Evaluation of donor and steric properties of anionic ligands on high valent transition metals

Article abstract of DOI:10.1021/ic202524r

Synthetic protocols and characterization data for a variety of chromium(VI) nitrido compounds of the general formula NCr(NPrⁱ₂) ₂X are reported, where X = NPrⁱ₂ (1), I (2), Cl (3), Br (4), OTf (5), 1-adamantoxide (6), OSiPh₃ (7), O ₂CPh (8), OBu^t_F6 (9), OPh (10), O-p-(OMe)C ₆H₄ (11), O-p-(SMe)C₆H₄ (12), O-p-(Bu^t)C₆H₄ (13), O-p-(F)C₆H ₄ (14), O-p-(Cl)C₆H₄ (15), O-p-(CF ₃)C₆H₄ (16), OC₆F₅ (17), κ(O)-N-oxy-phthalimide (18), SPh (19), OCH₂Ph (20), NO ₃ (21), pyrrolyl (22), 3-C₆F₅-pyrrolyl (23), 3-[3,5-(CF₃)₂C₆H₃]pyrrolyl (24), indolyl (25), carbazolyl (26), N(Me)Ph (27), κ(N)-NCO (28), κ(N)-NCS (29), CN (30), NMe₂ (31), F (33). Several different techniques were employed in the syntheses, including nitrogen-atom transfer for the formation of 1. A cationic chromium complex [NCr(NPrⁱ₂) ₂(DMAP)]BF₄ (32) was used as an intermediate for the production of 33, which was produced by tin-catalyzed degredation of the salt. Using spin saturation transfer or line shape analysis, the free energy barriers for diisopropylamido rotation were studied. It is proposed that the estimated enthalpic barriers, Ligand Donor Parameters (LDPs), for amido rotation can be used to parametrize the donor abilities of this diverse set of anionic ligands toward transition metal centers in low d-electron counts. The new LDPs do not correlate well to the pK_a value of X. Conversely, the LDP values of phenoxide ligands do correlate with Hammett parameters for the para-substituents. Literature data for ¹³C NMR chemical shifts for a tungsten-based system with various X ligands plotted versus LDP provided a linear fit. In addition, the angular overlap model derived e_- + e_π values for chromium(III) ammine complexes correlate with LDP values. Also discussed is the correlation with XTiCp*₂ spectroscopic data. X-ray diffraction has been used used to characterize 31 of the compounds. From the X-ray diffraction data, steric parameters for the ligands using the Percent Buried Volume and Solid Angle techniques were found.

Full text of DOI:10.1021/ic202524r

Article
pubs.acs.org/IC
Evaluation of Donor and Steric Properties of Anionic Ligands on High
Valent Transition Metals
Stephen A. DiFranco, Nicholas A. Maciulis, Richard J. Staples, Rami J. Batrice, and Aaron L. Odom*
Department of Chemistry, Michigan State University, East Lansing, Michigan 48824, United States
S
* Supporting Information
ABSTRACT: Synthetic protocols and characterization data for a variety of
chromium(VI) nitrido compounds of the general formula NCr(NPri2)2X
are reported, where X = NPri2 (1), I (2), Cl (3), Br (4), OTf (5), 1-
adamantoxide (6), OSiPh3 (7), O2CPh (8), OButF6 (9), OPh (10), O-p-
(OMe)C6H4 (11), O-p-(SMe)C6H4 (12), O-p-(But)C6H4 (13), O-p-
(F)C6H4 (14), O-p-(Cl)C6H4 (15), O-p-(CF3)C6H4 (16), OC6F5 (17),
κ(O)-N-oxy-phthalimide (18), SPh (19), OCH2Ph (20), NO3 (21),
pyrrolyl (22), 3-C6F5-pyrrolyl (23), 3-[3,5-(CF3)2C6H3]pyrrolyl (24),
indolyl (25), carbazolyl (26), N(Me)Ph (27), κ(N)-NCO (28), κ(N)-
NCS (29), CN (30), NMe2 (31), F (33). Several different techniques were
employed in the syntheses, including nitrogen-atom transfer for the
formation of 1. A cationic chromium complex [NCr(NPri2)2(DMAP)]BF4
(32) was used as an intermediate for the production of 33, which was
produced by tin-catalyzed degredation of the salt. Using spin saturation transfer or line shape analysis, the free energy barriers for
diisopropylamido rotation were studied. It is proposed that the estimated enthalpic barriers, Ligand Donor Parameters (LDPs),
for amido rotation can be used to parametrize the donor abilities of this diverse set of anionic ligands toward transition metal
centers in low d-electron counts. The new LDPs do not correlate well to the pKa value of X. Conversely, the LDP values of
phenoxide ligands do correlate with Hammett parameters for the para-substituents. Literature data for 13C NMR chemical shifts
for a tungsten-based system with various X ligands plotted versus LDP provided a linear fit. In addition, the angular overlap
model derived eσ + eπ values for chromium(III) ammine complexes correlate with LDP values. Also discussed is the correlation
with XTiCp*2 spectroscopic data. X-ray diffraction has been used used to characterize 31 of the compounds. From the X-ray
diffraction data, steric parameters for the ligands using the Percent Buried Volume and Solid Angle techniques were found.
published in Chemical Reviews in 1977 on steric and electronic
effects in phosphine chemistry.1 Tolman’s cone angle gave a
reasonable one-parameter metric for sterics. The energy of the
totally symmetric carbonyl vibration in Ni(CO)3(PR3)
complexes gave a useful single-parameter metric for donor
properties.
Using CO stretching frequencies to parametrize later
transition metal ligand effects predates this Tolman review,
however. For example, Wilkinson and co-workers in 1959
reported the IR stretching frequencies of transition metal
carbonyls bearing amine/phosphine donors and reported “the
resulting negative charge on the metal atom R3N+−M− may...be
dissipated by increasing the bond orders in the M−C−O
system”.2 They reported a steady rise in carbonyl stretching
frequencies on replacement of phenyl groups in (Ph3P)3Mo-
(CO)3 with chlorine until reaching (Cl3P)3Mo(CO)3.3
Parameterization methods have been extensively used in a
large variety of low-valent and late transition metal catalytic
reactions.4 Similar quantitative measures are a mainstay of
physical organic chemistry; for example, the reactivity of
INTRODUCTION
■
One of the most important methods for controlling reaction
pathways in many transition metal catalyzed systems is through
the steric and electronic adjustment of ancillary ligands.
Choosing from the extensive gallery of possible ligand choices
is often done by (1) analogy with similar reactions already in
the literature, (2) picking readily available ancillaries in the
investigator’s laboratory, or (3) making an educated guess
based on experience in the field. Once some of the desired
reactivity is found, reactions are optimized through similar
procedures involving available ligand sets and trying to
encourage hypothesized processes with slow reaction rates.
Finally, a reaction may be deemed interesting enough to
warrant full mechanistic investigations through experimental
and computational techniques that can often lead to improved
catalyst designs.
In this process of taking new reactions from conception to
fruition, the donor properties and steric profiles of the ancillary
ligands, along with reaction conditions, provide the major tools
for optimization. Simple metrics for donor and steric properties
have proven to be powerful tools for catalyst optimization,
especially in late transition metal chemistry. Perhaps one of the
most familiar citations in chemistry is by Chadwick Tolman
Received: November 22, 2011
Published: December 21, 2011
© 2011 American Chemical Society
1187
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
compounds with pendant aryl groups is often predicted or
explained by parameters developed by Hammett, Taft, and
others.5 Quantitative structure−activity relationships (QSAR)
have developed into a powerful tool for other areas as well, e.g.,
pharmaceutical design.6
In contrast, methods for determining donor properties of
ligands on metal complexes in higher formal oxidation states
are less well-known. This is despite the fact that the donor
properties of common ligands on earlier, higher-valent metals
are likely to be quite different from later, lower valent metals in
many cases because of differences in the number and type of
empty acceptor orbitals.
If the ligands in question are members of a closely related
series, e.g., para-substituted phenoxides, one can try to draw
analogies to pKa or Hammett parameters; however, the
investigator is left wondering if these are good measures for
the transition metal system in question. This problem is only
exacerbated if the ligands are more dissimilar, such as
comparing phenoxide to iodide to indolyl. Different donor
atoms, e.g., oxygen versus nitrogen, or even different hybrid-
ization of the same donor atom may affect radial extensions for
the orbitals even if the frontier orbitals are of similar shape,
which could lead to quite different bonding properties because
of the changes in overlap integrals and energies.
While QSAR has been done on early to middle transition
metal complexes, these are often studies of specific systems
with limited applicability to high valent metals in general. For
example, extensive QSAR has also been done in recent
computationally driven studies like the one published by
Jensen and co-workers on Grubbs’s catalyst.7 Steric and
electronic parametrizations have been applied to metallocene8
and nonmetallocene9 polymerization catalysts using a variety of
techniques varying from simply categorizing ligand types to
numerical quantization of ligand properties. In addition,
electronic influence of substituents on ansa-metallacene
complexes has been examined in great detail.10
Section. In addition, the amido ligands display variable rotation
rates dependent upon the donor properties of X. The rotation
of the diisopropylamido ligand in these systems has a rate that
1
is readily measured by H NMR spectroscopy. In this study,
spin saturation transfer was the standard method for ligand
rotation rate determination; however, line-shape analysis can
also be used and was used for some compounds (vide infra).
Since the compounds are pseudo-tetrahedral, the system is
not orthoaxial, and the σ- and π-orbitals mix during the
bonding interactions with the ligands. This is exemplified in the
Angular Overlap Model (AOM) Parameters for a rigorously
tetrahedral compound by the energy of the t2* orbital being
parametrized as et * = 4/3 eσ + 8/9 eπ where the ligand’s σ- and
2
π-donor parameters both contribute to the energy of the triply
degenerate orbital. Lowering the symmetry, as is done here, will
lead to further mixing, and a single parameter for ligand donor
properties is the result.
The highest symmetry available in the nitrido compounds
here would be C3v with a formula of NCrX3, where the
2
2
chromium-nitrido bond is along the z-axis. In C3v, the dxy/dx −y
-
orbitals comprise an e-set, with the px/py-orbitals having the
same symmetry. These e-sets act as both σ- and π-acceptor
orbitals for the basal amido ligands. In addition to the π-
accepting e-set near the xy-plane, there is an e-set composed of
the dxz/dyz-orbitals that are involved in strong π-interactions
with the nitrido.13
In other words, rotation of the amido ligands 90° from where
they π-donate into acceptor orbitals near the xy-plane to where
they could donate into the dxz/dyz orbitals along the nitrido
vector causes them to compete with the very strongly donating
nitrido. As a result, there is an electronic barrier to rotation
around the Cr−NPri2 bond determined by the energy
difference between the geometries where the amido CrNR2
plane is parallel with the Cr−N(nitrido) vector and where it is
perpendicular. Increasing the donor abilities of the ligands in
the basal set reduces this difference somewhat, decreasing the
barrier for amido rotation through ground-state destabilization.
In short, the stronger a donor X is in a compound like
NCr(NPri2)2X, the smaller the barrier to amido rotation is
expected to be. Because the σ- and π-systems are strongly
mixed, the σ- and π-donor properties of X both contribute to
the size of the diisopropylamido rotational barrier. We propose
that this isomerization barrier can be used as a measure of the
donor ability of X in high valent transition metal systems.
The above arguments can be illustrated using Density
Functional Theory (DFT) on the model system NCr(NH2)3
using B3LYP as the functional. Our initial exploration with this
molecule used the 6-31G** basis set, but the calculations were
extended to the much larger aug-cc-pVQZ basis set, which
provided similar results. Optimization provided a ground state
structure where all of the amido ligands are planar and the Cr−
NH2 planes are parallel to the nitrido vector as expected
(Figure 1, bottom).
For our investigations in titanium catalysis,11 we sought a
method for the comparison of a large variety of monodentate
ligands on early metals to aid in ligand design and for
understanding spectroscopic and reactivity trends within
various transition metal systems.
In this study, we discuss the use of a large selection of
monodentate ligands on the d0 metal complex, NCr(NPri2)2X,
where X is an adjustable monodentate ligand. As will be shown,
the synthetic versatility of this framework allows synthesis of a
series of compounds for evaluation. In this manuscript, we limit
the discussion to monoanionic X;12 however, these range from
common ancillaries used in organometallic chemistry like
amido and alkoxide to classical Werner-type ligands such as
halides, cyanide, and thiocyanate. The system’s design lends
itself to a one parameter quantification of donor properties
similar to the Ni(CO)3L system commonly used for late
transition metal ligands. Steric metrics for the ligands are
provided, and possible steric interference is discussed.
If NCr(NH2)3 is reoptimized while restricting the dihedral
angle in one of the amido ligands to induce rotation, the
rotating nitrogen pyramidalizes as the lone pair on the amido
approaches the nitrido π-orbitals.14 In other words, the nitrido
prohibits significant π-donation from the rotating amido when
it would donate into the same orbital. The energy of the
complex increases with increasing amido nitrogen hybridization
parameter (λ in spλ) from λ = 2, that is, sp2, in the ground state
to around λ = 2.8 (Figure 1, plot) at the transition state for the
rotation.15 From the calculations an enthalpic barrier, ΔH⧧, of
RESULTS AND DISCUSSION
■
I. System Used for Ligand Parameterization. The
method chosen here for the experimental parametrization of
ligand donor properties on high valent metal centers involves
the use of chromium(VI) nitrido complexes with diisopropyl-
amido ancillaries. All of the compounds in this study are of the
type NCr(NPri2)2X, where X is a monoanionic ligand. These
complexes are readily prepared, as will be shown in the next
1188
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
of compensating effects around the metal that we are measuring
in this study.
In the actual diisopropylamido complexes used in the
experimental studies, this degree of pyramidalization may not
be possible for steric reasons. However, the hybridization of the
amido nitrogen in the model is illustrative of the type of
electronic changes expected on rotation. Donation into the
same orbitals as the strongly donating nitrido is energetically
unfavorable, and, in the model at least, this competition for the
metal’s acceptor orbitals manifests as amido rehybridization.
In this study, the NCr(NPri2)2 fragment is held constant, and
the barrier to rotation of the diisopropylamido ligands in this
constant fragment are what is being measured. The X
substituents affect the amido barrier to rotation only indirectly,
and the only changes from one complex to another are the
electronic and steric components of X in NCr(NPri2)2X.
This system has several advantages for this type of study.
First, the compounds prepared thus far have good to excellent
thermal stability. Second, the complexes are diamagnetic,
allowing easy use of NMR for evaluation. Third, the Cr(VI)
nitrido compounds tend to be pseudotetrahedral; we have not
observed dimers with bridging X ligands in this system, for
example. Fourth, ligands tend to be monodentate on the metal
allowing a more uniform comparison between various ligand
sets. Even ligands such as carboxylate, with a strong tendency to
have higher hapticity in most complexes, only show what
appear to be weak secondary interactions with the metal if any
(vide infra). Fifth, the NCr(NPri2)2X complexes are readily
prepared from inexpensive reagents with an extraordinary
variety of X as will be described next.
Figure 1. DFT B3LYP/aug-cc-pVQZ computational results for
NCr(NH2)3. On the bottom are two views of the computed ground
state, with the right view looking down one of the three equivalent
amido-chromium vectors. In the middle is a plot of the hybridization
parameter (λ) vs the calculated enthalpic energy of the complex with
the ground state set to 0 kcal/mol. On the top are two views of the
transition state structure found for rotation of one amido ligand, with
the right view looking down the rotated amido-chromium vector.
II. Preparation of NCr(X)(NPri2)2 Complexes and
Characterization. Here, we begin by discussing a new
synthetic protocol based on nitrogen-atom transfer for the
formation of NCr(NPri2)3 (1). All other complexes are
prepared by modification of 1. The synthetic protocols for
the production of the other NCr(X)(NPri2)2 complexes will be
divided into 7 categories: protonolysis with lutidinium halides,
protonolysis with HX, exchange using thallium salts, exchanges
between lithium salts and the chromium phenoxide, metathesis
using sodium salts, ligand exchange with lithium to zinc
transmetalation, and tin(IV)-catalyzed decomposition of a
cationic BF4 salt.
5.7 kcal/mol was found using the aug-cc-pVQZ basis set.
Meanwhile, the Cr−N(nitrido) bond distance seems virtually
unaffected by the rotation, varying by less than 0.01 Å over the
entire course of the rotation.
The transition state for amido rotation, which had a single
negative vibration, was found at 61° in the N(nitrido)-Cr−
N(amido)−H dihedral. Figure 1 has images of the ground state
and transition state structures. Also in Figure 1 is a plot of λ
versus enthalpy. The hybridization parameter increases fairly
smoothly up to the transition state, consistent with competition
with the nitrido π-donation being the cause of amido
pyramidalization.
The amido distance does seem to change slightly with
rotation because of bond order effects with the chromium, but
the relationship is complicated by electronic adjustments made
by the other amido ligands. The average Cr−NH2 distance in
the ground state from the calculations is 1.83 Å. In the
transition state, the pyramidalized (rotating) amido distance
increases to 1.90 Å, but the amido ligands not undergoing
rotation shorten their distance to chromium to 1.81 Å. As a
result, the average Cr−NH2 distance in the transition state is
1.84 Å, essentially identical to the ground state. In other words,
rotating one ligand causes ripples of change through the other
ligands. It is these indirect changes in the amido ligands because
A. Synthesis of NCr(NPri2)3 (1) by Nitrogen Atom Transfer.
The starting material for all of this chemistry is the Bradley
complex, Cr(NPri2)3, prepared on large scales from CrCl3 and
LiNPri2 in ethereal solvent.16 The 3-coordinate compound is
soluble in hydrocarbons and crystallizes as large black plates. In
the previously reported synthesis of nitrido NCr(NPri2)3 (1),
black solutions of Cr(NPri2)3 reacted with NO gas to form
orange ONCr(NPri2)3, which was deoxygentated with
vanadium(III) to form the terminal nitrido.17
For this work, a somewhat more straightforward synthesis
was used where Cr(NPri2)3 was treated with NCr(OBut)3 to
give NCr(NPri2)3 (1) through a nitrogen-atom transfer (eq 1).
Yellow NCr(OBut)3 is available from chromyl chloride in a
one-pot procedure published by Chiu and co-workers.18 The
19
nitrido was readily separated from the oily Cr(OBut)3
byproduct by washing with acetonitrile.
1189
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
B. Syntheses Using Protonolysis with Lutidinium Halides.
Dark beet-red 1 was converted to orange NCr(I)(NPri2)2 (2)
with 2,6-lutidinium iodide using the published procedure.20
Using methods similar to the iodide synthesis, we prepared the
chloride (3) and bromide (4) complexes. The syntheses
involved the addition of anhydrous 2,6-lutidenium halide to 1
in chloroform at 60 °C (eq 2).
C. Syntheses Using Direct Protonolysis with HX. For this
study, a total of 15 complexes were prepared using the direct
addition of HX, where X is the new desired ancillary ligand (eq
3). For the synthesis of most alcohols, silanols, carboxylates,
and thiolates, direct protonolysis on 1 turned out to be the
most convenient and highest yielding methodology. The
reactions were carried out using toluene as the solvent for all
of these cases, with the exception of triflate where DME/
pentane was employed.
The other conditions required for the syntheses varied widely
depending on the substrate HX. Some reactions, such as with
triflic acid, worked best when started at near frozen
temperatures with short stirring times at room temperature.
Other substrates were heating for several days to get good
conversion; for example, the reaction with 1-adamantanol
(HOAd) required heating at 90 °C for 3 days.
Protonolysis of 1 with HOBn proved to be slow and not very
clean. Using TlOBn, NCr(NPri2)2(OBn) (19) was prepared in
good yield from iodide 2. Hexanes or toluene were used for the
majority of these transmetalation reactions.
Likewise, commercially available TlNO3 gave NCr-
(NPri2)2(NO3) (20); in this case, tetrahydrofuran (THF) was
advantageous because of the low solubility of the thallium salt
in most other solvents.
Thallium was especially useful for pyrrolyl and pyrrolyl
derivatives. The thallium salts were readily available by simple
reaction of TlOEt with the NH-pyrrole. The thallium pyrrolyls
seemed at best sparingly soluble in any solvent with which they
1
did not react, as evinced by being H NMR silent as saturated
solutions in several solvents, but the compounds reacted readily
and cleanly with the iodide 2. The chromium complexes of
pyrrolyl (22) and two different 3-aryl-pyrroles (23 and 24, eq
4) were prepared using this procedure.
E. Syntheses Using Exchange with Lithium Salts. When
using lithium reagents, reduction of iodide 2 to the known μ-
nitrido chromium(V) dimer20 [NCr(NPri2)2]2 was evident.
Transmetalation using the phenoxide 10 was often more
successful with these reagents, and this method was used to
prepare the indolyl (25), carbazolyl (26), and N-methylanilide
(27) complexes (eq 5) from their respective lithium salts. A
similar method was used in the conversion of NCr-
(OPh)2(NPri2) to NCr(CH2SiMe3)2(NPri2) in work from the
Cummins laboratory.21
F. Syntheses Using Exchanges with Sodium Salts. The
three complexes NCr(NPri2)2(X), where X = NCO (28), NCS
(29), and CN (30), were prepared (eq 6) from the
commercially available NaX salts and NCr(NPri2)2(I) (2).
The main difficulty with the reactions was the low solubility of
these reagents in organic solvents. Acetonitrile was used as the
solvent, and reactions required relatively long reaction times
and/or mild heating. In the case of NaCN, 1 equiv of 15-crown-
5 was advantageous.
Two of the compounds in eq 3, 9 and 10, have been
previously reported.21
D. Syntheses Using Exchange with Thallium Salts.
Thallium salts were advantageous in several cases for the
production of new NCr(NPri2)2X complexes. The reactions
with iodide (2), led to rapid precipitation of TlI, which is
readily removed by filtration. The reactions were generally
clean, and the use of thallium avoids unwanted reduction
processes found using some other reagents (vide infra). Some
obvious disadvantages for thallium are the toxicity of the metal
and lack of stability with some X substituents. Thallium was
employed in the preparation of five of the complexes (eq 4).
G. Ligand Exchange with Zinc Transmetalation. For
probing steric effects in the barriers to rotation (vide infra), it
was desirable to synthesize the dimethylamido complex
1190
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
NCr(NPri2)2(NMe2) (31) for comparison with NCr(NPri2)3
(1). The most fruitful route (eq 7) we discovered to 31
involves treating ZnCl2 with LiNMe2 in DME/THF solvent.
Presumably, a Zn(NMe2)2 solvate or perhaps an amido-
containing zincate complex is prepared. This mixture does
transmetalation with iodide 2 more cleanly than with the
lithium salt alone. Attempts to prepare 31 directly from 2 with
LiNMe2 resulted largely in reduction.
For example, the nitrido distance in the poorly donating triflate
(5), strongly donating and relatively small benzyloxy (19), and
the large and strongly donating diisopropylamido (1) were
found to be 1.543(3), 1.543(2), and 1.544(3) Å, respectively.
The full range of Cr−N(nitrido) values is 1.524(3) in nitrate 21
to 1.553(4) Å in O-p-(CF3)C6H4 16. However, there is no
obvious correlation between this distanceor, for that matter,
any other metric parameters investigatedand any of the steric
or electronic parameters derived.
The two chromium-diisopropylamido bond distances were
generally the same within error. The only exception in this list
of compounds was for carbazolyl 26, which had Cr−
N(diisopropylamido) distances of 1.796(2) and 1.833(2) Å.
The carbazolyl plane is tilted from the Cr−N(nitrido) vector
(Figure 2) and, according to the space filling models, there is
H. Tin(IV)-Catalyzed Decomposition of a Cationic BF4 Salt.
Many different methods were tried in attempts to prepare the
fluoro complex. Success was finally found when we generated a
cationic complex by treatment of the iodo 2 with AgBF4 in the
presence of DMAP to form [NCr(NPri2)2(DMAP)]BF4 (32).12
Thermal decomposition of this complex does form small
amounts of fluoride NCr(NPri2)2(F) (33) but also gives a large
amount of unidentified side-products. In one attempt to form
the fluoride by transmetalation, we reacted 32 with FSnBun ,
3
which gives fluoro 33 relatively cleanly (Scheme 1).
Scheme 1. Synthesis of NCr(NPri2)2(F) (33)
Subsequently, we found that FSnBun could be used catalyti-
3
cally. It appears that Sn(IV) complexes can catalyze the
decomposition of 32, as can some other mild Lewis acids.22
The expected byproduct, DMAP·BF3, was easily detected in the
19F NMR spectrum of a reaction to form 33 carried out in an
NMR tube.23
III. Single-Crystal X-ray Diffraction Studies. All of the
compounds of the formula NCr(NPri2)2(X), where X is an
anionic substituent, have been structurally characterized.24 As
might be expected, the Cr−N distances, both nitrido and
amido, are not exceedingly sensitive to changes in X outside the
error limits of the X-ray diffraction experiment. All of the
compounds exhibit diisopropylamido ligands with the Cr−NC2
amido planes parallel to the Cr−N(nitrido) vector, as expected
from the electronic structure (vide supra). Discussion of the X-
ray structure of each compound would be gratuitous; however,
a few of the more salient features will be addressed in this
section.
The structural characterization on so large a number of
derivatives was carried out predominately to facilitate the steric
analyses discussed below, to determine if there were any
secondary interactions (e.g., bidentate ligands), and to ascertain
if any of the structural parameters might reliably correlate with
the electronic and steric features. The presentation of the
results here will be brief; cif files and crystallographic tables can
be found in the Supporting Information.
Figure 2. Spacefilling views of NCr(NPri2)2(carbazolyl) (26). The top
view is looking down the Cr−N(carbazolyl) bond showing the tilting
of the heterocyclic framework. The bottom view is anti to the nitrido
and shows the tilted carbazolyl ring’s close contacts with one of the iso-
propyl groups.
steric clash between the aromatic ring of the carbazolyl anti to
the nitrido and a diisopropylamido ligand. The longer Cr−
NPri2 distance is associated with the amido closer to the tilted
carbazolyl on the side anti to the nitrido (the left NPri2 group in
the top of Figure 2). However, the average Cr−NPri2 distance
The donor abilities of the X ligands did not have a large
impact on the chromium-nitrido distance. Indeed, the nitrido
distance is very similar for all the complexes measured thus far.
1191
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
Figure 3. Plot of average Cr−NPri2 distance (Å) vs the donor ability of X (LDP in kcal/mol).
in 26 is similar to the other compounds, and it appears that the
steric influence of the carbazolyl is mostly to differentiate the
two diisopropylamidos in the solid state. (The two
diisopropylamido groups are equivalent in solution.)
nitrate containing 21, the Cr−O distance is 1.973(3) Å, and
there is a possible weak interaction with a second oxygen of the
nitrate nearly trans to the nitrido, which is quite long at over 2.7
Å. For benzoate 8, which has a similar structure near the metal
center, the short Cr−O distance is 1.924(1) Å with a possible
interaction with the second carboxylate oxygen that is around
3.0 Å away. The contributions from these secondary
interactions of X to the measured donor abilities are unlikely
to be large at those distances but are not known.
The NCr(NPri2)2 molecular fragment showed some
variability in its metric parameters. For example, the amido-
chromium-amido angle varied from 116.1(5)° (X = N(Me)Ph
27) to 124.9(1)° (X = CN 30), and the average Cr−N(amido)
distances varied from 1.805(3) Å (X = OTf 5) to 1.842(2) Å
(X = NPri2 1). While attempted correlations with these metric
parameters and the donor parameters are suggestive, plots of
N(amido)−Cr−N(amido) angles and Cr−N(amido) distances
with either the steric parameters or the donor properties
gleaned from the NMR data showed no strong correlations.
Steric factors seem evident in the tris(diisopropylamido)
complex 1 according to other data (vide infra), but it is difficult
to discern this from the X-ray diffraction studies alone. The
average Cr−N(amido) distance in the published structure for 1
is 1.842(3) Å. This distance in 1 is somewhat larger than many
of the derivatives prepared. For example, the average Cr−
N(diisopropylamido) distances for a few derivatives are Cl 3
1.813(2), OBn 20 1.823(1), OAd 6 1.822(7), N(Me)Ph 27
1.830(2), and OTf 5 1.805(3) Å. However, the average
diisopropylamido distance in 1 is very much in line with the
sterically less encumbered NMe2 31 with average Cr−N
distances of 1.842(4) Å; incidentally, 31 was one of the
compounds examined that displayed full molecule disorder in
the X-ray diffraction experiments. The disorder was fully
modeled.
IV. Measurement of Amido Rotational Barriers and
the Ligand Donor Parameters (LDP). Most of the
NCr(NPri2)2(X) complexes employed in this study exhibit
two distinct methyne peaks for the diisopropylamido ligands at
room temperature assigned as being syn and anti to the nitrido
substituent. The methyne that is anti is assigned as being
deshielded relative to the syn methyne resonance on the basis of
2D NMR experiments on iodo 2 (see the Supporting
Information). A few of the compounds where X is a strong
donor ligand, for example, dimethylamido and 1-adamantoxide,
have the methynes at or near the coalescence point at room
temperature. One of the reasons this system was chosen was
that the rate of diisopropylamido rotation was easily measured
by 1H NMR spectroscopy. Once the rate constants were
known, the Eyring equation was used to determine the free
energy barriers to rotation, ΔG⧧rot, relative to X.
The rate constant for the exchange of the two methynes in
the isopropyls, in the majority of cases, was measured using
26
1
Spin Saturation Transfer (SST) in the H NMR. The ideal
temperature for the SST experiment was found to be between
−56 °C and +27 °C, depending on the rate of rotation for the
particular complex being studied. Detailed descriptions of how
the SST experiments and error analyses were done can be
found in the Supporting Information. In addition, there is a
detailed discussion in the Supporting Information of how the
T1 values were found and the types of T1 values to use.
For one complex, NCr(NPri2)2(NMe2) (31), because of
instrument limitations, we were unable to reach the slow
exchange temperature. Line Shape Analysis (LSA) was used to
determine the barrier for the isopropyl exchange rather than
SST.
A plot (Figure 3) of the average diisopropylamido distance
versus the Ligand Donor Parameter (LDP) described below
shows no clear correlation. This may be due to the errors in the
structural parameters considering the less donating ligands
(toward the right in the plot) do, generally speaking, seem to
have shorter Cr−N(amido) distances and the more donating
ligands seem to have generally longer distances. However, the
scatter in the data is far too large to make anything resembling
an accurate correlation. In fact, the shortest Cr−N(amido)
averages are found for two aryloxide derivatives with moderate
LDP values (vide infra) for this series.25
Two compounds, benzoate 8 and nitrate 21, possibly show
weak secondary interactions between Cr and the X ligand. For
1192
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
Figure 4. Ligand Donor Parameters (kcal/mol) for various X in NCr(NPri2)2X with the associated errors.
The X = 1-adamantoxide 6 complex was studied using both
Under the assumption that entropy differences are minimal, a
LSA and SST for comparison. Both gave the same value to the
set of values approximating ΔH⧧ is obtained. Considering
rot
nearest tenth of a kcal/mol at ΔG⧧ = 12.8 kcal/mol.
that the values are an approximation based on the assumption
of ΔS⧧rot = −9 cal/mol K and their uses for the purposes of this
study are more dependent on their relative rather than absolute
magnitudes, we call each value a Ligand Donor Parameter
(LDP). The LDPs are collected in Figure 4 with horizontal
error bars allowing quick distinguishing of ligands that are
different outside error limits. The current best numerical
parameters are collected in Table 1.
The values in Figure 4 and Table 1 constitute our current
best measurements on this particular series for the enthalpies of
amido rotation at this time. There are a number of interesting
series that one can look at qualitatively. Quantitative
comparisons will be discussed in Section VI after discussion
of steric influences on rotational barriers.
rot
It was expected that the entropy associated with the
diisopropylamido rotation would be more or less constant
over the series. To investigate this assertion, Eyring plots were
done on several of the compounds. The plots were done over
as large a temperature range allowable by the kinetics of
rotation and our instrumentation. In addition, we explored
compounds that varied in ΔG⧧ and sterics as much as
rot
possible. Consequently, the Eyring plots using SST were
determined for iodo 2, benzyloxy 20, NPri2 1, and O-p-SMe-
C6H4 12. For these four compounds, ΔS⧧ was found to be
rot
−9, −6, −5, and −3 cal/mol K, respectively. Consequently, the
entropy values appear to be small and negative. The entropy
values were found from variable temperature (VT) experiments
over temperature ranges of 47, 36, 26, and 44 K, respectively.
Additional information on the entropy measurements is found
in the Supporting Information.
The halides are in the expected order with iodide being the
least donating and fluoride the most donating.
Looking at some monoanionic nitrogen-based heterocycles,
it was found that pyrrolyl was a far poorer donor than indolyl,
which was a poorer donor than carbazolyl. This is consistent
with the expected availability of the nitrogen lone pair for
donation in these particular heterocycles. The pyrrolyl ring’s
aromaticity is dependent upon use of the nitrogen lone pair to
It is most desirable to place the rotational barriers in terms of
ΔH⧧rot, to remove some of the temperature dependence
associated with the measurements. Each compound had to be
measured at a temperature best suited for its particular rotation
kinetics to measure the rate constant as accurately as possible.
Consequently, the SST and LSA data were collected at different
temperatures for each complex.
reach the 6 π-electrons required by the Huckel rule for
̈
aromaticity. As a consequence, the aromatic stabilization energy
of pyrrole directly competes with π-donation, which leads to
pyrrolyl being a poor π-donor.30 For indolyl and even more so
for carbazolyl, the aromaticity of the 5-membered heterocycles
must compete with the 6-membered carbocycle(s) in resonance
form contributions to the aromaticity.31 As a result, the
nitrogens in indolyl and carbazolyl seem to donate more
strongly to the metal center than pyrrolyl because of the greater
availability of their nitrogen-based lone pairs.
Here, we are always measuring the kinetics for the rotation of
a diisopropylamido ligand in a NCr(NPri2)2(X) complex. We
assume that the ΔS‡rot values for the compounds will all be similar.
Even in cases where the X ligand is quite large and steric effects
are likely, that is, for X = NPri2, we have not observed large
deviations in ΔS⧧ values.27
rot
The most reliable measurement of entropy, based on where
the slow and fast exchange limits occur relative to our available
instrumentation, appears to be the value for iodo 2, which was
done over a 47 K interval. As a result, −9 cal/mol K was used as
the entropy barrier for most of the compounds.28 The only
compound calculated differently is X = NMe2 31, where the
activation barriers were found using a different technique
(LSA). The experimental barriers for 31 were determined to be
ΔS⧧ = −4 cal/mol K and ΔH⧧ = 9.3 kcal/mol.29
The strongest donors explored thus far are dialkylamido and
alkoxides, which were thought previously to be strong σ- and π-
donors. The weakest donors in the series are those with poor
overlap due, in all likelihood, to poor size matches between
orbitals such as in iodo and thiophenolate, cf. phenolate, or
where the X ligand has a competing π-system that limits π-
donation such as NCS, benzoate, and pyrrolyl.
rot
rot
1193
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
radius of 3.5 Å is used,34 which is the default in Cavallo’s
program as well as the distance used by Nolan and co-workers.
In Figure 5, there are two examples showing space filling
Table 1. Values for LDP (kcal/mol) and ΔS⧧ (cal/mol K)
rot
for 1−31 and 33
a
e
‡
X =
NMe2 (31)29
LDP
ΔS rot
b
b
9.34 0.32
−4
1
OAd (6)
10.83 0.24
10.86 0.23
11.12 0.23
11.15 0.23
12.04 0.25
12.14 0.24
12.18 0.25
12.38 0.25
12.51 0.26
12.64 0.23
12.81 0.23
13.00 0.28
13.28 0.27
13.35 0.23
13.39 0.27
13.40 0.25
13.89 0.26
14.15 0.29
14.16 0.28
14.22 0.27
14.32 0.28
14.33 0.28
14.36 0.28
14.40 0.27
14.45 0.28
14.51 0.29
14.86 0.30
15.05 0.29
15.45 0.30
15.75 0.29
15.80 0.30
N(Me)Ph (27)
NPri2 (1)
d
c
c
−5
−6
2
5
OBn (20)
Carbazolyl (26)
O-p-(OMe)C6H4 (11)
O-p-(But)C6H4 (13)
OPh (10)
c
4
O-p-(SMe)C6H4 (12)
O-p-(F)C6H4 (14)
O-p-(Cl)C6H4 (15)
O-p-(CF3)C6H4 (16)
OSiPh3 (7)
OPht (18)
−3
F (33)
Indolyl (25)
OButF6 (9)
NO3 (21)
Pyr (22)
SPh (19)
OC6F5 (17)
PyrC6F5 (23)
PyrC6H3(CF3)2 (24)
CN (30)
O2CPh (8)
NCO (28)
Figure 5. Space filling model of relatively small chloro 3 (top) and
large indolyl 25 (bottom). The inscribed orange sphere shows the 3.5
Å radius limit.
NCS (29)
Cl (3)
Br (4)
models; the orange sphere is 3.5 Å and shows how much of
each ligand is included in the calculation. For a discussion and a
plot of sphere radius’s effect on %Vbur for a selection of ligands,
see the Supporting Information. The results of the ligand
parametrization using this method are shown in Figure 6.
In addition to %Vbur, we also investigated another steric
parametrization, the Solid Angle Steric Parameter using the
Solid G program.35 The computational technique works
directly from the X-ray diffraction data. The central metal is
viewed in some respects as a point source of light, and the
ligands block conal access to a sphere around the molecule. An
example of the Solid Angle Model is shown in Figure 7 for
indolyl 25, where the molecule is in a similar orientation as in
the bottom of Figure 5. The Solid Angle Steric Parameters for
the series of X ligands are shown in Figure 8. The x-axis values
are in percentage of the sphere occupied by the X ligand.34
The orderings for many of the ligands change somewhat
using the different methods. For example, the Solid G method
gives a halide ordering of F > Br > I > Cl because of a mixture
of bond distance and radii effects. The halide steric ordering
from %Vbur is I > Br > Cl > F and seems to be most greatly
affected by atomic radius.36
OTf (5)
c
I (2)
−9
5
a
Average value from at least 3 measurements for the free energy
barrier to rotation. Entropy values were assumed to be −9 cal/mol K
except for 31 where the variable temperature LSA value was used.
Experimental value from VT LSA. Experimental value from VT SST.
Steric effects are quite likely contributing to this LDP. Experimental
entropy values in cal/mol K. Errors are from the fits to the Eyring
plots.
b
c
d
e
V. Steric Properties of the Ligands. Coming up with a
single parameter for the steric properties of a diverse set of
ligands is an inherently inaccurate exercise. One is using a single
parameter to describe a 3-dimensional object, which unless that
object is a perfect sphere is an incomplete description.
However, the Tolman cone angle1 has been quite successful
and gives researchers a parameter for initial optimization of
reactions. More recently, Percent Buried Volume (%Vbur)
calculations have proven useful in determining steric parame-
ters for ancillary ligand sets.32 Encouragingly, the %Vbur of
phosphine ligands agrees nicely with Tolman’s cone angles for
many standard phosphines. Furthermore, %Vbur are easily
calculated from crystallographic data using Cavallo and co-
workers web-based utility, SambVca, and have been used
extensively for ligand types like N-heterocyclic carbenes.33
To determine the %Vbur, the ligands are placed in a sphere so
that the ligand is the Cr−X bonding distance away from the
center. The sphere size is an adjustable parameter meant to
approximate the size where ligands affect the primary
coordination sphere of the metal. In most instances, a sphere
That the complex with X = NPri2 1 has steric influences on
its rotational barrier seems likely. One would expect NPri2 to be
a similar or better donor than NMe2, and yet it has a higher
LDP by almost 2 kcal/mol. It appears that sterics are raising the
barrier for rotation in this system, and X = NPri2 is the largest
ligand investigated by far using both steric metrics.
1194
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
Figure 6. %Vbur for the ligands used in this study. Values are for the percentage volume occupied by the ligand in a sphere of radius 3.5 Å from the
chromium center.
The only compound where steric effects seem certain to be
playing a role in the measurement of LDP is NPri2 1. All other
compounds are assumed to have LDPs predominately
associated with the electronic barrier to amido rotation as
there is currently no compelling evidence to suggest otherwise.
Some of the most likely ligands to have unresolved steric effects
are OSiPh3, OButF6, OAd, N(Me)Ph, indolyl, and carbazolyl,
but all of these have steric metrics below NPri2. If other ligands
do have steric effects on the amido rotational barrier, their
observed LDPs are likely upper limits, and they are electroni-
cally more donating than observed.
VI. Applications and Comparisons with These New
Electronic Parameters. We conclude the Results and
Discussion Section by comparing the LDP data quantitatively
with different systems from the literature. In these types of
applications, the LDP values in Figure 4 and Table 1 are used
much as CO stretching frequencies may be used in alternative
late transition metal applications, as arbitrary numbers (in the
case of LDP inversely) proportional to the donor ability of the
ligand of interest.
Figure 7. Solid Angle Model from the Solid G program for 25 with
indolyl (green), diisopropylamido (yellow and blue), and nitrido
(red).
A. Comparison of LDPs with pKa Values of the HX
Compounds. First, we investigated if the LDPs found in this
Figure 8. Percentage of the chromium coordination sphere shielded, GM(L), from the Solid G program for the ligands used in this study.
1195
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
study correlated to the pKa of the substituents used. As shown
in Figure 9, there is no strong relationship. This is to be
Figure 10. Plot of LDP vs Hammett parameters for the aryloxide
complexes.
Figure 9. Plot of pKa in water versus LDP. The inset is an expansion of
the region containing the phenoxides.
expected considering the numerous size and orbital makeup
differences between the proton and the transition metal system
under study.
At best, pKa can give one a sense of the σ-donor ability of the
X ligand when attached to a proton. Perhaps this is the reason
for the linear correlation for pKa and LDP that seems to exist in
the plot between the heavier halides I, Br, and Cl where π-
effects to the metal are likely minimal. However, pKa would be
expected to give no information on π-donor ability, which is
perhaps why fluoride is not on the same line with its heavier
congeners.
Other closely related series, such as the phenoxides (inset in
Figure 9) may show some trend with pKa. However, relating
ligand acidity to ligand donor ability is likely to be of dubious
quality, especially if strong π-effects are present.
B. Phenoxides: An LDP Comparison with Hammett
Parameters. Hammett parameters are a reliable and well-
worn method for examining electronic effects in a variety of
systems with, in large part, para-substitution on an arene.5 To
determine if there was a correlation between LDP and
Hammett σp, we generated the set of para-substituted
phenoxides 10−16. The full range of LDP differences in the
series studied from para-OMe to para-CF3 is 12.14 to 13.00
kcal/mol with errors around 0.25 kcal/mol. While values at the
extremes of this subset are different outside the errors, many of
the values within the series are not distinguishable with these
error limits. However, plotting this series of LDP values versus
the Hammett Parameters (σp) for the substituents shows a
good linear correlation considering the error bars on the LDP
(Figure 10).
Figure 11. Plot of the 13C NMR chemical shift for the carbon directly
bonded to tungsten in W(C8H12C8H12NAr)(NAr)X2 versus
the donor ability of X found in this work with a linear fit.37
.
−8
chlorides for other X ligands leads to structural changes in
the complexes and changes in the 13C NMR chemical shift of
the carbon bonded to tungsten. These changes may be viewed
as being the result of differing contributions between the
alkylidene-imine (left) and alkyl-amido (right) resonance
forms. In the paper, we simply stated that the values for the
chemical shifts changed as “might be expected” with OEt higher
than O-p-(OMe)C6H4 higher than OC6F5.37 Whereas, for
chloride and triflate, also included as X ligands in the paper, it
was more difficult to discern their donor properties versus these
alkoxides.
C. Evaluation of 13C NMR Chemical Shifts in Tungsten
Metallacycles using LDP. In some cases, spectroscopic data
known to change with donor properties of ligands may be
correlated with LDP. In 2008, our research group published a
study on the reactivity and properties of an unusual class of
metallacycles with tungsten−carbon double bond character.37
Included in this study were NMR spectroscopic data for the
series and reactivity in carbonyl olefination reactions for the
chloride complex with various additives.
The addition of 2 equiv of cyclooctyne to W-
(NAr)2Cl2(DME) results in the formation of W(C8H12
C8H12NAr)(NAr)Cl2 (Figure 11). Substitution of the
Using the LDP and plotting versus the 13C chemical shifts,
one sees a good correlation between the ligand donor ability as
measured in this work across the available ligand sets in the
tungsten system and the NMR data reported (Figure 11).38
The chemical shifts of the α-carbons in the metallacycles do
1196
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
indeed correlate with the donor abilities of the X ligands across
the entire range of compounds produced in the tungsten study.
The linear relationship between LDP and the 13C NMR
chemical shift is exceptionally good (R2 = 0.996).
of d1 titanium complexes. Of specific interest in the context of
this paper, Andersen and co-workers report the singly occupied
a1 to b2 energy gap, which “depends directly upon the π-donor
ability of X”. Mach and co-workers have since extended the
system to include additional alkoxide ligands.42
A plot of the energy gap between a1 (approximately
nonbonding)43 and the b2 π-antibonding orbital (ΔExz) in
Cp*2TiX Andersen complexes versus LDP for all X in common
between the two studies is shown in Figure 13 (blue and red
D. Comparison Between the AOM of Cr(III) Complexes
and the SST Determined Donor Values. In this study, we were
able to include several classic Werner-type ligands and
determine their donor properties in this Cr(VI) system. In
these traditional coordination compounds, the donor properties
are usually determined using visible absorption data in
conjunction with Ligand Field Theory. The values can also
be parametrized using the Angular Overlap Model as σ- and π-
donor energies, eσ and eπ respectively,39 of individual X.
In Figure 12 is a plot of eσ + eπ for chromium(III) complexes
from the experimentally determined AOM values40 versus the
Figure 13. Plot of ΔExz in wavenumbers (cm−1) [Andersen data41 (red
and blue circles), Mach’s data42 (green squares)] versus LDP (kcal/
mol) for X. For the data represented by circles, methylcyclohexane was
the solvent. The data represented by green squares were taken in
either hexane (⧧) or toluene (†).44.
circles). In the case of X = OMe, the value for ΔExz was
correlated with the LDP value for OBn in the plot (blue line).
The obvious outlier is X = N(Me)Ph (red circle), which is well
away from what seems to be a linear correlation between the
Cp*2TiX spectroscopic data and LDP. Andersen and co-
workers centered much of their discussion on the differences
between X = N(Me)Ph and the other compounds, and this is
quite obvious in Figure 13 as well. Also plotted in Figure 13 are
Mach’s data (green squares) on Cp*2TiX, where we used our X
= OAd data for their X = OBut example.
There are several indications that the X = N(Me)Ph in
Cp*2TiX has little or no π-effects to the nitrogen; although
there are indications of agostic effects to the methyl.41 In the
structure from X-ray diffraction, the Cp*(centroid)−Ti−N−
Me average dihedral in the X = N(Me)Ph complex is 86.9°. In
other words, the large N(Me)Ph ligand rests in the plane
bisecting the Cp*−Ti−Cp* unit, and the nitrogen lone pair is
orthogonal to the empty orbital of appropriate symmetry to act
as an acceptor. Consequently, the experimental Ti−N bond
distance is quite long at 2.054(2) Å. This is similar to Ti−
N(pyrrolyl) distances,30 usually a much weaker donor than
N(Me)Ph (vide supra). This distance is also much closer to the
Ti−N single bond distance of 2.07 Å than the TiN distance
Figure 12. Plot of e + e for chromium(III) complexes from
σ
π
experimentally determined AOM values versus the LDP for X (top).
Plot of e (blue) and e (red) parameters versus the LDP for X
σ
π
(bottom).
donor ability of X found in this study, which shows a good
linear correlation between the two parametrization systems.
Also in Figure 12 is a plot of the individual eσ + eπ parameters
versus LDP (bottom). The correlation with either the eσ or the
eπ parameter alone is not nearly as good as their sum.
E. Comparison Values from Electronic Spectra of Cp*2TiX
Complexes. In 1996, Lukens, Smith, and Andersen41 reported a
“π-donor spectrochemical series for X” in Cp*2TiX titanium-
(III) compounds with a large number of X ligands. The study
employed electron paramagnetic resonance (EPR) and
absorption spectroscopy to elucidate the electronic structure
of 1.77 Å using Pyykko’s radii.45 In contrast, Ti−NMe2
̈
distances, where there is a strong dative π-bond, are typically
∼1.90 Å.46
It can be concluded that the lack of correlation for X =
N(Me)Ph is due to a deficiency of π-bonding in the Cp*2TiX
system because of steric effects that do not allow the amido to
reach the electronically preferred geometry, a fact readily seen
1197
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
in both the X-ray diffraction study and in correlations with
LDP.
ASSOCIATED CONTENT
* Supporting Information
■
S
If one examines the X = N(H)Me complex of Cp*2TiX, the
amide is rotated much closer to where maximal overlap with
the π-acceptor orbital (b2) would be possible. The Cp*-
(centroid)−Ti−N−Me dihedral for this compound is 13.2°.
However, the Ti−N bond, 1.955(5) Å, is slightly longer than
the average Ti−NMe2 bond in the CSD database.46 This
lengthened bond may be due to steric clash with the Cp*
ligands.47
Overall, the LDPs correlate fairly well with the Cp*2TiX
spectroscopic data in cases where steric effects are not apparent,
that is, all X ligands in common between the two studies except
for where the X ligand is an amido derivative.
Experimental details and characterization for the production of
all the compounds used in the study. 1H and 13C NMR spectra
for all new compounds. Data for the X-ray diffraction
experiments in the form of a cif file. Plot of radius versus %
Vbur for several ligands. Details on the spin saturation transfer
experiments. Details on propagation of error to find LDP
values. 2D NMR on 2 and assignments. Details on %Vbur and
Solid G calcs. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: odom@chemistry.msu.edu.
■
CONCLUDING REMARKS
■
ACKNOWLEDGMENTS
■
Changing metals, changing formal oxidation state, and other
ligands on the metal can greatly alter donor properties of
ligands. These types of single parameter studies should not
replace full mechanistic and computational studies for systems;
instead, this is a quick technique that will hopefully be useful in
the discussion of properties and mechanisms for metal
complexes at low d-electron counts. If a series of ligands for
a particular system correlate well to LDP and one does not, it
might indicate steric influences, hapticity differences between
the chromium system here and the system under study, or
other effects are important for that particular compound (for an
example see X = N(Me)Ph in Section VI E above). If the
system under study does not correlate at all with LDP, there are
any number of possible explanations ranging from differences
in ligand donor properties, differences in metal acceptor
properties, steric interactions, or simply a lack of correlation of
the property being measured with ligand donor ability.
The authors, and N.A.M. in particular, express gratitude for the
assistance of Dan Holmes and Kermit Johnson with the SST
and LSA experiments. Dan Mindiola is thanked for his role in
the development for the new synthesis of 1. We also thank Ilia
Guzei for aid in the calculation of Solid G parameters. Mitch
Smith is thanked for helpful discussion. The financial support of
the National Science Foundation CHE-1012537 for this work is
greatly appreciated.
REFERENCES
(1) Tolman, C. A. Chem. Rev. 1977, 77, 313.
(2) Abel, E. W.; Bennett, M. A.; Wilkinson, G. J. Chem. Soc. 1959,
2323.
■
(3) For other seminal work on CO stretches being used for electronic
evaluation see: (a) Nyholm, R. S.; Short, L. N. J. Chem. Soc. 1953,
2670. (b) Cotton, F. A. Inorg. Chem. 1964, 3, 702. (c) Abel, E. W.;
Stone, F. G. A. Q. Rev. Chem. Soc. 1969, 23, 325. (d) Poilblanc, R.;
Bigorgne, M. Bull. Soc. Chim. Fr. 1962, 1301. (e) Strohmeier, W.;
Muller, F. J. Chem. Ber. 1967, 100, 2812. For a recent paper showing
effects other than metal electron density can effect these CO stretching
frequencies see: (f) Valyaev, D. A.; Brousses, R.; Lugan, N.; Fernandez,
I.; Sierra, M. A. Chem.Eur. J. 2011, 17, 6602.
In the last segment of the Results and Discussion Section, we
attempted to correlate these new LDP values with data from
the literature. In the cases discussed above, the values that do
correlate, i.e., those other than pKa, did so linearly. It should be
kept in mind, however, that there is no reason to assume that all
correlations between LDP parameters and data determined
using numerous techniques on various systems will always be
linear. In addition, more involved methods than the single
parameter %Vbur and Solid Angle methods may be required for
accurate comparisons of sterics in some systems as well.
The absolute values above consititute our current best
evaluations of these ligands. Improved techniques and
instrumentation for the determination of ligand donor abilities
in this system may lead to improved values in the future.
There are obvious ligand types that would be useful to
include in a series of this type that have not yet been prepared
for parametrization of their donor properties. We are
continuing to expand the series presented here. Current plans
include the characterization of cationic chromium(VI) nitrido
systems with neutral X ligands and a selection of organometallic
ligands, which are being prepared for evaluation.
(4) For a few representative examples see: (a) Maldonado, A. G.;
Hageman, J. A.; Mastroianni, S.; Rothenberg, G. Adv. Synth. Catal.
2009, 351, 387. (b) Hageman, J. A.; Westerhuis, J. A.; Fruhauf, H.-W.;
̈
Rothenberg, G. Adv. Synth. Catal. 2006, 348, 361. (c) Burello, E.;
Rothenberg, G. Int. J. Mol. Sci. 2006, 7, 375. (d) Maldonado, A. G.;
Rothenberg, G. Chem. Soc. Rev. 2010, 39, 1891. (e) Mikhel, I. S.;
Garland, M.; Hopewell, J.; Mastroianni, S.; McMullin, C. L.; Orpen, A.
G.; Pringle, P. G. Organometallics 2011, 30, 974. (f) Birkholz, M.-N.;
Freixa, Z.; van Leeuwen, P. W. N. M. Chem. Soc. Rev. 2009, 38, 1099.
(g) Gillespie, J. A.; Dodds, D. L.; Kamer, P. C. J. Dalton Trans. 2010,
39, 2751. (h) Fey, N. Dalton Trans. 2010, 39, 296. (i) Droge, T.;
̈
Glorius, F. Angew. Chem., Int. Ed. 2010, 49, 6940.
(5) For a review see: Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev.
1991, 91, 165.
(6) For some selected recent reviews see: (a) Sprous, D. G.; Palmer,
R. K.; Swanson, J. T.; Lawless, M. Curr. Top. Med. Chem. 2010, 10,
619. (b) Verma, R. P.; Hansch, C. Chem. Rev. 2009, 109, 213.
(c) Michaelidou, A. S.; Hadjipavou-Litina, D. Chem. Rev. 2005, 105,
3235. (d) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Adv. Drug Delivery Rev. 2001, 46, 3.
Parameterizations of this type have seen some historical
success in explaining reaction mechanisms and trends in
reactivity. For example, Basolo, Pearson, Burdett, and many
others have used the Angular Overlap Model extensively in this
regard especially for later transition metal systems.48 It is hoped
that this method of ligand parametrization will be useful in
catalysis studies ongoing in our group and others on high valent
metal complexes.
(7) Occhipini, G.; Bjørsvik, H.-R.; Jensen, V. R. J. Am. Chem. Soc.
2006, 128, 6952.
(8) For a few examples see: (a) Mohring, P. C.; Coville, N. J. Coord.
̈
́
Chem. Rev. 2006, 250, 18. (b) Silveira, F.; de Sa, D. S.; da Rocha, Z. N.;
Alves, M. C. M.; dos Santos, J. H. Z. X-Ray Spectrom. 2008, 37, 615.
(c) Kozimor, S. A.; Yang, P.; Batista, E. R.; Boland, K. S.; Burns, C. J.;
Christensen, C. N.; Clark, D. L.; Conradson, S. D.; Hay, P. J.; Lezama,
1198
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
J. S.; Martin, R. L.; Schwarz, D. E.; Wilkerson, M. P.; Wolfsberg, L. E.
Inorg. Chem. 2008, 47, 5365.
(9) Redshaw, C. Dalton Trans. 2010, 39, 5595.
systems where alkoxide bond angle does seem to correlate with bond
distance see: (d) Tomaszewski, R.; Arif, A. M.; Ernst, R. D. J. Chem.
Soc., Dalton Trans. 1999, 1883−1890. (e) Huffman, J. C.; Moloy, K.
G.; Marsella, J. A.; Caulton, K. G. J. Am. Chem. Soc. 1980, 102, 3009.
(26) (a) Jarek, R. L.; Flesher, R. J.; Shin, S. K. J. Chem. Educ. 1997,
74, 978. (b) Perrin, C. L.; Thoburn, J. D.; Kresge, A. J. J. Am. Chem.
Soc. 1992, 114, 8800. (c) Dahlquist, F. W.; Longmuir, K. J.; Du Vernet,
R. B. J. Magn. Reson. 1975, 17, 406.
(10) Zachmanoglou, C. E.; Docrat, A.; Bridgewater, B. M.; Parkin,
G.; Brandow, C. G.; Bercaw, J. E.; Jardine, C. N.; Lyall, M.; Green, J.
C.; Keister, J. B. J. Am. Chem. Soc. 2002, 124, 9525.
(11) (a) Swartz, D. L.; Staples, R. J.; Odom, A. L. Dalton Trans. 2011,
40, 7762. (b) Swartz, L. D.; Odom, A. L. Organometallics 2006, 25,
6125.
(12) Some cationic complexes using this framework with neutral
ligands in place of X have already been prepared, and many others are
being developed. It is hoped that this series will allow more overlap
with the Tolman ligand maps, which involve mostly neutral ligands.
One such complex, 32, is used as an intermediate to the fluoride
reported here. Unpublished results, DiFranco, S.; Maciulis, N.; Odom,
A. L.
(27) We measured the variable temperature SST data for X = Pri in
two different solvents. In toluene, the activation parameters for
rotation were found to be ΔH⧧ = 12.4 kcal/mol and ΔS⧧ = −2 cal/mol
K over a temperature range of 30 °C. In CDCl3, the usual solvent
employed for the studies in this paper, the activation parameters for
rotation were found to be ΔH⧧ = 11.6 kcal/mol and ΔS⧧ = −5 cal/mol
K over a temperature range of 26 °C. The value of ΔH⧧ = 11.12 0.23
kcal/mol in Table 1 is from multiple measurements, consistent with
the majority of data in that table.
(28) The amido entropic barriers found here are quite similar to
values in the literature for organic amides of various types. For
example, the amide rotational barrier in nicotinamide was measured
using several different techniques and values from −3 to −10 cal/mol
K were obtained. Olsen, R. A.; Liu, L.; Ghaderi, N.; Johns, A.;
Hatcher, M. E.; Mueller, L. J. J. Am. Chem. Soc. 2003, 125, 10125. For
(13) The orbital labels are an approximation. In actuality, because the
2
2
dxy/dx −y and dxz/dyz e-sets have the same symmetry, they will mix as
well.
(14) In the calculations, two minima can sometimes be found in the
model: one where the lone pair is syn to the nitrido and one where it is
anti. The anti isomer is typically the global minimum, which is what is
used here.
(15) The hybridization at nitrogen was calculated by taking the
calculated angles around the nitrogen and dividing by 3 to get an
average hybridization for all the bonds. From the average angle,
Coulson’s equation for the Directionality Theorem, cos ωij = −1/
((λiλi)1/2), can be used to get the hybridization parameter. In this case
where we are simply looking at the average angle, effectively average
orbital hybridization, the equation can be simplified to λ = −1/(cos ω)
for looking at equivalent hybrids. (a) McWeeny, R. Coulson’s Valence;
Oxford University Press: Oxford, U.K., 1979. (b) Weinhold, F.;
Landis, C. R. Valency and Bonding; Cambridge University Press:
Cambridge, U.K., 2005.
some other examples see: Drakenberg, T.; Dahlqvist, K.-I.; Forsen
Phys. Chem. 1972, 76, 2178.
́
, S. J.
(29) Because the value for the NMe2 derivative was found using a
different technique from all the other values, we view its absolute value
relative to all the other LDP numbers with some suspicion. However,
we have little doubt in its placement as the most donating ligand
investigated in this study.
(30) This can be seen inthe metal-nitrogen bond distances of pyrrole
versus dialkylamidos.See: Harris, S. A.; Ciszewski, J. T.; Odom, A. L.
Inorg. Chem. 2001, 40, 1987, and references therein for examples..
(31) For a mixed experimental theoretical treatment of aromaticity in
indoles and pyrroles see: (a) Hou, D.-R.; Wang, M.-S.; Chung, M.-W.;
Hsieh, Y.-D.; Tsai, H.-H. G. J. Org. Chem. 2007, 72, 9231. For a
review on aromaticity of heterocycles see: (b) Balaban, A. T.; Oniciu,
D. C.; Katritzky, A. R. Chem. Rev. 2004, 104, 2777.
(16) (a) Bradley, D. C.; Hursthouse, M. B.; Newing, C. W.; Welch, A.
J. Chem. Commun. 1971, 411. (b) Bradley, D. C.; Copperthwaite, R. G.
Inorg. Synth. 1978, 18, 112. (c) Bradley, D. C.; Newing, C. W. Chem.
Commun. 1970, 219.
(17) (a) Odom, A. L.; Cummins, C. C.; Protasiewicz, J. D. J. Am.
Chem. Soc. 1995, 117, 6613. For related nitrosyl deoxygenation
studies see: (b) Veige, A. S.; Slaughter, L. M.; Lobkovsky, E. B.;
Matsunaga, N.; Decker, S. A.; Cundari, T. R.; Wolcanski, P. T. Inorg.
Chem. 2003, 42, 6204.
(32) For a reviewconcerningthe application of %Vbur see: Clavier, H.;
Nolan, S. P. Chem. Commun. 2010, 46, 841.
(33) (a) Hillier, A. C.; Sommer, W. J.; Yong, B. S.; Petersen, J. L.;
Cavallo, L.; Nolan, S. P. Organometallics 2003, 22, 4322. (b) Poater,
A.; Cosenza, B.; Correa, A.; Giudice, S.; Ragone, F.; Scarano, V.;
Cavallo, L. Eur. J. Inorg. Chem. 2009, 1759. (c) http://www.molnac.
unisa.it/OMtools/sambvca.php
(18) Chiu, H.-T.; Chen, Y.-P.; Chuang, S.-H.; Jen, J.-S.; Lee, G.-H.;
Peng, S.-M. Chem. Commun. 1996, 139.
(19) The byproduct Cr(OBut)3 is believe to be a soluble dimer.
Bradley, D. C.; Mehrotra, R.; Rothwell, I.; Singh, A. Alkoxo and
Aryloxo Derivatives of Metals; Academic Press: San Diego, CA, 2001.
(20) Odom, A. L.; Cummins, C. C. Organometallics 1996, 15, 898.
(21) Odom, A. L.; Cummins, C. C. Polyhedron 1998, 17, 675.
(22) For a similar decompositionof a DMAP containing BF4 salt see:
Huynh, K.; Rivard, E.; Lough, A. J.; Manners, I. Chem.Eur. J. 2007,
(34) It is certainly possible that no single parameter steric scheme
will accurately encompass the influences of a diverse set of ligands like
the one found here even if the methodologies are applicable to a
relatively narrow group like phosphines and N-heterocyclic carbenes.
Included in the Supporting Information is a plot of sphere radius
versus %Vbur for some ligands. As might be expected, the ligands show
different maxima and some of the lines do cross, suggesting the ideal
radius is likely system and application dependent. The values here are
intended to give a rough measure of sterics for the various ligands.
(35) Guzei, I. A.; Wendt, M. Dalton Trans. 2006, 3991.
(36) The online program for %Vbur, SambVca, does not include all of
the halides, and some of these were calculated using an overlapping
spheres methods at the appropriate distances. See Supporting
Information for details.
(37) (a) Lokare, K. S.; Odom, A. L. Inorg. Chem. 2008, 47, 11191.
See also (b) Lokare, K. S.; Ciszewski, J. T.; Odom., A. L.
Organometallics 2004, 23, 5386.
13, 3431. In addition, the tetraalkyltin SnBun (allyl), which we had on
3
hand, was an active catalyst.It is likely that Sn(IV) is acting as a mild
Lewis acid in the reaction,and other acids may catalyzed the
decomposition of 32.
(23) The chemical shift in the 19F NMR of the byproduct in the
reaction to produce 33 was identical to the product of the reaction
between 1 equiv each of BF3•OEt2 and DMAP in the same solvent.
(24) In this work, 31 of the compounds have been structurally
characterized; this number includes the previously reported structures
for NCr(NPri2)3 (1) and NCr(NPri2)2(I) (2). The only compound
not structurally characterized is compound 32, which was only used as
an intermediate in production of the fluoride 33.
(38) The value used in the plot for the mixed X ligand set (Cl)(OTf)
is the average of the two SST values for the two different ligands. The
SST value for OBn was used as the parameter for the ethoxide ligand.
(39) A note of caution for the use of AOM values: the eπ values are
(25) For related studies on aryloxides where no correlation was
found between M−O−Ar angle and M−O bond distance see:
(a) Steffey, B. D.; Fanwick, P. E.; Rothwell, I. P. Polyhedron 1990, 9,
963−968. (b) Howard, W. A.; Trnka, T. M.; Parkin, G. Inorg. Chem.
1995, 34, 5900. (c) Wolczanski, P. T. Polyhedron 1995, 14, 3335. For
often found using the assumption that e for NH3 is equal to zero
π
because of overparameterization within the AOM model. The validity
of this assumption has been called into question for some coordination
1199
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200

Inorganic Chemistry
Article
complexes, most notably Pt(II) where the e value for NMe3 was
determined to be well above zero. It is not believed that NH3 actually
(b) Burdett, J. K. Adv. Inorg. Chem. Radiochem. 1978, 21, 113.
(c) Jordan, R. B. Reaction Mechanisms of Inorganic and Organometallic
Systems, 2nd ed.; Oxford University Press: New York, 1998.
π
does have π-effects; instead, this is attributable to the e parameter not
π
being truly zero within the model. For further discussion see ref 39 and
(a) Chang, T.-H.; Zink, J. I. J. Am. Chem. Soc. 1987, 109, 692.
(b) Bridgeman, A. J.; Gerlock, M. Prog. Inorg. Chem. 1997, 45, 179.
(c) Schaffer, C. E.; Anthon, C.; Bendix, J. Coord. Chem. Rev. 2009, 253,
575.
(40) AOM values were takenfrom Hoggard, P. E. Struct. Bonding
(Berlin) 2004, 106, 37, whichwere calculated from splittings in spin-
allowed transitions of thecorresponding Cr(III) ammine complexes.
(41) Lukens, W. W. Jr.; Smith, M. R.; Andersen, R. A. J. Am. Chem.
Soc. 1996, 118, 1719.
(42) (a) Gyepes, R.; Varga, V.; Horacek, M.; Kubista, J.; Pinkas, J.;
Mach, K. Organometallics 2010, 29, 3780. (b) Varga, V.; Cisarova, I.;
Gyepes, R.; Horacek, M.; Kubista, J.; Mach, K. Organometallics 2009,
28, 1748.
(43) That the quantitymeasured is exclusively determined by the π-
donor ability ofX is of course an approximation. The a1 orbital is
weaklyσ-bonding but largely nonbonding. The b2 orbitalis orthogonal
and its energy is largely determined by the π-bondingof X. For a
discussion of the bonding see: Lauher, J. W.; Hoffmann, R. J. Am.
Chem. Soc. 1976, 98, 1729.
(44) The data for X = N(Me)Ph was not used in determining the
linear fit to the Andersen data (blue line) shown in Figure 10. The
exact cause for the two different slopes of the Andersen and Mach data
fits is unknown; however, there are several small differences in the data
sets like different solvents, slightly different instrumentation, and
possible differences in concentration. In addition, the Andersen data
are relative to Cp*2TiH as ΔExz = 0, and we did not adjust the baseline
in the Mach data similarly. If the data from the two groups are taken all
together and linearly fit, the line obtained is y = 20761 − 1205x with a
much worse R2 of 0.93 versus the R2 = ∼0.99 obtained fitting the data
from the two groups independently. This fit for Mach’s data is for only
four points from the four new alkoxides in common between our study
and Mach’s reports. In addition, one of the four points was done with a
substitution in the LDP value, tert-butoxide for 1-adamantoxide.
(45) Calculated using Year 2009 single-, double-, and triple-bond
covalent radii from a private communication. (a) http://www.chem.
helsinki.fi/∼pyykko/ (b) Pyykko, P.; Atsumi, M. Chem.Eur. J. 2009,
̈
15, 12770.
(46) The average Ti−NMe2 bond in CSD version 5.32 Nov 2010 is
1.897 Å. It should be noted, however, that the majority of those
complexes are Ti(IV), and the distance in titanium(III) Me(H)
NTiCp*2 may be somewhat longer just because of the difference in
oxidation state. There are a few Ti−NH2 cyclopentadienyl complexes
in the CSD database, which should allow the Ti−N bond to shorten if
electronically favorable. These do tend to have slightly shorter
titanium-amido bond distances. For example, amidobis[1,3-bis(trime-
thylsilyl)cyclopentadienyl]titanium(III) has a Ti−NH2 distance of
1.933(3) Å. (a) Sofield, C. D.; Walter, M. D.; Andersen, R. A. Acta
Crystallogr., Sect. C 2004, 60, M465. The compound H2NTiCp*2 has a
Ti−N distance of 1.944(2) Å. (b) Brady, E.; Telford, J. R.; Mitchell,
G.; Lukens, W. Acta Crystallogr., Sect C 1995, 51, 558. Neitherof these
NH2 complexes has a statistically significantdifference in the Ti−N
distance from the value in Me(H)NTiCp*2, however.
(47) The methyl group isrotated in the Me(H)N−TiCp*2 complex
so that theamido methyl is directed at one of the Cp* ligands. The
Solid-G programcalculates the G(gamma) value to be 3.03 in this
compound, which isa measure of how much of the sphere is shielded
by more than one ligand.For the complex Me(Ph)N−TiCp*2, were
the amido methyland phenyl groups rest between the Cp*s, the
G(gamma) value is 2.19,suggesting there is less ligand overlap. In
other words, the smallamido is nesting with the methyls of the Cp*
groups to reach thiselectronically preferred but sterically encumbered
position. For thestructure of Me(Ph)NTiCp*2 see: Feldman, J.;
Calabrese, J. C. Chem. Commun. 1991, 1042.
(48) For examples see: (a) Basolo, F.; Pearson, R. G. Mechanisms of
Inorganic Reactions, 2nd ed.; Wiley & Sons: New York, 1967.
1200
dx.doi.org/10.1021/ic202524r | Inorg. Chem. 2012, 51, 1187−1200