn
complex 2ox‚THF. Hydrogen atoms and the B(ArF)4 counteranion have been
omitted.
Figure 11. Thermal ellipsoid representation (50%) of the cationic Fe
complex 3ox‚THF. Hydrogen atoms and the B(ArF)4 counteranion have been
omitted.
antiferromagnetic coupling of one ligand radical with a divalent
high-spin metal.
The Ni 5ox and the Zn 6ox complexes should both be S ) 1/2
systems. The unpaired electron is located at the metal center in
the former case and is ligand-based in the latter. Indeed, the
µeff values of Ni 5ox and Zn 6ox are 1.93 and 1.64 µB,
respectively, which are near the expected spin-only value of
1.73 µB. For Ni 5ox, the data is simulated using g ) 2.23 (gNi)
with øTIP ) 55 × 106 cm3 mol-1 and θ ) -1.0 K. The X-band
EPR spectrum for Ni 5ox reflects a composite of two species
with similar g values of (2.23, 2.22, 2.09) and (2.24, 2.23, 2.05)
for the major and minor species, respectively (SI Figure 13).
Akin to the neutral Ni 5 compound, we suggest that two
geometric isomers are present in the frozen EPR solution. This
has been previously described for four-coordinate Ni(I) cations
coordinated by two R-diimines.15 The fit for Zn 6ox uses g )
1.90 with øTIP ) 20 × 106 cm3 mol-1 and θ ) -0.1 K. The
obtained g value is lower than expected for an organic radical,
although a difference of 0.05 is within the error of the magnetic
susceptibility measurement (see SI Figure 6). The EPR spectrum
for Zn 6ox clearly shows an organic radical with g ) 2.00 (SI
Figure 14).
For Co 4ox‚THF, the µeff is strongly temperature dependent.
In the lower temperature regime (from 50 to 180 K), Co 4ox‚
THF has a temperature independent µeff of 3.73µB, which can
be tentatively treated as a triplet ground state. This is akin to
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3189

A R T I C L E S
Lu et al.
Figure 14. X-band EPR spectrum (dX′′/dB) of Fe 3ox‚THF in frozen CH3-
CN (2.4 mM, 10.0 K, frequency ) 9.45 GHz, modulation ) 16.0 G, power
) 503 µW). The spectrum was simulated as a composite of two species.
See text for details.
Figure 12. Histogram of all the individual C-C (white) and C-Nimine
(gray) bond distances (Å, (3 esd) of the ligand backbone in 1ox‚THF-6‚
THF.
190 K. Above 190 K, the µeff becomes temperature independent
and can be fitted as an Stot ) 3/2 system with gtot and gFe values
of ∼2.1 (øTIP ) 20 × 106 cm3 mol-1). This spin system suits
the general electronic model of a high-spin divalent metal (SFe
) 2) coupled antiferromagnetically to one ligand radical (SL )
3
1/2) to provide an overall spin state of S ) /2. The dramatic
drop in µeff below 190 K shows that this model is no longer
valid at low temperatures. Recall that the 120 K solid-state
structure of 3ox‚THF revealed one ligand radical. Thus, whatever
the changes in the electronic structure are occurring, the ligand
radical does not appear to be affected.
We propose that the Fe center undergoes a spin transition
from high spin (SFe ) 2) to intermediate spin (SFe ) 1) with
decreasing temperature. To investigate whether Fe is in spin
equillibrium, EPR and Mo¨ssbauer experiments were conducted.
The X-band EPR spectrum of 3ox‚THF in frozen CH3CN at 10.0
K reveals the presence of two species. The predominant species
Figure 13. Temperature dependence of the magnetic moment, µeff, of 1ox‚
THF-6ox. The solid lines represent the spin-Hamiltonian simulations. See
text for fitting parameters.
Table 6. Values for S and g of the Cationic Complexes
1ox‚THF-4ox‚THFa
1
(58%) corresponds to the nearly axial signal of an Stot ) /2
system (with g ) (1.98, 2.07, 2.60) and linewidths Γ ) (134,
112, 270) G). This signal is consistent with an intermediate spin
Fe (SFe ) 1) that is coupled antiferromagnetically to a ligand
radical (SL ) 1/2). The minor species (42%) is interpreted as an
b
M
Stot
gtot
gtot
)
a (gmetal
)
−
b (gL)c
Smetal
gmetal
3
1d
2d
3e
4f
Cr
Mn
Fe
/
1.93
2.00
a ) 6/5, b ) 1/5
a ) 7/6, b ) 1/6
a ) 6/5, b ) 1/5
a ) 5/4, b ) 1/4
2
1.94
2.00
2
5
2
/
2
3
/
∼2.1
2
∼2.1
2
3
Co
1
2.75
/
2.60
3
2
Stot ) /2 system based on its g values of (2.03, 2.22, 4.80)
(linewidths Γ ) (141, 467, 500) G). Thus, in a frozen CH3CN
solution, both the intermediate and high spin Fe centers are
observable by EPR spectroscopy.59 The Mo¨ssbauer data were
collected on powder and wet crystalline samples of 3ox‚THF at
80 and 200 K. No differences were found between the dry
powder and the wet crystals, confirming that THF is present in
both samples. At 80 K, δ is 0.56 mm s-1 and |∆EQ| is 1.04
mm s-1 (Figure 15). Both Mo¨ssbauer parameters change at 200
K, with a δ of 0.77 mm s-1 and a |∆EQ| of 2.12 mm s-1. The
isomer shift at 200 K is nearly equivalent to that observed for
neutral Fe 3 (δ ) 0.75 mm s-1) and is fully consistent with a
high-spin Fe(II) center. The smaller isomer shift of 0.56 mm
s-1 recorded at 80 K is expected for an intermediate-spin Fe-
(II) due to increased s-density at the nucleus.
a Based on magnetic susceptibility measurements (0.01 T). b Error )
(0.05; See Supporting Information, Figure 6. c gL ) 2 is assumed. d Based
on data from 30 to 290 K. e 190 to 290 K. f 50 to 180 K.
the case for Cr 1ox‚THF and Mn 2ox‚THF, wherein the total
spin state (Stot ) 1) is the result of an antiferromagnetic
interaction between a high-spin Co(II) (SCo ) 3/2) and a ligand
1
radical (SL ) /2). The plot from 50 to 180 K can be well-
simulated by adopting a large gtot value of 2.75, from which
we derive a gCo of 2.60.58 At temperatures above 170 K, the
µeff of Co 4ox‚THF increases slowly. This slight rise in µeff is
reproducible for different batches, but we are puzzled by its
cause. One possible interpretation is that the metal and ligand
electrons are uncoupling at higher temperatures. However,
broken symmetry DFT calculations predict that this coupling
is quite strong (Vide infra). Moreover, the uncoupling of a
metal-ligand magnetic pair is unprecedented in this and related
systems.
Intermediate-spin Fe(II) is uncommon and are typically four-
coordinate,23,60 but a few examples of five-coordinate SFe ) 1
(59) It is almost certain that CH3CN replaces THF in the iron’s first coordination
sphere. The EPR spectrum of 3ox‚THF in frozen THF shows a similar
spectrum albeit with the superposition of many small features ranging from
240 to 320 mT.
(60) Debrunner, P. G. In Iron Porphyrins Part 3; Lever, A. B. P., Gray, H. B.,
Eds.; VCH: New York, 1983; p 139.
The most prominent feature in the µeff versus T plot for Fe
3ox‚THF is the gradual spin transition that occurs from 50 to
(58) For Co 4ox‚THF, øTIP ) 1000 × 106 cm3 mol-1 and θ ) -5.2 K.
9
3190 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
Figure 15. Zero-field Mo¨ssbauer spectrum of 3ox‚THF at 80 K in black
and 0.77 (2.12) mm s-1 at 200 K in red. The fits are shown as solid lines,
with δ (∆EQ): 0.56 (1.04) mm s-1 at 80 K and 0.77 (2.12) mm s-1 at
200 K.
Figure 17. Comparison of important bond distances (Å) from the solid-
state structures of 3ox‚THF collected at 120 and 220 K. The parameters
from the 220 K measurement are italicized.
upon isolating the iron A tensor by the relation, AFe ) 0.75-
(Atot) ) -186 T.
Another method to study the temperature dependence of the
overall spin state is through X-ray crystallography. In addition
to the solid-state structure at 120 K described previously, a
single-crystal structure determination of 3ox‚THF was performed
at 220 K.62 Figure 17 is a simplified depiction of 3ox‚THF with
the pertinent bond distances from the two molecular structural
solutions at 120 and 220 K. Notably, the ligands in the two
structures are identical, and the ligand radical is present and
delocalized at low and high temperatures. Only the Fe-O and
Fe-N bond distances increase significantly with increasing
temperature. From the 120 K structure to that of the 220 K, the
subtle elongation of the bonds (e0.08 Å) within the first
coordination sphere of Fe is expected upon changing from
intermediate-spin to high-spin iron.
2.7. DFT Calculations. To garner a better understanding of
the electronic structures, DFT calculations for 1-6 and 1ox‚
THF - 6ox‚THF were conducted at the B3LYP level. All the
calculations used spin-unrestricted methods. Broken symmetry
states, with n spin-up electrons and m spin-down electrons, are
denoted by the label BS(n,m). The broken symmetry approach
is pertinent to this study because it permits the n and m electrons
to be localized on different areas of the molecule, i.e., metal
versus ligand, so they can couple magnetically but are not
forcibly paired.63,64 The robustness of all BS solutions was
checked by complementary calculations without broken sym-
metry. These latter calculations, unless otherwise noted, resulted
in a higher energy state.
Figure 16. Applied field Mo¨ssbauer spectra of 3ox‚THF with variable
temperature (4.2, 12.0 K) and fields (4 and 7 T). The fits are shown as
solid lines, with δ (∆EQ): 0.54 (+1.05) mm s-1. Other fitting parameters
are as follows: asymmetry η ) 0.40; line width ) 0.25 mm s-1; g ) (2.0,
2.6, 2.0) fixed; Axx/gNâN ) -244.24 T; Ayy/gNâN ) -32.44; Azz/gNâN
)
-17.97; Axx/gNâN(Fe) ) -183.18; Ayy/gNâN(Fe) ) -24.33; Azz/gNâN(Fe)
) -13.48 T.
complexes are known.23,61 Due to the rarity of such species,
applied-field Mo¨ssbauer studies of 3ox‚THF were undertaken
at low temperatures (4.2, 12.0 K). Fixing the g-values to those
obtained from the frozen EPR spectrum of 3ox‚THF in CH3CN
(g ) (1.98, 2.60, 2.07)), the applied field Mo¨ssbauer plots in
Figure 16 were simulated using the following fit parameters:
δ ) 0.54 mm s-1; ∆EQ ) +1.05 mm s-1; asymmetry η ) 0.40;
line width ) 0.25 mm s-1; fixed; Axx/gNâN ) -247.5 T; Ayy/
gNâN ) -32.7 T; Azz/gNâN ) -17.7 T. The A tensor is
approximately axial with notably one large negative component
of -248 T. This component remains quite large and negative
(62) Crystallographic Data for 3ox‚THF at 220 K: Monoclinic, P21/n (No. 14);
a ) 12.8349(2) Å, b ) 29.9514(5), c ) 18.9948(3), â ) 94.795(3)°; V )
7276.5(2) Å3; Z ) 4; calcd F ) 1.391 g cm-3; µ ) 3.14 cm-1; λ ) 0.710 73
Å; reflns collected 106 596; unique reflns 16 674 (12 891 for I > 2σ(I));
2Θmax ) 55.0°; No. param/restraints 1014/534; GOF 1.022; R1 ) 0.0646,
wR2 ) 0.1656 (I > 2σ(I)); residual density 0.711/-0.714 e Å-3. See Table
5 for observation criterion.
(63) Noodleman, L.; Davidson, E. R. Chem. Phys. 1986, 109, 131.
(64) Noodleman, L. J. Chem. Phys. 1981, 74, 5737.
(61) Bacci, M.; Ghilardi, C. A.; Orlandini, A. Inorg. Chem. 1984, 23, 2798.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3191

A R T I C L E S
Lu et al.
Table 7. Optimized Geometric and Electronic Parameters for DFT-Calculated Bis(ligand)metal Models of 1-6a
metal
Cr
Mn
Fe
Co
Ni
Zn
3
1
Stot
BS(n,m)
1
/
1
/
0
1 (and 0)b
BS(1,1)
2
2
BS(4,2)
BS(5,2)
BS(4,2)
BS(3,2)
BS(2,2)
M-Nim (Å)
2.075
2.018(4)
2.076
2.046(4)
1.346
1.351(3)
1.408
2.082
2.051(4)
2.136
2.09(1)
1.342
1.342(2)
1.415
2.032
1.986(4)
2.105
2.041(7)
1.341
1.340(2)
1.418
2.009
1.954(4)
2.030
1.968(5)
1.337
1.336(2)
1.419
1.984
1.912(6)
2.029
1.94(1)
1.332
1.332(2)
1.423
2.028
1.992(6)
2.060
2.03(1)
1.341
1.339(5)
1.417
M-Npyr (Å)
C-Nim (Å)
C-C (Å)
θ (deg)
1.3978(4)
40.8
1.410(3)
77.1
1.410(2)
73.4
1.409(2)
75.9
1.416(4)
68.5
1.409(4)
77.6
41.4
84(2)
80.4(0.7)
80(1)
70.9
84(2)
c
J (cm-1
)
-460
0.53, 0.03
-492
0.37, 0.36
-517
0.43, 0.38
-696
0.46, 0.42
-1002
0.66, 0.42
-32
0.01
Sc
metal spin density
ligand spin density
+3.98
-1.98
+4.76
-1.76
+3.67
-1.67
+2.58
-1.58
+1.38
-1.38
0
+1.97(0.00)d
a Average experimental values are given in italics with standard deviation in parentheses. b Both the S ) 0 and S ) 1 ground states were calculated for
Zn 6 and found to be isoenergetic (∆ ≈ 1 × 10-5 eV). All the calculated structural parameters are identical. c Coupling constant J or spatial overlap S
between the two magnetic pairs of metal-ligand spin vectors (except in the case of Zn). For Zn, the values correspond to the magnetic pair of ligand-ligand
spin vectors. d In parentheses, the total spin density of the S ) 0 Zn model, comprising -1 and +1 for the two ligands separately.
2.7.1. Geometry Optimizations of the Neutral [(L•)2M]
Species. Taking the initial coordinates from the solid-state
structure of 6, the geometry optimizations of the bis(ligand)-
metal complexes 1-6 converged successfully for the given Stot
and BS state (Table 7). For the Cr complex 1, we decided to
model the non-solvento species instead of the structurally
characterized 1‚Et2O to better follow any theoretical trends in
the neutral series. Notably, the calculated model of 1 reproduces
the dihedral angle in 1‚Et2O despite its lesser coordination
number, reinforcing our earlier discussion regarding the pro-
pensity of a d4 ion in a tetrahedral environment to undergo
Jahn-Teller distortion.
In all the converged structures, the two ligands are exactly
identical. In general, we observe excellent agreement between
the experimental structures and their calculated analogues. For
example, all the experimental bond distances in the ligand are
reproducible within 0.01 Å. Selected bond distances are shown
in Table 7 alongside their experimental values. Even the dihedral
angles do not deviate by more than 7°. The only poor agreements
are the metal-ligand bond distances, which are overestimated
by 0.03 to 0.09 Å, but this is typical for the B3LYP functional.65
2.7.2. Electronic Structures of the Neutral [(L•)2M] Spe-
Figure 18. Qualitative MO diagram of the magnetic orbitals derived from
cies. Given the faithful reproduction of the geometrical param-
eters, the theoretical models were further investigated for their
electronic structures. The molecular orbital (MO) diagrams for
1-6 have in common the feature that the metal d-orbitals are
energetically below two high-lying ligand-based π* orbitals. The
d-orbital splitting patterns vary within this series. This is not
surprising for 1 given its drastically different dihedral angle (see
SI Figure 15). We would expect the pattern to be similar for
2-5, and this is valid for Mn 2, Fe 3, and Co 4, which all have
an approximate “two-over-three” pattern. For simplicity, we
show only the MO diagram for the Mn BS(5,2) model in Figure
18. As stated previously, we observe three energetically low-
lying d-orbitals and two high-lying d-orbitals. This pattern is
the inVerse expected for a tetrahedral ligand field, but recall
that the ligand bite angles are significantly smaller than the
the BS(5,2) calculation of the Mn complex 2. The spatial overlaps (S) of
the corresponding alpha and beta orbitals are given.
idealized tetrahedral angles and consequently the antibonding
interactions for one d-orbital of the Td t2* set are greatly
relieVed.
For the Mn 2 model, there are five occupied metal-based
SOMOs in the spin-up manifold. In the spin-down manifold,
two ligand-based SOMOs are occupied which are primarily the
LUMO π* orbitals of the neutral R-iminopyridine. The ligand-
based SOMOs couple magnetically to metal-based SOMOs with
spatial overlaps of 0.37 and 0.36. The spin-spin coupling
constant, J, between the Mn spin vector and the total ligand
spin vector can be determined using Yamaguchi’s equation,66
(66) Soda, T.; Kitagawa, Y.; Onishi, T.; Takano, Y.; Shigeta, Y.; Nagao, H.;
(65) Neese, F. J. Biol. Inorg. Chem. 2006, 11, 702.
Yoshioka, Y.; Yamaguchi, K. Chem. Phys. Lett. 2000, 319, 223.
9
3192 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
EHS - EBS
J ) -
S2 HS - S2
BS
for a two-spin system described by the Hamiltonian, Hˆ )
-2JSˆA ‚ SB. The coupling constant is based on the energies (E)
and spin-expectation values ( S2 ) of the broken symmetry and
the high-spin states. For Mn, J is -460 cm-1, corroborating
the strong antiferromagnetic coupling between the metal center
and ligand radicals previously indicated by the temperature-
independent µeff of 2.
To extend the MO picture to the later metals in the neutral
series, one simply needs to add additional d-electron(s) to the
metal-based orbitals. The overall d-orbital splitting scheme
remains consistent for the models of Fe 3 and Co 4. Two ligand-
based SOMOs are consistently observed, reaffirming the
generality of the presence of two ligand radicals in the neutral
series. This is also true for the model of Ni 5; however, a rather
different situation is encountered for the d-orbital splitting.
Unlike the others, the Ni model reveals altogether seven MOs
with a strong d-orbital contribution. Three are predominantly
metal-based (75 to 90%); however the remaining four have only
40 to 47% metal character. These latter MOs are the result of
strong metal-ligand mixing as depicted in the MO diagram
for Ni 5 shown in Figure 19.
Figure 19. Qualitative MO diagram of the magnetic orbitals derived from
the BS(2,2) calculation of the Ni complex 5. The spatial overlaps (S) of
the corresponding alpha and beta orbitals are given.
The spin-density plot for the Mn species also depicts a picture
of a high-spin Mn(II) center and two ligand radicals (Figure
20). The spin densities of +4.76 and -1.76 for the metal and
ligands, respectively, are near the expected values of five
d-electrons and two unpaired radicals. Interestingly, the DFT
calculations reveal an interesting correlation between the
coupling constant J, the spatial overlap S, and spin-density
values (Table 7). In the instance of Zn 6 (S ) 1), two full ligand
radicals are observed (+1.97) in the spin-density map. The
radicals interact very weakly with one another as indicated by
a small J and a nearly zero spatial overlap of the ligand SOMOs.
In the case of the other metals, a strong periodic trend is
observed in that J and S increase dramatically across the period.
Moreover, the spin-density values show a gradual progression
from an electronic formulation of a divalent metal center with
two ligand radicals (Cr, Mn) toward that of a monovalent metal
center with one ligand radical (Ni). Though the absolute values
for J and S may not be accurate, the trend is unmistakable. One
explanation to account for these DFT observations is an
increasing covalent character between the ligand π* orbitals and
the metal d-orbitals, which would reflect in higher J and S values
as well as perturbed spin densities. Indeed, a closer inspection
of the ligand π* SOMOs reveals a growing metal contribution
traversing from Cr (6.4%) to Ni (21.0%) (Table 8). Metal-
ligand mixing was also considered for the ligand SOMOs of
σ* symmetry, but no clear periodic trend was seen.
Figure 20. Spin density plot of the bis(ligand)Mn complex 2 shown with
values based on a Mulliken spin population analysis. Only spin densities
for one ligand are shown (the two ligands have equivalent spin-density
distribution), and only atoms with spin densities g 0.10 are labeled.
high spin and intermediate spin due to the observed spin
equilibrium between these two spin states. The doublet state
(SFe ) 1) is found at lower energy relative to the quartet state
(SFe ) 2), which is in accord with our experimental data, though
the calculated stabilization of 43.5 kcal/mol is likely overesti-
mated.
Recall that in the cationic series 1ox‚THF-6ox‚THF the
equivalency of the two ligands depends on whether the ligand-
based π* radical is delocalized or localized. We had hoped DFT
calculations could reproduce the metal-dependency of the ligand
bond distances in the geometry-optimized models. However,
in all the converged structures, the two ligands were structurally
identical (full delocalization of the ligand radical), which is
inconsistent with experimental data. We were able to favor
localization of the ligand radical by considering solvation effects
using the conductor-like screening model (COSMO), which is
a continuum dielectric approach.67 In the specific case of Zn
6ox‚THF, the effect of various solvents was studied, including
2.7.3. Geometry Optimizations of the Cationic [(Lx)2M-
(THF)]+ Species. For comparative purposes, all the calculated
models of the monocations 1ox‚THF-6ox‚THF contain a bound
THF, and the initial coordinates for each calculated complex
were taken from their solid-state structures. The geometry
optimizations converged successfully for the given Stot and BS
states (Table 9). For Ni 5ox‚THF, the BS(2,1) calculation
converged to a non-BS solution, which is consistent with the
experimental finding of a Ni(I) center and the absence of a
ligand radical. For Fe, the metal center was modeled as both
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3193

A R T I C L E S
Lu et al.
Table 8. Percentage (%) of Metal Character in Ligand-Based σ* and π* SOMOs Determined from the DFT-Calculated Models of 1-5 and
1ox‚THF-5‚THFa
a Values for the neutral (top) and cationic (bottom) model complexes are given.
Table 9. Optimized Geometrical and Electronic Parameters for DFT-Calculated [(Lx)2M(THF)]+ Models of 1ox‚THF-6ox‚THFa
metal
Cr
Mn
Fe (HS)
Fe (IS)
Co
Ni
Zn
3
3
1
1
1
Stot
/
2
/
/
1
/
/
2
2
2
2
2
BS(n,m)
BS(4,1)
BS(5,1)
BS(4,1)
BS(2,1)
BS(3,1)
-
-
M-Nim (Å)
2.055
1.994(2)
2.180
2.111
2.067(2)
2.271
2.098
2.018(2)
2.133
2.026
1.980(2)
2.037
2.094
1.900(2)
2.095
2.059
1.970(2)
2.131
2.064
1.994(1)
2.189
2.118(2)
2.068
2.215(2)
2.187
2.054(2)
2.176
1.999(2)
1.986
1.920(2)
2.148
2.008(2)
2.111
2.121(1)
2.108
M-Npy (Å)
2.020(2)
2.160
2.153(2)
2.306
2.068(2)
2.190
2.001(2)
1.987
1.935(2)
2.149
2.058(2)
2.131
2.077(1)
2.205
2.111(2)
2.308
2.169(2)
2.268(2)
2.184
2.126(1)
2.086(2)
2.139
2.077(2)
2.007(2)
2.173
2.060(1)
1.941(2)
2.138
2.115(1)
2.067(2)
2.226
2.138(1)
2.199(1)
2.150
2.076(1)
M-O (Å)
∑ M-L (Å)
C-Nim (Å)
10.77
10.41
11.06
10.83
10.74
10.30
10.21
10.05
10.62
9.81
10.66
10.24
10.72
10.47
1.286
1.286(3)
1.346
1.282
1.276(2)
1.337
1.303
1.295(4)
1.314
1.315
1.311(3)
1.319
1.300
1.314(2)
1.300
1.285
1.288(3)
1.299
1.285
1.276(2)
1.333
1.349(3)
1.408
1.343(2)
1.420
1.314(3)
1.436
1.319(3)
1.424
1.324(2)
1.447
1.304(3)
1.449
1.351(2)
1.423
C-C (Å)
1.402(3)
1.462
1.415(3)
1.469
1.424(4)
1.445
1.421(3)
1.428
1.424(3)
1.447
1.445(3)
1.461
1.415(2)
1.463
1.464(3)
32.2
1.472(3)
45.2
1.430(4)
49.0
1.428(3)
40.1
1.427(3)
52.5
1.457(3)
39.5
1.470(2)
42.0
θ (deg)
45.4
50.9
52.8
50.3
46.5
48.9
47.4
b
J (cm-1
)
-637
0.50
-330
0.37
-436
0.49
-1370
0.43
-794
0.67
-
-
-
-
Sb
metal spin density
ligand spin density
+3.96
-0.96
+4.80
-0.78
+3.71
-0.71
+2.06
-1.06
+2.50
-0.50
+1.26
-0.26
+0.01
-0.99
a Experimental values are given in italics with esd in parentheses. b Coupling constant J or spatial overlap S between the magnetic pair of metal and
ligand spin vectors.
CH2Cl2 (dielectric constant, ꢀ ) 9.08), THF (ꢀ ) 7.25), pyridine
(ꢀ ) 12.5), MeOH (ꢀ ) 32.63), and H2O (ꢀ ) 80.4). Not
surprisingly, the calculated bond distances in the two ligand
backbones become more inequivalent with increasing dielectric
constants (Figure 21). Similarly, the extent of localization of
the ligand-based radical correlates with the dielectric constant
as seen in their corresponding spin-density distribution plots
(SI Figure 16).
Encouraged by these results, we used the COSMO approach
to model the solvent THF in the DFT calculations of the cationic
series. In the instance of Co 4ox‚THF, the predicted localization
of the ligand radical was contrary to reality, and thus, for this
Figure 21. Bond distances (Å) in the ligand backbone versus dielectric
complex only, we used the DFT-calculated model without any
solvent considerations. Overall, good agreement between the
experimental and calculated bond parameters for the ligands
constants for the COSMO DFT-calculated models of Zn 6ox‚THF. The two
independent ligands are denoted by either a black solid line or a red dashed
line.
was obtained (mostly within 0.01 Å for the selected bond
distances shown in Table 9). The calculated and experimental
(67) Sinnecker, S.; Rajendran, A.; Klamt, A.; Diedenhofen, M.; Neese, F. J.
Phys. Chem. A 2006, 110, 2235.
9
3194 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
Figure 23. Spin density plot of [(Lx)2Mn(THF)]+ 2ox‚THF shown with
values based on a Mulliken spin population analysis. Only atoms with spin
densities g 0.10 are labeled.
essentially uncoupled in these distorted tetrahedral systems due
to the nearly orthogonal arrangement of the two ligand π* sys-
tems. This is contrary to square planar systems, wherein ligand
radicals are known to couple strongly with one another through
an Smetal ) 0 metal center.38 The weakness of the ligand-ligand
interaction is evident in Zn 6, where the singlet and triplet states
for the two ligand radicals are essentially degenerate. In the
remaining neutral complexes, the ligand radicals remain parallel
with respect to one another and can be treated as a combined
ligand spin vector with SL ) 1. However, this is a consequence
not of ligand-ligand ferromagnetic coupling but of a strong
metal-ligand antiferromagnetic interaction.
Of central interest is the unraveling of the electronic structures
of the mixed-valent complexes, 1ox‚THF-6ox. Collectively, they
comprise a unique series in that the localization or delocalization
of a ligand radical is determined entirely by the nature of the
bridging metal center. The ligand radical is localized (Class I)
for Zn(II), Cr(II), and Mn(II) and fully delocalized (Class III)
for the later metals, Fe(II) and Co(II). The diversity of electronic
structures in this series is surprising given that all the complexes
are essentially isostructural in crystals grown from THF. We
also emphasize that electronic communication between the two
ligand sites is symmetry-allowed because the observed dihedral
angles (45° to 51°) do not approach 90°. Hence, the ligand
π*-systems can interact via d-orbitals that are not orthogonal
to one another.
We do not yet fully understand the origins of the mechanisms
of delocalization in Fe 3ox‚THF and Co 4ox‚THF versus Cr 1ox‚
THF and Mn 2ox‚THF. One enticing possibility is that delocal-
ization arises from a greater orbital overlap between the ligands’
π*-systems and the metal d-orbital(s) of appropriate symmetry.
In other words, metal-ligand π covalency increases across the
period from Cr to Ni. The growing metal-ligand covalency
might arise from the shortening of the metal-ligand bond
distances down the period. The ionic radii of divalent metal
ions are known to increase from Cr to Mn (due to increased
shielding provided by five d-electrons) but then decrease
continually from Mn to Ni.68 Indeed, the metal-ligand bond
distances in the solid-state structures and the DFT models reflect
this trend. Only the cationic Ni complex 5ox‚THF breaks the
Figure 22. Qualitative MO diagram of the magnetic orbitals derived from
BS(5,1) calculation of [(Lx)2Mn(THF)]+ 2ox‚THF. The spatial overlap (S)
of the corresponding alpha and beta orbitals is given.
dihedral angles differ by less than 12°. Again, the only poor
agreements are those concerning metal-ligand bond lengths.65
2.7.4. Electronic Structures of the Cationic [(Lx)2M-
(THF)]+ Species. The d-orbital splitting scheme is basically
consistent from Mn 2ox‚THF to Ni 5ox‚THF, and a representative
scheme (for Mn 2ox‚THF) is shown in Figure 22. (The MO
diagram for Cr 1ox‚THF is different in this series and is shown
in SI Figure 17.) The calculated structures are depicted viewing
down the Mn-O bond. The d-orbital splitting pattern is
approximately “one-three-one,” wherein three slightly antibond-
ing orbitals lie between a relatively nonbonding d-orbital and a
strongly antibonding d-orbital. In the calculated cationic Mn
species, there are five occupied metal-based SOMOs in the spin-
up manifold and a ligand-based SOMO in the spin-down
manifold, which fits the electronic picture of a divalent, high-
spin Mn center with one ligand radical.
Based on the spin-density maps (Mn 2ox‚THF: Figure 23,
SI Figure 18), a ligand radical is present in the complexes of
Cr, Mn, both spin states of Fe, and Zn. No radical is predicted
in the case of Ni, and for Co, the ligands are reduced by only
half an electron. As in the neutral series, the spin densities
gradually decrease from the expected values for a high-spin,
divalent metal center with one ligand radical toward those
expected for a monovalent metal and no ligand radicals (Table
9). The spin-spin coupling constant J and spatial overlap S
decrease from Cr to Mn but then continually increase across
the period from Mn to Co. As in the neutral series, we invoke
an increasing metal-ligand π covalency to explain these
observed theoretical trends (Table 8).
3. Summary and Concluding Remarks
A series of bis(R-iminopyridine)metal complexes of the first-
row transition ions has been synthesized and characterized. The
neutral series is formulated as M(L•)2, comprising a divalent
metal center and two ligand radicals. The ligand radicals are
(68) Cotton, F. A.; Wilkinson, G.; Gaus, P. L. Basic Inorganic Chemistry, 3rd
ed.; John Wiley & Sons, Inc.: Chicester, 1995; p 538.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3195

A R T I C L E S
Lu et al.
of sodium metal (0.0501 g, 2.179 mmol), and 2 equiv of ligand (0.5663
g, 2.126 mmol). A dark-brown powder was obtained. Single crystals
of X-ray quality can be grown from vapor diffusion of pentane into a
benzene solution. Yield: 0.506 g (81%). Anal. Calcd for C36H44FeN4:
C, 73.46; H, 7.53; N, 9.52. Found: C, 73.35; H, 7.46; N, 9.39.
[(L•)2Co] (4). A similar procedure for the synthesis of 1 was used
to prepare this compound from CoCl2 (0.1479 g, 1.139 mmol), 2.05
equiv of sodium metal (0.0537 g, 2.336 mmol), and 2 equiv of ligand
(0.6070 g, 2.279 mmol). A dark maroon-brown powder was obtained.
Single crystals of X-ray quality can be grown from vapor diffusion of
pentane into a benzene solution. Yield: 0.513 g (76%). Anal. Calcd
for C36H44CoN4: C, 73.08; H, 7.50; N, 9.47. Found: C, 73.18; H, 7.48;
N, 9.38.
expected trend by showing a small increase in the metal-ligand
bond distances relative to Co, but recall that the Ni center is no
longer divalent.
The experimental and theoretical data for both zinc complexes
6 and 6ox‚THF consistently reveal that they bear a greater
resemblance to the earlier metals Cr and Mn than the later
metals. A simple explanation is that the Zn(II) center in these
compounds is relatively redox inert and, thus, cannot partake
in metal-ligand covalency. Consequently, the Zn complexes 6
and 6ox‚THF are the standards in their respective series as classic
Werner coordination complexes.
Indeed, the experimental data and the complimentary DFT
results both point to a periodic trend of metal-ligand covalency.
The trend matches those observed in the DFT calculations for
the neutral series, wherein the coupling constant J and spatial
overlap S between the metal-based and ligand-based SOMOs
increase going down the period. More importantly, the experi-
mental data are congruent with this hypothesis, including the
increasingly large potential separations between the two redox
events observed in the cyclic voltammograms. Also consistent
is the presence of LLIVCT bands in the electronic absorption
spectra of the later cationic species, Fe 3ox‚THF and Co 4ox‚
THF. Finally, we note that metal-ligand covalency makes
assignment of formal oxidation states quite inappropriate for
the later metals, and the calculated MOs indicate that this is
especially true for the neutral nickel compound 5.
[(L•)2Ni] (5). The ligand (2 equiv, 95 mg, 35.7 mmol) and Ni(COD)2
(50 mg, 17.8 mmol) were added to a glass vessel, and n-pentane (10
mL) was added. After stirring for 18 h, the dark royal-purple solution
was evaporated in Vacuo, washed with CH3CN (5 mL), and extracted
with benzene (15 mL). After removing the solvent under reduced
pressure, a black-purple powder was obtained. Single crystals of X-ray
quality were grown from the CH3CN filtrate at -20 °C. Yield: 0.102
1
g (97%). By H NMR spectroscopy, two isomers are resolved at -60
1
°C in an exact ratio of 1.3 to 1. H NMR (400 MHz, toluene-d8, 213
K): δ ) 10.16 (1.3H, imine CH), 10.03 (d, 1.3H, J ) 5.4 Hz, pyridine
CH), 9.42 (1H, imine CH), 9.02 (d, 1H, J ) 5.4 Hz, pyridine CH),
7.81 (t, 1.3H, J ) 5.7 Hz, aryl H), 7.44 (t, 1.3H, J ) 7.5 Hz, aryl H),
7.31 (t, 1H, J ) 7.5 Hz, aryl H), 7.24 (t, 1H, J ) 5.7 Hz, aryl H), 6.92
(d, 1.3H, J ) 8.0 Hz), 6.74 (d, 1H, J ) 8.3 Hz), 5.26 (quin, 1.3H, J )
6 Hz, Me2CH), 5.10 (br s, 1H, Me2CH), 2.42 (m, 2.3H, two overlapping
Me2CH), 1.34 (d, 3.9H, J ) 6 Hz, CH3), 1.27 (br s, 3H, CH3), 1.20 (br
s, 3H, CH3), 1.05 (m, 6.9H, overlapping CH3), 0.94 (d, 3.9H, J ) 5.6
Hz, CH3), 0.85 (br s, 3H, CH3), 0.63 (d, 3.9H, J ) 5 Hz, CH3). Some
aryl resonances are obscured by the residual peaks of toluene-d8. Anal.
Calcd for C36H44NiN4: C, 73.11; H, 7.50; N, 9.47. Found: C, 73.00;
H, 7.39; N, 9.36.
[(L•)2Zn] (6). A similar procedure for the synthesis of 1 was used
to prepare this compound from ZnCl2 (0.1689 g, 1.239 mmol), 2.05
equiv of sodium metal (0.0584 g, 2.540 mmol), and 2 equiv of ligand
(0.6602 g, 2.478 mmol). A dark forest-green powder was obtained.
Single crystals of X-ray quality can be grown from vapor diffusion of
pentane into a benzene solution. Yield: 0.710 g (96%). Anal. Calcd
for C36H44ZnN4: C, 72.29; H, 7.41; N, 9.37. Found: C, 72.06; H, 7.28;
N, 9.25.
4. Experimental Section
All syntheses were carried out using standard glovebox and Schlenk
techniques in the absence of water and dioxygen, unless otherwise
noted. Dry solvents were purchased from Fluka and degassed prior to
use. Benzene-d6 and THF-d8 were purchased from Cambridge Isotope
Laboratories, Inc., degassed via repeated freeze-pump-thaw cycles
and dried over 3-Å molecular sieves. Sodium metal was purchased from
Aldrich and washed with hexanes prior to use. The reagents CrCl2,
MnCl2, FeCl2, CoCl2, Ni(COD)2, and ZnCl2 were purchased from Strem
or Aldrich and used without further purification. The following reagents
were prepared as described in the literature: the ligand, 2,6-bis(1-
methylethyl)-N-(2-pyridinylmethylene)phenylamine69 (abbreviated as
L), and [Cp2Fe][B(ArF)4],70 where ArF ) 3,5-(CF3)2C6H3.
[(L•)2Cr] (1). CrCl2 (0.1335 g, 1.086 mmol), 2.05 equiv of sodium
metal (0.0512 g, 2.227 mmol), and 2 equiv of ligand (0.5788 g, 2.173
mmol) were added to a glass vessel. DME (10 mL) was then added,
and the reaction was stirred vigorously. After 8 h, the dark-brown
solution was evaporated in Vacuo. The residue was washed with pentane
(3 × 3 mL), extracted with benzene (15 mL), and filtered through Celite.
After removing the solvent in Vacuo, a black-brown powder was
obtained. Single crystals of X-ray quality can be grown from Et2O at
-20 °C. Yield: 0.596 g (94%). Anal. Calcd for C76H98Cr2N8O2: C,
72.47; H, 7.84; N, 8.90. Found: C, 72.01; H, 7.53; N, 8.59.
[(L•)(L)Cr(THF)][B(ArF)4] (1ox‚THF). Compound 1 (50.7 mg, 86.7
µmol), [Cp2Fe][B(ArF)4] (91.0 mg, 86.7 µmol), and THF (5 mL) were
added to a glass vessel. After stirring overnight, a dark green solution
is obtained. The solvent was removed under reduced pressure, and the
resultant residue was washed liberally with hexanes. After extraction
with THF (8 mL), filtration through Celite, and evaporation of the
solvent in Vacuo, a dark blue-green powder was obtained. Single crystals
of X-ray quality can be grown from vapor diffusion of pentane into a
THF solution. Yield: 126 mg (95%). Anal. Calcd for C68H56BF24N4Cr
(without THF): C, 56.41; H, 3.90; N, 3.87. Found: C, 56.18; H, 4.12;
N, 3.78.
[(L•)(L)Mn(THF)][B(ArF)4] (2ox‚THF). A similar procedure for the
synthesis of 1ox‚THF was used to prepare this compound from 2 (50.5
mg, 85.9 µmol) and [Cp2Fe][B(ArF)4] (90.2 mg, 85.9 µmol). A dark
brown powder was obtained. Single crystals of X-ray quality can be
grown from vapor diffusion of pentane into a THF solution. Yield:
128 mg (98%). Anal. Calcd for C68H56BF24N4Mn (without THF): C,
56.29; H, 3.89; N, 3.86. Found: C, 56.31; H, 4.04; N, 3.80.
[(L•)2Mn] (2). A similar procedure for the synthesis of 1 was used
to prepare this compound from MnCl2 (0.1383 g, 1.099 mmol), 2.05
equiv of sodium metal (0.0518 g, 2.253 mmol), and 2 equiv of ligand
(0.5855 g, 2.198 mmol). A dark yellow-brown powder was obtained.
Single crystals of X-ray quality can be grown from vapor diffusion of
pentane into a benzene solution. Yield: 0.574 g (89%). Anal. Calcd
for C36H44MnN4: C, 73.57; H, 7.55; N, 9.53. Found: C, 73.42; H,
7.51; N, 9.40.
[(L•)2Fe] (3). A similar procedure for the synthesis of 1 was used to
prepare this compound from FeCl2 (0.1375 g, 1.063 mmol), 2.05 equiv
[(L•)(L)Fe(THF)][B(ArF)4] (3ox‚THF). A similar procedure for the
synthesis of 1ox‚THF was used to prepare this compound from 3 (58.6
mg, 99.5 µmol) and [Cp2Fe][B(ArF)4] (104.4 mg, 99.5 µmol). A green-
black powder was obtained. Single crystals of X-ray quality can be
grown from vapor diffusion of pentane into a THF solution. Yield:
(69) Laine, T. V.; Klinga, M.; Leskela¨, M. Eur. J. Inorg. Chem. 1999, 959.
(70) Cha´vez, I.; Alvarez-Carena, A.; Molins, E.; Roig, A.; Maniukiewicz, W.;
Arancibia, A.; Arancibia, V.; Brand, H.; Manr´ıquez, J. M. J. Organomet.
Chem. 2000, 601, 126.
9
3196 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Mixed Valency Involving a Redox Noninnocent Ligand System
A R T I C L E S
150 mg (98%). Anal. Calcd for C68H56BF24N4Fe (without THF): C,
56.26; H, 3.89; N, 3.86. Found: C, 56.38; H, 4.03; N, 3.77.
NMR spectra were recorded on Varian Mercury 400 MHz instruments
at ambient temperature. Mo¨ssbauer data were recorded on an alternating
constant-acceleration spectrometer. The minimum experimental line
width was 0.24 mm s-1 (full width at half-height). The sample
temperature was maintained constant in an Oxford Instruments Variox
or an Oxford Instruments Mo¨ssbauer-Spectromag 2000 cryostat, which
is a split-pair superconducting magnet system for applied fields (up to
8 T). The field at the sample is oriented perpendicular to the γ-beam.
The 57Co/Rh source (1.8 GBq) was positioned at room temperature
inside the gap of the magnet system at a zero-field position. Isomer
shifts are quoted relative to iron metal at 300 K.
Calculations. All calculations were done with the ORCA program
package.74 The geometry optimizations were carried out at the B3LYP
level75-77 of DFT. The all-electron Gaussian basis sets used were those
reported by the Ahlrichs group.78,79 For metal, nitrogen, and oxygen
atoms, the triple-ú-quality basis sets with one set of polarization
functions were used (TZVP).78 The carbon and hydrogen atoms were
described by smaller polarized split-valence SV(P) basis sets (double-
ú-quality in the valence region with a polarizing set of d-functions on
the non-hydrogen atoms).79 The SCF calculations were tightly con-
verged (1 × 10-8 Eh in energy, 1 × 10-7 Eh in the density change,
and 1 × 10-7 in the maximum element of the DIIS error vector). The
geometries were considered converged after the energy change was
less than 5 × 10-6 Eh, the gradient norm and maximum gradient
element were smaller than 1 × 10-4 and 3 × 10-4 Eh/Bohr, respectively,
and the root-mean-square and maximum displacements of the atoms
were smaller than 2 × 10-3 and 4 × 10-3 Bohr, respectively. In certain
calculations (specified in the text), solvation effects were included using
the conductor-like screening model (COSMO).67 In these cases, the
solvent THF was modeled with a dielectric constant of 7.25 and a
refractive index of 1.407. The MO pictures were generated using
corresponding orbitals and depicted using the program Molekel (version
4.3).
[(L•)(L)Co(THF)][B(ArF)4] (4ox‚THF). A similar procedure for the
synthesis of 1ox‚THF was used to prepare this compound from 4 (50.4
mg, 85.2 µmol) and [Cp2Fe][B(ArF)4] (89.4 mg, 85.2 µmol). A dark
brown powder was obtained. Single crystals of X-ray quality can be
grown from vapor diffusion of pentane into a THF solution. Yield:
128 mg (98%). Anal. Calcd for C68H56BF24N4Co (without THF): C,
56.14; H, 3.88; N, 3.85. Found: C, 56.11; H, 3.95; N, 3.80.
[(L)2Ni][B(ArF)4] (5ox). A similar procedure for the synthesis of 1ox‚
THF was used to prepare this compound from 5 (50.1 mg, 84.7 µmol)
and [Cp2Fe][B(ArF)4] (88.9 mg, 84.7 µmol). A dark purple-brown
powder was obtained. Single crystals of X-ray quality can be grown
from vapor diffusion of pentane into a THF solution. A THF molecule
is bound to the nickel center in the solid-state structure, but the THF
is labile and can be removed by drying the sample in Vacuo for 2 h.
Yield: 120 mg (98%). Anal. Calcd for C68H56BF24N4Ni: C, 56.15; H,
3.88; N, 3.85. Found: C, 56.29; H, 3.95; N, 3.72.
[(L•)(L)Zn][B(ArF)4] (6ox). A similar procedure for the synthesis
of 1ox‚THF was used to prepare this compound from 6 (51.0 mg, 85.3
µmol) and [Cp2Fe][B(ArF)4] (89.5 mg, 85.3 µmol). A dark forest-green
powder was obtained. Single crystals of X-ray quality can be grown
from vapor diffusion of pentane into a THF solution. A THF molecule
is bound to the zinc center in the solid-state structure, but the THF is
labile and can be removed by drying the sample in Vacuo for 2 h.
Yield: 120 mg (96%). Anal. Calcd for C68H56BF24N4Zn: C, 55.89; H,
3.86; N, 3.83. Found: C, 56.02; H, 4.00; N, 3.74.
X-ray Crystallographic Data Collection and Refinement of the
Structures. Single crystals of 1-6 and 1ox‚THF-6ox were coated with
perfluoropolyether, picked up with nylon loops, and were mounted in
the nitrogen cold stream of the diffractometer. A Bruker-Nonius Kappa-
CCD diffractometer equipped with a Mo-target rotating-anode X-ray
source and a graphite monochromator (Mo KR, λ ) 0.710 73 Å) was
used. Final cell constants were obtained from least-squares fits of all
measured reflections. The structures were readily solved by direct
methods and subsequent difference Fourier techniques. The Siemens
ShelXTL71 software package was used for solution and artwork of the
structure, and ShelXL9772 was used for the refinement. All non-
hydrogen atoms were refined anisotropically. Hydrogen atoms were
placed at calculated positions and refined as riding atoms with isotropic
displacement parameters. Crystallographic data of the compounds are
listed in Tables 2 and 5.
Physical Measurements. Electronic spectra of complexes were
recorded with a Perkin-Elmer double-beam photometer (300 to 2000
nm). Cyclic voltammograms were recorded with an EG&G potentiostat/
galvanostat. Variable-field (0.01 or 1 T), variable temperature (4 to
300 K) magnetization data were recorded on a SQUID magnetometer
(MPMS Quantum Design). The experimental magnetic susceptibility
data were corrected for underlying diamagnetism using tabulated
Pascal’s constants. X-band EPR spectra were recorded on a Bruker
ESP 300 spectrometer and simulated with the XSophe program, which
is written by Hanson et al.73 and distributed by Bruker Biospin GmbH.
Acknowledgment. The authors thank Prof. Frank Neese
(University of Bonn) and Dr. Marat Khusniyarov for insightful
discussions and Heike Schucht, Bernd Mienert, Andreas Go¨bels,
Frank Reikowski, and Petra Ho¨fer for technical assistance.
C.C.L. is funded by the Alexander von Humboldt Foundation.
Supporting Information Available: Additional spectroscopic
and theoretical data, including the 1H NMR spectrum of 5, solid-
state structures, EPR spectra, and DFT-calculated spin-density
distributions. This material is available free of charge via the
Internet at http://pubs.acs.org.
JA710663N
(73) Hanson, G. R.; Gates, K. E.; Noble, C. J.; Griffin, M.; Mitchell, A.; Benson,
S. J. Inorg. Biochem. 2004, 98, 903.
(74) Neese, F. ORCA, an Ab Initio, Density Functional and Semiempirical
Electronic Structure Program Package, version 2.6, revision 4 ed.; Max-
Planck-Institut fu¨r Bioanorganische Chemie: Mu¨lheim/Ruhr, Germany,
May 2005.
(75) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(76) Becke, A. D. J. Chem. Phys. 1986, 84, 4524.
(77) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(78) Scha¨fer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829.
(79) Scha¨fer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571.
(71) SHELXTL, version 5; Siemens Analytical X-ray Instruments Inc.: Madison,
WI, 1994.
(72) Sheldrick, G. M. SHELXL97; University of Go¨ttingen: Go¨ttingen, Germany,
1997.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3197