Four-coordinate cobalt pincer complexes: Electronic structure studies and ligand modification by homolytic and heterolytic pathways

Article abstract of DOI:10.1021/ja504334a

A family of cobalt chloride, methyl, acetylide and hydride complexes bearing both intact and modified tert-butyl substituted bis(phosphino)pyridine pincer ligands has been synthesized and structurally characterized and their electronic structures evaluated. Treatment of the unmodified compounds with the stable nitroxyl radical, TEMPO (2,2,6,6-tetramethylpiperidin-1-yloxidanyl) resulted in immediate H- atom abstraction from the benzylic position of the chelate yielding the corresponding modified pincer complexes, ( ^tBumPNP)CoX (X = H, CH₃, Cl, CCPh). Thermolysis of the methyl and hydride derivatives, (^tBuPNP)CoCH₃ and ( ^tBuPNP)CoH, at 110 °C also resulted in pincer modification by H atom loss while the chloride and acetylide derivatives proved inert. The relative ordering of benzylic C-H bond strengths was corroborated by H atom exchange experiments between appropriate intact and modified pincer complexes. The electronic structures of the modified compounds, (^tBumPNP)CoX were established by EPR spectroscopy and DFT computations and are best described as low spin Co(II) complexes with no evidence for ligand centered radicals. The electronic structures of the intact complexes, (^tBuPNP)CoX were studied computationally and bond dissociation free energies of the benzylic C-H bonds were correlated to the identity of the X-type ligand on cobalt where pure σ donors such as hydride and methyl produce the weakest C-H bonds. Comparison to a rhodium congener highlights the impact of the energetically accessible one-electron redox couple of the first row metal ion in generating weak C-H bonds in remote positions of the supporting pincer ligand.

Full text of DOI:10.1021/ja504334a

Article
pubs.acs.org/JACS
Four-Coordinate Cobalt Pincer Complexes: Electronic Structure
Studies and Ligand Modification by Homolytic and Heterolytic
Pathways
Scott P. Semproni,† Carsten Milsmann,† and Paul J. Chirik*
Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States
S
* Supporting Information
ABSTRACT: A family of cobalt chloride, methyl, acetylide
and hydride complexes bearing both intact and modified tert-
butyl substituted bis(phosphino)pyridine pincer ligands has
been synthesized and structurally characterized and their
electronic structures evaluated. Treatment of the unmodified
compounds with the stable nitroxyl radical, TEMPO (2,2,6,6-
tetramethylpiperidin-1-yloxidanyl) resulted in immediate H-
atom abstraction from the benzylic position of the chelate
yielding the corresponding modified pincer complexes,
(
(
tBumPNP)CoX (X = H, CH3, Cl, CCPh). Thermolysis of the methyl and hydride derivatives, (tBuPNP)CoCH3 and
tBuPNP)CoH, at 110 °C also resulted in pincer modification by H atom loss while the chloride and acetylide derivatives proved
inert. The relative ordering of benzylic C−H bond strengths was corroborated by H atom exchange experiments between
appropriate intact and modified pincer complexes. The electronic structures of the modified compounds, (tBumPNP)CoX were
established by EPR spectroscopy and DFT computations and are best described as low spin Co(II) complexes with no evidence
for ligand centered radicals. The electronic structures of the intact complexes, (tBuPNP)CoX were studied computationally and
bond dissociation free energies of the benzylic C−H bonds were correlated to the identity of the X-type ligand on cobalt where
pure σ donors such as hydride and methyl produce the weakest C−H bonds. Comparison to a rhodium congener highlights the
impact of the energetically accessible one-electron redox couple of the first row metal ion in generating weak C−H bonds in
remote positions of the supporting pincer ligand.
INTRODUCTION
■
Neutral tridentate, “pincer” type chelates have emerged as a
privileged class of ligands due to their widespread use in
homogeneous catalysis and small molecule activation pro-
cesses.1−8 Base metal catalysis is no exception as pincer ligated
iron and cobalt complexes are prevalent motifs with a diverse
range of applications from energy science to asymmetric
catalysis.9,10 At the core of many of these applications is metal−
ligand cooperativity, whereby redox or chemical events between
the transition metal and an ancillary ligand work in concert.
This concept is now established as a key mechanistic
component of iron-11−16 and ruthenium-catalyzed hydro-
genation of carbonyl compounds,17,18 iridium-19 and iron-
catalyzed20 CO2 hydrogenation to formate, and the ruthenium-
catalyzed dehydrogenation of alcohols to esters, amides and
imines21,22 and dehydrogenation of ammonia-borane.23
Our laboratory has been interested in the application of iron
and cobalt complexes containing redox-active ligands, those
that undergo reversible electron transfer events with the
transition metal,24,25 to catalysis. Aryl-substituted bis(imino)-
pyridine cobalt methyl complexes,26 (ArPDI)CoCH3 are
exemplary as well-established redox-active compounds best
described as having low-spin Co(II) centers engaged in
antiferromagnetic coupling interactions with chelate-centered
radical anions (Figure 1).27,28 Compounds with this electronic
Figure 1. Electronic structures of cobalt methyl complexes bearing
tridentate pincer ligands.
structure are active precatalysts for alkene hydrogenation,29
including asymmetric variants,10 as well as for alkene
hydroboration.30 With this class of compounds, introduction
of electron-donating groups into bis(imino)pyridine iron and
cobalt catalysts increased activity in alkene hydrogenation,31
hydrosilylation,32 and hydroboration.30
Inspired by these observations in iron and cobalt catalysis
where more electron-donating ligands improved catalytic
performance, aryl-substituted bis(arylimidazol-2-ylidene)-
pyridine cobalt alkyl complexes, (ArCNC)CoCH3, with more
Received: April 30, 2014
Published: June 4, 2014
© 2014 American Chemical Society
9211
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
electron-donating pyridine dicarbene ligands,33 were studied.
These compounds are highly active precatalysts for the
hydrogenation of sterically demanding, unactivated alkenes
such as trans-methylstilbene, 1-methyl-1-cyclohexene, and 2,3-
dimethyl-2-butene, representing one of the most active cobalt
hydrogenation catalysts reported to date.34 A combination of
structural, spectroscopic, and computational studies established
that the pincer ligands in (ArCNC)CoCH3 compounds are also
redox-active, and the complexes are best described as low-spin
cobalt(II) compounds engaged in antiferromagnetic coupling
with one electron reduced bis(arylimidazol-2-ylidene)pyridine
chelates to account for the observed diamagnetic ground states
(Figure 1). Alkyl and hydride migration to the 4-position of the
central pyridine provided experimental support for this view of
the electronic structure. These results in conjunction with the
alkene, ketone, and imine hydrogenation activity reported by
Hanson and co-workers with amine-based PNP pincer ligated
cobalt complexes,35−37 suggested that four-coordinate cobalt
alkyl complexes supported by pincer ligands offer the
opportunity to exploit metal−ligand cooperativity and enable
new reactivity and catalytic chemistry.
Scheme 1. Oxidative Addition Reactivity of (iPrPNP)CoCH3
As a Route to Co(III) Complexes
Neutral, tridentate meridionally coordinating bis-
(phosphino)pyridine (“PNP”) pincer ligands38 are among the
most widely studied and aptly demonstrated examples for
metal−ligand cooperativity in homogeneous catalysis. The
acidity of the methylene spacers in the benzylic positions of the
ligands often results in facile deprotonation and enables the
dearomatization−aromatization behavior that is the origin of
the chemical noninnocence observed during catalytic turn-
over.39−41 One open question surrounding this class of pincers
is their redox activity and ability to undergo reversible one
electron transfer with a first row transition metal. The presence
of a central pyridine ligand, much like in the CNC complexes,
may be the origin of such behavior. It is well established that
C−H bonds adjacent to organic radicals are significantly
weakened as compared to those in the parent closed-shell
molecule,42 suggesting that one electron reduction of PNP-type
pincer ligands may lead to chemical noninnocence.
In 2006, our laboratory reported the synthesis and
spectroscopic investigations of reduced [(iPrPNP)Fe] com-
pounds.43 Our results, in conjunction with those of Goldman
and co-workers, established these complexes as more electron
rich as compared to analogous bis(imino)pyridine and
bis(arylimidazol-2-ylidene)pyridine iron complexes but ob-
tained no evidence to suggest a redox-active pincer.43−46 In
line with these observations, we recently reported the synthesis
of a related cobalt alkyl complex, (iPrPNP)CoCH2SiMe3 and its
propensity to undergo two-electron reactivity such as the
oxidative addition of H2 and the C−H bonds of terminal
alkynes and arenes to provide rare examples of organometallic
Co(III) complexes derived from oxidative addition (Scheme
1).47,48 Extending this chemistry to boranes, including HBPin
and B2Pin2 (Pin = pinacolato), resulted in highly active catalysts
for the C−H borylation of arenes and various heterocycles,49
suggesting that changes in preferred electronic structure may
enable new catalytic chemistry.
SOMO suggesting that following H atom loss, the pincer
supports an organic radical. As an alternative possibility,
concomitant intramolecular metal-to-ligand one-electron trans-
fer could lead to a modified pincer that adopts a closed-shell
anionic form, resulting in a low spin Co(II) center and, hence, a
metal-based SOMO. This view of the electronic structure is
frequently encountered with second and third row transition
metals.2,41
These discrepancies raised several questions for future study
such as: (i) What are the electronic structures of cobalt halide
complexes with the modified pincer ligands? (ii) Can alkyl
complexes with intact PNP pincers be isolated and what effect
does the X-type ligand coordinated to cobalt have on the
stability of the compounds? (iii) Under what conditions does
ligand modification occur and are homolytic versus heterolytic
pathways dominant? To answer these questions, a systematic
study of the electronic structures of four coordinate cobalt
halide, alkyl, and hydride complexes bearing both intact and
modified pincer ligands has been carried out. Specifically, the
preparation and structural characterization of cobalt halide,
alkyl, acetylide, and hydride derivatives is described. Systematic
studies into the origin and pathways of H atom loss from the
benzylic positions of the pincer were conducted and the bond
dissociation free energies as a function of X-type ligand were
computed and experimentally evaluated. The electronic
structures of the resulting cobalt products with modified
pincers have also been studied to distinguish between ligand
and metal centered radicals. Comparisons to a rhodium
congener are made and highlight the unique role of the base
metal and its accessible one electron chemistry in enabling
metal−ligand cooperativity.
The two-electron chemistry enabled by the [(RPNP)Co]
platform contrasts a recent communication by Milstein and co-
workers who reported the kinetic instability of (RPNP)CoCH3
toward loss of an H atom from the benzylic position of the
pincer ligand (Scheme 2).50 The resulting cobalt methyl
complex with a modified pincer, (RmPNP)CoCH3 has an S =
1/2 ground state and was described as having a ligand based
RESULTS AND DISCUSSION
■
Synthesis and Characterization of [(tBuPNP)Co] Chlor-
ide, Alkyl, and Hydride Compounds. Our studies
commenced with the synthesis of chloride, methyl, and hydride
complexes of the tert-butyl variant of the PNP pincer. This
specific substitution was chosen to compare the reaction
9212
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
Scheme 2. Ligand Modification via Proposed H-Atom Loss from (RPNP)CoCH3 Compounds As Previously Reported (R = iPr,
tBu).50
Scheme 3. Synthesis of (tBuPNP)CoCl, (tBuPNP)CoCH3, and (tBuPNP)CoH
Figure 2. Representation of the molecular structures of (tBuPNP)CoCl (left) and (tBuPNP)CoCH3 (right) with 30% probability ellipsoids.
Hydrogens, except those on the benzylic positions, are omitted for clarity.
chemistry with our previous studies with (iPrPNP)CoX
compounds and because Milstein and co-workers previously
reported improved thermal stability for the methyl derivative,
(iPrPNP)CoCl and is likely a result of the enforcement of a
planar geometry imparted by the large tert-butyl substituents
(vide infra).
(
(
tBuPNP)CoCH3.50 The synthetic route used to prepare
tBuPNP)CoCl, (tBuPNP)CoCH3, and (tBuPNP)CoH is detailed
The solid state structure of (tBuPNP)CoCl was determined
by X-ray diffraction and a representation of the molecule is
presented in Figure 2. Selected metrical parameters are
reported in Table 1. The overall molecular geometry is best
described as distorted planar with one of the phosphine ligands
lifted slightly above the idealized metal−ligand plane. The
N(1)−Co(1)−Cl(1) bond angle of 178.36(3)° is nearly linear
while the P(1)−Co(1)−P(2) angle is comparatively large at
172.086(12)°. This contrasts the nearly tetrahedral geometry
observed with (iPrPNP)CoCl47 and suggests that the planarity
of (tBuPNP)CoCl is a result of the steric profile of the
in Scheme 3. As reported for the isopropyl variant,47 treatment
of a toluene suspension of (tBuPNP)CoCl2 with one equivalent
of NaBEt3H followed by filtration and recrystallization
furnished (tBuPNP)CoCl as a red-brown solid in 66% isolated
31
yield. The benzene-d6 1H, 13C, and P NMR spectra exhibit
the number of resonances for a diamagnetic C2v symmetric
compound, confirming that the pincer ligand remains intact
and has not undergone modification. The observed diamag-
netism contrasts the S = 1 magnetic ground state observed for
9213
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
Table 1. Selected Bond Distances (Å) and Angles (Deg) for (tBuPNP)CoCl, (tBuPNP)CoCH3, (tBuPNP)CoH, and
(
tBuPNP)CoCCPh
(
tBuPNP)CoCl
(
tBuPNP)CoCH3
(
tBuPNP)CoH
(
tBuPNP)CoCCPh
Co(1)−Cl(1)/C(24)
Co(1)−N(1)
Co(1)−P(1)
Co(1)−P(2)
C(2)−C(3)
C(3)−C(4)
C(4)−C(5)
C(5)−C(6)
N(1)−C(2)
N(1)−C(6)
C(1)−C(2)
C(6)−C(7)
N(1)−Co(1)−Cl(1)/C(24)
P(1)−Co(1)−P(2)
2.2490(6)
1.8972(10)
2.1912(9)
2.1868(9)
1.3896(18)
1.3817(19)
1.3819(19)
1.3816(17)
1.3740(16)
1.3863(16)
1.4983(18)
1.5007(17)
1.9994(13)
1.9033(10)
2.1834(4)
1.8579(12)
1.9103(10)
2.1574(3)
1.889(5)
2.127(2)
2.127(2)
1.377(11)
1.405(11)
1.385(11)
1.391(10)
1.387(9)
1.383(9)
1.502(10)
1.505(10)
2.1816(4)
2.1599(3)
1.3793(17)
1.3945(19)
1.392(2)
1.3862(16)
1.3856(18)
1.3875(18)
1.3861(16)
1.3785(14)
1.3807(14)
1.5013(16)
1.4996(16)
178.25(4)
1.3796(18)
1.3892(16)
1.3909(15)
1.4974(17)
1.4977(18)
178.76(5)
178.36(3)
172.086(12)
172.942(14)
173.20(8)
174.557(13)
Figure 3. Representation of the molecular structures of (tBuPNP)CoH (left) and (tBuPNP)CoCCPh (right) at 30% probability ellipsoids. Hydrogens,
except those on the benzylic postions and the cobalt-hydride, are omitted for clarity.
phosphine substituents where the larger tert-butyl groups
inhibit the chloride ligand from distorting from the idealized
metal−ligand plane.
resonance. A single 31P NMR signal was observed at 85.2
ppm, corroborating formation of a diamagnetic C2v symmetric
product. A medium intensity band was also identified at 1746
cm−1 in the solid-state (KBr) infrared spectrum, consistent with
a terminal cobalt-hydride.51
Methylation of (tBuPNP)CoCl was accomplished by treat-
ment with one equivalent of a 1.6 M diethyl ether solution of
LiCH3 at −100 °C and warming to ambient temperature
followed by recrystallization (Scheme 3). This procedure
furnished (tBuPNP)CoCH3 as dark brown crystals in 85%
yield and the NMR spectroscopic features are identical to those
previously reported by Milstein.50 To our knowledge, the solid-
state structure of (tBuPNP)CoCH3 has not been reported. X-ray
quality crystals of the compound were obtained from a diethyl
ether solution stored at −35 °C. A representation of the
molecular structure is reported in Figure 2 and established an
overall idealized planar geometry about the cobalt. Selected
bond distances are reported in Table 1.
The solid state structure of (tBuPNP)CoH was determined by
X-ray diffraction. A representation of the molecule is presented
in Figure 3 and selected bond distances and angles are reported
in Table 1. The hydride ligand was located in the difference
map. As with the chloride and methyl derivatives, (tBuPNP)-
CoH is planar. The metrical parameters of the cobalt-pincer
linkages are similar to those observed for (tBuPNP)CoCl.
Notably, introduction of larger tert-butyl substituents on the
phosphines of the pincer stabilize the cobalt(I) hydride; the
corresponding isopropyl-substituted compound, (iPrPNP)CoH,
has thus far eluded observation due to rapid decomposition by
P−C bond cleavage of the pincer. Placing a benzene-d6 solution
of (tBuPNP)CoH under four atmospheres of H2 resulted in
broadening of the hydride signal into the baseline and no
observation of free dihydrogen in solution. The methylene
hydrogens shift downfield to 2.46 ppm and the para-pyridine
upfield to 7.79 ppm. Both values approach those observed for
(iPrPNP)CoH3,47 suggesting (tBuPNP)CoH3 is formed and
exchanges with (tBuPNP)CoH. Even higher pressures of H2
may allow observation of (tBuPNP)CoH3, conditions not
conveniently or safely accessed in standard NMR tubes.
The preparation of (tBuPNP)CoH was also explored (Scheme
3). Addition of NaBEt3H to a toluene solution of (tBuPNP)-
CoCl followed by filtration and recrystallization yielded red-
brown crystals identified as (tBuPNP)CoH in 69% yield. This
compound was also prepared by addition of 4 atm of H2 to
(
tBuPNP)CoCH3. With the isopropyl variant, addition of 4 atm
of H2 to (iPrPNP)CoCH3 resulted in loss of methane and
isolation and crystallographic characterization of a rare example
of a Co(III) trihydride, (iPrPNP)CoH3.47 The benzene-d6 1H
NMR spectrum of (tBuPNP)CoH exhibited a triplet (2JPH
=
65.3 Hz) at −29.0 ppm assigned as the cobalt-hydride
9214
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
The cobalt acetylide, (tBuPNP)CoCCPh was prepared by
Table 2. Measured Redox Potentials for (tBuPNP)CoX
Compounds
addition of the terminal alkyne to a benzene-d6 solution of
(
tBuPNP)CoH. The desired product was isolated as diamag-
compound
E° (V vs Fc/Fc+)
netic, red-brown crystals in 94% yield and characterized by
multinuclear NMR spectroscopy and X-ray diffraction (eq 1).
No evidence was obtained for alkyne insertion products.
(
(
(
(
tBuPNP)CoCl
−1.32
−1.54
−1.79
−2.14
tBuPNP)CoCCPh
tBuPNP)CoH
tBuPNP)CoCH3
CoCH3 in this study and thermal stability of (iPrPNP)CoCH3
reported in our prior work47 contrasts Milstein’s claim that
complexes of this type are kinetically unstable toward
spontaneous H atom loss.50 Because it is well established that
PNP-based pincer ligands often have acidic methylene
spacers,39 heterolytic C−H bond cleavage by deprotonation
of the benzylic positions of (tBuPNP)CoCl was explored.
Addition of one equivalent of NaOtBu to a THF solution of
The electronic properties of the series of (tBuPNP)CoX
compounds were investigated by cyclic voltammetry (CV). In a
standard experiment, a 1 mM solution of the cobalt compound
was prepared in tetrahydrofuran with 0.1 M of [NBu4][PF6] as
electrolyte. The obtained CVs are presented in Figure 4. At 100
(
tBuPNP)CoCl at −35 °C followed by removal of the volatiles
and extraction of the product into pentane yielded red crystals
identified as the cobalt dinitrogen complex, (tBumPNP)CoN2,
with a modified pincer ligand (eq 2). Similar chemistry has
previously been reported for the analogous bis(imino)pyridine
cobalt chloride compound where treatment with KOtBu
resulted in deprotonation of an imine methyl group with
concomitant salt elimination and yielded the modified cobalt
dinitrogen complex.28,53 The rhodium congener of (tBumPNP)-
CoN2 has also been isolated by Goldberg and co-workers from
treatment of [(tBuPNP)RhN2][OTf] with strong bases such as
LiCH3 or KOtBu in THF.41 Proton transfer from the benzylic
positions was also used to account for isotopic exchange with
D2 gas in related PNP-pincer iridium phenyl complexes.54
The diamagnetic cobalt dinitrogen complex was character-
ized by a combination of multinuclear NMR and IR
spectroscopies as well as X-ray diffraction. A strong N−N
stretch was observed at 2021 cm−1 for the coordinated
dinitrogen ligand. The solid-state structure, presented in Figure
5, confirms the identity of the molecule as an essentially planar,
Figure 4. Cyclic voltammagrams of (tBuPNP)CoX complexes in
tetrahydrofuran with 0.1 M [NBu4][PF6] as electrolyte and 100 mV/s
scan rate.
mV/s scan rate, all complexes display a single reversible wave
assigned as the Co(I)/Co(II) redox couple. The measured
redox potentials for the four compounds are presented in Table
2. Consistent with the relative electron density at the Co(I)
center in each complex, the compounds become progressively
easier to oxidize as we move from (tBuPNP)CoCl to
(
tBuPNP)CoCH3. Milstein and co-workers have reported the
electrochemistry of (tBuPNP)RhCl in acetone and located the
Rh(I)/Rh(II) redox couple at 0.37 V vs Fc/Fc+, shifted by over
1.6 V from the analogous cobalt complex.52
Figure 5. Representation of the molecular structure of (tBumPNP)-
CoN2 at 30% probability ellipsoids. Hydrogen atoms, except on
benzylic positions, are omitted for clarity. Only one crystallographically
independent molecule is shown.
Exploration of Ligand Modification: Heterolytic and
Homolytic C−H Bond Cleavage. The isolation of (tBuPNP)-
9215
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
cobalt(I) dinitrogen compound. This electronic structure is
confirmed by DFT calculations. A qualitative MO diagram is
shown in Supporting Information Figure S3. Selected bond
distances and angles are reported in Table 3. The metrical
establishing a homolytic bond dissociation energy of less than
67 kcal/mol for the methylene positions of the PNP pincer.55
Similar experiments were conducted with (tBuPNP)CoCl,
(
tBuPNP)CoH, and (tBuPNP)CoCCPh and cleanly yielded the
corresponding (tBumPNP)CoX derivatives in good to high yield
in each case (eq 3).
Table 3. Selected Bond Distances (Å) and Angles (Deg) for
(
tBumPNP)CoN2, (tBumPNP)CoCl, and (tBumPNP)CoH
(
tBumPNP)
CoN2
(
tBumPNP)
CoCl
(
tBumPNP)
CoH
Co(1A)-N(1)
Co(1A)-P(1)
Co(1A)-P(2A)
Co(1A)-Cl(1A)
C(2)−C(3)
C(3)−C(4)
C(4)−C(5)
C(5)−C(6)
N(1)−C(2)
1.9439(12)
2.2240(4)
2.1975(4)
1.962(5)
2.248(6)
2.264(13)
2.197(7)
1.420(3)
1.362(3)
1.395(3)
1.382(3)
1.390(2)
1.380(3)
1.399(3)
1.459(13)
1.953(6)
2.1999(16)
2.1999(16)
1.437(2)
1.360(2)
1.406(2)
1.374(2)
1.4049(18)
1.3714(17)
1.378(2)
1.499(2)
1.411(7)
1.385(7)
1.385(7)
1.411(7)
1.388(7)
1.388(7)
1.41(2)
All three modified pincer cobalt complexes are paramagnetic
with effective magnetic moments consistent with S = 1/2
1
ground states. Each compound exhibits relatively sharp H
N(1)−C(6)
C(1)−C(2)
C(6)-C(7A)
NMR spectra in benzene-d6 solution at 23 °C with the number
of paramagnetically shifted resonances consistent with Cs
symmetric compounds with a mirror plane of symmetry that
contains the idealized metal-chelate plane. Two examples,
1.52(3)
P(1)-Co(1A)-P(2A)
169.688(16)
167.9(3)
174.36(9)
(
tBumPNP)CoCl and (tBumPNP)CoH were characterized by X-
parameters of the solid state structure clearly demonstrate the
dearomatization of the pyridine ring, with alternating bond
lengths of C(2)−C(3) (1.437(2) Å), C(3)−C(4) (1.360(2)
Å), C(4)−C(5) (1.406(2) Å), and C(5)−C(6) (1.374(2) Å).
The difference between the benzylic C−C bond lengths of
C(1)−C(2) (1.378(2) Å) and C(6)−C(7) (1.499(2) Å)
indicate the presence of a monoanionic chelate. The facile
deprotonation associated with (tBuPNP)CoCl established the
acidity of the methylene positions of the PNP positions when
coordinated to a reduced cobalt center, suggesting heterolytic
C−H bond cleavage pathways are possible in the formation of
modified chelates similar to known rhodium complexes.
Because of the reported H atom loss chemistry from
(RPNP)CoCH3 complexes,50 homolytic pathways for mod-
ification of the PNP ligand were also explored. The isolation of
a family of alkyl, chloride, acetylide, and hydride complexes
allows for the systematic study of homolytic C−H bond
scission as a function of the X-type ligand in the cobalt
coordination sphere. Milstein has previously reported that
addition of TEMPO to (iPrPNP)CoCH3 resulted in H atom
abstraction and formed (iPrmPNP)CoCH3 and TEMPO-H,
ray diffraction and representations of the molecular structures
are presented in Figure 6. Although the hydrogens on the
pincer ligand were located in both cases and confirmed ligand
modification, the Co−H of (tBumPNP)CoH was not. A
medium intensity band was located at 1819 cm−1 in the
solid-state infrared spectrum of the compound, establishing the
presence of the Co−H. In both structures, the C(1)−C(2)
distances of 1.399(3) (Cl) and 1.41(2) (H) Å are significantly
shorter than the values of 1.459(13) and 1.52(3) Å for the
corresponding Cipso−Cmethylene positions, signaling ligand
modification. Alternating bond lengths are also observed in
the pyridine rings, indicating dearomatization (Table 3).
Each of the (tBumPNP)CoX compounds was additionally
characterized by X-band EPR spectroscopy. The electronic
structure of (iPrmPNP)CoCH3 was previously described as a
low spin Co(I) compound with a ligand-centered radical arising
from a redox-active modified pincer.50 Throughout the
manuscript, we have represented the modified pincer as an
anionic ligand thereby rendering the metal as low-spin Co(II)
to accommodate the measured magnetic data. For all
complexes, the obtained magnetic moments are significantly
Figure 6. Representation of the molecular structure of (tBumPNP)CoCl (left) and (tBumPNP)CoH (right) at 30% probability ellipsoids. Hydrogen
atoms, except on benzylic positions, are omitted for clarity. Only one of two disordered orientations of each molecule is shown.
9216
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
Figure 7. EPR spectra of (iPrmPNP)CoCH3 (upper left), (tBumPNP)CoCl (upper right), (tBumPNP)CoCH3 (lower left), and (tBumPNP)CoH
(lower right). Each spectrum was recorded in toluene/diethyl ether glass at 10 K. Conditions for (iPrmPNP)CoCH3, microwave frequency = 9.384
GHz, power = 0.63 mW, modulation amplitude = 1 mT/100 kHz. Conditions for (tBumPNP)CoCl, microwave frequency = 9.384 GHz, power =
0.63 mW, modulation amplitude = 1 mT/100 kHz. Conditions for (tBumPNP)CoH, microwave frequency = 9.375 GHz, power = 2.00 mW,
modulation amplitude = 1 mT/100 kHz. Conditions for (tBumPNP)CoCH3, microwave frequency = 9.379 GHz, power = 0.63 mW, modulation
amplitude = 1 mT/100 kHz.
higher than the spin-only value expected for S = 1/2
compounds, indicating significant contributions from spin−
orbit coupling. The two descriptions differ by the parentage of
the SOMO (Scheme 2); in the published case, the unpaired
spin is pincer-based, which should result in small orbital
contributions to the magnetic moment, whereas in the latter,
the electron is metal-centered opening a pathway for strong
spin−orbit coupling.
Measurements were initially conducted in toluene solution at
ambient temperature as organic radicals are typically observed
under these conditions. No signal was observed for any
(RmPNP)CoX compound at this temperature. However,
preparing samples of each compound in toluene−diethyl
ether glasses and subsequent cooling to 10 K produced the
spectra presented in Figure 7. In each case, an axial signal was
observed with one large g value at approximately g = 4.3−4.5,
which is split into eight lines due to hyperfine coupling to the
59Co nucleus (I = 7/2, 100% natural abundance), and two
smaller g values below 2. These signals do not exhibit any
resolved hyperfine coupling.
spin cobalt(II) complexes are well known to exhibit highly
anisotropic EPR spectra.56 Tonzetich and co-workers have
recently reported similar EPR spectral properties for planar
Co(II) complexes bearing N-heterocyclic carbenes.57 The
origin of the large g anisotropy is a result of strong spin−
orbit coupling arising from a nearly degenerate set of d orbitals
in a planar ligand field, which results in large orbital
contributions to the electronic g values. A theoretical analysis
by Nishida and co-workers demonstrated that both g and A
values are highly sensitive to small changes in energy differences
between the four low-lying d orbitals in the planar ligand
field.56c On the basis of these findings, the spectra presented in
Figure 7 were readily simulated using the parameters reported
in Table 4. Accurate reproduction of the line shapes was
accomplished using a combination of g-strain and A-strain
parameters to model a distribution of slightly different
molecular geometries frozen out upon formation of the solvent
glass resulting in highly anisotropic line broadening.
The possibility of thermally induced C−H bond scission was
investigated for all four (tBuPNP)CoX compounds (Table 5).
Each thermolysis reaction was conducted in benzene-d6 and
The observation of highly anisotropic signals and very large
cobalt hyperfine coupling clearly supports a cobalt rather than
[PNP]-based radical and definitively establishes that the
modified pincer is a closed-shell, monoanionic ligand. Low-
1
monitored by H NMR spectroscopy. Both the diamagnetic
starting compounds and the paramagnetic products exhibit
sharp, readily assigned NMR spectra facilitating determination
9217
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
Equilibration studies were also conducted to gain additional
experimental insight into the relative C−H bond strengths of
the benzylic positions of the pincer ligand as a function of the
X-type ligand. In each experiment, equimolar mixtures of
Table 4. Summary of EPR Spectroscopic Parameters for
a
(RmPNP)CoX Compounds
(
tBumPNP)
CoCl
(
tBumPNP)
CoMe
(
iPrmPNP)
CoMe
(
tBumPNP)
CoH
(
tBumPNP)CoX and (tBuPNP)CoY derivatives were mixed and
gx
4.50
1.15
1.06
1450
4.66
0.75
0.65
1600
4.35
1.32
1.15
1270
4.30
1.68
1.34
1560
the propensity to undergo H atom transfer assessed by NMR
spectroscopy as both the diamagnetic and paramagnetic
compounds are readily observable. A summary of the results
of these experiments is reported in Table 6.
gy
gz
Axx
(MHz)
Mixing cobalt compounds with two X-type ligands that both
Astrain
gstrain
150, 0, 0
70, 0, 0
160, 0, 0
40, 0, 0
undergo thermal H atom loss, (tBuPNP)CoCH3 and
(
0.26, 0.26,
0.30
0.14, 0.14,
0.14
0.25,0.22,0.31 0.20, 0.30,
0.30
tBumPNP)CoH, yielded (tBumPNP)CoCH3 and (tBuPNP)CoH
a
A values reported in MHz.
as products. At 23 °C, all four compounds were present in near
equal amounts immediately after mixing (Table 6). The same
equilibrium mixture was attained from mixing (tBumPNP)-
CoCH3 with (tBuPNP)CoH (Entry 1). Preparation of a
combination of compounds where one member undergoes
thermal H atom loss, (tBuPNP)CoCH3, with a modified
compound that does not, (tBumPNP)CoCl, resulted in
immediate formation of (tBumPNP)CoCH3 and (tBuPNP)CoCl
(Entry 4). Allowing the resulting solution to stand at 23 °C or
mixing fresh solutions of the products produced no change,
demonstrating the irreversibility of the process. Performing
similar experiments with other combinations support the
observations from the thermolysis experimentsthose com-
pounds that undergo thermal ligand modification undergo H
atom transfer to modified complexes that do not (Table 6).
These results establish that the compounds with strong field,
pure σ-donating X-type ligands such as hydride and methyl
have weaker benzylic C−H bonds than analogues with weaker
field ligands, such as chloride and acetylide.58
To gain additional insight into ligand modification by C−H
bond homolysis, attempts were made to measure the kinetic
isotope effect for conversion of the pincer complexes into their
modified counterparts. These experiments required the
deuterated isotopologue, (tBuPNP-d5)CoD, a compound
conveniently prepared by exposure of a benzene-d6 solution
of (tBuPNP)CoH to 4 atm of D2 for 48 h. The deuteria are
located in the four benzylic positions and in the para pyridine.
Performing a control experiment whereby a benzene-d6
solution of (tBuPNP)CoH was treated with an equimolar
amount of (tBumPNP-d4)CoD at 23 °C resulted in immediate
(<5 min) exchange of the deuterium and hydrogen between the
methylene positions of the two compounds. The converse
experiment whereby (tBuPNP-d5)CoD and (tBumPNP)CoH
were mixed produced an identical result, demonstrating that
Table 5. Evaluation of the Thermal Stability of (RPNP)CoX
Compounds
compound
tBuPNP)CoCl
tBuPNP)CoCCPh
tBuPNP)CoH
tBuPNP)CoCH3
T (°C)
time (hrs)
outcome
no reaction
(
(
(
(
120
120
110
120
24
24
16
18
no reaction
(
(
tBumPNP)CoH
tBumPNP)CoCH3
of the progress of each reaction. Heating a solution of either
tBuPNP)CoCl or(tBuPNP)CoCCPh to 120 °C for 24 h
produced no change. By contrast, heating (tBuPNP)CoCH3 or
tBuPNP)CoH to 110 °C for 18 h resulted in conversion to the
(
(
corresponding modified compounds. Analysis of the volatile
products following (tBuPNP)CoH thermolysis by Toepler
pump established formation of H2 gas as a product of the
reaction. Approximately 30% of the expected combustible gas
was collected. The fate of the balance of the H atoms is
unknown, but a significant quantity of free ligand and metallic
cobalt was also observed. Performing the thermolysis of
(
tBuPNP)CoH on a preparative scale allowed for the isolation
of (tBumPNP)CoH in 46% yield, suggesting decomposition of
(
tBuPNP)CoX complexes is competitive with H atom loss. It is
possible the observed H2 is a result of an unidentified
decomposition process.
Table 6. Ladder Experiments to Establish Experimental Estimates of Pincer C−H Bond Dissociation Free Energies (BDFEs)
entry
Y
X
Keq (23 °C)
1
2
3
4
5
6
H
CH3
Cl
∼1
H
>100
>100
>100
>100
∼1
H
CCPh
Cl
CH3
CH3
Cl
CCPh
CCPh
9218
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
interpretation of kinetic isotope effect by this method is
compromised by “self exchange” between like isotopologues.
Notably, the 1H NMR spectra of mixtures of (tBuPNP)CoX and
loss chemistry is not operative at ambient temperature and
more traditional hydride insertion and two-electron oxidative
addition type reactivity is preferred.
(
tBumPNP)CoX are unperturbed from their consistuent single
component spectra, indicating that the H atom self-exchange
process, although rapid synthetically, is not observable on the
NMR time scale. Attempts to obtain reliable KIE data on the
thermolysis of (tBuPNP)CoH to (tBumPNP)CoH were also
frustrated by the rapid self-exchange process as well as from
complications from the formation of metallic cobalt and free
ligand that accompanied the H atom loss event.
How then is Milstein’s isolation of (RmPNP)CoCH3
compounds explained? Recall that the synthesis of these
molecules relied on treatment of (RPNP)CoCl2 with a large
(∼10 equiv) excess of MeLi.50 A proposed sequence to account
for the observed products is presented in Scheme 4. The
sequence begins with deprotonation of a C−H bond of the
methylene position with LiCH3 to liberate methane and
precipitate lithium chloride. Similar reactivity has been
observed upon addition of KOtBu to a related PNN-pincer
complex of platinum,59 PNP derivatives of rhodium,52
palladium,60 nickel,61 copper,62 and in this work in the
synthesis of (tBumPNP)CoN2 by deprotonation and salt
elimination from (tBuPNP)CoCl. Once (RmPNP)CoCl com-
pounds are present, straightforward salt metathesis with LiCH3
yields (RmPNP)CoCH3 (Scheme 4). It is likely that the excess
alkylating agents used in the Milstein synthesis accounts for the
observation of the modified methyl complexes as heterolytic,
deprotonation pathways are preferred and that C−H homolysis
reactions are not operative at ambient temperature.
Reaction Chemistry with Alkynes. In Milstein’s
communication describing spontaneous ligand modification of
(RPNP)CoCH3 compounds at ambient temperature,50 diphe-
nylacetylene was used as a trap of H atoms ejected from the
benzylic positions of the pincer and isomers of stilbene were
detected. Because our observations demonstrate that C−H
bond homolysis does not occur at room temperature and has a
much higher barrier, we sought to reconcile this discrepancy.
Addition of PhCCPh to a benzene-d6 solution of either
(
tBuPNP)CoH or (tBuPNP)CoCH3, resulted in insertion of the
alkyne into the cobalt-hydride or alkyl followed by
orthometalation of one of the phenyl substituents (eq 4). For
the cobalt hydride, the reaction was complete after 18 h at 23
°C, whereas for the cobalt methyl, heating to 60 °C was
required and produced significant quantities of free pincer
ligand and metallic cobalt as side products. Both products were
characterized by multinuclear NMR spectroscopy and the
molecular structures were also confirmed by X-ray diffraction.
One example, (tBuPNP)Co(C6H4−CHCHPh) is presented
in Figure 8. All of the data suggests that one electron, H atom
Support for this hypothesis derives from the observation that
treatment of (tBuPNP)CoCl2 with 1.5 equiv of NaOtBu
exclusively yielded (tBumPNP)CoCl, demonstrating the acidity
of the benzylic position. Addition of 1 equiv of MeLi to a
diethyl ether or benzene slurry of (tBuPNP)CoCl2 also yielded
(
tBumPNP)CoCl along with (tBuPNP)CoCl and (tBumPNP)-
CoCH3 as judged by NMR spectroscopy. This experiment
demonstrates that MeLi acts as a base for benzylic
deprotonation. The observed side products are likely due to
the poor solubility of (tBuPNP)CoCl2. Also consistent with the
proposal in Scheme 4 is the clean alkylation of isolated
(
tBumPNP)CoCl with MeLi to yield (tBumPNP)CoCH3.
Computational Studies: Electronic Structure of
(RmPNP)CoX Compounds. In the original report of the
(RmPNP)CoCH3 derivatives, the electronic structure of the S =
1/2 modified complex was described as low spin cobalt(I) with
a ligand-centered radical as the SOMO.50 If such a description
is correct, then modified PNP ligands are redox active and
potentially offer unique reactivity pathways. The EPR spectra
reported in Figure 7 suggest otherwise as signals with large g
anisotropy and cobalt hyperfine coupling were observed,
consistent with metal-based rather than ligand-based spin.
Unrestricted, full molecule DFT calculations were also
Figure 8. Molecular structure of (tBuPNP)Co(C6H4−CHCHPh) at
30% probability ellipsoids. Hydrogen atoms, except those attached to
C30 and C31, are omitted for clarity.
Scheme 4. Proposed Pathway for the Generation of Modified Pincer Complexes from (RPNP)CoCl2 with Excess Methyl
Lithium
9219
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
performed on (tBumPNP)CoCH3 at the B3LYP level of theory
to corroborate the electronic structure proposed based on the
EPR data.63 A qualitative molecular orbital diagram and
associated spin density plot corresponding to the optimized
structure are presented in Figure 9. The electronic structure
Information Figure S4), demonstrating that such species are
chemically accessible with sufficiently potent reductants.
The S = 0 ground states for the (tBuPNP)CoX compounds
present an experimental challenge for establishing the redox
activity of the pincer. Comparison of the metrical parameters of
the crystallographically characterized cobalt compounds to
those with pincers undoubtedly in their neutral form, (e.g.,
(iPrPNP)FeBr2) revealed no statistically significant differences
in the bond lengths in the central pyridine ring. No anomalous
1H NMR shifts were observed.34 Because of these experimental
challenges, we relied on the output of DFT calculations for
electronic structure determination. Full molecule geometry
optimization computations on (tBuPNP)CoCl and (tBuPNP)-
CoCCPh established preferred closed-shell singlet ground
states corresponding to traditional low-spin Co(I), d8
complexes with redox innocent pincers. A representative
qualitative molecular orbital diagram for (tBuPNP)CoCl is
presented in the left portion of Figure 10, and that for
(
tBuPNP)CoCCPh is reported in Supporting Information
Figure S1.
Figure 9. Qualitative molecular orbital diagram (left) and computed
spin density plot (right) for (tBumPNP)CoCH3 at the B3LYP level of
DFT.
corresponds to a low spin Co(II) complex with the SOMO of
the molecule principally a dz molecular orbital on the metal.
2
The Mulliken spin density plot confirms this view of the
electronic structure with almost all of the unpaired spin in the
molecule located on the cobalt center. The highest doubly
occupied orbital is principally ligand-based, further supporting a
closed-shell form of the modified pincer. Importantly, there is
no unpaired spin density on the pincer ligand. Analogous
calculations for the remaining complexes (tBumPNP)CoX (X =
H, Cl, CCPh) produced the same electronic structure.
Figure 10. DFT-computed qualitative molecular orbital diagrams for
(
tBuPNP)CoCl (left) and (tBuPNP)CoCH3 (right).
Analogous computations on both (tBuPNP)CoCH3 and
(
tBuPNP)CoH produced broken symmetry solutions that are
best described as low-spin Co(II), d7 complexes engaged in
antiferromagnetic coupling interactions with a pincer radical
anion. A qualitative molecular orbital diagram for (tBuPNP)-
CoCH3 is presented in Figure 10, and that for (tBuPNP)CoH is
reported in Figure S2 of the Supporting Information. The
antiferromagnetic coupling interaction (S = 0.67, X = CH3; S =
0.77, X = H) occurs between an essentially dxz orbital on Co
with a ligand b2 orbital in the SOMO of the pincer. The spin
density plots for both (tBuPNP)CoCH3 and (tBuPNP)CoH are
presented in Figure 11. It is important to note, however, that
the energy differences between the broken symmetry and
closed-shell solutions for (tBuPNP)CoCH3 and (tBuPNP)CoH
are very small (∼1.5 kcal, Table 7) and perhaps within the error
limits of the computation. Recall that when RKS solutions are
favored, as in the case of (tBuPNP)CoCl and (tBuPNP)CoCCPh,
the energy of the broken symmetry solution can not be
obtained and the relative energy between the redox active and
redox innocent cases for these compounds is unknown. On the
Electronic Structure of (RPNP)CoX Compounds and
Computation of Benzylic C−H Bond Dissociation Free
Energies (BDFEs) as a Function of the X-Type Ligand.
The electronic structure of the family of (tBuPNP)CoX (X = Cl,
CCPh, Me, H) compounds was also explored to determine
(i) if the PNP in its unmodified form can support ligand
centered radicals and (ii) if there is a correlation between
preferred electronic ground state and propensity to undergo H
atom loss. Generation of a PNP radical anion was initially
targeted to determine under what, if any, conditions the chelate
could be reduced. Cyclic voltammetry of the free ligand,
tBuPNP showed no oxidation or reduction waves in the THF
solvent window. Chemical reduction of the free pincer was
accomplished by stirring a THF solution of tBuPNP with excess
sodium metal and the resulting radical anion was characterized
by EPR spectroscopy at ambient temperature (Supporting
9220
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
each complex undergoes H atom abstraction regardless of the
identity of X to form the corresponding modified compound
and TEMPO-H.
Comparison of Benzylic C−H Bond Dissociation Free
Energies (BDFEs) in First Row versus Second Row
Transition Metals. The remarkably low BDFEs of the
benzylic C−H bonds in (tBuPNP)CoX compounds prompted
further investigation into the origin of heterolytic vs homolytic
bond cleavage in (RPNP)MX complexes. Although ligand
modification is commonly observed in second row transition
metal complexes supported by PNP pincer ligands,2 heterolytic
C−H bond cleavage via deprotonation of the ligand framework
in conjunction with two-electron redox transformations is the
accepted pathway. During the preparation of this manuscript, a
rare example of homolytic C−H bond cleavage in a PNN-
ligated rhodium complex was reported.64 To explore this
possibility in more detail, the electronic structures of (tBuPNP)-
RhCl and its modified derivative (tBumPNP)RhCl were
computed. The homolytic bond dissociation free energies
(BDFE) for the benzylic C−H bond of the [tBuPNP] pincer
was also calculated. Similar to the cobalt congener, full
molecule calculations of (tBuPNP)RhCl established a diamag-
netic closed-shell ground state with no indication of a broken-
symmetry solution indicating that the compound is best
described as a Rh(I) complex with a neutral closed-shell pincer.
For the modified complex, (tBumPNP)RhCl, the computa-
tions yielded a doublet ground state in which the spin density is
located predominantly on the metal center (Figure 12). This is
Figure 11. DFT-computed Mulliken spin density plots corresponding
to broken-symmetry solutions for (tBuPNP)CoCH3 (left) and
(
tBuPNP)CoH (right).
Table 7. Electronic Structure and BDFE Correlations for
(
tBuPNP)CoX Compounds
electronic
structure
ΔE (kcal) (BS-
calculated BDFE
(kcal)
compound
RKS)
(
tBuPNP)
N/A
RKS
RKS
N/A
N/A
N/A
79.8
49.9
47.2
(
(
tBuPNP)CoCl
tBuPNP)
CoCCPh
(
(
(
tBuPNP)CoCH3
tBuPNP)CoH
tBuPNP)RhCl
BS(1,1)
BS(1,1)
RKS
1.4
44.6
42.7
73.7
1.0
N/A
basis of the combined experimental and computational results,
definitive evidence for the radical character on the tBuPNP
ligand is lacking but, as suggested by DFT, should be
considered for cases where strong σ-donating ligands are
present. Importantly, however, the compounds where broken
symmetry solutions were favored undergo thermal H atom loss
and have the weakest benzylic C−H bond as established by H
atom transfer experiments (Table 6).
Figure 12. DFT-computed Mulliken spin density plot for (tBumPNP)-
RhCl.
again analogous to the corresponding cobalt congener with the
compound best described as a rhodium(II) derivative with an
anionic closed-shell pincer. In contrast to the cobalt analogue,
however, the SOMO possesses mostly rhodium dxz character
The apparent experimental correlation between thermal H
atom loss and favored electronic structure prompted a more
detailed computational investigation into the homolytic bond
dissociation free energies (BDFE) of the benzylic positions of
the pincer and the identity of the X-type ligand on cobalt. Table
7 presents the results of DFT-computed (B3LYP functional)
BDFEs for the benzylic C−H bond for the four (tBuPNP)CoX
complexes. As a reference point, the BDFE of the benzylic
positions of the free ligand (tBuPNP) was also calculated and
gave a value of 79.8 kcal/mol. For (tBuPNP)CoCl and
(dz is the parentage for Co) and exhibits strong covalent
2
interactions with the now extended π system of the PNP ligand.
The increased covalency in metal−ligand bonding is expected
for the second row metal rhodium, because of its more diffuse
4d orbitals compared to the 3d metal cobalt. Similar trends
have previously been reported for a series of isostructural iron
and ruthenium dithiolene complexes.65,66
(
tBuPNP)CoCCPh where classical Co(I), d8 electronic
After establishing the electronic structures of the relevant
compounds, the BDFEs of the benzylic C−H bonds in
descriptions were computationally preferred, the strongest
C−H bonds, 49.9 and 47.2 kcal/mol were calculated. In the
cases where broken symmetry descriptions were favored,
(
tBuPNP)RhCl were computed as 73.7 kcal/mol. This value is
slightly lower than the calculated value for the free ligand but is
significantly higher than for the cobalt analogue. These results
clearly indicate that the nature of the central metal ion has a
dramatic influence on the BDFEs of the remote backbone C−
H bonds and showcases the differences between first and
(
tBuPNP)CoCH3 and (tBuPNP)CoH, significantly weaker C−
H BDFEs of 44.6 and 42.7 kcal/mol were computed, consistent
with the observed thermolysis studies. The DFT-computed C−
H BDFEs also rationalize the observations with TEMPO where
9221
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
second row transition metals. During the H atom transfer step,
the accessible Co(I)−Co(II) one-electron redox couple of the
first-row metal engages in single-electron transfer to the ligand,
resulting in an oxidized metal center and an overall one-proton,
two-electron transformation (net deprotonation) of the ligand.
In other words, the heterolytic C−H bond cleavage proceeds
via concerted proton−electron transfer (CPET)67 in which the
proton is supplied by the ligand, whereas the electron is
supplied by the metal center. In contrast to the first-row metal
cobalt, the second-row congener rhodium prefers two-electron
chemistry (Rh(I)−Rh(III)) and does not easily engage in
single-electron transfer to generate Rh(II). This is clearly
reflected in the Rh(I)−Rh(II) redox potential reported
previously for (tBuPNP)RhCl, which is shifted by more than
1.5 V compared to the potential reported in this work for
(
tBuPNP)CoX complexes were computed by DFT and range
between 42.7 and 49.9 kcal/mol, depending on the identity of
the X-type ligand where pure σ donors such as hydride and
methyl generate the weakest C−H bonds. These values were
experimentally corroborated by abstraction with TEMPO and
by H atom transfer experiments between cobalt complexes. The
origin of the relatively weak C−H bonds in these compounds
was traced to the energetically accessible one-electron redox
couple, for example, Co(I)−Co(II) of the first row transition
metal. In contrast, the second row congener, (tBuPNP)RhCl,
has a much higher thermodynamic barrier to one-electron
redox chemistry, and as such much, stronger pincer C−H
bonds are observed.
EXPERIMENTAL SECTION68
(
tBuPNP)CoCl rendering oxidation more difficult. These results
■
Preparation of (tBuPNP)CoCl. A 100 mL round-bottom flask was
charged with 0.409 g (0.779 mmol) of (tBuPNP)CoCl2 and 25 mL of
toluene. The solution was frozen in the cold well of the glovebox. To
the thawing suspension was added 780 μL (0.779 mmol) of a 1.0 M
NaHBEt3 solution in toluene dropwise via syringe. A color change
from dark purple to red-orange was observed immediately upon
warming to 23 °C. The reaction mixture was stirred for 12 h at 23 °C
followed by filtration through Celite. Recrystallization of the crude
material from a toluene-pentane mixture at −35 °C furnished 0.252 g
(0.514 mmol, 66% yield) of red-brown crystals suitable for X-ray
diffraction. Anal. Calcd for C23H43ClCoNP2: C, 56.39; H, 8.85; N,
show that the dramatic difference in C−H BDFEs between Co
and Rh can, therefore, be directly attributed to the tendency of
first-row metals to undergo one-electron redox processes.
Experimental support was sought for the higher computed
C−H BDFE of (tBuPNP)RhCl. As anticipated, treatment of a
benzene-d6 solution of (tBuPNP)RhCl with one equivalent of
TEMPO produced no reaction, even upon heating to 80 °C. In
contrast, addition of the sterically encumbered 2,4,6-tri(tert-
butyl)phenoxy radical with a BDFE of 76.7 kcal/mol in
benzene,56 proved capable of abstracting a H atom from the
PNP chelate, generating one equivalent of 2,4,6-tri(tert-
butyl)phenol and (tBumPNP)RhCl (eq 5).
1
2.86. Found: C, 56.23; H, 8.58; N, 2.93. H NMR (benzene-d6, 23
°C): δ 1.59 (br s, 36 H, P−CMe3), 1.69 (br s, 4H, P−CH2), 5.60 (d,
3JHH = 7.2 Hz, 2H, meta aryl CH), 7.69 (t, 1H, 3JHH = 7.2 Hz, para aryl
CH). {1H}13C NMR (benzene-d6, 23 °C): δ 30.0 (P−CMe3), 34.6 (br
s, P-CH2), 35.6 (br s, P-CMe3), 116.3 (meta aryl C), 122.3 (br m,
para-aryl C), 160.6 (br m, ipso aryl C). {1H}31P NMR (benzene-d6, 23
°C): δ 54.9 (br m, P(tBu)2).
Preparation of (tBumPNP)CoCl. A 20 mL scintillation vial was
charged with 0.148 g (0.302 mmol) of (tBuPNP)CoCl and 3 mL of
diethyl ether. To the stirring solution was added dropwise a solution
containing 0.045 g (0.302 mmol) of 2,2,6,6-tetramethylpiperidine-N-
oxide (TEMPO) in 1 mL of diethyl ether. An immediate lightening of
the solution from red-brown to orange was observed upon addition of
TEMPO. The solution was stirred overnight at room temperature
before the solution was filtered through Celite to remove excess 1-
hydroxy-2,2,6,6-tetramethylpiperidine (TEMPO-H). The filtrate was
concentrated in vacuo and recrystallized at −35 °C and yielded 0.118 g
(0.242 mmol, 80% yield) of dark red-orange crystals suitable for X-ray
diffraction identified as (tBumPNP)CoCl. Anal. Calcd for
These data confirm the computed BDFE and provide
convincing experimental support for the influence of a first
versus a second row transition metal on the properties of a
coordinated ligand. The reverse reaction, H atom addition to
the benzylic position of (tBumPNP)RhCl was also studied.
Heating a benzene-d6 solution of (tBumPNP)RhCl in the
presence of excess TEMPO-H to 80 °C for 12 h resulted in
reconstitution of (tBuPNP)RhCl, also in agreement with the
computed BDFE for this position. Repeating this procedure
with TEMPO−D furnished (tBuPNP-d1)RhCl with the isotopic
label located exclusively at the benzylic position in the
diamagnetic product.
C23H42ClCoNP2: C, 56.50; H, 8.65; N, 2.86. Found: C, 56.54; H,
1
8.70; N, 2.68. H NMR (benzene-d6, 23 °C): δ −55.6 (1H, Δν1/2
=
562 Hz), −47.8 (1H, Δν1/2 = 312 Hz), −30.4 (1H, Δν1/2 = 293 Hz),
−0.87 (1H, Δν1/2 = 26 Hz), 1.36 (2H, Δν1/2 = 33 Hz, P−CH2), 9.22
(18H, Δν1/2 = 514 Hz, P−CMe3), 12.55 (18H, Δν1/2 = 514, P−
CMe3). Magnetic susceptibility: (benzene-d6, 23 °C) μeff = 1.6(5) μB.
Preparation of (tBuPNP)CoH. This compound was prepared in a
similar manner to (tBuPNP)CoCl using 0.250 g (0.476 mmol) of
CONCLUDING REMARKS
tBuPNP)CoCl2 and 952 μL (0.952 mmol) of a 1.0 M NaHBEt3
■
(
A series of four-coordinate, tert-butyl substituted pincer cobalt
complexes, (tBuPNP)CoX, were isolated and their electronic
structures and thermal stability systematically evaluated.
Modification of the benzylic positions of the pincer by both
heterolytic and homolytic C−H bond cleavage proved feasible
with deprotonation pathways likely operative in previous
literature reports. The electronic structure of the resulting
modified pincer cobalt complexes, (tBumPNP)CoX was
established by EPR spectroscopy, magnetometry and computa-
tional studies as low-spin cobalt(II) complexes with a closed-
shell anionic form of the modified chelate. The homolytic bond
dissociation free energies of the benzylic C−H bonds of the
solution in toluene. Upon warming to 23 °C, a color change from dark
purple to red-orange was observed. Following work-up, recrystalliza-
tion of the crude material from diethyl ether at −35 °C furnished
0.150 g (0.328 mmol, 69% yield) of red-brown crystals suitable for X-
ray diffraction identified as (tBuPNP)CoH. Anal. Calcd for
C23H44CoNP2: C, 60.65; H, 9.74; N, 3.08. Found: C, 60.33; H,
1
2
9.59; N, 3.05. H NMR (benzene-d6, 23 °C): δ −29.0 (t, JPH = 65.3
Hz, 1H, Co-H), 1.45 (AA′XX′, 36 H, P−CMe3), 1.84 (m, 4H, P−
CH2), 5.77 (d, 3JHH = 7.2 Hz, 2H, meta aryl CH), 8.93 (br m, 1H, para
aryl CH). {1H}13C NMR (benzene-d6, 23 °C): δ 30.2 (m, P−CMe3),
33.8 (m, P-CH2), 38.0 (m, P-CMe3), 110.2 (br m, meta aryl C), 124.1
(para aryl C), 155.6 (br m, ipso aryl C). {1H}31P NMR (benzene-d6,
23 °C): δ 85.2 (br s, P(tBu)2). IR (KBr): vCo−H = 1746 cm−1.
9222
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
(22) (a) Zhang, J.; Leitus, G.; Ben-David, Y.; Milstein, D. J. Am.
Chem. Soc. 2005, 127, 10840. (b) Gunanathan, C.; Ben-David, Y.;
Milstein, D. Science 2007, 317, 790. (c) Gnanaprakasam, B.; Zhang, J.;
Milstein, D. Angew. Chem., Int. Ed. 2010, 48, 1468. (d) Gnanaprakasam,
B.; Milstein, D. J. Am. Chem. Soc. 2011, 133, 1682.
ASSOCIATED CONTENT
■
S
* Supporting Information
Additional experimental procedures, including general consid-
erations and characterization data, as well as computational and
spectroscopic results. Crystallographic data for (tBuPNP)-
CoCH3, (tBuPNP)CoCl, (tBuPNP)CoH, (tBumPNP)CoCl,
(23) Kass, M.; Friedrich, A.; Dress, M.; Schneider, S. Angew. Chem.,
̈
Int. Ed. 2009, 48, 905.
(
tBumPNP)CoH, (tBumPNP)CoN2, (tBuPNP)Co(C6H4−CH
(24) Knijnenburg, Q.; Gambarotta, S.; Budzelaar, P. H. M. Dalton
Trans. 2006, 5442.
(25) Chirik, P. J. Inorg. Chem. 2011, 50, 9737.
(26) Tellman, K. P.; Humphries, M. J.; Rzepa, H. S.; Gibson, V. C.
CHPh), and (tBuPNP)CoCCPh in .txt format. This material is
available free of charge via the Internet at http://pubs.acs.org.
Organometallics 2004, 23, 5503.
(27) Knijnenburg, Q.; Hetterscheid, D.; Kooistra, T. M.; Budzelaar,
P. H. M. Eur. J. Inorg. Chem. 2004, 1204.
AUTHOR INFORMATION
■
Corresponding Author
pchirik@princeton.edu
(28) Bowman, A. C.; Milsmann, C.; Atienza, C. C. H.; Lobkovsky, E.;
Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2010, 132, 1676.
(29) Knijnenburg, Q.; Horton, A. D.; van der Heijden, H.; Kooistra,
T. M.; Hetterscheid, D. G. H.; Smits, J. M. M.; de Bruin, B.; Budzelaar,
P. H. M.; Gal, A. W. J. Mol. Catal. A: Chem. 2005, 232, 151.
(30) Obligacion, J. V.; Chirik, P. J. J. Am. Chem. Soc. 2013, 135,
19107.
(31) Yu, R. P.; Darmon, J. M.; Hoyt, J. M.; Margulieux, G. W.;
Turner, Z. R.; Chirik, P. J. ACS Catal. 2012, 2, 1760.
(32) Atienza, C. C. H.; Tondreau, A. M.; Weller, K. J.; Lewis, K. M.;
Cruse, R. W.; Nye, S. A.; Boyer, J. L.; Delis, J. G. P.; Chirik, P. J. ACS
Catal. 2012, 2, 2169.
(33) Danopoulos, A. A.; Wright, J. A.; Motherwell, W. B.; Ellwood, S.
Organometallics 2004, 23, 4807.
(34) Yu, R. P.; Darmon, J. M.; Milsmann, C.; Margulieux, G. W.;
Stieber, S. C. E.; DeBeer, S.; Chirik, P. J. J. Am. Chem. Soc. 2013, 135,
13168.
Author Contributions
†These authors contributed equally to this work.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Princeton University is acknowledged for financial support.
Initial studies were enabled by financial support from the U. S.
National Science Foundation and the Deutsche Forschungsge-
meinschaft for a Cooperative Activities in Chemistry between
U. S. and German Investigators Grant (CHE-1026084).
Support for C.M. was provided by the Air Force Office of
Scientific Research (FA9550-11-1-0252) and we thank Dr.
Eckhard Bill (Max Planck Institute for Chemical Energy
Conversion) for assistance in interpreting the EPR spectra.
(35) Zhang, G.; Scott, B. L.; Hanson, S. K. Angew. Chem., Int. Ed.
2012, 51, 12102.
REFERENCES
(36) Zhang, G.; Hanson, S. K. Org. Lett. 2013, 15, 650.
(37) Zhang, G.; Vasudevan, K. V.; Scott, B. L.; Hanson, S. K. J. Am.
Chem. Soc. 2013, 135, 8668.
■
(1) The Chemistry of Pincer Compounds; Morales-Morales, D., Jensen,
C. M., Eds.; Elsevier: Amsterdam, 2007.
(38) (a) Moulton, C. J.; Shaw, B. L. J. Chem. Soc., Dalton Trans. 1976,
1020. (b) Gupta, M.; Hagen, C.; Flesher, R. J.; Kaska, W. C.; Jensen,
C. M. Chem. Commun. 1996, 2083. (c) van der Boom, M. E.; Milstein,
D. Chem. Rev. 2003, 103, 1759.
(2) Gunanathan, C.; Milstein, D. Acc. Chem. Res. 2011, 44, 588.
(3) Haneline, M. R.; Heyduk, A. F. J. Am. Chem. Soc. 2006, 128, 8410.
(4) Chirik, P. J.; Weighardt, K. Science 2010, 327, 794.
(5) Mazzacano, T. J.; Mankad, N. P. J. Am. Chem. Soc. 2013, 135,
17258.
(39) van der Vlugt, J. L.; Reek, J. N. H. Angew. Chem., Int. Ed. 2009,
(6) Grutzmacher, H.-J. Angew. Chem., Int. Ed. 2008, 47, 1814.
(7) Vogt, M.; Gargir, M.; Iron, M. A.; Diskin-Posner, Y.; Ben-David,
Y.; Milstein, D. Chem.Eur. J. 2012, 18, 9194.
48, 8832.
(40) Gunanathan, C.; Gnanaprakasam, B.; Iron, M. A.; Shimon, L. J.
W.; Milstein, D. J. Am. Chem. Soc. 2010, 132, 14763.
(41) Kloek, S. M.; Heinekey, D. M.; Goldberg, K. I. Angew. Chem.,
Int. Ed. 2007, 46, 4736.
(42) Blanksby, S. J.; Ellison, G. B. Acc. Chem. Res. 2003, 36, 255.
(43) Trovitch, R. J.; Lobkovsky, E.; Chirik, P. J. Inorg. Chem. 2006,
45, 7252.
(44) Pelczar, E. M.; Emge, T. J.; Krogh-Jespersen, K.; Goldman, A. S.
Organometallics 2008, 27, 5759.
(45) Danopoulos, A. A.; Wright, J. A.; Motherwell, W. B. Chem.
Commun. 2005, 784.
(46) For related iron carbonyl compounds with bis(phosphine)
(8) Bichler, B.; Holzhacker, C.; Stoger, B.; Puchberger, M.; Veiros, L.
F.; Kirchner, K. Organometallics 2013, 32, 4414.
(9) Alberico, E.; Sponholz, P.; Cordes, C.; Nielsen, M.; Drexler, H.-J.;
Baumann, W.; Junge, H.; Beller, M. Angew. Chem., Int. Ed. 2013, 52,
14162.
(10) Monfette, S.; Turner, Z. R.; Semproni, S. P.; Chirik, P. J. J. Am.
Chem. Soc. 2012, 134, 4561.
(11) Prokopchuk, D. E.; Morris, R. Organometallics 2012, 31, 7375.
(12) Mikhailine, A. A.; Maishan, M. I.; Lough, A. J.; Morris, R. H. J.
Am. Chem. Soc. 2012, 134, 12266.
(13) Bauer, G.; Kirchner, K. A. Angew. Chem., Int. Ed. 2011, 50, 5798.
(14) Langer, R.; Iron, M. A.; Lonstantinovski, L.; Diskin-Posner, Y.;
Leitus, G.; Ben-David, Y.; Milstein, D. Chem.Eur. J. 2012, 18, 7196.
(15) Casey, C. P.; Guan, H. J. Am. Chem. Soc. 2007, 129, 5816.
(16) Zou, W.; Lough, A. J.; Li, Y. F.; Morris, R. H. Science 2013, 342,
1080.
pyridine pincers with amino rather than methylene linkers see: Bichler,
B.; Holzhacker, C.; Stoger; Puchberger, M.; Veiros, L. F.; Kirchner, K.
Organometallics 2013, 32, 4114.
(47) Semproni, S. P.; Atienza, C. C. H.; Chirik, P. J. Chem. Sci. 2014,
̈
5, 1956.
(48) For related H2 oxidative addition chemistry with cobalt pincers
see: Rozenel, S. S.; Padilla, R.; Camp, C.; Arnold, J. Chem. Commun.
2014, 50, 2612.
(49) Obligacion, J. V.; Semproni, S. P.; Chirik, P. J. J. Am. Chem. Soc.
2014, 136, 4133.
(17) Clapham, S. E.; Hadzovic, A.; Morris, R. H. Coord. Chem. Rev.
2004, 248, 2201.
(18) Ikariya, T.; Blacker, A. J. Acc. Chem. Res. 2007, 40, 1300.
(19) Tanaka, R.; Yamashita, M.; Nozaki, K. J. Am. Chem. Soc. 2009,
131, 14168.
(50) Khaskin, E.; Diskin-Posner, Y.; Weiner, L.; Leitus, G.; Milstein,
D. Chem. Commun. 2013, 2771.
(51) Kaesz, H. D.; Saillant, R. B. Chem. Rev. 1972, 72, 233.
(20) Langer, R.; Diskin-Posner, Y.; Leitus, G.; Shimon, L. J. W.; Ben-
David, Y.; Milstein, D. Angew. Chem., Int. Ed. 2011, 50, 9948.
(21) Milstein, D. Top. Catal. 2010, 53, 915.
9223
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224

Journal of the American Chemical Society
Article
(52) Feller, M.; Ben-Ari, E.; Gupta, T.; Shimon, L. J. W.; Leitus, G.;
Diskin-Posner, E.; Weiner, L.; Milstein, D. Inorg. Chem. 2007, 46,
10479.
(53) Atienza, C. C. H.; Milsmann, C.; Semproni, S. P.; Turner, Z. R.;
Chirik, P. J. Inorg. Chem. 2013, 52, 5403.
(54) Ben-Ari, E.; Leitus, G.; Shimon, L. J. W.; Milstein, D. J. Am.
Chem. Soc. 2006, 128, 15390.
(55) Warren, J. J.; Tronic, T. A.; Mayer, J. M. Chem. Rev. 2010, 110,
6961.
(56) (a) Maki, A. H.; Edelstein, N.; Davison, A.; Holm, R. H. J. Am.
Chem. Soc. 1964, 86, 4580. (b) Nishida, Y.; Sumita, A.; Kida, S. Bull.
Chem. Soc. Jpn. 1977, 50, 759143. (c) Nishida, Y.; Kida, S. Bull. Chem.
Soc. Jpn. 1978, 51, 143. (d) Nishida, Y.; Hayashi, K.; Sumita, A.; Kida,
S. Bull. Chem. Soc. Jpn. 1980, 53, 271.
(57) Przyojski, J. A.; Arman, H. D.; Tonzetich, Z. J. Organometallics
2013, 32, 723.
(58) Lazzaro, D. P.; Fanwick, P. E.; McMillin, D. R. Inorg. Chem.
2012, 51, 10474.
(59) Vuzman, D.; Poverenov, E.; Shimon, L. J. W.; Diskin-Posner, Y.;
Milstein, D. Organometallics 2008, 27, 2627.
(60) (a) van der Vlugt, J. I.; Siegler, M. A.; Janssen, M.; Vogt, D.;
Spek, A. L. Organometallics 2009, 28, 7025. (b) Feller, M.; Ben-Ari, E.;
Iron, M. A.; Diskin-Posner, Y.; Leitus, G.; Shimon, L. J. W.;
Konstantinovski, L.; Milstein, D. Inorg. Chem. 2010, 49, 1615.
(61) (a) Vogt, M.; Rivada-Wheelaghan, O.; Iron, M. A.; Leitus, G.;
Diskin-Posner, Y.; Shimon, L. J. W.; Ben-David, Y.; Milstein, D.
Organometallics 2013, 32, 300. (b) van der Vlugt, J. I.; Lutz, M.; Pidko,
E. A.; Vogt, D.; Spek, A. L. Dalton Trans. 2009, 1016.
(62) van der Vlugt, J. I.; Pidko, E. A.; Bauer, R. C.; Gloaguen, Y.;
Rong, M. K.; Lutz, M. Chem.Eur. J. 2011, 17, 3850.
(63) Neese, F. Orca - an ab initio, DFT and Semiempirical Electronic
Structure Package, Version 2.8, Revision 2287; Institut fur Physikalische
̈
und Theoretische Chemie, Universitat Bonn: Bonn (Germany), 2010.
̈
(64) Gloaguen, Y.; Rebreyend, C.; Lutz, M.; Kumar, P.; Huber, M.;
Schneider, S.; de Bruin, B. Angew. Chem., Int. Ed. 2014, No. 10.1002/
anie.201403445.
(65) Milsmann, C.; Patra, G. K.; Bill, E.; Weyhermuller, T.; DeBeer
̈
George, S.; Wieghardt, K. Inorg. Chem. 2009, 48, 7430.
(66) Milsmann, C.; Bill, E.; Weyhermuller, T.; DeBeer George, S.;
̈
Wieghardt, K. Inorg. Chem. 2009, 48, 9754.
(67) Warren, J. J.; Tronic, T. A.; Mayer, J. M. Chem. Rev. 2010, 110,
6961.
(68) Only representative procedures are reported here. Full
experimental details, including General Considerations, are reported
in the Supporting Information.
9224
dx.doi.org/10.1021/ja504334a | J. Am. Chem. Soc. 2014, 136, 9211−9224