Quantifying the Electron Donor and Acceptor Abilities of the Ketimide Ligands in M(N=CtBu2)4 (M = V, Nb, Ta)

Article abstract of DOI:10.1021/acs.inorgchem.5b02017

Addition of 4 equiv of Li(N=C^tBu₂) to VCl₃ in THF, followed by addition of 0.5 equiv of I₂, generates the homoleptic V(IV) ketimide complex, V(N=C^tBu₂)₄ (1), in 42% yield. Similarly, reaction of 4 equiv of Li(N=C^tBu₂) with NbCl₄(THF)₂ in THF affords the homoleptic Nb(IV) ketimide complex, Nb(N=C^tBu₂)₄ (2), in 55% yield. Seeking to extend the series to the tantalum congener, a new Ta(IV) starting material, TaCl₄(TMEDA) (3), was prepared via reduction of TaCl₅ with Et₃SiH, followed by addition of TMEDA. Reaction of 3 with 4 equiv of Li(N=C^tBu₂) in THF results in the isolation of a Ta(V) ketimide complex, Ta(Cl)(N=C^tBu₂)₄ (5), which can be isolated in 32% yield. Reaction of 5 with Tl(OTf) yields Ta(OTf)(N=C^tBu₂)₄ (6) in 44% yield. Subsequent reduction of 6 with Cp₂Co in toluene generates the homoleptic Ta(IV) congener Ta(N=C^tBu₂)₄ (7), although the yields are poor. All three homoleptic group 5 ketimide complexes exhibit squashed tetrahedral geometries in the solid state, as determined by X-ray crystallography. This geometry leads to a dx2-y2¹ (²B₁ in D_2d) ground state, as supported by DFT calculations. EPR spectroscopic analysis of 1 and 2, performed at X- and Q-band frequencies (～9 and 35 GHz, respectively), further supports the ²B₁ ground-state assignment, whereas comparison of 1, 2, and 7 with related group 5 tetra(aryl), tetra(amido), and tetra(alkoxo) complexes shows a higher M-L covalency in the ketimide-metal interaction. In addition, a ligand field analysis of 1 and 2 demonstrates that the ketimide ligand is both a strong π-donor and strong π-acceptor, an unusual combination found in very few organometallic ligands.

Full text of DOI:10.1021/acs.inorgchem.5b02017

Article
pubs.acs.org/IC
Quantifying the Electron Donor and Acceptor Abilities of the
Ketimide Ligands in M(NCtBu2)4 (M = V, Nb, Ta)
Peter L. Damon,† Cameron J. Liss,‡ Richard A. Lewis,† Simona Morochnik,† David E. Szpunar,‡
Joshua Telser,*,‡ and Trevor W. Hayton*,†
†Department of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106, United States
‡Department of Biological, Chemical and Physical Sciences, Roosevelt University, 430 South Michigan Avenue, Chicago, Illinois
60605-1394, United States
S
* Supporting Information
ABSTRACT: Addition of 4 equiv of Li(NCtBu2) to VCl3
in THF, followed by addition of 0.5 equiv of I2, generates the
homoleptic V(IV) ketimide complex, V(NCtBu2)4 (1), in
42% yield. Similarly, reaction of 4 equiv of Li(NCtBu2) with
NbCl4(THF)2 in THF affords the homoleptic Nb(IV)
ketimide complex, Nb(NCtBu2)4 (2), in 55% yield. Seeking
to extend the series to the tantalum congener, a new Ta(IV)
starting material, TaCl4(TMEDA) (3), was prepared via
reduction of TaCl5 with Et3SiH, followed by addition of
TMEDA. Reaction of 3 with 4 equiv of Li(NCtBu2) in THF results in the isolation of a Ta(V) ketimide complex, Ta(Cl)(N
CtBu2)4 (5), which can be isolated in 32% yield. Reaction of 5 with Tl(OTf) yields Ta(OTf)(NCtBu2)4 (6) in 44% yield.
Subsequent reduction of 6 with Cp*2Co in toluene generates the homoleptic Ta(IV) congener Ta(NCtBu2)4 (7), although the
yields are poor. All three homoleptic group 5 ketimide complexes exhibit squashed tetrahedral geometries in the solid state, as
21
determined by X-ray crystallography. This geometry leads to a dx −y (2B1 in D2d) ground state, as supported by DFT
2
calculations. EPR spectroscopic analysis of 1 and 2, performed at X- and Q-band frequencies (∼9 and 35 GHz, respectively),
2
further supports the B1 ground-state assignment, whereas comparison of 1, 2, and 7 with related group 5 tetra(aryl),
tetra(amido), and tetra(alkoxo) complexes shows a higher M−L covalency in the ketimide−metal interaction. In addition, a
ligand field analysis of 1 and 2 demonstrates that the ketimide ligand is both a strong π-donor and strong π-acceptor, an unusual
combination found in very few organometallic ligands.
orthogonal to the aforementioned nitrogen lone pair. More-
over, an energy decomposition analysis calculation on Co(N
CtBu2)4 suggests that the Co−N π-back-donation interaction
accounts for a sizable (ca. 25%) portion of the total Co−N
bonding energy in this complex.6 This finding was somewhat
unanticipated, yet it is significant because it suggests that the
combined donor/acceptor properties of the ketimide ligand
could engender unique structure and reactivity.
To better understand the electronic properties of the
ketimide ligand, we endeavored to synthesize the tetrakis-
(ketimide) group 5 complexes, M(NCtBu2)4 (M = V, Nb,
Ta), as the d1 electronic configuration within these complexes
would render them amenable to characterization by a wide
variety of spectroscopic techniques, including UV−vis and EPR
spectroscopies, which could shed further light on the π-bonding
and π-accepting properties of this class of ligand. Herein, we
describe the synthesis and comprehensive spectroscopic
characterization of M(NCtBu2)4 (M = V, Nb, Ta). Most
significantly, this study provides experimental confirmation that
the ketimide ligand is, in fact, a good π-acceptor, knowledge
INTRODUCTION
■
The ketimide ligand, [NCR2]−, has proven to be adept at
stabilizing high oxidation states, both for transition metal ions
and also the actinides.1−7 For example, in 2010, we reported
the synthesis of an isolable Fe(IV) ketimide complex, Fe(N
CtBu2)4,3 a rare example of an MX4-type complex of the late
first-row transition metals.8 Our group also synthesized the
homoleptic Mn(IV) and Co(IV) ketimides, M(NCtBu2)4 (M
= Mn, Co),4,6 whereas Hoffman and co-workers reported the
isolation of Cr(NCtBu2)4.9 In addition, Kiplinger and co-
workers reported the formation of a rare U(V) organometallic,
Cp*2U(NDipp)(NCPh2) (Dipp = 2,6-iPr2C6H3), stabilized
by inclusion of the ketimide ligand.2 The ability of the ketimide
ligand to stabilize high oxidation states likely stems from the
orbitals available for the metal−ketimide interaction. DFT
calculations on Cr(NCtBu2)4 and Co(NCtBu2)4 reveal
that the ketimide ligand is a strong σ- and π-donor and, as such,
should be adept at satisfying the 4+ charge of an M(IV) ion.
Interestingly, however, the ketimide ligand also appears to be a
good π-acceptor, at least according to DFT calculations.6,7,9
The unusual combination of π-donor and π-acceptor abilities is
made possible by the presence of a π-donating nitrogen lone
pair and a π-accepting CN antibonding orbital, which is
Received: September 2, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Scheme 1
Scheme 2
that is essential for the further development of the ketimide
moiety for use as a coligand in metal-centered catalysis. In
addition, we describe the synthesis of TaCl4(TMEDA), a
potentially useful synthon for Ta(IV) chemistry.
a convenient Ta(IV) synthon, we explored the reduction of
TaCl5 with a readily available silane, Et3SiH. Reaction of TaCl5
with 1 equiv of HSiEt3 in toluene, followed by addition of
TMEDA, affords TaCl4(TMEDA) (3), which can be isolated as
an orange crystalline solid in 87% yield after workup (Scheme
2). The synthesis of 3 mirrors that of TaCl3 from TaCl5 and
BTCD (BTCD = 3,6-bis(trimethylsilyl)-1,4-cyclohexadiene),
which also generates R3SiCl as a byproduct.13 Complex 3 is
soluble in CH2Cl2, THF, and acetonitrile, but it is insoluble in
hexanes and Et2O and only sparingly soluble in toluene. Its 1H
NMR spectrum in CD2Cl2 exhibits two broad singlets, at 2.72
and 8.76 ppm, in a 3:1 ratio. These are assignable to the methyl
and methylene resonances of the TMEDA moiety, respectively,
and are shifted significantly from those observed for free
TMEDA, consistent with coordination of TMEDA to a
paramagnetic metal center. Complex 3 was also characterized
by X-ray crystallography (see Supporting Information for full
details). Interestingly, the reduction of TaCl5 with Et3SiH also
affords a second, minor product, which can be isolated as a pale
blue, CH2Cl2-insoluble powder in very low yield. This material
was subsequently identified as [{(TMEDA)TaCl2(μ-Cl)}2]-
[TaCl6] (4) by X-ray crystallography (see Supporting
Information for more details).
RESULTS AND DISCUSSION
■
Synthesis. Addition of 4 equiv of Li(NCtBu2) to VCl3, in
THF, affords a deep green solution. Subsequent addition of 0.5
equiv of I2 to this solution yields the homoleptic V(IV)
ketimide complex, V(NCtBu2)4 (1), in moderate yields, after
workup (Scheme 1). Its niobium congener is accessible via a
similar metathetical protocol. Thus, addition of 4 equiv of
Li(NCtBu2) to NbCl4(THF)2,10 in THF, affords a dark
brown solution, which, after workup, yields the analogous
homoleptic Nb(IV) ketimide complex, Nb(NCtBu2)4 (2), in
moderate yields (Scheme 1). As a solid, complex 1 is dark
orange-brown; however, it is dark green in solution. Complex 2
is dark brown, both in the solid state and in solution. Both
complexes are very soluble in nonpolar solvents, such as
hexanes, toluene, and Et2O. They are also very soluble in THF
but are insoluble in acetonitrile. Complexes 1 and 2 decompose
in the presence of CH2Cl2.
1
The H NMR spectrum of 1 in C6D6 exhibits a broad
resonance centered at 3.84 ppm (fwhm = 900 Hz), which is
assignable to the tert-butyl protons of the ketimide ligand.
Similarly, the 1H NMR spectrum of 2 in C6D6 exhibits a broad
resonance centered at 6.80 ppm (fwhm = 850 Hz), which is
assignable to the tert-butyl protons of the coordinated ketimide
ligand. Interestingly, unlike previously reported transition metal
ketimide complexes from our group,3,4,6,7 complex 1 is
thermally stable: thermolysis of a C6D6 solution of 1 at 70
°C for 2 h does not result in any signs of decomposition
(Figure S2).
We next sought to synthesize the tantalum congener to
complexes 1 and 2; however, unlike Nb, an easily prepared
Ta(IV) starting material has yet to be reported. For example,
TaCl4(THF)2 is unknown, and although TaCl4 is known, its
preparation is not straightforward.11,12 In an effort to synthesize
Addition of 4 equiv of Li(NCtBu2) to 3, in THF, affords a
dark yellow-brown solution. Extraction into hexanes, filtration,
and crystallization affords the Ta(V) ketimide, Ta(Cl)-
(NCtBu2)4 (5), as yellow blocks in 32% yield (Scheme 2).
The isolation of a Ta(V) product from this reaction suggests
that addition of Li(NCtBu2) to 3 results in Ta(IV)
disproportionation. However, efforts to identify and isolate
the Ta(III) byproduct from the reaction have proven to be
1
unsuccessful. The H NMR spectrum of 5 in C6D6 exhibits a
sharp singlet at 1.36 ppm, assignable to the tert-butyl protons of
the ketimide ligand. Only one ketimide environment is
observed, suggestive of rapid exchange between equatorial
and axial ketimide ligand environments about the Ta5+ center
(see below).
B
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Curiously, a direct synthetic route to complex 5 from TaCl5
has not been forthcoming. For example, reaction of TaCl5 with
4 equiv of Li(NCtBu2), in THF or DME, results in the
1
formation of complex reaction mixtures, as determined by H
NMR spectroscopy. Complex 5 is present in the reaction
mixture; however, it is but one of many components and is
present only in low yield. To rationalize these results, we note
that TaX5 (X = Cl, Br, I) readily reacts with a variety of ethereal
solvents, including DME and THF, to form alkoxides.14,15
Thus, it seems likely that the in situ formation of a tantalum
alkoxide during the reaction with Li(NCtBu2) in THF
inhibits the desired salt metathesis.
Complex 5 crystallizes in the monoclinic space group P21/c
with two independent molecules in the asymmetric unit. The
solid-state molecular structure of one independent molecule is
shown in Figure 1. This complex exhibits a distorted square
pyramidal structure, as determined by continuous shape
measure (CSM = 1.14) developed by Alvarez and co-workers,16
wherein the chloride ligand and three ketimide ligands occupy
the equatorial positions, and one ketimide ligand (N3) occupies
the axial position. The Ta−N bond lengths in 5 range from
1.918(4) to 1.990(5) Å and are similar to those observed in
other tantalum ketimides.17,18 In addition, the average Ta−N−
C angle (172.7°) is indicative of sp hybridization at nitrogen
and is suggestive of significant π-donation from the ketimide to
the metal, consistent with our ligand field analysis (see
below).3,4,6,19 Finally, the Ta−Cl bond lengths (2.498(1),
2.496(1) Å) are similar to those observed for other structurally
characterized Ta(V) chlorides.14,15,20−22
Complex 5 proved to be amenable to further synthetic
manipulation. For example, reaction of 5 with 1 equiv of
Tl(OTf) in hexanes resulted in a color change to deep red.
Filtration, followed by crystallization from concentrated
hexanes, resulted in deposition of Ta(OTf)(NCtBu2)4 (6)
as red blocks in 44% yield (Scheme 3). Complex 6 crystallizes
in the P21/c space group, and, as observed for 5, it exhibits a
distorted square pyramidal geometry about the metal center
(Figure 1). Not surprisingly, the metrical parameters of 6 are
very similar to those of 5.
With complexes 5 and 6 in hand, we probed their utility as
precursors to the Ta(IV) ketimide complex, Ta(NCtBu2)4 (7).
However, neither 5 nor 6 proved to be very amenable to
chemical reduction. For example, reduction of either 5 or 6
with KC8 results in the observation of a mixture of products,
whereas reductions with sodium metal or Na/Hg amalgam did
not go to completion, even over long reaction times. In
contrast, the reaction of 6 with Cp*2Co (Cp* = pentam-
ethylcyclopentadienide) appeared to be much more promising.
Thus, addition of 1 equiv of Cp*2Co to a toluene-d8 solution of
6 results in the formation of a broad singlet at 7.42 ppm (fwhm
Figure 1. Solid-state molecular structures of Ta(Cl)(NCtBu2)4 (5)
(left) and Ta(OTf)(NCtBu2)4 (6) (right) with 50% probability
ellipsoids. Hydrogen atoms and a second molecule of 5 in the
asymmetric unit cell are omitted for clarity. Selected bond distances
(Å) and angles (deg) for 5: Ta1−N1 = 1.958(4), Ta1−N2 = 1.967(4),
Ta1−N3 = 1.922(4), Ta1−N4 = 1.982(4), Ta1−Cl1 = 2.498(1), N2−
Ta1−Cl1 = 79.0(1), N4−Ta1−Cl1 = 79.1(1), N1−Ta1−Cl1 =
156.4(1), N3−Ta1−Cl1 = 103.6(1). Selected bond distances (Å) and
angles (deg) for 6: Ta1−N4 = 1.921(2), Ta1−N2 = 1.931(2), Ta1−
N1 = 1.955(2), Ta1−N3 = 1.959(3), Ta1−O3 = 2.218(2), N4−Ta1−
O3 = 95.27(9), N3−Ta1−O3 = 79.99(9), N1−Ta1−O3 = 80.76(8),
N2−Ta1−O3 = 166.11(9).
1
= 860 Hz) in the H NMR spectrum, assignable to the tert-
butyl protons of Ta(NCtBu2)4 (7) (Scheme 3). Also present in
the sample are resonances assignable to [Cp*2Co][OTf].
Unfortunately, most attempts to isolate complex 7 proved to be
unsuccessful, in part because 7 appears to be quite temperature-
sensitive. However, in a few instances, a few crystals of 7 were
isolable, which permitted its characterization by X-ray
crystallography (see below).
orthorhombic space group Pnna and exhibits a squashed
tetrahedral geometry about the vanadium center, as evidenced
by the two largest N−V−N bond angles (N1−V1−N1* =
133.1(1)° and N2−V1−N2* = 132.9(1)°). This corresponds
to a τ4 value of 0.67, where a τ4 value of 1 indicates an idealized
tetrahedron and a τ4 value of 0 indicates an idealized square
X-ray Crystallographic Analysis of M(NCtBu2)4 (M = V,
Nb, Ta). Complexes 1, 2, and 7 were characterized by X-ray
crystallography, and their solid-state molecular structures are
shown in Figure 2. A selection of relevant metrical parameters
can be found in Table 1. In the solid state, 1 crystallizes in
C
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Scheme 3
plane.23 Complex 1 features V−N bond lengths of 1.837(1)
and 1.834(1) Å. For comparison, these values are slightly
shorter than those of the homoleptic V(IV) amide, V(NMe2)4,
which displays V−N bond lengths of 1.866(1)−1.871(1) Å;24
however, they are within the range exhibited by other V(V)
ketimides (1.787−1.847 Å).25−29 For further comparison, the
average V−C bond length in V(Mes)4 (Mes = 2,4,6-Me3C6H2)
is substantially longer (2.08 Å).30 Finally, the V−N−C angles
(V1−N1−C1= 177.2(1)° and V1−N2−C10 = 176.7(1)°) are
suggestive of π-donation to the vanadium center from the
ketimide ligand.
at least one d electron must occupy an antibonding or partially
antibonding orbital, which results in a decrease of the crystal
field stabilization. This assessment was confirmed by DFT
calculations on Co(NCtBu2)4, which was found to have a
2
21
dx −y 2(dxz,yz) dz (4A2 in D2d) ground-state electronic config-
2
2
uration. These calculations demonstrated that the partially
2
occupied orbitals (dxz,dyz,dz ) all possess considerable π* Co−N
character.6 Several group 5 tetra(amido) complexes (e.g.,
M(NR2)4) have also been structurally characterized, and a
comparison of their metrical parameters with those of 1, 2, and
7 is similarly informative. Most interestingly, this class of
materials does not exhibit the squashed tetrahedral geometry
observed in M(NCtBu2)4 and, instead, exhibits only slightly
distorted tetrahedral geometries. For example, the two largest
N−V−N interligand angles in V(NMe2)4 are 115.28(6) and
111.42(6)°, substantial smaller than those exhibited by 1 and
close to the 109.5° expected for a perfect tetrahedron.24
Similarly, Ta(NPh2)2(NEt2)2 features a slightly distorted
tetrahedral geometry,35 whereas Nb(NPh2)4 and Ta-
(NCy2)2(NEt2)2 feature some modest flattening, but it is not
of the magnitude observed by complexes 1, 2, and 7. In
particular, the two largest N−Nb−N interligand angles for
Nb(NPh2)4 are 120.8(1) and 116.7(1)°,31 and the two largest
N−Ta−N interligand angles for Ta(NCy2)2(NEt2)2 are
123.9(1) and 123.6(1)°.36 This difference in coordination
geometry can be rationalized by arguing that the ketimide
ligand is a stronger field ligand than the amide ligand and that
the larger CFSE extant in the ketimide complex is enough to
overcome the unfavorable steric repulsion engendered by
flattening the geometry about the metal center.35 However, this
explanation cannot discriminate between the σ- and π-
contributions to the overall ligand field, which differ
significantly between amide and ketimide ligands (see below).
Thus, to better understand this aspect of metal−ketimide
bonding we performed a thorough spectroscopic analysis of
complexes 1, 2, and 7.
Complexes 2 and 7 feature similar squashed tetrahedral
geometries in the solid state as that observed for 1. For
example, the largest N−M−N bond angles [2: N1−Nb1−N1*
= 129.39(9)° and N2−Nb1−N2* = 129.81(9)°; 7: N1−Ta1−
N1* = 128.2(1)° and N2−Ta1−N2* = 128.5(1)°] for 2 and 7
correspond to τ4 values of 0.72 and 0.73, respectively.23 As
anticipated, the Nb−N bond lengths in 2 (Nb1−N1 =
1.937(2), Nb1−N2 = 1.939(2) Å) are longer than those
observed in 1, consistent with the presence of the larger Nb4+
ion, but are shorter than those found in Nb(NPh2)4
(1.985(3)−2.029(3) Å).31 The Ta−N bond lengths in 7
(Ta1−N1 = 1.934(2), Ta1−N2 = 1.931(2) Å) are identical to
those exhibited by 2 and also similar to those exhibited by
complex 5, despite the difference in oxidation states between
the two complexes. For further comparison, these values are
comparable to those exhibited by MCl(tmkh)(tBu2pz)2 (M =
Nb, Ta; tmkh3− = 2,2,6,6-tetramethyl-5-ketimidohept-3-en-3-
imide; tBu2pz− = 3,5-di-tert-butylpyrazolate),32 but they are
longer than that in Cp*Ta(η4-C4H6)(κ2-C,N-C6H4-2-C(Me)-
N).33 Finally, the M−N−C angles in 2 (Nb1−N1−C4 =
176.5(2) and Nb1−N2−C1 = 176.5(2)°) and 7 (Ta1−N1−C1
= 176.3(2) and Ta1−N2−C10 = 174.1(2)°) are indicative of a
substantial π-donation from the ketimide to the metal center.
Complex 1 is the sixth first-row transition metal M(N
CtBu2)4-type complex to be structurally characterized (Table 1;
the present study provides the complete group 5 series;
Hoffman and co-workers have previously reported the
complete Group 6 series,9 these being the only other second-
and third-row M(NCtBu2)4 complexes). This series of
complexes exhibits an interesting correlation between the
coordination geometry about the metal center and the d
electron count. Specifically, we observe a gradual planarization
of the geometry about the metal ion (as indicated by the τ4
value) as the electronic configuration changes from d0 (Ti) to
d4 (Fe). Previously, we argued that this trend was a
consequence of increased crystal field stabilization that would
be achieved by flattening to a D2d structure.34 Interestingly, the
trend reverses at Co (τ4 = 0.59), which features a d5 electronic
configuration. Presumably, once a d5 configuration is achieved,
EPR Spectroscopy. The solution-phase X-band EPR
spectrum of 1 in hexanes (2 mM) exhibits eight major features,
consistent with a single unpaired electron coupling to the 51V
nucleus (I = 7/2, ∼ 100%) with a(51V) = 124 MHz (Figure 3).
The large a(51V) value, along with the g value (giso = 1.988 <
ge), are consistent with a predominantly metal-centered radical.
In addition, hyperfine coupling to the four 14N nuclei (I = 1,
100%) of the ketimide ligand is also partly resolved (a(14N) =
10.5 MHz). The line widths show a strong dependence on |mI|,
as has been described in detail for [VO(acac)2]37 and was also
seen in several V(IV) phosphine complexes.38 For comparison
with other organovanadium(IV) complexes, Cp2V(SPh)2 and
Cp2V(S5) exhibit nearly identical isotropic g and a(51V) values
as those exhibited by complex 1.39,40 Also relevant are the
D
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 1. Comparison of the Metrical Parameters for M(N
CtBu2)4 (M = Ti, V, Nb, Ta, Cr, Mn, Fe, Co)
M−N
distances
(Å)
M−N−C
angles
N−M−N
angles
a
b
M
d count
d0
(deg)
(deg)
τ4
ref.
Ti
av. 1.87
av. 175
114.9
0.96 19
111.3
V (1)
d1
d1
d1
d2
d3
d4
d5
1.834(1)
1.837(1)
1.937(2)
1.939(2)
1.931(2)
1.934(2)
1.784(2)
1.785(2)
av. 1.79
176.7(1)
177.2(1)
176.5(2)
176.5(2)
176.3(2)
174.1(2)
178.4(2)
179.1(2)
av. 176
132.94(9)
133.12(9)
129.39(9)
129.81(9)
128.5(1)
128.2(1)
136.1(1)
136.2(1)
151.1(1)
150.3(1)
167.1(2)
167.6(2)
137.6(2)
139.3(2)
0.67 this
work
Nb
(2)
0.72 this
work
Ta
(7)
0.73 this
work
Cr
0.62
0.42
0.18
0.59
9
4
3
6
Mn
Fe
1.771(3)
1.775(3)
av. 1.80
165.5(3)
166.6(3)
av. 167
Co
a
Defined as the two largest N−M−N angles observed in the complex.
As defined in ref 23.
b
Figure 3. Room-temperature X-band EPR spectrum of 1 (2 mM in
hexanes). Simulated spectrum shown in blue. Experimental conditions:
microwave frequency = 9.856 GHz, microwave power = 1.0 mW, field
modulation = 1.0 G, time constant = 0.32 ms, average of three 250 s
scans. Simulation parameters: g = 1.9885, a(51V) = 124 MHz, a(14N)4
= 10.5 MHz, Gaussian line width = 5 MHz, to match the resolved
splitting in the center of the spectrum (an expansion of this central
region is shown in Figure S20, Supporting Information). The
increased line broadening that occurs with higher magnitude 51V mI
transitions (e.g., ⟨mS, mI| = ⟨−1/2, 7/2| → ⟨+1/2, 7/2|; see |mI|
labels on simulated trace; absolute values are given because the sign of
aiso is unknown from experiment) is reproduced using the model of
Froncisz and Hyde,45 with their A-strain factor, c1 (here, isotropic) =
1.5 MHz. When the ligand hyperfine coupling was not resolved, it was
possible to match all transitions even more exactly (see Figure S19,
Supporting Information).
Figure 2. Solid-state molecular structures of V(NCtBu2)4 (1),
Nb(NCtBu2)4 (2), and Ta(NCtBu2)4 (7) with 50% probability
ellipsoids. Hydrogen atoms are omitted for clarity. Atoms with an
asterisk are generated by symmetry.
E
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 2. Spin Hamiltonian Parameters for Group 5 Complexes Relevant to This Study
giso (fluid
solution)
gavg (frozen
solution)
g∥, g⊥ (frozen
solution)
aiso (fluid solution, Aavg (frozen solution,
A∥, A⊥ (frozen
solution, MHz)
complex
MHz)
MHz)
ref.
a
[V(NCtBu2)4], 1
1.9885(5)
1.9840(5)
1.955(5)
1.965
1.983
1.988
1.979
1.979(1),
1.985(5)
124(2)
125
245(5), 65(15)
this
work
b
[Nb(NCtBu2)4], 2
[Ta(NCtBu2)4], 7
1.966 (2),
1.999(5)
185(5)
250(10)
185
150
250
300(10), 80(20)
335, 210
this
work
b
c
1.998, 1.970
this
work
d
e
[V(C6Cl5)4]
1.925, [1.985]
385, [85]
41
46
42
47
[Nb(C6Cl5)4]
1.940
1.968
1.973
1.883, 1.969
1.925, 1.989
1.948, 1.986
371
598, 258
370.5, 99.5
458, 158
f
V(Mes)4
1.972
1.973
172.4
260
189.8
260
V(N(C6H4-2-tBu))
g
(NHMe2)2Cl2
V(NEt2)4
V(OtBu)4
1.976
1.976
1.957, 1.985
180
180
372, 84
43
44
1.964(5)
1.969(5)
1.940(5),
1.984(5)
192(6)
197(15)
375(15), 108(12)
Cp2V(SPh)2
VCl4(PEt3)2
NbCl4(PEt3)2
TaCl4(PEt3)2
2.00
175
40
1.981
1.972, 1.985
1.959, 1.912
1.831, 1.695
252
450, 147
599, 276
730, 413
38
1.927
381
38
1.740
514
38
h
[(MeCp)Nb(C7H7)]
1.986
94.5
171
48
h
[CpTa(C7H7)]
1.944
1.944
1.943
1.948
1.989, 1.921
1.987, 1.921
1.989, 1.927
157
192
306
428, 22
465, 56
582, 168
49
h
[(MeCp)Ta(C7H7)]
1.945
199
49
h
[Cp*Ta(C7H7)]
1.949
316
49
i
[
X,R[NNNsq•]TaCl3]−
1.958−1.964
81−92
50, 51
a
These parameters are based on X-band spectra. Frozen-solution Q-band spectra gave g∥ = 1.979, g⊥ = 1.987, A∥(51V) = 250 MHz, and A⊥(51V) = 65
b
MHz. These parameters are for the majority species. The frozen-solution spectrum of 2 is very sensitive to freezing artifacts. Frozen-solution Q-
band spectra gave g∥ = 1.979, g⊥ = 1.995, A∥(93Nb) = 310 MHz, and A⊥(93Nb) = 80 MHz. However, species with A∥ = 365 MHz were also observed
c
at X-band under slower freezing conditions, which give Aavg = 175 MHz, a value much closer to the fluid-solution value. These parameters roughly
reproduce one species found in frozen solution. No error ranges are given because of the approximate nature of this simulation and the lack of
d
e
corroborating 35 GHz EPR data, in contrast to 1 and 2. Not reported but calculated here using the reported giso and g∥ values. Not reported but
f
calculated here using the reported aiso and A∥ values. In addition to a fluid solution of V(Mes)4 (Mes =2,4,6-Me3C6H2), a single crystal of Ti(Mes)4
doped with V(Mes)4 was studied so that the full g and 51V hyperfine coupling tensors were determined with high precision: gx = 1.984, gy = 1.994,
g
and gz = 1.925; Ax = 80.6, Ay = 118.4, and Az = 370.5 MHz; these are converted to axial tensors for easier comparison with the other entries. This is
but one of a large number of five- and six-coordinate V(IV) complexes reported in this article; the one presented here is more relevant to the
ketimides. It is representative of a series of five-coordinate V(IV) imido complexes, V(=NR)(NHMe2)2Cl2 where R = five different bulky aryl or alkyl
substituents. Their EPR parameters vary minimally as a function of R group as seen in Table 5 of Bigmore et al.47 Their A values have been
h
converted from G to MHz using their reported g values. These are considered as Nb(IV) or Ta(IV) complexes by virtue of the cycloheptatrienyl
i
ligand being counted as a trianion (10 π electrons, Huckel aromatic), along with the cyclopentadienyl monoanion. The X,R[NNNsq•
]
2− ligand is the
̈
semiquinone form of X,R[NNNcat]3−, where, e.g., OMe,iPr[NNNcat]3− is the trianion of bis(2-isopropylamino-4-methoxyphenyl)amine (X = 4-OMe; F,
H, Me, and tBu; R = 2-iPr; 3,5-C6H3Me2). These complexes are neither truly Ta(V) nor Ta(IV) as a result of the noninnocent ligand. Simulations of
room-temperature spectra of four complexes were presented, but no parameters were given; the range of giso and aiso values presented here are taken
from legends to a composite figure in the Supporting Information of ref 50.
homoleptic V(IV) complexes that feature tetra(aryl), tetra-
(amido), and tetra(alkoxido) ligand sets.41−44 One of the
tetra(aryl) complexes was observed only in situ by Alonso et
al.,41 but another, V(Mes)4, was studied not only in fluid
solution but also as a doped single crystal (diamagnetic
Ti(Mes)4 host),42 likely affording the most accurate and
complete spin Hamiltonian parameters for any complex
relevant to the present work. The EPR investigations and
analysis of the tetra(amido)43 and tetra(alkoxido)44 complexes
were also very thorough, with a variety of solvents and
diamagnetic hosts (the Ti(IV) congeners) employed. These
and other EPR parameters of interest are summarized in Table
2.
The room-temperature EPR spectrum of 2 in hexanes (2
mM) consists of a 10 line pattern, consistent with a single
unpaired electron coupling to one 93Nb center (I = 9/2, 100%)
with aNb = 185 MHz (Figure 4). Hyperfine coupling to 14N was
not resolved in this spectrum. Again, the g value falls within the
range of a transition metal-centered radical (giso = 1.984). Most
relevant is the structurally characterized (in contrast to its V
congener) homoleptic tetra(aryl) Nb(IV) complex, [Nb-
(C6Cl5)4].46 The spin Hamiltonian parameters for this complex
are given in Table 2 and show that the 93Nb hyperfine coupling
constant in [Nb(C6Cl5)4] is much larger than in 2. This
difference, which is also seen for the V complexes, can be
qualitatively explained by the greater covalency (via σ- and π-
bonding; see below) in 2, as opposed to the more ionic σ-only
bonding in the pentachlorophenyl complex.
The solution-phase EPR spectrum of 7, generated in situ in
toluene-d8 by reaction of 6 with 1 equiv of Cp*2Co, exhibits the
expected eight line signal, consistent with a single unpaired
7
electron coupling to 181Ta (I = /2, 100%). In the spectrum
shown in Figure 5, complex 7 exhibits a hint of resolved
hyperfine coupling to the 14N nuclei of the ketimide ligand. As
shown on an expanded field scale in Figure S20 (Supporting
Information), we estimate this coupling at a(14N) ≈ 30 MHz.
For comparison, the corresponding well-resolved central signals
for 1 are also shown in Figure S20, demonstrating the good fit
1
to that region (|mI| = /2), which is unaffected by hyperfine-
dependent line broadening effects. The observed giso value (giso
F
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
= 1.955) again falls in the range of transition metal-centered
radicals, and the aiso value (aiso = 250 MHz) continues the trend
of increasing aiso values of the group 5 ketimides on going down
the column, despite the decrease in gN from V to Nb to Ta. A
similar trend was seen in the MCl4(PEt3)2 (M = V, Nb, Ta)
series, in which the aiso values increased from 252 to 381 to 514
MHz on going down the group.38 As seen in Table 2, the 181Ta
hyperfine coupling is also in the range reported for the
hydrocarbyl complexes, (CpR)Ta(C7H7) (R = H, Me, Me5)49
and is much larger than that reported for a series of tantalum
complexes with a noninnocent pincer ligand.51
Being the most stable of the three, the frozen-solution EPR
spectra of 1 were investigated under a variety of experimental
conditions. As shown in Figure 6, in hexanes at 125 K, complex
Figure 4. Room-temperature X-band EPR spectrum of 2 (2 mM in
hexanes). Experimental conditions: microwave frequency = 9.290
GHz, microwave power = 1.0 mW, field modulation = 0.2 G, time
constant = 20 ms, average of twenty-five 20 s scans. Simulation
parameters: g = 1.9840, a(93Nb) = 185 MHz, Gaussian line width = 20
MHz, to match the central transitions. The increased line broadening
that occurs with higher magnitude 93Nb mI transitions is reproduced
using the model of Froncisz and Hyde,45 with their A-strain factor, c1
(here, isotropic) = 2.5 MHz.
Figure 6. Frozen-solution X-band EPR spectrum (black =
experimental, red = simulated) of 1 (2 mM in hexanes) at 125 K.
Experimental conditions: microwave frequency = 9.296 GHz,
microwave power = 2.0 mW, modulation = 0.5 G, time constant =
2.56 ms, average of thirty-three 20 s scans. Simulation parameters: g =
[1.985, 1.985, 1.979], A(51V) = [65, 65, 245] MHz, Gaussian line
widths = 30 MHz.
1 displays average g and A values in good agreement with the
room-temperature isotropic (i.e., rotationally averaged) values
(see Table 2) but with significant uniaxial anisotropy, defined
here as the z-direction (Ax = Ay = A⊥ = 65 MHz, Az = A∥ = 245
MHz). Similar EPR anisotropy has been observed in the
pseudo-four-coordinate V(IV) complex [Cp2V(dbm)][ClO4]
(dbm = dibenzoylmethanate)52 and in V(Mes)4.42 Indeed, the
ratio of A∥/A⊥ in V(Mes)4 equals 3.72, and in 1, this ratio is
3.77. The origin of this anisotropy will be discussed below. The
X-band EPR spectrum of 1 recorded in toluene solution at 20 K
(Figure S21, Supporting Information) was similar, but not
identical, to that shown here. This may be a consequence of the
tetrakis(ketimide) complexes being rather flexible, as next
shown more convincingly for the Nb congener.
We also recorded the 35 GHz (Q-band) EPR spectrum of 1
at 2 K in toluene (Figure 7). Under these experimental
conditions, the spectra are in rapid passage mode and appear as
the absorption line shape.53 This presentation mode is helpful
for observing broad features, but it is less so for the relatively
narrow lines of 1. Nevertheless, the use of a second EPR
resonant frequency is very helpful in showing the validity of a
Figure 5. Room-temperature X-band EPR spectrum of 7 (0.02 mM in
toluene-d8). Experimental conditions: microwave frequency = 9.858
GHz, microwave power = 1.0 mW, field modulation = 1.0 G, time
constant = 10 ms, average of twenty 10 s scans. Simulation parameters:
g = 1.955, a(181Ta) = 250 MHz, Gaussian line width = 50 MHz. No
attempt was made to model any A-strain effects, in contrast to the
situation for 1 (Figure 3) and 2 (Figure 4), but the 14N hyperfine
coupling is shown on an expanded scale in Figure S20 (Supporting
Information).
G
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
spectra, recorded only at g∥, which thus provide only A∥(14N)
(i.e., the coupling along the parallel (z) direction assuming
colinearity of A and g). Nevertheless, A∥(14N) ≈ 14 MHz is
consistent with aiso(14N) = 10.5 MHz and also with DFT
calculations described below.
Frozen-solution EPR spectra of complex 2 proved to be
more difficult to obtain as a homogeneous species, despite
multiple attempts. A representative spectrum, recorded at 20 K
in toluene solution, is shown in Figure 8. This spectrum clearly
Figure 7. Low-temperature (2 K) Q-band EPR spectrum of 1 (0.1
mM in toluene). The lower pair of traces are the experimental
spectrum (black trace), together with a simulation (blue trace), and
the upper pair is a digital derivative with a simulation. Experimental
conditions: microwave frequency = 35.2076 GHz, microwave power =
2 mW, field modulation = 1.0 G, time constant = 16 ms, single 120 s
scan. Simulation parameters: g = [1.987, 1.987, 1.979], A(51V) = [65,
65, 250] MHz, W = 30, 30, 35 MHz.
set of simulation parameters, as the g values are field-
dependent, whereas the hyperfine couplings are not. As can
be seen from the figure caption, the simulation parameters are
essentially unchanged from those determined from X-band
spectra (which were, in addition, recorded at higher temper-
ature and under slow passage, i.e., normal conditions). The
increase in frequency to 35 GHz, however, is not sufficient to
separate the parallel from the perpendicular transitions. To
achieve this, as has been shown in vanadyl complexes,
microwave frequencies at 95 GHz (W-band)54 or higher55
would be needed.
Figure 8. Frozen-solution X-band EPR spectrum of 2 (2 mM in
toluene) at 20 K. Several A∥ features of a second species are indicated
by arrows in the upper, expanded vertical scale spectra; no attempt was
made to simulate or quantify this minor species. Experimental
conditions: microwave frequency = 9.373 GHz, microwave power =
2 mW, field modulation = 15 G, time constant = 160 ms, single 90 s
scan. Simulation parameters (magenta traces): g = [1.999, 1.999,
1.966], A(93Nb) = [80, 80, 300] MHz, Gaussian line widths = 35
MHz.
ENDOR spectroscopy of V(IV) complexes is rather limited
and has been chiefly performed on vanadyl complexes, such as
in pioneering work by van Willigen and co-workers56,57 and
more recently by Britt and co-workers58,59 and others.60,61 We
therefore recorded 35 GHz CW ENDOR spectra of complex 1,
partly to test the feasibility of ENDOR on such a system. We
reveals that there are two species present in the sample, which
have similar EPR signatures, suggesting that they are only
slightly different structurally. Only one species is present in
fluid solution (Figure 4), albeit with no resolution of ligand
hyperfine, which suggests that these two species can
interconvert at room temperature. We made several attempts
to generate a homogeneous sample by freezing the solution as
quickly as possible; however, this did not completely mitigate
the heterogeneity. Nonetheless, the majority species gives
average g and A values in reasonable agreement with the room-
temperature isotropic values (see Table 2). The 35 GHz 2 K
spectrum of 2 proved to be slightly more homogeneous (Figure
9), which may be a function of the smaller sized EPR tubes,
allowing for faster freezing. The simulation parameters were
essentially the same as those of the major species observed at X-
band.
1
indeed readily observed H, 14N, and 51V ENDOR from this
1
complex, as shown in Figure S23. The H ENDOR signals are
centered at the proton Larmor frequency (∼53 MHz) and
show very small hyperfine coupling, as expected for the tert-
butyl hydrogen atoms, which are relatively distant from the
paramagnetic metal center. The 51V ENDOR signals are of
qualitative interest only, but they are consistent with A∥ = 240
MHz (i.e., features centered at A∥/2 = 120 MHz). The 14N
signals also confirm the results of EPR spectroscopy. In the
absence of a complete ENDOR study, it is not appropriate to
make a definitive correspondence between the 14N hyperfine
splitting observed by EPR and that by 14N ENDOR. This is
because one is comparing fluid-solution X-band EPR spectra,
which provide aiso(14N) (i.e., an average coupling value of all
molecular orientations), to frozen-solution Q-band ENDOR
In keeping with the trend in stability, complex 7 gave the
most problematic frozen-solution EPR behavior. There are
multiple species present in these samples, albeit with similar
EPR parameters. It was possible to simulate only approximately
one of the species (see Figure S22, Supporting Information),
H
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
⎡
⎤
4
7
11
14
5
14
A − aiso = −P
+
(ge − g ) −
(g − g )
e iso
⎢
⎣
⎥
⎦
(4)
allowing calculation of P in the system of interest; eq 3 then
yields |κ|, since the sign of aiso is not known experimentally.
This method, applied to [V(C6Cl5)4], gave |P| = 323 MHz and
|κ| = 0.54,41 and for [Nb(C6Cl5)4], it gave |P| = 352 MHz and
|κ| = 0.99.46 This method had not been previously applied to
V(Mes)4, yet this is among the most relevant complexes to ours
and that for which the best EPR data are available.42 We thus
find here for V(Mes)4 that |P| = 291.5 MHz and |κ| = 0.617.63
The tetra(amido) complex, V(NEt2)4, was reported to have |P|
= 330 MHz,43 and substituting that value and their EPR data
(see Table 2) into eq 3 gives |κ| = 0.50. The tetra(alkoxido)
complex, V(OtBu)4, had |P| = 324 MHz, giving |κ| = 0.55.44
Concerning the ketimides, we calculate here for 1 that |P| = 207
MHz and |κ| = 0.59 and for 2 that |P| = 194 MHz and |κ| =
0.94.64 If one takes the ratio of the calculated P value to the
free-ion P value, then the tetra(aryl), amido, and alkoxido
vanadium(IV) complexes all give values of ∼0.6, whereas the V
and Nb ketimides give this ratio as roughly 0.4, indicating the
much higher covalency in the latter complexes.
Figure 9. Low-temperature (2 K) Q-band EPR spectrum of 2 (0.1
mM in toluene). The lower pair of traces are the experimental
spectrum (black trace), together with a simulation (magenta trace),
and the upper pair is a digital derivative with a simulation.
Experimental conditions: microwave frequency = 35.2650 GHz,
microwave power = 2 mW, field modulation = 0.5 G, time constant
= 16 ms, single 120 s scan. Simulation parameters: g = [1.995, 1.995,
1.979], A(93Nb) = [80, 80, 310] MHz, W = 40, 40, 45 MHz.
UV−Vis Spectroscopy. Both complexes 1 and 2 exhibit
rich UV−vis spectra (Figure 10). For example, the electronic
although it gave an average 181Ta hyperfine coupling consistent
with the room-temperature value (see Table 2).
The results of fluid- and frozen-solution EPR studies can be
combined, as was done by Alonso et al.41 This makes use of the
fact that in the frozen-solution spectra of these systems only g∥
and A∥ can be easily determined (single-crystal studies can
provide the full tensors for both g and A, as done by Kirmse et
al.),42 but here, for 2 and especially 7, the g⊥ and A⊥ values are
relatively uncertain. The starting point relevant equations for an
21
2
axial hyperfine interaction for a dxy/x −y ground state are as
follows:
Figure 10. Room-temperature UV−vis−NIR absorption spectra of 1
(blue traces) and 2 (magenta traces) in hexanes solution. For both
compounds, the bands in the approximate range 300−450 nm are due
to ligand-to-metal charge transfer (LMCT) transitions. The inset
shows an expansion of the vis−NIR region with d−d band assignments
as described in the text.
⎡
⎤
4
7
3
7
A = −P κ +
+
+
(g − g ) + (g − g ) ,
⎢
⎥
e
⊥
e
⎣
⎦
⎡
⎤
2
7
11
14
A = −P κ −
(g − g )
e ⊥
⎢
⎥
⎦
⊥
⎣
(1a,b)
where P is the intrinsic hyperfine coupling constant, P =
gegNβeβN⟨r−3⟩av, which for free-ion 51V4+ equals 516 MHz62 and
for free-ion 93Nb4+ equals 576 MHz62 (the value for 181Ta4+ is
unavailable), and κ is a unitless scaling factor to account crudely
for covalency. Using the relationships
absorption spectrum of 1 features two intense absorptions at
360 (ε = 12 000 M−1·cm−1) and 460 nm (ε = 6000 M−1·cm−1),
which are both assignable to ligand-to-metal charge transfer
processes, i.e., crudely represented by [MIV(−NCtBu2)] →
[MIII(•NCtBu2)]. Additionally, two absorptions are observed
at 620 (ε = 560 M−1·cm−1) and 764 nm (ε = 440 M−1·cm−1).
We have tentatively assigned these less intense bands to d−d
transitions, as will be discussed in detail below. The UV−vis
spectrum of 2 features an intense absorption at 320 nm (ε =
12 000 M−1·cm−1), which is likewise assignable to ligand-to-
metal charge transfer. In addition, this spectrum features four
transitions at 432 (ε = 2700 M−1·cm−1), 596 (ε = 1200 M−1·
cm−1), 698 (ε = 1400 M−1·cm−1), and 830 nm (ε = 1800 M−1·
cm−1), which are assignable to d−d transitions. Unfortunately,
(g + 2g⊥)
(A + 2A⊥)
giso
=
,
aiso =
(2a,b)
3
3
By substituting eqs 1a,b into eq 2b and making use of eq 2a to
substitute for g∥ and g⊥, one obtains
aiso = −P[κ + (ge − giso)]
(3)
Subtracting eq 3 from eq 1a and substituting for g⊥ from eq 2a
gives
I
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
it was not possible to prepare UV−vis samples of 7 that were
free of other optically absorbing components.
forbidden in D2d point group symmetry,65 but the B1 → E
(dx1z,yz) transition is allowed with xy polarization.66 It is
calculated to occur at 13 470 cm−1, reasonably close to
experiment (12 800 cm−1, based on the band centers on an
energy scale). For comparison, VCl4 has Δtet = 7900 cm−1,67
and V(Mes)4 has Δtet = 13 370 cm−1,30 giving εσ = 10 030
cm−1. The higher energy transition is 2B1 → 2B2 (dx1y), which is
calculated to occur at 15 410 cm−1, also reasonably close to
experiment (16 000 cm−1). Although this transition is
forbidden,68 it may become partially allowed by vibronic
coupling, and its intensity perhaps also benefits from underlying
charge transfer band(s).69 Vibronic coupling would be
reasonable for 1 due to the Jahn−Teller effect. Tetrahedral
nd1 is a classic Jahn−Teller active system, being e1t02, the
electron analogue to the hole configuration t62e3, as exemplified
by octahedral Cu(II), for which vibronic coupling of an eg
mode (ν2) provides a mechanism for Jahn−Teller distortion.70
V(Mes)4 and VCl4 are complexes with, respectively, only σ-
bonding and cylindrical π-bonding, which makes them better
candidates for this effect than ketimides with their unsym-
metrical π-bonding (i.e., making the distortion a pseudo Jahn−
Teller effect). Although a (true) Jahn−Teller effect was not
invoked to justify the D2d symmetry of V(Mes)4 (this complex
may have significant steric effects),30 VCl4 may exhibit this
effect,67 and it may be related to the observation of a nominally
forbidden band in 1.
2
2
Ligand Field Theory Analysis. To employ a simple ligand
field theory for the [M(NCtBu2)4] complexes, we need to
establish a physical model. In D2d point group symmetry, the
2
2
dx −y AO is represented by B1, and its parentage (via ascent in
2
symmetry) is E (together with dz ) in Td. The dxy orbital is
represented by B2, and its parentage (along with dxz and dyz) is
T2 in Td. This symmetry and coordinate system are identical to
that used by some of us in our previous study of [Co(N
CtBu2)4].6 To achieve this description, we define the x and y
axes as lying between the M−L bonds (bisecting the L−M−L
angles) (Figure 11), which is, in any case, more appropriate for
Figure 11. Structure of generic M(NCtBu2)4 complex showing
coordinate system used for the AOM analysis. Two ketimide ligands
are in the plane of the paper: one projects from the paper and the
other is behind. The z axis (shown in blue) corresponds to the
approximate S4 axis; the x (green) and y (red) axes are rotated 45°
from the ligands planes. For the AOM analysis, average θ, θ′ values are
used (ideal θ = 54.7356°), with an ideal S4 axis (ϕ = 45° (as shown),
135°, 225°, 315°). For M = Ti, θ = 56.53°; for M = V (1), θ = 66.5°
(as shown); for M = Nb (2), θ = 64.4°; for M = Ta (7), θ = 64.16°;
the angle to the other pair of ligands, θ′ ≡ 180° − θ.
However, a problem arises when we use the actual geometry
of 1 (Figure 11). In this case, the greater distortion from
tetrahedral geometry makes it impossible to fit both electronic
transitions simultaneously. Either the lower or the higher band
can be fit exactly using the B1 → E assignment (with εσ =
16 000 cm−1 or εσ = 20 000 cm−1, respectively), but not both,
unless the symmetry is lowered such that the four ketimide
ligands are divided into two pairs: two with the lower εσ value
and two with the higher. This C2v split is large but not
inconceivable.71 The problem is not so much symmetry but
that these εσ values are extremely high.72,73 Thus, a model with
only σ-bonding is unsatisfying.
2
2
an axial (x = y) system. If the x and y axes were defined along
the M−L bonds, as done by Soriaga et al.,9 then the
Clearly, π-bonding must be considered, as was the case with
the more electron-rich Co(IV) congener.6 Soriaga et al.
identified the ketimide MOs having π-interactions with an
electron-poor metal ion.9 These are shown in Figure 12, which
is inspired by their work and is also a cartoon version of the
more recent DFT results depicted in Figure 2 of Lewis et al.6
2
2
representation of dx −y and dxy would be reversed. Such a
definition would be appropriate for C2v symmetry, where x and
y are symmetry inequivalent. Given the axial nature of the EPR
spectra and a desire for the simplest useful model to describe
[M(NCtBu2)4] complexes, we shall use primarily D2d
symmetry in our simple ligand field model.
2
The occupied, π-donor MO that interacts strongly with dz (and
2
2
not at all with dx −y in ideal tetrahedral geometry) is labeled π−
c,72 with interaction parametrized by επ−c, increasing the energy
2
We can now begin to describe the ketimide MOs involved in
bonding using the angular overlap model (AOM) with an ideal
S4 axis defining z and the x and y axes between the bonds, as
described above (see Figure 11). The angle θ is given by the
relevant crystal structures. This geometrical model is sufficient
for the simple nature of our analysis. Soriaga et al. did not
mention σ-bonding in this class of complexes,9 but we note that
the ketimide ligand is likely a strong σ-donor. This interaction
alone could potentially lead to a large splitting between the
2
2
of dz . The unoccupied, π-acceptor MO that interacts with dx −y
2
(and not at all with dz in ideal tetrahedral geometry) is labeled
π−s, with interaction parametrized by επ−s, decreasing the
2
2
energy of dx −y . The dxz,dyz AOs are moderately affected by
these π-bonding interactions using the AOM (see Table S2,
Supporting Information), but in V(IV) (and Cr(IV)), these
empty AOs are relatively unimportant compared to their role in
Co(IV).6
2
2
2
e(dz ,dx −y ) and t2(dxz,dyz,dxy) metal ion orbitals. This splitting,
4
We also note that the electronic absorption spectra of the
homoleptic V(IV) complex with π-donor ligands, namely,
V(NEt2)4, has been recorded, both as a pure solid and in a
Ti(NEt2)4 host.43,74 These spectra gave shoulders at 17 300
cm−1 (17 065 cm−1 in the Ti(IV) host) and 13 000 cm−1
(13 120 cm−1 in Ti(IV)), which are quite close to those
absorptions observed for 1, but they were assigned oppositely,
for a Td ML4 complex, is given by Δtet = /3εσ. For illustrative
purposes, we begin with only σ-bonding and the structure of
the Ti(IV) congener, which has θ only ∼2° away from ideal
tetrahedral (τ = 0.96).19 We can moderately well fit the two
observed putative d−d electronic transitions for 1 using this
nearly tetrahedral model with εσ = 10 610 cm−1. The ground
2
21
2
21
2
state is B1(dx −y ), with the A1(dz ) excited state barely
higher in energy (∼100 cm−1). The transition B1 → A1 is
i.e., as B1 → E and B1 → B2, respectively. However, the
2
2
2
2
2
2
J
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Figure 12. Qualitative MO diagram for V(IV) (d1, blue electron) and Cr(IV) (d2, red electrons) tetrakis(ketimide) complexes. On the far left is
shown two views of the relevant ketimide π-MOs; on the far right is shown the most important metal 3d AOs, with their representations in Td and
D2d symmetry. In the center is shown qualitatively the MOs resulting from interactions between the relevant ligand and metal orbitals, with
representations in D2d symmetry; quantitative results for V(IV) are shown in Figure 13. At the bottom is shown two of these resulting ketimide−
metal MOs; for clarity, only the nitrogen p AO of the ketimide π-MO is shown in these diagrams. For completeness, at the top, the σ* (b2) MO is
also shown, but interactions involving the unfilled dxz, dyz AOs are not shown.
authors were not adamant in this assignment as opposed to the
reverse. These workers did not have a crystal structure;
however, many years later, the structure of V(NMe2)4 was
reported, and its geometry is much closer to tetrahedral
symmetry (although still D2d) than the geometry of the present
group 5 ketimide complexes.24 The application of modern
computational methods, unavailable at the time of the EPR
studies on tetra(amido) and (alkoxido) V(IV) complexes,43,44
is therefore of interest but must be the subject of future
investigations.
εσ = 6150 cm−1, επ = −330 cm−1 for pyridine coordinated to
Cr(III).72 Pyrazole nitrogens, as found in the well-known
“scorpionate” ligand (hydridotris(pyrazol-1-yl)borate, Tp−),
give εσ = 8350 cm−1, επ−s = 1300 cm−1 (επ−c = 0) for
[Tp2Cr]+,77 with similar parameters estimated for Mn(III) in
[Tp2Mn]+.78 The extreme among N (or likely any other)
donors is nitride, which exhibits εσ = 25 000 cm−1, επ = 18 000
cm−1 for Cr(V) and Mn(V).75,76 Phosphines bound to late
transition metal ions, e.g., Ni(II), can be described as having επ
≈ −1500 cm−1,79 comparable to what is seen here for ketimide,
and are good σ-donors (εσ ≈ 5000 cm−1), but they have no
simultaneous π-donor ability. Bonding parameter character-
ization of ketimides bound to a wider variety of metal ions is
needed, but it is clear that, while being only moderate σ-donors,
their combination of π-donor and π-acceptor abilities makes
them unusual, if not unique, among N (or P) donor ligands.
We can also calculate EPR transitions using the LFT software
DDN (see Supporting Information),79 which also includes the
electronic spin and orbital Zeeman interaction. Thus,
application of an external magnetic field (B0 = 350 mT, to
correspond to X-band) and inclusion of spin−orbit coupling, ζ
= 175 cm−1 (70% of the free-ion value, 250 cm−1)80 yields g∥ =
1.914 and g⊥ = 1.974. These g values are more anisotropic and
with lower gavg than experiment, but the relation g∥ < g⊥ < 2.00
is reproduced. This is as expected from perturbation theory,81
where the g values are described by
With this π-bonding model, the fit results are completely
different. With three parameters εσ, επ−s, and επ−c, it is possible
to fit the two transitions exactly. The difference from the above-
mentioned σ-only analysis, however, is that while the transition
at 12 800 cm−1 is still assigned to (allowed) 2B1 → 2E and that
at 16 000 cm−1 to (vibronically allowed) 2B1
→
2B2, the
2
2
forbidden B1 → A1 transition is now calculated to occur at
14 800 cm−1. This is fully possible, as it would lie under existing
bands. The fit parameters are εσ = 3410 cm−1, επ−c = 4780
cm−1, and επ−s = −1840 cm−1, so the ketimide ligands could be
considered to be moderate σ-donors, strong π-donors, and
strong π-acceptors. These ketimide bonding parameters can be
put into the context of other ligands of interest, although such
comparison data are relatively scarce. To our knowledge, these
data are not available for the amide, [NR2]−, ligand, which is
probably the most relevant comparison for this study. However,
they are known for classical neutral N-donor ligands, such as
NH3, RNH2 (as in ethylenediamine), and RR′NH (as in
1,4,8,11-tetraazacyclotetradecane (cyclam)) when ligated to
first-row early transition metal ions such as Cr, mainly in the
common 3+ oxidation state72,73 but also in higher oxidation
states such as Cr(V)75 and Mn(V, VI).76 Briefly, these σ-only
donors exhibit εσ ≈ 7000 cm−1 (higher for the highly oxidized
ions). Imine nitrogens donors come in many forms, but they
can be good σ-donors with π-acceptor (or donor) ability, e.g.,
8ζ
g = ge −
,
2
2
(E − E
)
B2
B1
2ζ
g⊥ = ge −
2
2
(E − E
)
E
B1
(5)
where (E − E B ) = 12 800 cm−1 and (E − E B ) = 16 000
2
2
2
2
E
B2
1
1
cm−1, so that g∥ = 1.915 and g⊥ = 1.975, exactly the same values
as the LFT exact calculation. The value for ζ could, of course,
K
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
be reduced to improve the correspondence, e.g., ζ ≈ 110 cm−1
(44% of the free-ion value) matches the experimental g⊥,
although ζ ≈ 50 cm−1, a meaninglessly low value for V(IV), is
needed to match g∥. As shown below, quantum calculations are
much more successful at not overstating the g anisotropy in this
complex.
We can apply the same model to the Nb(IV) complex, 2. In
this case, there may be rhombic splitting resolved in the visible
region, with bands at 12 300, 14 300, 18 700 cm−1 and an
additional band at 23 400 cm−1. This is similar to a band at
21 400 cm−1 observed for [V(C6Cl5)4] by Alonso et al.,41 but
we do not attempt to assign it, as we doubt that these violet/
blue absorption bands are d−d in character. We note that
[V(Mes)4] exhibits several bands in this region (23 870, 21 480,
18 350 cm−1), but all were assigned to CT transitions.30 In any
case, using the same fit procedure, the bands match exactly (the
center of the first two bands is the fit result, 13 300 cm−1).82
The fit parameters for 2 are εσ = 5210 cm−1, επ−c = 4670 cm−1,
and επ−s = −1700 cm−1, so there is a stronger σ-bonding
interaction with the larger Nb(IV) ion than with V(IV), with
the π-bonding being roughly the same. This observation is
consistent with what would be expected on going down a group
in the d block.
Figure 13. Frontier molecular orbitals for [V(NCMe2)4] with
representations in D2d symmetry. Isosurface value = 0.06.
HOMO−LUMO gap) is much larger as well (∼4 eV for V(IV)
versus ∼2 eV for Co(IV)). Nevertheless, the extreme closeness
in energy among the e, a1 MOs (which are half-filled in
[Co(NCtBu2)4]) is again found for the V(IV) complex.
Although we have not specifically performed calculations for
the Cr(IV) congener, it is clear from our MO diagram that this
complex would have a filled b1 HOMO with the triplet excited
state(s) significantly higher in energy, in agreement with earlier
work.9
In summary, with the ligand field analysis in hand, it is now
apparent that the geometry differences observed for M(amide)4
and M(ketimide)4 are not due to differences in σ- or π-donor
ability, which are probably similar between the two ligand types,
but are instead due to the π-acceptor ability of the ketimide
ligand, as the amide ligand is unable to act in this capacity.
Quantum Chemical Theory Analysis. Analogously to
what was done previously for [Co(NCtBu2)4],6 we have used
density functional theory (DFT) to probe the series of group 5
ketimides. In the interest of computational efficiency,
complexes 1 and 2 were both truncated to M(NCMe2)4.
The program ORCA by Neese83,84 was employed with the
B3LYP functional and def2-TZVP basis set,85,86 which are
widely used for transition metal complexes.87,88 Further details
are given in Supporting Information, including representative
input/output files, Tables S3 and S4, respectively, for [V(N
CMe2)4] and [Nb(NCMe2)4]. In the latter case, our
computational methodology was still effective, with the only
qualitatively relevant point being a greater delocalization onto
the ketimide nitrogen orbitals. The key result for both the V
and Nb ketimide complexes is that the appearance of the
The corresponding MO diagram for [Nb(NCMe2)4] is
shown in Figure S26 (Supporting Information). The results are
overall very similar to those for the V(IV) complex, with the
Nb-based MOs being higher in energy. As a result, these are
more in range with high-energy ligand-based MOs. Thus, there
is a purely ketimide π* MO (#86) lying in between two MOs
2
each with Nb 4dz character, the latter of which (#87) appears
2
to be a more classical 4dz orbital. Again, we have not
performed calculations for the Mo(IV) congener, nor did
Soriaga et al.,9 but our results for [Nb(NCMe2)4] give a
HOMO−LUMO gap of ∼3 eV, so the singlet ground state for
[Mo(NCtBu2)4] should be greatly favored.
In addition to providing an MO picture of these group 5
ketimide complexes, ORCA software can calculate EPR
parameters. In this case, this includes the g tensor, with its
orientation, and the hyperfine coupling tensors both for these
magnetically active metal ions and the 14N atoms of the
ketimide ligands. As seen in Table 3, which compares the
experimental and calculated EPR parameters, the g tensor
anisotropy in these complexes is relatively small, making
computational reproduction of the experimental data challeng-
ing. What is gratifying, however, is that for both [V(N
CMe2)4] and [Nb(NCMe2)4] an essentially axial g tensor is
calculated with g∥ < g⊥ < 2.00. This is the result expected from
simple LFT, as described above, but what is significant is that
the calculated g values themselves are quite close to experiment,
more so than from LFT, particularly for gz (g∥, gmin), which is
the best determined among the various experimental
parameters. For both V and Nb complexes, the g tensors
derived from optimized geometries give slightly better
agreement with experiment than those using the crystal
structures. This may reflect the possibility that the solution
structure of these complexes is less squashed than in the solid
2
2
HOMO is clearly that which we have defined above as dx −y
,
with a π* interaction with the ketimide ligands, which is
consistent with the ligand field analysis. This can be seen in
Figure S25, which presents side-by-side the spin up (α)
HOMOs for [V(NCMe2)4] (MO #73) and [Nb(N
CMe2)4] (MO #82).
The frontier MOs for [V(NCMe2)4] with predominant 3d
character (MOs #73−76 and 78) are shown in Figure 13, with
their calculated energies and representations in D2d symmetry.
Immediately below the HOMO in energy are two degenerate
π-bonding orbitals (MOs #71 and 72) that do have significant
V 3d character. Their π-antibonding counterpart MOs with
higher d character are shown in Figure 13. Overall, the MO
diagram is very similar to that determined previously for the
Co(IV) congener.6 The chief difference between the two cases,
besides the spin quartet d5 configuration for Co(IV), is that the
3d-based MOs are overall higher in energy for V(IV), as
expected from its being in group 5 as opposed to group 9, and
the energy separation between the b1 and e, a1 MOs (i.e., the
L
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
Table 3. Experimental and Calculated g Values and Metal (51V, 93Nb) and Ligand (14N) Hyperfine Coupling Constants (in
a
MHz) for Group 5 Ketimides
aiso (fluid), Aavg
f
d
e
f
giso (fluid), gavg (frozen)
isotope
gz, gx, gy (frozen)
(frozen)
Az, Ax, Ay (frozen)
51V/93Nb
Adip (z, x, y)
51V/ 93Nb
complex
51V/93Nb
14N
[V(NCtBu2)4], 1
expt.
XRD
1.9885, 1.983
1.9852
1.979, 1.985, 1.985
124, 125
−85.0
−114.2
185, 150
−190.3
10.5,
245, 65, 65
120, 60, 60
b
[V(NCMe2)4]
1.9824, 1.9866, 1.9864
1.9756, 1.9903, 1.9903
1.966, 1.999, 1.999
−12.8
−11.5
−214.7, −19.8, −20.8
−255.4, −43.3, −43.8
300, 80, 80
−129.7, +64.1, +65.6
−141.2, +70.8, +70.4
150, 75, 75
c
DFT
expt.
1.9854
[Nb(NCtBu2)4], 2
[Nb(NCMe2)4]
1.9840, 1.988
1.9803
b
XRD
1.9741, 1.9841, 1.9827
1.9646, 1.9882, 1.9882
−8.2
−8.4
−315.0, −127.9, −128.0 −124.7, +62.4, +62.3
−340.1, −135.6, −135.6 −136.4, +68.2, +68.2
c
DFT
1.9803
−203.8
b
a
No sign information on hyperfine coupling constants is available experimentally. XRD = DFT calculation using X-ray diffraction (experimental)
c
d
structure. DFT = DFT calculation using DFT-optimized geometry. The value presented for calculated aiso(14N) is an average of those for the four
e
nitrogen ligands; however, the maximum variation among these calculated values is <0.05 MHz, less than experimental precision. 14N hyperfine
coupling was not resolved in frozen-solution EPR spectra; however, ENDOR spectroscopy recorded at a field corresponding to parallel (z) EPR
transitions provided a value for Az(14N) ≈ 14 MHz (1) and 11 MHz (2). Quadrupole coupling was also estimated from these spectra: Pz(14N) = 0.75
MHz (1) and 1.0 MHz (2). The maximum calculated quadrupole coupling was Pmax(14N) = 1.7 MHz, with a very rhombic tensor (η = |Pmid| − |Pmin|/
|Pmax| = 0.5 (1) and 0.7 (2)); if this were indeed the case, then analysis of a full field-dependent set of experimental ENDOR spectra would be
f
problematic. The calculated orientation of the g and metal nucleus A tensors is with the unique value (i.e., smallest g value, largest magnitude A
value) along the molecular z axis, so gz ≡ g∥ and Az ≡ A∥, with the remaining, essentially equal components exactly in between the molecular x and y
axes (gx ≈ gy ≡ g⊥ and Ax ≈ Ay ≡ A⊥), as defined in Figure 11 and in agreement with LFT.
state, but the agreement may be fortuitous. Regarding the
hyperfine coupling, which is an even greater computational
challenge, the results are also very satisfying. In this case, the
experimental values for A⊥(51V, 93Nb) are at best approximate,
so the agreement in aiso and Adip may well be better than it
appears. The calculated direction of the g and A(51V, 93Nb)
tensors is as expected, with the unique (parallel) value aligned
with the molecular approximate S4 axis, in agreement with LFT.
As seen in Table 3, the calculations reproduce the components
of the axial A(51V) tensor reasonably well, noting that the only
well-determined experimental value is Az (A∥). Here, the results
of the optimized geometry are clearly better than those from
the experimental geometry, again supporting the idea that the
solution structure is less squashed. However, the situation is
less clear for A(93Nb), where each geometry has strengths and
weaknesses in terms of matching experiment. We also note that,
since 2 did not give a single species upon freezing (and 7 was
even more heterogeneous), the expectations for experimental
and calculated correspondence for these complexes are
inherently lower. Nevertheless, the calculated values for the
14N hyperfine couplings are in reasonable agreement with the
experimental values (see Table 3). Even the 14N quadrupole
couplings, of which only one component was estimated from
ENDOR, are consistent between experiment and theory (Table
3). Interestingly, the calculated P(14N) tensor was very rhombic
[P(14N) = [−0.40, −1.26, +1.66] MHz (1); [−0.25, −1.42,
+1.67] MHz (2)], which may be a consequence of the complex
σ-donation/π-donation/π-accepting nature of the metal−
ketimide interaction, as described above, leading to a very
nonspherical electron distribution around the nitrogen atoms.
The quadrupole coupling tensors for 51V and 93Nb were also
calculated and found to be small (e.g., Pmax(51V) = −0.17
MHz). For comparison, vanadyl complexes such as VO(salen)
and VO(OEP) gave Pmax(51V) = −0.29 MHz.58,59 One might
expect a larger quadrupole in heteroleptic vanadyl complexes
than in homoleptic 1, but such generalizations are speculative
given the paucity of such data. Overall, the calculated spin
Hamiltonian parameters for 1 and 2 are fully consistent with
experiment, especially given the deficiencies in experimental
data and in computational modeling.
CONCLUSIONS
■
In summary, we have synthesized a series of homoleptic group
5 ketimide complexes, M(NCtBu2)4 (M = V, Nb, Ta). With
their syntheses, a nearly complete set of first-row M(N
CtBu2)4 complexes (M = Ti−Co), as well as two triads, the
group 6 triad (which was reported by Hoffman and co-
workers)9 and now the group 5 triad, have been isolated and
fully characterized. In the solid state, the group 5 ketimide
complexes feature squashed tetrahedral (D2d) geometries. Both
EPR spectroscopic results and DFT calculations support a
21
dx −y (2B1 in D2d) ground state for this series of complexes.
2
EPR spectroscopy also reveals that the M−L interaction in
M(NCtBu2)4 is far more covalent than the M−L interactions
in the related tetra(aryl), tetra(amido), and tetra(alkoxo)
vanadium(IV) complexes. A ligand field analysis of the
vanadium and niobium congeners also supports the presence
of a strongly covalent metal−ketimide interaction. Most
importantly, however, this comprehensive investigation pro-
vides experimental verification that the ketimide ligand is a
good σ-donor, strong π-donor, and strong π-acceptor, a
combination that is not found in amine, imine, or phosphine
ligands. The combination of strong π-donor and π-acceptor
abilities places the ketimide ligand within a select group of
organometallic ligands. For comparison, both the cyclo-
pentadienyl anion and benzene can act as simultaneous π-
donors and δ-acceptors, which is a similar bonding situation to
that seen for ketimide.89 Although it should be noted that Cp−
is not considered to be a particularly good δ-acceptor
ligand.89,90 We suggest that the combined acceptor and
donor abilities of the ketimide ligand should make it capable
of stabilizing both high and low oxidation states at the same
metal center, potentially permitting the development of unusual
catalytic cycles.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.5b02017.
Crystallographic data (CIF).
M
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Experimental and computational details, spectral data,
Article
(19) Martins, A. M.; Marques, M. M.; Ascenso, J. R.; Dias, A. R.;
Duarte, M. T.; Fernandes, A. C.; Fernandes, S.; Ferreira, M. J.; Matos,
I.; Conceica̧ o Oliveira, M.; Rodrigues, S. S.; Wilson, C. J. Organomet.
̃
and additional figures and tables (PDF).
Chem. 2005, 690, 874−884.
AUTHOR INFORMATION
Corresponding Authors
*(J.T.) E-mail: jtelser@roosevelt.edu.
*(T.W.H.) E-mail: hayton@chem.ucsb.edu.
(20) Heinselman, K. S.; Miskowski, V. M.; Geib, S. J.; Wang, L. C.;
Hopkins, M. D. Inorg. Chem. 1997, 36, 5530−5538.
(21) Marchetti, F.; Pampaloni, G.; Zacchini, S. Dalton Trans. 2007,
4343−4351.
(22) Marchetti, F.; Pampaloni, G.; Zacchini, S. Eur. J. Inorg. Chem.
2008, 2008, 453−462.
(23) Yang, L.; Powell, D. R.; Houser, R. P. Dalton Trans. 2007, 955−
964.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
(24) Dubberley, S. R.; Tyrrell, B. R.; Mountford, P. Acta Crystallogr.,
Sect. C: Cryst. Struct. Commun. 2001, 57, 902−904.
(25) Yamada, J.; Fujiki, M.; Nomura, K. Organometallics 2005, 24,
2248−2250.
■
We thank the National Science Foundation (CHE 1361654)
for financial support of this work. We thank Prof. Brian M.
Hoffman, Northwestern University, for use of the X- and Q-
band EPR spectrometers in his laboratory, which is supported
by the NIH (GM 111097 to B.M.H.). We also thank Prof. T.
David Harris and Kang Du, Northwestern University, for use of
their glovebox and assistance with air-sensitive EPR sample
preparation. We thank Dr. Andrew Ozarowski, National High
Magnetic Field Laboratory, for providing the program
CRYSTAL and assisting with its use. We thank Prof. Jesper
Bendix, Copenhagen University, for providing the program
Ligfield and helpful comments. Finally, we thank Dr. Guang
Wu, University of California, Santa Barbara, for assistance in
collecting X-ray crystallographic data and Anthony DeMartino
for preliminary investigations in this area.
(26) Zhang, W.; Nomura, K. Inorg. Chem. 2008, 47, 6482−6492.
(27) Yamada, J.; Fujiki, M.; Nomura, K. Organometallics 2007, 26,
2579−2588.
(28) Onishi, Y.; Katao, S.; Fujiki, M.; Nomura, K. Organometallics
2008, 27, 2590−2596.
(29) Zhang, W.; Yamada, J.; Nomura, K. Organometallics 2008, 27,
5353−5360.
(30) Głowiak, T.; Grobelny, R.; Jezowska-Trzeblatowska, B.; Kreisel,
̇
G.; Seidel, W.; Uhlig, E. J. Organomet. Chem. 1978, 155, 39−46.
(31) Bott, S. G.; Hoffman, D. M.; Rangarajan, S. P. Inorg. Chem.
1995, 34, 4305−4310.
(32) Perera, T. H.; Lord, R. L.; Heeg, M. J.; Schlegel, H. B.; Winter,
C. H. Organometallics 2012, 31, 5971−5974.
(33) Mashima, K.; Matsuo, Y.; Tani, K. Organometallics 1999, 18,
1471−1481.
REFERENCES
■
(34) Krishnamurthy, R.; Schaap, W. B. J. Chem. Educ. 1969, 46, 799−
810.
(1) Kiplinger, J. L.; Morris, D. E.; Scott, B. L.; Burns, C. J.
Organometallics 2002, 21, 3073−3075.
(2) Graves, C. R.; Vaughn, A. E.; Schelter, E. J.; Scott, B. L.;
Thompson, J. D.; Morris, D. E.; Kiplinger, J. L. Inorg. Chem. 2008, 47,
11879−11891.
(3) Lewis, R. A.; Wu, G.; Hayton, T. W. J. Am. Chem. Soc. 2010, 132,
12814−12816.
(4) Lewis, R. A.; Wu, G.; Hayton, T. W. Inorg. Chem. 2011, 50,
4660−4668.
(35) Suh, S.; Hoffman, D. M. Inorg. Chem. 1996, 35, 5015−5018.
(36) Lehn, J.-S. M.; van der Heide, P.; Wang, Y.; Suh, S.; Hoffman, D.
M. J. Mater. Chem. 2004, 14, 3239−3245.
(37) Wilson, R.; Kivelson, D. J. Chem. Phys. 1966, 44, 154−168.
(38) Labauze, G.; Samuel, E.; Livage, J. Inorg. Chem. 1980, 19, 1384−
1386.
(39) Muller, E. G.; Petersen, J. L.; Dahl, L. F. J. Organomet. Chem.
1976, 111, 91−112.
(5) Seaman, L. A.; Wu, G.; Edelstein, N.; Lukens, W. W.; Magnani,
N.; Hayton, T. W. J. Am. Chem. Soc. 2012, 134, 4931−4940.
(6) Lewis, R. A.; George, S. P.; Chapovetsky, A.; Wu, G.; Figueroa, J.
S.; Hayton, T. W. Chem. Commun. 2013, 49, 2888−2890.
(7) Lewis, R. A.; Smiles, D. E.; Darmon, J. M.; Stieber, S. C. E.; Wu,
G.; Hayton, T. W. Inorg. Chem. 2013, 52, 8218−8227.
(8) Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M.
Advanced Inorganic Chemistry, 6th ed.; John Wiley & Sons, Inc.: New
York, 1999.
(9) Soriaga, R. A. D.; Nguyen, J. M.; Albright, T. A.; Hoffman, D. M.
J. Am. Chem. Soc. 2010, 132, 18014−18016.
(10) Manzer, L. E. Inorg. Chem. 1977, 16, 525−528.
(11) Deutscher, R. L.; Kepert, D. L. Inorg. Chem. 1970, 9, 2305−
2310.
(40) Muller, E. G.; Watkins, S. F.; Dahl, L. F. J. Organomet. Chem.
1976, 111, 73−89.
(41) Alonso, P. J.; Fornies
́
, J.; García-Monforte, M. A.; Martín, A.;
Menjon
́
, B. Chem. - Eur. J. 2005, 11, 4713−4724.
(42) Kirmse, R.; Stach, J.; Kreisel, G. J. Organomet. Chem. 1981, 210,
73−82.
(43) Holloway, C. E.; Mabbs, F. E.; Smail, W. R. J. Chem. Soc. A
1968, 2980−2984.
(44) Kokoszka, G. F.; Allen, H. C., Jr.; Gordon, G. Inorg. Chem. 1966,
5, 91−93.
(45) Froncisz, W.; Hyde, J. S. J. Chem. Phys. 1980, 73, 3123−3131.
(46) Alonso, P. J.; Ara, I.; Arauzo, A. B.; García-Monforte, M. A.;
́
Menjon, B.; Rillo, C. Angew. Chem., Int. Ed. 2010, 49, 6143−6146.
(47) Bigmore, H. R.; Zuideveld, M. A.; Kowalczyk, R. M.; Cowley, A.
R.; Kranenburg, M.; McInnes, E. J. L.; Mountford, P. Inorg. Chem.
2006, 45, 6411−6423.
(12) Schafer, H.; Kahlenberg, F. Z. Anorg. Allg. Chem. 1960, 305,
̈
178−189.
(13) Arteaga-Muller, R.; Tsurugi, H.; Saito, T.; Yanagawa, M.; Oda,
̈
(48) Green, J. C.; Green, M. L. H.; Kaltsoyannis, N.; Mountford, P.;
Scott, P.; Simpson, S. J. Organometallics 1992, 11, 3353−3361.
(49) Noh, W.; Girolami, G. S. Inorg. Chem. 2008, 47, 535−542.
S.; Mashima, K. J. Am. Chem. Soc. 2009, 131, 5370−5371.
(14) Marchetti, F.; Pampaloni, G.; Zacchini, S. Dalton Trans. 2008,
7026−7035.
́
(50) Munha, R. F.; Zarkesh, R. A.; Heyduk, A. F. Inorg. Chem. 2013,
(15) Marchetti, F.; Pampaloni, G.; Zacchini, S. Inorg. Chem. 2007, 47,
365−372.
52, 11244−11255.
(51) Nguyen, A. I.; Blackmore, K. J.; Carter, S. M.; Zarkesh, R. A.;
Heyduk, A. F. J. Am. Chem. Soc. 2009, 131, 3307−3316.
(52) Petersen, J. L.; Griffith, L. Inorg. Chem. 1980, 19, 1852−1858.
(53) Mailer, C.; Taylor, C. P. S. Biochim. Biophys. Acta, Protein Struct.
1973, 322, 195−203.
(16) Casanova, D.; Alemany, P.; Bofill, J. M.; Alvarez, S. Chem. - Eur.
J. 2003, 9, 1281−1295.
(17) Mashima, K.; Tanaka, Y.; Nakamura, A. Organometallics 1995,
14, 5642−5651.
(18) Strauch, H. C.; Erker, G.; Frohlich, R. Organometallics 1998, 17,
̈
5746−5757.
(54) Baute, D.; Goldfarb, D. J. Phys. Chem. A 2005, 109, 7865−7871.
N
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
(55) Mustafi, D.; Galtseva, E. V.; Krzystek, J.; Brunel, L.-C.; Makinen,
M. W. J. Phys. Chem. A 1999, 103, 11279−11286.
(83) Neese, F. ORCAan ab initio, Density Functional and
Semiempirical Program Package, 3.0.3; Max Planck Institut fur
̈
Chemische Energiekonversion: Mulheim an der Ruhr, Germany,
(56) Kirste, B.; Van Willigen, H. J. Phys. Chem. 1982, 86, 2743−2749.
(57) Mulks, C. F.; Kirste, B.; Van Willigen, H. J. Am. Chem. Soc. 1982,
104, 5906−5911.
̈
2014. https://orcaforum.cec.mpg.de/.
(84) Neese, F. Wiley Interdisciplinary Reviews: Computational
Molecular Science 2012, 2, 73−78.
(58) Aznar, C. P.; Deligiannakis, Y.; Tolis, E. J.; Kabanos, T.; Brynda,
M.; Britt, R. D. J. Phys. Chem. A 2004, 108, 4310−4321.
(59) Grant, C. V.; Ball, J. A.; Hamstra, B. J.; Pecoraro, V. L.; Britt, R.
D. J. Phys. Chem. B 1998, 102, 8145−8150.
(60) Fallis, I. A.; Murphy, D. M.; Willock, D. J.; Tucker, R. J.; Farley,
R. D.; Jenkins, R.; Strevens, R. R. J. Am. Chem. Soc. 2004, 126, 15660−
15661.
(61) Mustafi, D.; Makinen, M. W. Inorg. Chem. 2005, 44, 5580−5590.
(62) McGarvey, B. R. J. Phys. Chem. 1967, 71, 51−66.
(63) These values are calculated using the reported frozen-solution
average values for aiso, giso; use of the fluid-solution values gives |P| =
318.8 MHz and |κ| = 0.510.
(85) Schafer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97,
̈
2571−2577.
(86) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7,
3297−3305.
(87) Roemelt, M.; Neese, F. J. Phys. Chem. A 2013, 117, 3069−3083.
(88) Maganas, D.; Sottini, S.; Kyritsis, P.; Groenen, E. J. J.; Neese, F.
Inorg. Chem. 2011, 50, 8741−8754.
(89) Rayon
3308.
́
, V. M.; Frenking, G. Organometallics 2003, 22, 3304−
(90) Crabtree, R. H. The Organometallic Chemistry of the Transition
Metals, 3rd ed.; John Wiley & Sons: New York, 2001.
(64) For 7, |P| = 152 MHz and |κ| = 1.6. This value of |κ| > 1 is not
physically reasonable (i.e., a larger coupling than in the free ion). The
lack of definitive EPR low-temperature data for 7 makes this analysis
inapplicable compared to the V and Nb ketimides.
(65) Even if C2v symmetry is used, this transition, which becomes 2A2
2
→ A1, is still forbidden.
(66) In Td symmetry, this transition directly gives Δtet; in C2v
2
2
symmetry, this would simply split into A2 → B1 (dxz, y allowed)
2
2
and A2 → B2 (dyz, x allowed).
(67) Clark, R. J. H.; Machin, D. J. J. Chem. Soc. 1963, 4430−4433.
2
(68) Note that if the ground state were A1, then this transition
would be allowed with z polarization. Again, even if C2v symmetry is
2
2
used, this transition, which becomes A2 → A1, is still forbidden.
(69) Considering only the MN4 core, the nine normal vibrational
modes in D2d are: a1 (×2, IR forbidden), b1 (IR forbidden), b2 (×2, IR
allowed), and e (×2, IR allowed). Note that lower case is used for
vibrations to distinguish from electronic transitions. Vibronic coupling
with the e modes would make this electronic transition allowed: a1 ×
2
2B1 × (E) × B2 × (e) = A1 + ....
(70) Riley, M. J.; Hitchman, M. A.; Mohammed, A. W. J. Chem. Phys.
1987, 87, 3766−3778.
(71) Alonso et al., in their study of in situ generated [V(C6Cl5)4],
reported bands at 14 950 and 16 680 cm−1 (there appears to be no
resolution between these bands; the higher one could charitably be
called a shoulder). See ref 41.
(72) Figgis, B. N.; Hitchman, M. A. Ligand Field Theory and Its
Applications; Wiley-VCH: New York, 2000.
(73) Miessler, G. L.; Fischer, P. J.; Tarr, D. A. Inorganic Chemistry;
Pearson: Upper Saddle River, NJ, 2014.
(74) The UV−vis spectrum of the related homoleptic alkoxido
complex, V(OtBu)42, consists of a single broad band at 13 500 cm−1,
2
simply assigned to E → T2 in Td symmetry. See ref 44.
(75) Meyer, K.; Bendix, J.; Bill, E.; Weyhermuller, T.; Wieghardt, K.
̈
Inorg. Chem. 1998, 37, 5180−5188.
(76) Meyer, K.; Bendix, J.; Metzler-Nolte, N.; Weyhermuller, T.;
Wieghardt, K. J. Am. Chem. Soc. 1998, 120, 7260−7270.
(77) Fujihara, T.; Schonherr, T.; Kaizaki, S. Inorg. Chim. Acta 1996,
̈
̈
249, 135−141.
(78) Forshaw, A. P.; Smith, J. M.; Ozarowski, A.; Krzystek, J.;
Smirnov, D.; Zvyagin, S. A.; Harris, T. D.; Karunadasa, H. I.; Zadrozny,
J. M.; Schnegg, A.; Holldack, K.; Jackson, T. A.; Alamiri, A.; Barnes, D.
M.; Telser, J. Inorg. Chem. 2013, 52, 144−159.
(79) Desrochers, P. J.; Telser, J.; Zvyagin, S. A.; Ozarowski, A.;
Krzystek, J.; Vicic, D. A. Inorg. Chem. 2006, 45, 8930−8941.
(80) Bendix, J.; Brorson, M.; Schaffer, C. E. Inorg. Chem. 1993, 32,
̈
2838−2849.
(81) McGarvey, B. R. In Transition Metal Chemistry; Carlin, R. L.,
Ed.; Marcel Dekker: New York, 1966; pp 89−201.
(82) It is possible to match all of the bands exactly by dividing the
ligands into pairs, but this allows for six fit parameters and is thus
overparametrized.
O
DOI: 10.1021/acs.inorgchem.5b02017
Inorg. Chem. XXXX, XXX, XXX−XXX