Radical mechanisms in the reaction of organic halides with diiminepyridine cobalt complexes

Article abstract of DOI:10.1021/om300182c

The formally Co(0) complex LCo(N₂) (L = 2,6-bis(2,6- dimethylphenyliminoethyl)pyridine) can be prepared via either Na/Hg reduction of LCoCl₂ or hydrogenolysis of LCoCH₂SiMe₃. In the latter reaction, LCoH could be trapped by reaction with N≡CC ₆H₄-4-Cl to give LCoN=CHC₆H₄-4-Cl. LCo(N₂) reacts with many alkyl and aryl halides RX, including aryl chlorides, to give a mixture of LCoR and LCoX in a halogen atom abstraction mechanism. Intermediacy of free alkyl and aryl radicals is confirmed by the ring-opening of cyclopropylmethyl to crotyl, and the rearrangement of 2,4,6-^tBu₃C₆H₂ to 3,5- ^tBu₂C₆H₃CMe₂CH ₂, before binding to Co. The organocobalt species generated in this way react further with activated halides R′X (alkyl iodides; allyl and benzyl halides) to give cross-coupling products RR′ in what is most likely again a halogen abstraction mechanism. DFT studies support the proposed radical pathways for both steps. MeI couples smoothly with LCoCH₂SiMe ₃ to give LCoI and CH₃CH₂SiMe₃, but the analogous reaction of ^tBuI leads in part to radical attack at the 3 and 4 positions of the pyridine ring to form (^tBu ₂-L)CoI and (^tBu₂-L)CoI₂.

Full text of DOI:10.1021/om300182c

Article
pubs.acs.org/Organometallics
Radical Mechanisms in the Reaction of Organic Halides with
Diiminepyridine Cobalt Complexes
Di Zhu,† Ilia Korobkov,‡ and Peter H. M. Budzelaar*,†
†Department of Chemistry, University of Manitoba, Winnipeg, MB R3T 2N2, Canada
‡Department of Chemistry, University of Ottawa, Ottawa, ON K1N 6N5, Canada
S
* Supporting Information
ABSTRACT: The formally Co(0) complex LCo(N2) (L =
2,6-bis(2,6-dimethylphenyliminoethyl)pyridine) can be pre-
pared via either Na/Hg reduction of LCoCl2 or hydrogenolysis
of LCoCH2SiMe3. In the latter reaction, LCoH could be
trapped by reaction with NCC6H4-4-Cl to give LCoN
CHC6H4-4-Cl. LCo(N2) reacts with many alkyl and aryl
halides RX, including aryl chlorides, to give a mixture of LCoR
and LCoX in a halogen atom abstraction mechanism. Inter-
mediacy of free alkyl and aryl radicals is confirmed by the ring-opening of cyclopropylmethyl to crotyl, and the rearrangement of
2,4,6-tBu3C6H2 to 3,5-tBu2C6H3CMe2CH2, before binding to Co. The organocobalt species generated in this way react further
with activated halides R′X (alkyl iodides; allyl and benzyl halides) to give cross-coupling products RR′ in what is most likely again
a halogen abstraction mechanism. DFT studies support the proposed radical pathways for both steps. MeI couples smoothly with
LCoCH2SiMe3 to give LCoI and CH3CH2SiMe3, but the analogous reaction of tBuI leads in part to radical attack at the 3 and 4
positions of the pyridine ring to form (tBu2-L)CoI and (tBu2-L)CoI2.
(a) via SN2-like nucleophilic attack by an electron-rich metal
center (mostly for alkyl halides)
INTRODUCTION
■
In recent years, redox-active ligands have been intensively
studied due to the unusual properties they confer on their metal
complexes. Popular classes of such ligands are α-diimines,1
α-imino-ketones,2 and diiminepyridines (DIP);3 the latter are
especially attractive because they bind strongly to transition
metals and provide considerable steric protection.
(b) via concerted addition involving a three-center transition
state (mostly for aryl halides)
One intriguing aspect of the chemistry of redox-active ligands
is the possibility of “ennobling” light transition metals,4 making
them behave more like their heavier congeners. For example,
DIP complexes of iron catalyze hydrogenation and hydro-
silylation,5 in a manner more typical for the platinum metals.
Whereas second- and third-row transition metals mostly exhibit
low-spin states and 2e redox reactions, first-row transition
metals tend to display high-spin states and 1e redox steps.
However, the possibility of having transition-metal-centered
unpaired electrons antiferromagnetically (AF) coupled to ligand-
centered electrons opens up the possibility of having 2e steps
where both ligand and metal are oxidized or reduced in a
“coupled” fashion. On the other hand, the flexible electronic
structure of complexes of redox-active ligands might also allow
for new reaction modes with significant potential in synthesis and
catalysis.
(c) via radical mechanisms (usually for activated alkyl
halides)
In each of these, one typically obtains a product that has the
halide and the carbon fragment attached to the same metal
atom (“mononuclear oxidative addition”) in an overall 2e
oxidation step (Scheme 1, A).
Cases where the halide and the organic fragment end up on
two different metal atoms in two separate 1e oxidation steps
(“binuclear oxidative addition”: Scheme 1, B) are much more
rare. Most of them involve alkyl halides,7−10 presumably
because the Csp3−X bond is weaker than the Csp2−X bond as
One of the most important redox reactions in organic syn-
thesis is the breaking of carbon−halogen bonds. This is a key
step in the catalyzed formation of C−C, C−N, and C−S bonds;
in addition, C−X cleavage is required in the disposal of CFCs
and similarly harmful environmental contaminants. The most
common mechanisms of C−X cleavage are6
Received: March 5, 2012
Published: May 15, 2012
© 2012 American Chemical Society
3958
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
(R = CH2SiMe3) in THF or benzene with a mixture of H2 and
N2.11 Formation of (1)Co(N2) from (1)CoCl2 and Na/Hg
likely involves straightforward reduction andat some stage
capture of a dinitrogen molecule. In contrast, the mechanism by
which (1)Co(N2) forms from (1)CoR and H2/N2 seems less
obvious. (3)CoR reacts quickly with H2 to give (3)CoH, which
is fairly stable at room temperature and is efficient at catalyz-
ing olefin hydrogenation,17 a reaction that likely involves
hydrogenolysis of (3)Co(alkyl) intermediates. Thus, one would
expect treatment of (1)CoR with H2 to initially produce
(1)CoH. This hydride appears to be much less stable than
(3)CoH,17b and we have never directly observed it in 1H NMR
spectra. However, a hydride intermediate could be trapped
(Scheme 3) by treating (1)CoR with H2 in the presence of
Scheme 1. Schematic Representation of (A) Mononuclear
and (B) Binuclear Oxidative Additions
well as for steric reasons. In fact, prior to our own recent
communication,11 the only example of binuclear oxidative
addition of an aryl halide was reported about 45 years ago,9a
when Halpern and co-workers observed the reaction of the
3−
activated halide 2-iodopyridine with Co(CN)5 to give a
3−
3−
mixture of ICo(CN)5 and 2-C5H4NCo(CN)5
.
Redox-active ligands can make electrons available during
metal-centered reactivity. This has been exploited, for example,
by Soper in his study of Co−C and C−C bond formation at
CoIII centers bearing amidophenolate ligands.12 The DIP ligand
has two low-lying π-acceptor orbitals and can accept up to three
electrons in its extended π-system.13 Thus, it should similarly
be able to function as an electron reservoir during reactions at
the metal center. Indeed, the group of Chirik observed
oxidative addition of C−X and even C−O bonds to (DIP)Fe(0)
complexes in what seems to be a traditional mononuclear
oxidative addition reaction, although in many products the
carbon-containing fragment had been “lost” from the metal.7b
We decided to explore in some detail the potential of the
related (DIP)Co(0) and (DIP)CoI complexes in oxidative
addition reactions. For (DIP)Co(0), we found that surprisingly
binuclear oxidative addition predominates.11 Interestingly,
the reaction works best for aryl chlorides, which are normally
more difficult to activate than the corresponding bromides or
iodides. We then reported that (DIP)CoI(aryl) complexes
undergo C−C coupling with benzyl halides.14 In the present
paper, we provide full details of this remarkable radical-mediated
chemistry, including an exploration of the scope for C−C
coupling reactions. Interestingly, the group of Chan very recently
reported binuclear oxidative addition to iridium(III) porphyrin
complexes,15 so it appears that this type of reaction is not
restricted to first-row transition metals.
Scheme 3. Trapping of (1)CoH with PhCCPh and
4-NCC6H4Cl
PhCCPh [to give (1)CoC(Ph)CHPh] or 4-NCC6H4Cl
[to give (1)CoNCHC6H4Cl].
Both species were easily identified by 1H NMR spectroscopy.
The former could not be obtained pure but always contained
either some unreacted (1)CoR or (1)CoCH(Ph)CH2Ph18 result-
ing from over-reduction. However, ketimide complex (1)CoN
CHC6H4Cl could be isolated and characterized by single-crystal
X-ray diffraction. The structure (Figure 1) shows the square-planar
Co coordination environment typical of DIP CoI complexes, and a
near-linear Co−NC arrangement suggesting considerable ionic
character in the Co−N bond.19 The N4−C41−C42 angle of
124.3(6)° leaves little doubt that this is a ketimide rather than a
nitrile complex; in addition, the HCN resonance is prominently
1
visible in the H NMR spectrum at 9.62 ppm (Figure S6).
The above results strongly suggest that the initial product
formed from (1)CoR and H2 is (1)CoH or possibly
(1)Co(H)(N2). However, the route via which this transforms
into (1)Co(N2) is not clear at present. Steric hindrance appears
to slow formation of the N2 complex, since conversion of
(2)CoH to (2)Co(N2) takes about 30 min, while (3)CoH is
stable for a day or more at room temperature. This may be
taken as an indication for bimolecular elimination of H2, e.g.,
via the path shown in Scheme 4. Steric hindrance would
obviously slow the intermolecular H transfer step. Preliminary
DFT calculations on strongly simplified models support easy
transfer of hydrogen from LCo(N2)(H) to LCoH, but accurate
calculations of the complete disproportionation path for the
real ligands 1−3 are currently not feasible.
RESULTS AND DISCUSSION
■
The choice of DIP ligand turned out to be critical in the pre-
sent work. Most reactions have been carried out with the 2,
6-Me2C6H3-substituted ligand 1 (Scheme 2). In several cases,
Scheme 2. DIP Ligands Used in This Work
Cleavage of C−X Bonds by LCo(N2) Complexes. Many
alkyl and aryl chlorides, bromides, and iodides react smoothly
with green (1)Co(N2) to give a purple mixture of (1)CoR and
(1)CoX. While the idealized stoichiometry for this reaction
would be eq 1, we typically obtained the products in a ratio
(1)CoR:(1)CoX < 1:1, as detailed below.11
the analogous but more sterically hindered 2,6-Et2C6H3 (2) and
2,6-iPr2C6H3 (3) ligands were also tested. Unless otherwise
stated, reactions mentioned in the text refer to ligand 1.
Generation of (1)Co(N2). The starting material (1)Co(N2)
for this study can be generated via reduction of (1)CoCl2 with
Na/Hg as described by Chirik.16 However, we found it more
convenient to generate the same species via reaction of (1)CoR
2LCo(N2) + ArX → LCoAr + LCoX + 2N2
(1)
3959
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
Table 1. Influence of Ligand on C−Cl Cleavage
a
of 4-ClC6H4CF3 by LCo(N2)
b
ligand
LCoAr/LCoCl
completion time
1
2
3
0.77
0.43
<10 s
30 s
c
0.22
hours
a
LCo(N2) (27 μmol) in around 0.4 mL of benzene-d6, 4-ClC6H4CF3
b
(3.3 μL, 27 μmol) added, reaction monitored by 1H NMR. Qualitative
indication. (3)Co(N2) prepared via Na/Hg reduction of (3)CoCl2.16
c
ligand 3, apart from (3)CoAr and (3)CoCl, a small but significant
quantity of (3)CoH was also detected; the source of the hydride
has so far remained elusive.21
The “oxidative addition” is slower in THF-d8 than in C6D6,
probably because of the stronger donor properties of
the former solvent. However, neither the choice of solvent nor
the concentrations of cobalt complex and aryl chloride affected the
final LCoAr:LCoCl product ratio. Since ligand 1 produced the
highest yield of LCoAr product, we concentrated on this ligand
for further exploration of the scope of the reaction.
Table 2 summarizes the results obtained with a range of alkyl
and aryl halides. The reaction product is a mixture of (1)CoR and
(1)CoX, which have similar solubilities and could in most cases
not be completely separated, making it impossible to obtain
satisfactory elemental analyses for the new complexes (1)CoR.
The observation in most cases of a product ratio (1)CoR:(1)CoX
substantially smaller than 1 indicates some organic groups must be
“lost” along the way, but we could not ascertain the fate of these
groups. For 4-ClC6H4CF3, we observed about 8% of C6H5CF3
relative to (1)CoC6H4-4-CF3 by 19F NMR and could detect some
(C6H4CF3)2 by GC/MS. The former could be due to hydrogen
abstraction from solvent or ligand; the latter arises most likely
from aryl radical dimerization. Yields in Table 2 have been
calculated assuming quantitative formation of (1)Co(N2) from its
(1)CoCH2SiMe3 precursor, the idealized stoichiometry of eq 1,22
and the absence of other reactions consuming ArX; the latter
assumption is probably not valid for several alkyl halides.
From the data in the table, a few trends are clear: rates increase
in the order Cl < Br < I; activated alkyls (allyl, benzyl) are more
reactive; aryl halides bearing electron-withdrawing substituents react
faster; and the yield of (1)CoAr products is largest for chlorides.
In most cases, the organocobalt product formed corresponds
to the organic halide used, but there are a few exceptions:
Figure 1. X-ray structure of (1)CoNCHC6H4Cl (thermal ellipsoids
drawn at 30% probability, hydrogen atoms omitted for clarity). Selected bond
distances (Å) and angles (deg): Co(1)−N(1): 1.804(4); Co(1)−N(2):
1.896(4); Co(1)−N(3): 1.899(4); Co(1)−N(4): 1.726(4); C(12)−N(2):
1.336(6); C(12)−C(13): 1.423(7); C(18)−N(3): 1.330(6); C(17)−C(18):
1.453(7); N(1)− Co(1)−N(2): 81.46(17); N(4)−Co(1)−N(2): 98.79(19);
N(4)−Co(1)−N(1): 178.1 (2); C(41)−N(4)−Co(1): 169.9(5); N(4)−
C(41)−C(42): 124.3(6); N(2)−Co(1)−N(3): 162.9(17); C(18)−N(3)−
Co(1): 116.8(3); C(12)−N(2)−Co(1): 116.2 (3).
Scheme 4. Possible Path for Reaction of (1)CoR with H2/N2
to Produce (1)Co(N2)
1
These reactions can easily be followed by H NMR because
CoI complexes LCoZ (Z = H, alkyl, aryl, or halide) have a
characteristic pyridine H4 resonance in the range 8.0−11.0
ppm, which is rather sensitive to the nature of the group Z.17b,20
The separation of these characteristic triplets is nearly always
good enough to allow reliable, quantitative determination of the
relative amounts of LCoR and LCoX products. However, the
resonances of several unreacted halides are obscured by other
signals, since the spectra are rather crowded in both the aliphatic
and aromatic regions, so the complete stoichiometry could not be
established in all cases. We selected 4-ClC6H4CF3 (which can also
be observed by 19F NMR) for a more complete evaluation of the
influence of reaction conditions and ligand variation. Ligand
effects are summarized in Table 1. The reaction of (1)Co(N2)
with this aryl chloride is fast (complete within 10 s). Byproducts
detected in this reaction are ArH (about 10% relative to (1)CoCl)
and Ar2 (about 6%); for reactions carried out in benzene-d6, no
ArD or ArC6D5 could be detected by GC/MS. The reaction of
(2)Co(N2) is somewhat slower (∼30 s), while the reaction of
(3)Co(N2) is much slower (hours). Interestingly, also the yield of
LCoAr (relative to LCoCl) decreases in the order 1 > 2 > 3. For
t
t
iBuBr, BuCl, and BuI (entries 16, 19, and 18) all give
(1)CoiBu, although in different yields. For the latter two sub-
strates, this is most likely due to “chain walking” of initially
17
i
formed (1)CotBu. Similarly, PrCl gave (1)ConPr. We pre-
viously reported that (3)CoH adds to internal olefins to give
terminal alkyls, also via chain walking.17a
In contrast to tBuI, AdI (entry 20) did not produce any identifiable
organocobalt complex; the only recognizable product was (1)CoI.
3960
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
a b
,
Table 2. Reaction of (1)Co(N2) with Organic Halides
(1)CoR/(1)
c
d
entry
halide RX
CoX
yield (1)CoR, % rxn time
m
1
2
3
4
5
6
7
8
9
PhBr
PhCl
0.25
n.d.
n.d.
29
1 min
hours
m
0.59
e
2,4,6-tBu3C6H2Br
2,6-Me2C6H3Cl
p-MeCOC6H4Cl
p-CF3C6H4Cl
p-ClC6H4Cl
0.27
seconds
days
m
∼0
0.91
0.77
0.59
0.59
1.0
n.d.
71
30 min
seconds
seconds
hours
89
Unactivated C−F bonds do not react, but the allylic fluoride
C6F13CHCH2 produced (1)Co(CH2CHC6F12) in fair yield.
On the basis of 1H NMR parameters we believe this most likely
contains a σ-bound allyl moiety (σ-CH2CHC(F)C5F11): the
complex exhibits the low-field pyridine H4 resonance character-
istic of square-planar LCoZ complexes, whereas the η3-allyl
and η3-crotyl complexes mentioned above do not. Pure
(1)Co(η3-allyl) was prepared independently from (1)CoCl2
and allylmagnesium chloride, and the structure was determined
by single-crystal X-ray diffraction. Its molecular structure
(Figure 2) is best described as a distorted square pyramid, in
which the allyl group occupies an equatorial (C41) and an
apical (C43) position. The C41−C42−C43 angle of 125.0(5)°
is somewhat larger than usual for π-allyl complexes, but very
similar to that reported for (3)Fe(η3-allyl).7b In view of the
dynamic behavior of the complex in solution (vide infra), we
cannot rule out the presence of minor disorder in the allyl
group, and therefore the bond lengths and angles in the allyl
fragment should be treated with caution.
Solution NMR data for this π-allyl complex clearly reveal its
fluxional character. At low temperature (−60 °C), the 1H NMR
spectrum shows different “left” and “right” sides (pyridine H3
and H5 are inequivalent) but equivalent “top” and “bottom”
sides (only two Me signals for the 2,6-Me2C6H3 groups).
Also, the allyl group shows three resonances (center, syn, and
anti protons) in the ratio 1:2:2. If we assume the most stable
structure in solution corresponds to the solid-state one, this
means that even at low temperature there is fast exchange
between the two types of square pyramids (Scheme 5, reaction A).
On increasing the temperature, left−right exchange becomes
visible and the pyridine H3 and H5 signals broaden and
coalesce, as do the two xylyl Me peaks. At somewhat higher
temperature, also the allyl syn and anti signals begin to broaden
and coalesce. However, fitting of the exchange-broadened
spectra clearly shows that these two exchanges are distinct
processes, with rates that differ by a factor of 10−100 over the
range of temperatures where both can be fitted satisfactorily, the
syn/anti exchange always being slowest. This suggests that the
left−right exchange is an in-place rotation of the allyl group
maintaining its π-coordination (Scheme 5, reaction B), while the
syn/anti exchange involves interconversion of σ- and π-bound allyl
groups (Scheme 5, reaction C). The activation parameters
obtained from the fitted exchange rates are shown in Scheme 4
(for full details see the SI); not surprisingly, the resulting entropy
of activation for σ/π exchange is somewhat larger than for in-place
rotation, although the difference may not be statistically significant.
Mechanism of C−X Cleavage. As already mentioned
in our preliminary communication, we believe that the actual
C−X cleavage reaction follows a radical path for all or most
substrates tested. On the experimental side, this is supported by
the observation of radical-rearrangement products from
2,4,6-tBu3C6H2Br and cyclopropylmethyl chloride. DFT studies
also support such a mechanism. Despite extensive searches, we
could not locate any transition states for conventional “side-on”
48
p-MeC6H4Cl
38
2,6-Cl2C5H3N
88
seconds
10 n-C8H17F
n.r.
11 (n-C6F13)CHCH2
12 CH2CHCH2Cl
13 (cy-C3H5)CH2Cl
14 Br(CH2)4CHCH2
15 MeI
16 iBuBr
17 iPrCl
0.50
1.0
25
minutes
seconds
seconds
seconds
seconds
seconds
minutes
seconds
minutes
seconds
88
f
0.54
48
g
0.40
35
h
m
0.71
n.d.
44
i
0.48
j
m
m
0.20
0.17 ,
n.d.
n.d.
8
ik
18 tBuI
19 tBuCl
i
0.08
l
20 AdI
∼0
∼0
a
Reaction conditions: (1)Co(N2) generated from (1)CoCH2SiMe3 (27
μmol) and H2 (2.0 mL) in 0.4 mL of C6D6; combined with ArX (27
b
μmol) or RX (14 μmol). Entries 1−11 and 15 were already reported
in our communication.11 From 1H NMR, pyridine H4 peaks;
c
d
e
estimated error margin ≈ 5%. Qualitative indication. Organocobalt
f
product: R = CH2CMe2-3,5-tBu2C6H3 Organocobalt product: R = π-
g
CH2CHCHMe. Organocobalt product: R = primary alkyl; see SI for
h
discussion.26 Py H4 resonances of (1)CoMe and (1)CoI over-
i
lap; ratio determined from NCCH3 peaks. Organocobalt product:
j
R = CH2CHMe2. Organocobalt product: R = CH2CH2CH3
(a second diamagnetic cobalt(I) complex was also detected but
k
could not be identified with certainty). In addition, ligand attack
l
products were observed: (1-tBu2)CoI:(1)CoI ≈ 0.16. An unidentified
m
diamagnetic cobalt(I) complex was also detected. Conversion of
RX (and hence yield of (1)CoR) could not be determined from
1H NMR.
2,4,6-tBu3C6H2Br (entry 3) formed (1)CoCH2CMe2
(3,5-tBu2)C6H3.23 The nature of this product was verified by
comparison with (1)CoCH2CMe2C6H5, independently pre-
pared from (1)CoCl2 and C6H5CMe2CH2MgCl. Formation of
the alkyl likely involves rearrangement of the intermediate aryl
free radical.24
Cyclopropylmethyl chloride (entry 13) gave the η3-crotyl
complex as the only organocobalt complex. Its structure could
be assigned via comparison with the independently prepared
η3-allyl complex (entry 12; vide infra). Opening of an inter-
mediate cyclopropylmethyl radical25 appears to be a reasonable
explanation.
3961
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
Figure 2. X-ray structure of (1)Co(η3-allyl) (thermal ellipsoids drawn at 30% probability, hydrogen atoms omitted for clarity). Selected bond
distances (Å) and angles (deg): Co(1)−N(1): 1.826(3); Co(1)−N(2): 1.992(3); Co(1)−N(3): 1.925(3); Co(1)−C(41): 2.070(4); Co(1)−C(42):
2.016; Co(1)−C(43): 2.131(5); C(4)−C(9): 1.414(5); N(3)−C(9): 1.338(4); N(1)−C(4): 1.378(4); N(1)−C(8): 1.368(4); C(8)−C(10):
1.425(5); N(2)−C(10): 1.320(4); C(42)−C(43): 1.346(7); C(41)−C(42): 1.381(7); N(1)−Co(1)−N(3): 80.57(12); N(1)−Co(1)−N(2):
79.14(12); N(3)−Co(1)−N(2): 152.96(12); N(1)−Co(1)−C(43): 122.20(18); N(1)−Co(1)−C(42): 153.5(2); N(1)−Co(1)−C(41):
166.76(19); C(41)−Co(1)−C(43): 70.3(2); C(43)−C(42)−C(41): 125.0(5).
a
Scheme 5. Fluxional Behavior of (1)Co(η3-C3H5)
a
Activation parameters (1σ error limits) obtained from Arrhenius plots to fitted exchange rates.
attack of Co on the C−X bond of aryl halides. All searches
eventually converged to saddle points where only the halide is
close to the cobalt atom, and the imaginary mode is
predominantly a C−X stretch (Scheme 6).
3962
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
a
both reducing electrons come from the ligand, and the elec-
tronic structure is completely changed, with the product contain-
ing high-spin FeII and an “innocent” DIP ligand.29 In (1)Co(N2),
the metal is low-spin (diamagnetic) CoI, with only a small
amount of spin polarization at the metal (Figure 3, left); the
single unpaired electron resides in a ligand π*-orbital.28
Scheme 6. Schematic Representation of Br Abstraction TS
The halide abstraction product (1)CoX has low-spin CoII AF
coupled to a ligand-centered unpaired electron20 (Figure 3,
right). Therefore, to proceed from reactant to product one
could imagine simple transfer of a single Co d electron to the
aryl halide; its remaining partner could then AF couple to the
ligand-centered electron to smoothly give the product.
However, this is not what we find. At the transition state for
C−X bond cleavage, the metal has undergone a spin-flip to
high-spin CoI (two β electrons); an amount of spin density
corresponding to one α electron is spread out over both the
DIP ligand and the aryl halide, apparently in the process of
being transferred from one to the other (Figure 3, middle). The
change in metal spin state at this point is also evident from the
increase in Co−N bond lengths, by about 0.1 Å for Co−Npy
and nearly 0.2 Å for Co−Nim. Finally, the reduction in DIP
ligand π* population results in an increase of the Cpy−Cim bond
lengths and a shortening of the CN bonds. After the electron
has been transferred and the phenyl radical ejected, at some
point one of the two metal-centered unpaired electrons has to
flip and transfer to the ligand to produce the ground state of the
product LCoX. These movements and flips of electrons are
summarized in Scheme 8.
a
Bond lengths in Å, angle in deg; the double arrow represents the
imaginary mode.
Consistent with the observed relative reactivity of the
various halides, the free energy barrier calculated for PhBr
(19.0 kcal/mol) is substantially lower than that for PhCl
(23.2 kcal/mol),11 and the magnitude of both barriers is con-
sistent with reactions happening around room temperature
(1 min for PhBr, hours for PhCl). One possible mechanistic
picture for the complete reaction is shown in Scheme 7. The
Scheme 7. Proposed Reaction Mechanism for Reaction of
LCo(N2) with Aryl Halides
aryl halide first displaces N2 from the metal (the retarding effect
of THF, mentioned earlier, can be ascribed to competition
between THF and ArX for the empty site at Co). Then, C−X
cleavage results in expulsion of an aryl radical, which will indepen-
dently make its way to a second Co center. Such radicals are
highly reactive, and the low yield for some substrates may be due
to the ease with which these radicals undergo side reactions. Also,
the lower yields for bromides and iodides (compared to chlorides)
might be due to the larger concentration of radicals formed, lead-
ing to a higher probability of radical combination.27
Scheme 8. Movements and Flips of Electrons during
Reaction of (1)Co(N2) with PhBr
It is interesting to contrast this reaction mechanism with the
one for addition of vinyl acetate to (3)Fe(N2)2, where Chirik
explained formation of the “traditional” product (3)Fe(C2H3)-
(OAc) via standard mononuclear oxidative addition.7b In
(3)Fe(N2)2, the metal has the actual oxidation state +II and
is intermediate-spin;28 its two unpaired electrons each AF
couple to a ligand-centered electron. In the oxidative addition,
Thus, the ArX halide abstraction process appears to be
more complicated than one might at first imagine, involving a
Figure 3. Spin density plots for (1)Co(N2), the Ph−Br cleavage transition state, and the final products (1)CoBr and Ph·. The sign of the spin
population on Co (nCo) is given relative to that of density on the ligand. Bond lengths in Å.
3963
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
Figure 4. Free-energy profile (ΔG, kcal/mol, b3-lyp/TZVPP//b3-lyp/TZVP) for the reaction between (1)CoPh and benzyl chloride (BzCl).
temporary spin flip even though there is no obvious need for
one; this is unlike, for example, the halide abstractions descri-
bed by Soper,2a where the ligand functions as an electron
reservoir and the metal (CoIII) appears to be truly innocent. It
is not clear at present how essential these details are to the
different reaction outcomes.
C−C Coupling Reactions. Since “standard” oxidative
addition is so important in metal-catalyzed C−C coupling reac-
tions, we decided to investigate whether the alkyl and aryl com-
plexes studied here can also be used to effect C−C coupling
reactions. Complexes (1)CoAr and (1)CoR were found to be
much less reactive than (1)Co(N2) toward organic halides;
only activated halides (allyl, benzyl, or iodides) reacted at all
(Table 3).14 Complexes (1)CoAr bearing electron-withdrawing
substituents at the Ar group are further deactivated, to the
extent that, for example, complex 4 (formed from 2,6-dichloro-
pyridine) is slower to react with benzyl chloride (BzCl) than
even (1)CoCl.
Table 3. Reaction of Mixtures (1)CoR/(1)CoCl with
a b
,
Activated Alkyl Halides R′X
c
products detected (%)
entry
R
R′X
RR′
RR
R′R′
h
h
h
1
p-CF3C6H4
C6H5
CH2CHCH2Cl
CH2CHCH2Cl
CH2CHCH2Cl
C6H5CH2Cl
C6H5CH2Br
C6H5CH2Br
C6H5CH2Cl
C6H5CH2Cl
MeI
77
23
n.o.
n.o.
n.o.
50
77
6
2
84
16
3
p-MeOC6H4
p-ClC6H4
p-CF3C6H4
C6H5
94
6
4
50
n.o.
n.o.
39
5
23
6
55
7
C6H5
77
22
1
8
π-crotyl
100
100
93
n.o.
n.o.
<1
n.o.
n.o.
7
e
9
CH2SiMe3
g
10
11
12
13
14
15
16
p-MeC6H4
MeI
C6H5
C6H5I
n.r.
n.r.
n.r.
n.o.
n.o.
trace
f
C6H5
n-C6H13Br
nBuCl
p-CF3C6H4
p-MeOCOC6H4
6-Cl-2-C5H3N
6-Cl-2-C5H3N
C6H5CH2Cl
C6H5CH2Cl
n.o.
n.o.
n.o.
100
100
100
g
C6H5CH2Br
a
A mixture of (1)CoAr + (1)CoCl generated from (1)Co(N2) as
described in Table 2, then addition of 0.5 equiv of RX relative
b
to original ArX used. Entries 1−7, 9, 15, and 16 were already
In most cases where reaction did occur, the major product
was the heterocoupling product, but significant amounts of one
or both possible homocoupling products were observed as well,
indicating that also this coupling reaction most likely involves
free radicals. DFT calculations support this idea: we were able
to locate a transition state for abstraction of a chlorine atom
from benzyl chloride by (1)CoPh, with a calculated activation
energy of 27.1 kcal/mol (for the reaction profile, see Figure 4).
The complex (1)Co(Ph)(Cl) formed this way should easily
lose a phenyl radical, which can combine with the earlier
generated benzyl radical. Some product may also be formed
by a benzyl radical attacking intact (1)CoPh, to first form
(1)Co(Bz)(Ph) and then eliminate PhBz; the naked (1)Co left
this way could reduce the next benzyl chloride molecule. This
C−C coupling reaction is complicated by the fact that halides
(1)CoX, always formed together with (1)CoR, also react slowly
with allyl and benzyl halides.
reported in our mini-review.14 By GC/MS; not calibrated.
Calibrated against authentic samples of PhPh, PhBz, and BzBz.
Using separately prepared (1)CoCH2SiMe3 and 1.8 equiv of CH3I;
product identified by NMR and GC/MS. Using separately prepared
(1)CoPh. Using 2.0 equiv of BzBr relative to (1)Co(N2). Any R′R′
(1,5-hexadiene) formed would have been missed due to its low boiling
point.
c
d
e
f
g
h
find that MeI couples smoothly with (1)CoCH2SiMe3 to give
EtSiMe3. This led us to check whether our system also allows
coupling involving a tertiary alkyl group, which is normally one
of the more difficult couplings to achieve. Unfortunately,
t
reaction of (1)CoCH2SiMe3 with BuI did not produce any
tBuCH2SiMe3. Instead, we obtained a mixture containing one
1
major diamagnetic cobalt complex (with a complicated H
NMR spectrum) and one or more30 paramagnetic complexes. A
few crystals of the diamagnetic component were obtained, and
t
Ligand Attack with BuI. Metal-catalyzed C−C coupling
t
involving two sp3 carbons tends to be more difficult than
refinement indicated the presence of Bu substituents at posi-
tions 3 and 4 of the pyridine ring. However, the crystal exhibits
coupling involving at least one sp2 or sp carbon. However, we
3964
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
serious disorder or twinning problems, and the structure deter-
mination results cannot be taken as definitive evidence even for
the connectivity of the complex (for details, see the SI). Because
of these issues, we also studied the reaction of (2)CoCH2SiMe3
t
with BuI. This produces diamagnetic and paramagnetic com-
1
plexes with H MR spectra very similar to those obtained from
(1)CoCH2SiMe3. Again, complete separation proved impossible,
but several crystallization attempts eventually produced some
acceptable crystals of both components. The structures of
(tBu2-2)CoI2 (the paramagnetic component) and (tBu2-2)CoI
(the diamagnetic component, contaminated with about 6% of
cocrystallized (tBu2-2)CoCl) are shown in Figure 5 and Figure 6,
Figure 6. X-ray structure of (tBu2-2)CoI (40% thermal ellipsoids). In
(B), most carbon atoms of the ligand N-aryl groups have been omitted
for clarity. Selected bond distances (Å) and angles (deg): Co(1)−I(1):
2.5170(12); Co(1)−N(1): 1.955(3); Co(1)−N(2): 1.810(3); Co(1)−
N(3): 1.918(3); C(1)−C(2): 1.422(5); N(3)−C(1): 1.328(4); C(6)−
C(7): 1.460(5); N(1)−C(7): 1.296(4); N(2)−C(2): 1.330(4); C(2)−
C(3): 1.510(5); C(3)−C(4): 1.542(6); C(4)−C(5): 1.509(6); C(5)−
C(6): 1.337(5); N(2)−C(6): 1.396(5); N(1)−Co(1)−N(2): 82.30(12);
N(2)−Co(1)−N(3): 80.45(12); N(1)−Co(1)−N(3): 162.67(12);
N(2)−Co(1)−I(1): 175.94(10); C(4)−C(5)−C(6): 121.0(3); N(2)−
C(2)−C(3): 120.2(3); C(2)−C(3)−C(4): 110.2(3); C(3)−C(4)−
C(5): 111.4(3); C(61)−C(3)−C(4)−C(51): 144.9(4).
Figure 5. X-ray structure of (tBu2-2)CoI2 (40% thermal ellipsoids). In
(B), most carbon atoms of the ligand N-aryl groups have been omitted
for clarity. Selected bond distances (Å) and angles (deg): Co(1)−
N(1): 2.221(3); Co(1)−N(2): 2.034(3); Co(1)−N(3): 2.213(3);
C(2)−C(3): 1.501(5); N(1)−C(2): 1.281(5); C(7)−C(8): 1.472(6);
N(3)−C(8): 1.280(5); N(2)−C(3): 1.279(5); C(3)−C(4): 1.510(5);
C(4)−C(5): 1.542(5); C(5)−C(6): 1.512(6); C(6)−C(7): 1.334(6);
N(2)−C(7): 1.400(5); N(2)−Co(1)−N(1): 73.93(11); N(2)−
Co(1)−N(3): 75.78(11); N(3)−Co(1)−N(1): 147.79(12); N(2)−
Co(1)−I(1): 143.28(8); N(2)−Co(1)−I(2): 104.67(8); C(7)−
C(6)−C(5): 122.2(3); N(2)−C(3)−C(4): 122.3(3); C(3)−C(4)−
C(5): 108.8(3); C(6)−C(5)−C(4): 110.7(3); C(10)−C(5)−C(4)−
C(14): −146.3(4).
The most reasonable mechanism for formation of the modi-
t
t
fied ligands Bu2-L is sequential attack of Bu radicals (formed
via halide abstraction) on intact LCoR and/or LCoI complexes,
possibly accompanied by iodide redistribution between CoI and
CoII complexes. Alkylation of DIP ligands is quite common,31
but multiple alkylation is rare.32 It should be noted here that
t
the first Bu group acts as an oxidant, changing the formally
t
neutral ligand into a monoanionic one; the second Bu group
t
then acts as a reductant, so that the final ligands Bu2-L are
formally neutral again. It is not clear at this point whether the
initial attack occurs at carbon 3 or 4 of the pyridine ring, but since
attack at C3 is extremely rare33 initial C4 attack seems more likely.
Curiously, the corresponding reaction with 1-iodoadamantane did
not result in similar ring-alkylated products.
respectively. A comparison of the structures shows that the
ligand in (tBu2-2)CoI2 is “innocent” (imine bond lengths close
to 1.28 Å), whereas the ligand in (tBu2-2)CoI appears to have
accepted an electron (imine bond lengths 1.328(4) and 1.296(4) Å)
in a manner very similar to Co(I) complexes of the original DIP
ligand. Thus, the interruption of conjugation in the DIP skeleton
CONCLUSIONS
■
This work demonstrates the first well-defined binuclear oxida-
tive addition reactions applicable to a variety of alkyl and aryl
halides, including aryl chlorides. The required highly reactive
t
caused by the two Bu substituents does not appear to affect the
electron-accepting properties of the ligand much.
3965
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
room temperature unless otherwise noted. GC/MS instrument: Varian
3800 gas chromatograph with a 30 m VF-5 ms column coupled to a
Varian 320-MS operated in single quadrupole mode. KBr pellets for IR
spectra were prepared in a N2-filled drybox and measured in a Bruker
Tensor27 IR instrument prepurged with nitrogen. The IR data were
processed using OPUS6.5 software. Elemental analysis was done by
Guelph Chemical Laboratories Ltd.
starting material (1)Co(N2) can be obtained via either hydro-
genolysis of (1)CoCH2SiMe3 or reduction of (1)CoCl2 in the
presence of N2. The steric properties of the ligand are impor-
tant here, ligand 1 giving much better results than the more
hindered variations 2 and 3. We obtained strong indications
from both experiment and theorythat these reactions pro-
ceed via free radicals. Also, the resulting Co alkyls and aryls
could be used in further C−C coupling reactions with activated
alkyl halides, though not yet in a catalytic fashion. These results
are significant not only in themselves, but also because they
provide well-defined model systems for (presumably radical-
based) C−C coupling reactions involving much less well-defined
low-valent metal species, such as CoBr2(2,2′-bipyridine)/
Mn/pyridine, Co(acac)3/PPh3/Grignard reagent.34 We hope that
eventually our binuclear oxidative addition can be made cleaner and
more efficient by generating (1)Co(N2) in situ from (1)CoX, thus
allowing complete conversion to (1)CoAr; the trick here will be to
find a reductant that does not react with (1)CoAr (Scheme 9).
(2)CoCH2SiMe3. Complex (2)CoCl2 (0.44 g, 0.79 mmol) was
suspended in around 12 mL of dry toluene. In a N2-filled glovebox,
solid LiCH2SiMe3 (0.15 g, 1.59 mmol) was dissolved in 6 mL of
toluene, and the resulting clear solution was slowly dropped into the
above (2)CoCl2 suspension over 3 h. The resulting mixture turned
purple and was stirred overnight. After filtering over Celite, the purple
filtrate was evaporated to dryness. The resulting residue was dissolved
in toluene and layered with pentane at −35 °C over two days. Some
dark solid (0.044 g) precipitated, which was determined to be
(2)CoCl2. The mother liquor was evaporated to dryness, and the
residue was crystallized from Et2O/pentane at −35 °C, giving 0.15 g
(33%) of a dark purple solid.
1H NMR (benzene-d6, 300 MHz): δ 10.05 (1H, t, J 7.6, Py H4),
7.72 (2H, d, J 7.6, Py H3), 7.52 (2H, t, J 7.6, NAr p), 7.39 (4H, d, J 7.6,
NAr m), 2.49−2.69 (8H, m, CH2CH3), 1.09 (12H, t, J 7.6, CH2CH3),
0.72 (2H, s, CoCH2), −0.66 (9H, s, SiMe3), −1.09 (6H, s, NCMe).
13C NMR (benzene-d6, 75 MHz): δ 166.0, 157.0, 155.4, 135.6, 126.4
Scheme 9. Possible Route for Complete Conversion of
(1)CoX to (1)CoAr
(NAr p), 126.2 (NAr m), 123.4 (Py C3), 116.3 (Py C4), 24.7
(CH2CH3), 24.0 (NCMe), 13.3 (CH2CH3), 3.5 (SiMe3). The
CoCH2 resonance was not observed, probably due to broadening by
the quadrupolar Co. Anal. Calcd for C33H46CoN3Si (571.76): C,
69.32; H, 8.11; N, 7.35. Found: C, 69.59; H, 7.85; N, 7.08.
Hydrogenolysis of (2)CoCH2SiMe3 and Subsequent Reaction
with CF3C6H4-4-Cl. In a N2-filled drybox, (2)CoCH2SiMe3 (15 mg,
26 μmol) was dissolved in 0.4 mL of benzene-d6. Outside the drybox,
2 mL of H2 was injected into it. The sample first turned blue-purple
[5 min after the injection; at this stage, the 1H NMR spectrum showed
a mixture of (2)CoH and (2)Co(N2)], then gray, and finally green
(30 min after the injection of H2). The 1H NMR spectrum after
30 min showed mainly (2)Co(N2), but a trace amount of (2)CoH was
Making the C−C coupling described in the present work
fully catalytic, by using a mixture of a reductant, halides RX and
R′X, and a catalytic amount of (1)CoX, is likely to be difficult,
and one would normally expect preferential homocoupling of
the most reactive halide (this in addition to the reduction issue
mentioned above). However, the price of Co (relative to, for
example, Pd) is such that even a batch process involving a sepa-
rate recycling step might be acceptable, provided interesting
coupling products can be identified.
1
still observed. After one night, the H NMR showed there was no
(2)CoH left. The NMR sample was transferred into the drybox and
flushed with nitrogen to remove the excess of hydrogen. Into this
sample was injected 3.3 μL of 4-CF3C6H4Cl (24 μmol), and the tube
was immediately shaken well to obtain good mixing. The sample
1
turned purple in about 30 s. The H NMR showed that there was no
(2)Co(N2) left, but unreacted 4-CF3C6H4Cl could still be detected;
from the ratio of the pyridine H4 resonances, (2)CoC6H4-4-CF3:
(2)CoCl = 0.43:1.
EXPERIMENTAL SECTION
■
General Considerations. All experiments were done under an argon
atmosphere using standard Schlenk techniques or in a nitrogen-filled
drybox. Pentane, hexane, toluene, diethyl ether, tetrahydrofuran, THF-
d8, benzene, and benzene-d6 were distilled from sodium/benzophe-
none. Phenyl lithium (1.8 M in di-n-butyl ether) was purchased from
Aldrich and used as received. Ligands 1−3 were prepared according to
published procedures.35,36 (1)CoCH2SiMe3 was prepared according to
the literature procedure using (TMEDA)Co(CH2SiMe3)2.37 (3)-
CoCH2SiMe3 was prepared according to the reported procedure.38
Anhydrous chlorobenzene and anhydrous benzotrifluoride were
purchased form Aldrich and used as received. Other aryl halides and
alkyl halides used for oxidative additions were purchased from Aldrich
or Acros, degassed, and dried over 4 Å molecular sieves in a drybox
before use. Alkyl halides used in C−C coupling reactions were used as
received.
Tentative, partial assignments for (2)CoC6H4-4-CF3: 1H NMR
(benzene-d6, 300 MHz): δ 10.21 (1H, br, Py H4), 4.77 (2H, br, CoAr o),
−0.78 (6H, s, NCMe). 19F NMR (25 °C, benzene-d6, 282 MHz):
δ −61.2.
Hydrogenolysis of (3)CoCH2SiMe3 in the Presence of
Dinitrogen. In a N2-filled drybox, (3)CoCH2SiMe3 (17.1 mg, 27
μmol) was weighed into a small vial and dissolved in 0.4 mL of
benzene-d6. The resulting purple solution was transferred into an
NMR tube. Outside the drybox, 2.0 mL of H2 was injected into it.
After 5 min, all (3)CoCH2SiMe3 had been converted into (3)CoH; no
1
(3)Co(N2) could be detected by H NMR. After 1 h, some solid
1
precipitated at the bottom of the NMR tube and H NMR showed
only (3)CoH. After 24 h, the sample had turned violet and more solid
1
had precipitated. The H NMR spectrum at this stage showed the
1H NMR, 13C{H} NMR, and 19F NMR spectra were recorded on a
Bruker Avance 300 MHz or Bruker Avance 500 MHz spectrometers.
presence of a small amount of (3)Co(N2); the main complex in
solution was still (3)CoH. Longer standing resulted in the
precipitation of more black solids.
1
All NMR shifts (δ, ppm) of H NMR and 13C NMR spectra were
referenced to the solvent (benzene-d6, H NMR: C6D5H δ 7.16; 13C
1
Reaction of (3)Co(N2) with 4-CF3C6H4Cl (ref 14). In a N2-filled
drybox, (3)Co(N2)16 (11.8 mg, 20 μmol) was weighed and dissolved
in about 0.4 mL of dry benzene-d6; 4-CF3C6H4Cl (2.45 μL, 19.6
μmol) was then added. The mixture turned gray-blue. The
1
NMR: C6D6 δ 128.0; CDCl3, H NMR: CHCl3 δ 7.26; 13C NMR:
CHCl3 δ 77.0; THF-d8; 1H NMR: OCHD δ 3.62; 13C NMR: OCD2 δ
68.03). Coupling constants J are given in Hz. Where necessary, COSY
or HSQC NMR spectra were also acquired to assist 13C NMR and/or
1H assignments. 1H peaks detected only via COSY (hence with
unknown splitting patterns) are indicated by *. Data were collected at
1
immediately recorded H NMR spectrum showed that the reaction
was not complete [still (3)Co(N2) visible], and three diamagnetic
cobalt(I) complexes could be clearly observed: (3)CoH:(3)CoAr:
3966
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
1
(3)CoCl = 0.11:0.14:1.00. After 4 h, the H NMR spectrum showed
that there was no (3)Co(N2) left, and the product ratio was now
(3)CoH:(3)CoAr:(3)CoCl = 0.045:0.14:1. Assignments for (3)CoAr
are based on analogy with previously reported (1)CoAr.
the mother liquor and washing with dry hexane. One fragment of this
crystalline material was used for determination of the crystal structure
by single-crystal X-ray diffraction.
1H NMR (benzene-d6, 300 MHz): δ 9.62 (1H, s, NCH), 7.92
(1H, t, J 7.7, Py H4), 7.63 (2H, d, J 7.7 Hz, Py H3), 7.01 (2H, d, J 8.3
Hz, NCAr m), 6.77 (2H, d, J 8.3 Hz, NCAr o), 6.78−6.84 (6H, br,
NAr m,p), 2.04 (s, 12H, NAr o-Me), 1.09 (s, 6H, NCMe). 13C NMR
(benzene-d6, 75 MHz): δ 156.7 (Py C2), 152.0 (NAr i), 148.6 (br,
CoNC), 147.5 (NCAr i), 138.7 (NAr o), 132.9 (NCAr p), 130.0
(NCAr m), 127.6 (NAr C-m), 126.3 (NCAr C-o), 124.9 (NAr C-p),
119.8 (Py C3), 115.6 (Py C4), 18.7 (NAr o-Me), 17.1 (NCMe).
NCMe was not observed. Anal. Calcd for C32H32ClCoN4 (567.01):
C, 67.78; H, 5.69; N, 9.88. Found: C, 67.73; H, 5.45; N, 9.49.
General Procedure for the Reaction of (1)Co(N2) with
Organic Halides. In a drybox, (1)CoCH2SiMe3 (14 mg, 27 μmol)
was weighed and dissolved in around 0.4 mL of benzene-d6 in an
NMR tube. Outside of the drybox, 2 mL of H2 was injected into the
tube; the solution turned green within one minute. The NMR tube
was transferred back into the drybox, and the excess hydrogen was
removed by flushing with nitrogen. The organic halide was then
added: 1.0 equiv for aryl halides, 0.5 equiv for aliphatic halides. The
NMR tube was vigorously shaken to mix the reactants. The reaction
1
Tentative, partial assignments for (3)CoAr: H NMR (benzene-
d6, 300 MHz): δ 10.27 (1H, t, J 7.6, Py H4), 5.14 (2H, d, J 7.1 Hz,
CoAr o), −0.65 (6H, s, NCMe). 19F NMR (benzene-d6, 282 MHz):
δ −61.2.
(1)CoCH2CMe2Ph. Under an argon atmosphere, (1)CoCl2 (0.32 g,
0.64 mmol) was weighed into a 50 mL Schlenk tube, followed by
12 mL of dry toluene. To the resulting green suspension,
ClMgCH2CMe2Ph (0.5 M in Et2O, 0.80 mL) was added dropwise
in around 30 min. The suspension turned pink. About 10 min later,
another 0.85 mL of ClMgCH2CMe2Ph solution was added over
30 min, and the suspension turned blue. After another 10 min, a final
portion of 0.91 mL of the ClMgCH2CMe2Ph solution was added
dropwise (total amount of ClMgCH2CMe2Ph: 1.28 mmol). The result-
ing blue mixture was stirred for a further 1.5 h at room temperature. After
evaporation of all solvents to dryness, 22 mL of dry toluene was added to
dissolve most of the solid and the solution was filtered over Celite. The
filtrate was concentrated to 2 mL and layered with hexane at −35 °C
overnight. Pipetting off the mother liquor in the drybox left dark blue
flakes of the product (0.16 g, 44%).
1
was monitored by H NMR (and 19F NMR where possible). The
1H NMR (benzene-d6, 300 MHz): δ 10.82 (1H, t, J 7.4, Py H4),
7.79 (2H, d, J 7.4, Py H3), 7.45 (2H, t, J 7.4, NAr p), 7.29 (4H, d, J 8.0,
NAr m), 6.99 (2H, t, Jav 7.3, Ph m), 6.91 (1H, t, J 6.8, Ph p), 6.63 (2H,
d, J 7.4, Ph o), 2.02 (12H, s, NAr o-Me), 1.45 (2H, s, CoCH2), 0.40
(6H, s, CMe2), −1.93 (6H, s, NCMe). 13C NMR (benzene-d6,
75 MHz): δ 167.0, 159.8, 158.8, 154.3, 130.4, 129.2 (NAr m), 127.5
(Ph m), 125.9 (NAr p), 125.0 (Ph o), 124.4 (Py C3), 123.6 (Ph p),
117.2 (Py C4), 47.3 (CMe2), 31.6 (CMe2), 26.0 (NCMe), 19.6
(NAr o-Me), 0.9 (br, CoCH2). Anal. Calcd for C35H40CoN3 (561.65):
C, 74.85; H, 7.18; N, 7.48. Found: C, 74.59; H, 7.17; N, 7.32.
(1)Co(η3-allyl). Under an argon atmosphere, (1)CoCl2 (0.58 g,
1.16 mmol) was weighed into a 50 mL Schlenk tube, followed by
addition 20 mL of dry toluene. To the resulting green suspension was
added 1.35 mL of allyl magnesium chloride solution (1.7 M in THF)
in three portions (0.4 mL in 15 min; 0.80 mL in 1.5 h; 0.15 mL in 10
min). After the addition, the resulting orange mixture was stirred for
1.5 h at room temperature. After evaporation of all solvents, 24 mL of
dry toluene was added to dissolve the solid, and the resulting
suspension was filtered over Celite. The filtrated was concentrated and
layered with hexane at −35 °C overnight. The mother liquor was
pipetted off, leaving a dark orange solid (0.11 g, 20%).
products were not isolated; attempted separations of (1)CoAr and
(1)CoX were never successful. All reactions in Table 2 were done
according to this procedure. For several (1)CoAr and (1)CoR
complexes, 1H NMR data were already provided in our previous
communication.11 Data for new compounds are provided below.
(1)Co(N2) and Cyclopropylmethyl Chloride. The 1H NMR
spectrum recorded immediately after mixing showed that there was no
(1)Co(N2) left, but peaks were broad due to some suspended solids;
the spectrum quality was much improved by first centrifuging the
sample. Apart from a peak due to (1)CoCl, there were no triplets in
the range 9.8−11 ppm, but there was a doublet at 7.9 ppm. By
1
comparing to the H NMR parameters of (1)Co(η3-allyl), the new
complex was identified as (1)Co(η3-crotyl). In addition, a terminal
1
olefin (probably 1-butene) could be observed. On the basis of the H
NMR data, the product ratio is (1)CoCl:(1)Co(η3-crotyl):olefin:
(cyclopropylmethyl chloride) = 1:0.55:0.25:0.12.
1
For (1)Co(η3-crotyl): H NMR (benzene-d6, 300 MHz): δ 7.89
(2H, d, J 7.5, Py H3), 7.40 (1H, t, J 7.5, Py H4), 5.18 (1H, dt, Jd 11.2,
Jt 9.2, CH2CH), 2.48 (1H, dq, Jd 11.2, Jq 6.5, CH3CH), 1.82 (2H, *,
CH2CH), 1.82 (12H, s, Ar o-Me), 1.57 (6H, s, NCMe), 1.24 (3H, d,
J 6.5, CH3CH). Aryl hydrogens could not be unambiguously assigned.
1H NMR (benzene-d6, 300 MHz): δ 7.84 (2H, d, J 7.8, Py H3), 7.41
(1H, t, J 7.8, Py H4), 6.87−6.97 (6H, m, NAr p and m), 5.15 (1H,
quintet, J 10.2, allyl CH), 2.42 (4H, br, allyl CH2), 1.76 (12H, s, NAr
o-Me), 1.55 (6H, s, NCMe). 13C NMR (benzene-d6, 75 MHz):
δ 150.1, 149.8, 149.2, 130.6, 128.3 (NAr m), 125.2 (NAr p), 120.5
(Py C3), 115.5 (Py C4), 106.1 (allyl CH), 43.9 (allyl CH2), 18.5 (NAr
o-Me), 16.6 (NCMe). Anal. Calcd for C28H32CoN3 (469.51): C,
71.63; H, 6.87; N, 8.95. Found: C, 71.75; H, 6.81; N, 8.82. Variable-
1
(1)Co(N2) and Allyl Chloride. The H NMR spectrum of the
purple solution, recorded immediately after mixing, showed that there
was no (1)Co(N2) left, but peaks were broad; quality improved after
centrifugation. The spectrum showed (1)CoCl:(1)Co(η3-allyl) =
1:0.98; no biallyl (1,5-hexadiene) could be detected.
1
(3)Co(N2) and Allyl Chloride. The H NMR spectrum of the
purple solution showed the product ratio (3)CoCl:(3)Co(η3-allyl) =
1:1, and no other products.
1
temperature H NMR spectra were recorded in toluene-d8 (20 mg in
For (3)Co(η3-allyl): 1H NMR (benzene-d6, 300 MHz): δ 7.82 (2H,
d, J 7.4, Py H3), 5.11 (1H, quintet, J 10.2, allyl CH), 2.61−2.67 (4H,
br d, allyl CH2), 2.68−2.80 (4H, m, CHMe2), 1.68 (6H, s, NCMe),
1.19 (24H, br, CHMe2). The pyridine H4 signal overlaps with the NAr
peaks in the region 7.39−7.43 ppm. The remaining NAr peaks could
not be unambiguously assigned.
0.4 mL) over the range −60 to 70 °C; for an analysis of the dynamic
behavior, see the SI. Low-temperature NMR data: 1H NMR (toluene-
d8, 300 MHz, −60 °C): δ 7.84 (1H, d, J 7.8, Py H3), 7.51 (1H, d, J 7.6,
Py H5), 7.41 (1H, t, Jav 7.7, Py H4), 6.87−6.97 (6H, m, NAr p and m),
5.24 (1H, tt, J 12.6 and 7.8, allyl CH), 3.40 (2H, br d, J 7.8, allyl Hsyn),
1.57 (2H, br d, J 12.6, allyl Hanti), 1.85, 1.74 (6H each, s, NAr o-Me),
1.86, 1.21 (3H each, s, NCMe).
(1)Co(N2), (3)Co(N2), and 6-Bromohexene. Analysis of these
reaction mixtures was neither simple nor unambiguous, and we believe
that chain walking and reversible β-elimination are the main causes of
the observed complications. For a discussion, see the SI.
(1)CoNCH-4-C6H4Cl. In a N2-filled drybox, (1)CoCH2SiMe3
(0.103 g, 0.20 mmol) was weighed in a small vial, followed by 4-
chlorobenzonitrile (0.0278 g, 0.20 mmol). A 3 mL portion of dry
toluene was added to dissolve the two reactants, and the resulting
solution was transferred into a 25 mL Schlenk tube. A 20 mL amount
of H2 gas was injected into the stirred solution, and stirring of the
resulting deep purple mixture was continued for 30 min. All solvents
were evaporated to dryness. Then 3 mL of hexane and 0.5 mL of Et2O
were added, and the solution was cooled to −35 °C overnight. The
dark crystalline product (0.096 g, 85%) was isolated by pipetting off
(1)Co(N2) and Isobutyl Bromide. The 1H NMR spectrum
indicated formation of two (1)CoBr and (1)CoCH2CHMe2 (1:0.48).
The alkyl complex decomposes slowly; after 24 h, the ratio is 1:0.33.
Further separation of the two products was not tried.
For (1)CoCH2CHMe2: 1H NMR (benzene-d6, 300 MHz): δ 10.47
(1H, t, J 7.8, Py H4), 7.97 (2H, d, J 7.8, Py H3), 2.05 (12H, s,
Ar o-Me), 1.64 (1H, m, CHMe2), 0.86 (6H, d, CHMe2), −1.75 (6H, s,
3967
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
NCMe). Aryl hydrogen peaks and CoCH2 cannot be unambiguously
assigned
was also confirmed by GC/MS. A spectrum of the solid in CD2Cl2
indicated the presence of a mixture of (1)CoCl2, (1)CoClBr, and
(1)CoBr2.
(1)Co(N2) and Isopropyl Chloride. The 1H NMR spectrum
immediately after mixing (Figure S7) indicated the presence of three
diamagnetic cobalt(I) complexes: (1)CoCl, (1)CoCH2CH2CH3, and
an unknown complex (Z) in the ratio 1:0.20:0.50. (1)CoCH2CH2CH3
is not stable and decomposed over three days. However, unknown
complex Z is stable.
General Procedure for C−C Coupling Reactions. To the
product mixture of (1)CoAr and (1)CoX generated from (1)CoN2
and ArX (1:1) as described earlier was added 0.5 equiv of the organic
halide. For benzyl bromide and methyl iodide, the reaction was
instantaneous; for benzyl chloride and allyl chloride it took hours for
the reaction to complete. After addition of the alkyl halide, the mixture
slowly turned green and deposited a dark solid. After the sample had turned
gray, 0.5 mL of water was added. The organic layer was filtered over glass
wool and examined by GC/MS. The results are shown in Table 3.
(1)Co(Py-6-Cl) and BzBr. When a mixture of (1)Co(Py-6-Cl) and
(1)CoCl (1:1) generated from (1)Co(N2) and 1.0 equiv of 2,
6-dichloropyridine was reacted with 0.5 equiv of BzBr, a pink solution
with dark suspended material was observed. After centrifugation, the
pink supernatant solution was determined by NMR to be a mixture
of cobalt(I) complexes (1)Co(Py-6-Cl):(1)CoBr:(1)CoCl =
1.0:0.67:0.44; the dark-colored solid was identified to be mainly
(1)CoBrCl together with other DIP cobalt(II) dihalides. After addition of
excess BzBr to the above pink solution, a cross-coupled product suspected
to be 2-chloro-6-benzylpyridine could be detected by GC/MS.
For (1)CoCH2CH2CH3: 1H NMR (benzene-d6, 300 MHz): δ 10.26
(1H, t, J 7.6, Py H4), 8.03 (2H, d, J 7.6, Py H3), 2.01 (12H, s, Ar o-
Me), 1.14 (2H, t, CoCH2CH2CH3), 0.35 (3H, t, J 7.0,
CoCH2CH2CH3), −1.14 (2H, m, CoCH2CH2CH3), −1.61 (6H, s,
NCMe). Aryl hydrogen peaks and CoCH2 could not be
unambiguously assigned.
For complex Z: 1H NMR (benzene-d6, 300 MHz): δ 7.97 (1H, d, J
7.0), 7.44 (1H, d, J 5.5), 7.31 (1H, *), 2.14 (1H, *, CHMe2), 2.11
(3H, s), 1.96 (3H, s), 1.85 (3H, s), 1.56 (3H, s), 1.16 (3H, d, J 7.0,
CHMe2), 0.99 (3H, d, J 6.6, CHMe2), −0.81 (3H, s). A possible
structure for this complex is proposed in the SI.
(1)Co(N2) and tert-Butyl Iodide. The 1H NMR spectrum
immediately after mixing (Figure S8) showed four diamagnetic
cobalt(I) complexes: (1)CoI, (1)CoCH2CHMe2, (tBu2-1)CoI, and
an unknown cobalt(I) complex (Z′), in the ratio 1:0.18:0.16:0.32.
NMR data for (tBu2-1)CoI are given below. For unknown complex Z′:
1H NMR (benzene-d6, 300 MHz): δ 8.02 (1H, d, J 7.0), 7.25 (*),
−1.25 (3H, s).
(1)CoCH2SiMe3 and tert-Butyl Iodide. In a N2-filled drybox,
(1)CoCH2SiMe3 (70 mg, 136 μmol) was dissolved in 15 mL of dry
toluene, and the resulting solution was transferred into a 25 mL
Schlenk tube. Outside the drybox, 2-iodo-2-methylpropane (17 μL,
135 μmol) was injected into it. The resulting mixture was stirred at
room temperature overnight, during which time a lot of solid pre-
cipitated. An additional 17 μL of 2-iodo-2-methylpropane was injected.
After stirring at room temperature for another two days, the
suspension was centrifuged and the liquid was evaporated to dryness.
The resulting residue was dissolved in toluene and layered with pen-
tane at −35 °C overnight. Some powder settled at the bottom; 1H NMR
showed that this consisted mainly of paramagnetic compound(s).
The mother liquor was evaporated to dryness, and the residue
dissolved in toluene. The resulting solution was allowed to slowly
evaporate in a drybox at room temperature. After two days, some dark
cubes of a crystalline solid were obtained. A fragment broken from
one of these crystals was used in a single-crystal X-ray diffraction
measurement. The data were of poor quality, and the attempted solu-
tion indicated that the crystal suffered from severe disorder and/or twin-
ning problems. The structure appeared to correspond to (tBu2-1)CoI,
but even the connectivity cannot be considered to be definitive.
Further details are given in the SI.
(1)Co(N2) and tert-Butyl Chloride. The reaction mixture turned
purple after 5 min. The 1H NMR spectrum of this solution showed the
presence of (1)CoCl, (1)CoCH2CHMe2, and an unknown cobalt(I)
complex most likely identical to Z′ obtained with tert-butyl iodide, in
the ratio 1:0.08:0.32.
(1)Co(N2) and 1-Iodoadamantane. The mixture turned pink
immediately after mixing. The 1H NMR spectrum indicated the
generation of (1)CoI and an unknown cobalt(I) complex Z″ in the
ratio 1:0.46. Complex Z″ appears to be similar but not identical to Z′
above. No obvious cobalt(I) alkyl could be observed.
For unknown complex Z″: 1H NMR (benzene-d6, 300 MHz): δ 7.99
(1H, d, J 6.2), 7.33 (*), −1.29 (3H, s).
(1)CoCl and Benzyl Chloride. In a N2-filled drybox, (1)CoCl (14
mg, 29 μmol) was weighed and dissolved in about 0.4 mL of dry
benzene-d6. Benzyl chloride (3.4 μL, 29 μmol) was added. No obvious
1
color change was observed, and the H NMR spectrum immediately
after mixing showed that no new product had been generated. After
10 min, solid started to precipitate. After 5 days at room temperature,
1
the suspension was centrifuged. The H NMR spectrum of the solu-
1
For (tBu2-1)CoI: H NMR (benzene-d6, 300 MHz): δ 8.25 (1H, d,
1
tion did not show any bibenzyl, while the H NMR spectrum of the
J 6.5, Py CH), 7.80 (1H, t, J 7.3, Ar p), 7.45 (1H, d, J 7.3, Ar m), 7.31
(1H, d, J 7.3, Ar m), 6.60 (1H, t, J 7.4, Ar p), 6.51 (2H, d, J 7.4, Ar m),
3.23 (3H, s, NCMe), 2.57 (3H, s, Ar o-Me), 2.44 (3H, s, Ar o-Me), 1.96
(3H, s, Ar o-Me), 1.84 (3H, s, Ar o-Me), 0.30 (9H, s, CMe3), 0.27 (9H, s,
CMe3), −2.26 (1H, d, J 6.5, Py CHtBu), −2.79 (1H, s, Py CHtBu), −3.1
(3H, s, NCMe). 13C NMR (benzene-d6, 75 MHz; from C−H HSQC
spectrum): δ 128.2 (NAr C-m), 127.9 (NAr m), 127.6 (NAr m), 125.9
(NAr p), 125.3 (NAr p), 25.0 (CMe3), 23.9(CMe3), 20.7 (NCMe),
19.6 (NAr o-Me), 19.2 (NAr o-Me), 18.5 (NAr o-Me), 18.3 (NAr o-Me).
Since neither (tBu2-1)CoI nor (tBu2-1)CoI2 could be obtained pure,
no elemental analysis was attempted.
(2)CoCH2SiMe3 and tert-Butyl Iodide. In a N2-filled drybox,
(2)CoCH2SiMe3 (100 mg, 175 μmol) was dissolved in 5 mL of dry
toluene, and the resulting solution was transferred into a 25 mL Schlenk
tube. Outside the drybox, 2-iodo-2-methylpropane (0.12 mL, 987 μmol,
5.6 equiv) was injected into it. After 10 min the mixture turned orange. It
was stirred at room temperature overnight. After centrifuging, the liquid
was evaporated to dryness. The residue was dissolved in dry toluene and
layered with pentane at −35 °C overnight. Some oil together with some
solids settled at the bottom. The liquid was pipetted off and evaporated to
dryness. The residue was first washed with hexane, and then the residue
was extracted by hexane/toluene (2:1). The resulting solution was cooled
to −35 °C. After two days, the supernatant was pipetted off; the mother
liquid and the solid were processed separately.
solid in CD2Cl2 showed it to be mainly (1)CoCl2.
(1)CoCl and p-Trifuoromethylbenzyl Chloride. In a N2-filled
drybox, (1)CoCl (13 mg, 27 μmol) was weighed and dissolved in
about 0.4 mL of dry benzene-d6. p-Trifluoromethylbenzyl chloride (3.5
μL, 27 μmol) was added. No obvious color change was observed, and
1
the H NMR spectrum immediately after mixing showed that no new
product had been generated. After 3 days at room temperature, a lot of
dark solid had precipitated and the suspension was centrifuged. The
1H NMR spectrum of the solution showed new singlets at 2.40 ppm
(1H) and −61.9 ppm (19F) corresponding to 4,4′-trifluoromethylbi-
benzyl [(CF3C6H4CH2)2:CF3C6H4CH2Cl ≈ 1:8; a few other fluorine-
containing unidentified side products were also visible in the 19F
NMR]. After hydrolysis, analysis of the organic layer by GC/MS
showed the presence of (CF3C6H4CH2)2 and CF3C6H4CH2Cl as well
as a trace of CF3C6H4CH3.
(1)CoCl and Benzyl Bromide. In a N2-filled drybox, (1)CoCl
(14 mg, 29 μmol) was weighed and dissolved in about 0.4 mL of dry
benzene-d6. Benzyl bromide (3.5 μL, 29 μmol) was added. The
resulting mixture turned green immediately and a lot of solid
1
precipitated. After 10 min, the color turned purple. The H NMR of
this mixture showed that some benzyl bromide was left, but no
(1)CoCl. After 5 days at room temperature, the mixture was red and
contained suspended dark solid material. After centrifuging, the H
1
NMR of the liquid showed that bibenzyl had been generated, which
3968
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
Table 4. Details of Crystal Structure Determinations
(1)CoNCH-4-C6H4Cl
(1)Co(η3-allyl)
(tBu2-2)CoI (Cl)
(tBu2-2)CoI2
formula
mol wt
T (K)
C32H32ClCoN4
567.00
C28H32CoN3
469.50
C37H53CoN3I0.94Cl0.06
720.17
C37H53CoI2N3
852.55
293(2)
293(2)
orthorhombic
Pccn
293(2)
200(2)
cryst syst
space group
a/Å
orthorhombic
P212121
10.9905(8)
14.5602(11)
18.7056(15)
90
monoclinic
P21/n
orthorhombic
P212121
12.9841(8)
16.2444(10)
18.4949(11)
90.00
19.1003(11)
30.6877(18)
8.3627(5)
90
13.586(5)
16.687(7)
16.630(6)
90
b/Å
c/Å
α/deg
β/deg
90
90
101.567(9)
90
90.00
γ/deg
V/Å3
90
90
90.00
2993.3(4)
4
4901.7(5)
8
3694(2)
4
3900.9(4)
4
Z
Dc/g cm−3
abs coeff/mm−1
F000
1.258
1.272
1.295
1.452
0.689
0.720
1.282
2.050
1184
1984
1495
1716
index ranges
−13 ≤ h ≤ 13
−17 ≤ k ≤ 17
−22 ≤ l ≤ 22
51
−22 ≤ h ≤ 23
−37 ≤ k ≤ 37
−10 ≤ l ≤ 10
51
−16 ≤ h ≤ 16
−20 ≤ k ≤ 20
−20 ≤ l ≤ 20
51
−17 ≤ h ≤ 17
−20 ≤ k ≤ 21
−24 ≤ l ≤ 240
56.68
2θmax/deg
no. reflns
19 632
27 619
26 916
71 802
no. unique
5568
4558
6870
9576
no. > 2σ
3614
2816
5955
8231
GOF
0.923
0.992
1.159
1.015
no. params
R (Fo > 4σ(F))
R (all data)
wR2 (all data)
largest peak, hole/e Å−3
349
304
395
415
0.0523
0.0435
0.0408
0.0344
0.0784
0.0930
0.0494
0.0445
0.1443
0.1444
0.1427
0.0979
0.421, −0.350
0.363, −0.265
0.947, −0.765
1.472, −0.507
X-ray Structure Determinations. Crystal data and refinement
parameters for the complexes are listed in Table 4. Details of individual
structure determinations follow.
(a) The mother liquor was evaporated to dryness, and the residue
dissolved in toluene. The resulting solution was allowed to
slowly evaporate in a drybox at room temperature. After four
days, some dark cubes of a crystalline solid were obtained. A
fragment broken from one of these crystals was used for single-
crystal X-ray diffraction, which showed it to be (tBu2-2)CoI
contaminated with about 6% of (tBu2-2)CoCl.
(b) The 1H NMR spectrum of the solid showed mainly dia-
magnetic (tBu2-2)CoI (at least 70%) but also one or more para-
magnetic compounds. The solid was washed with 0.2 mL of
ether/hexane. The remaining solid was further extracted with
ether (∼0.5 mL)/toluene (6 drops), and the resulting solution
was slowly evaporated to dryness over two weeks at room
temperature to give a crystalline solid. A fragment broken off
this solid was used for single-crystal X-ray diffraction, which
showed it to be (tBu2-2)CoI2.
(1)CoNCH-4-C6H4Cl. A deep purple fragment (approximately
0.20 × 0.20 × 0.25 mm) broken from a large crystal cluster was
mounted in a thin glass capillary. Data were collected at 293 K in a
Bruker four-circle diffractometer with an APEX detector using Mo Kα
radiation (0.71073 Å). A sphere of data was collected with 0.2° scan
width and 45 s scan time. The crystal system and space group were
determined from the cell metrics and systematic absences. The data
were integrated using the SAINT program,39 and a semiempirical
absorption correction was done using SADABS.40 The structure was
solved by Patterson methods using SHELXS41 and refined using
SHELXL9741 (full-matrix least-squares refinement on F2); hydrogen
atoms were placed at calculated positions and refined in riding
mode. The structure was checked for solvent-accessible voids with
PLATON.42
(1)Co(η3-allyl). A deep-orange crystal fragment (0.05 × 0.10 ×
0.30 mm) was broken from a large piece of a crystalline aggregate
and was sealed in a thin glass capillary and mounted on a Bruker D8
three-circle diffractometer equipped with a rotating anode generator
(Mo Kα X-radiation), multilayer optics incident beam path, and
an APEX-II CCD detector. A hemisphere of X-ray diffraction data
(81 337 reflections) was collected to 60° 2θ using 25 s per 0.2° frame
with a crystal-to-detector distance of 5 cm. A semiempirical absorption
correction (SADABS) was applied, and identical data were merged to
give 44 010 reflections covering the Ewald hemisphere, of which all
data with up to 2θ = 51° were used for further refinement. The unit-
cell parameters were obtained by least-squares refinement of 4833
reflections with I > 10σ(I).
For (tBu2-2)CoI: 1H NMR (benzene-d6, 300 MHz): δ 8.37 (1H, br
d, Py CH), 7.96 (1H, t, J 7.2, Ar p), 7.60 (1H, d, J 7.5, Ar m), 7.41
(1H, d, J 7.1, Ar m), 6.77 (1H, t, J 7.3, Ar p), 6.65 (2H, t, J 7.0, Ar m),
4.63 (1H, m, NAr CH2CH3), 3.46 (1H, m, NAr CH2CH3), 3.16 (1H,
m, NAr CH2CH3), 2.73 (2H, m, NAr CH2CH3), 2.53 (1H, m, NAr
CH2CH3), 2.21 (2H, m, NAr CH2CH3), 1.71 (t, 3H, Jav 7.4, NAr
CH2CH3), 1.31 (t, 3H, Jav 7.4, NAr CH2CH3), 0.99 (t, 3H, Jav 7.4, NAr
CH2CH3), 0.96 (t, 3H, Jav 7.5, NAr CH2CH3), 0.36 (s, 9H, CMe3),
0.31 (s, 9H, CMe3), −2.32 (1H, br d, Py CHtBu), −2.80 (1H, s, Py
CHtBu), −3.01 (s, 3H, NCMe).
For (tBu2-2)CoI2 (incomplete and tentative assignments; all peaks
1
(tBu2-1)CoI. For details of this (inconclusive) structure determi-
nation, see the SI.
are broad): H NMR (benzene-d6, 300 MHz): δ 109.1 (1H, Py H3),
15.4 (9H, CMe3), 12.2, 11.1, −16.1 (9H, CMe3).
Since neither (tBu2-2)CoI nor (tBu2-2)CoI2 could be obtained pure,
no elemental analysis was attempted.
(tBu2-2)CoI (Cl). A deep-brown fragment (0.3 × 0.4 × 0.6 mm) was
broken from a large piece of a crystalline aggregate and was sealed in a
3969
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
thin glass capillary and mounted on a Bruker D8 three-circle
diffractometer equipped with a rotating anode generator (Mo Kα X-
radiation), multilayer optics incident beam path, and an APEX-II CCD
detector. A full sphere of X-ray diffraction data (132 039 reflections)
was collected to 2θ = 60° using 2 s per 0.3° frame with a crystal-to-
detector distance of 5 cm. A semiempirical absorption correction
(SADABS) was applied, and identical data were merged to give 42 999
reflections covering the Ewald hemisphere, of which all data up to 2θ =
51° were used for further refinement. The unit-cell parameters were
obtained by least-squares refinement of 9940 reflections with
I > 10σ(I). From the refinement results it became clear some
(tBu2-2)CoCl had cocrystallized with the (tBu2-2)CoI (the Cl occupa-
tion refined to 6.3%). The source of this impurity is likely some
(tBu2-2)CoCl present in the starting material (tBu2-2)CoCH2SiMe3.
(tBu2-2)CoI2. A deep purple crystal block (0.16 × 0.19 × 0.26 mm)
broken from a large piece of crystal was selected under an inert
atmosphere and mounted on a glass fiber. Unit cell measurements and
intensity data collections were performed on a Bruker-AXS SMART
1K CCD diffractometer using graphite-monochromatized Mo Kα
radiation (λ = 0.71073 Å). The measurement was done at 200 K. The
data reduction included a correction for Lorentz and polarization
effects, with an applied multiscan absorption correction (SADABS).
The crystal structures were solved and refined using the SHELXTL
program suite. Direct methods yielded all non-hydrogen atoms, which
were refined with anisotropic thermal parameters. All hydrogen atom
positions were calculated geometrically and were riding on their
respective atoms. Thermal parameters of carbon atoms C26, C27, and
C37 suggested the presence of positional disorder. Disorder for C26
and C27 was modeled as oscillation of the ethyl substituent moiety
with partial occupancies of 50%:50%. Disorder for C37 was modeled
as oscillation of the methyl end group of another ethyl substituent with
partial occupancies of 50%:50%. A set of “rigid-bond” and “thermal
parameters” restraints was applied to the disordered fragments to
improve refinement of thermal parameters.
REFERENCES
■
(1) (a) Johnson, L. K.; Killian, C. M.; Brookhart, M. J. Am. Chem. Soc.
1995, 117, 6414−6415. (b) Wissing, E.; Kaupp, M.; Boersma, J.; Spek,
A. L.; Vankoten, G. Organometallics 1994, 13, 2349−2356.
(c) Fedushkin, I. L.; Tishkina, A. N.; Fukin, G. K.; Hummert, M.;
Schumann, H. Eur. J. Inorg. Chem. 2008, 483−489. (d) Scarborough,
C. C.; Wieghardt, K. Inorg. Chem. 2011, 50, 9773−9793.
(2) (a) Smith, A. L.; Clapp, L. A.; Hardcastle, K. I.; Soper, J. D.
Polyhedron 2010, 29, 164−169. (b) Verma, P.; Weir, J.; Mirica, L.;
Stack, T. D. P. Inorg. Chem. 2011, 50, 9816−9825.
(3) (a) Gibson, V. C.; Redshaw, C.; Solan, G. A. Chem. Rev. 2007,
107, 1745−1776. (b) Gibson, V. C.; Solan, G. A. Iron-based and
Cobalt-based Polymerisation Catalysts. In Metal Catalysts in Olefin
Polymerization; Guan, Z., Ed.; Springer: Berlin, 2009; Vol. 26, pp 107−
158.
(4) Chirik, P. J.; Wieghardt, K. Science 2010, 327, 794−795.
(5) Bart, S. C.; Lobkovsky, E. B.; Chirik, P. J. Abstr. Pap. Am. Chem.
Soc. 2004, 228, U898−U898.
(6) Spessard, G. O.; Miessler, G. L. Organometallic Chemistry, 2nd
ed.; Oxford University Press: Oxford, 2010.
(7) (a) Trovitch, R. J.; Lobkovsky, E.; Chirik, P. J. J. Am. Chem. Soc.
2008, 130, 11631−11640. (b) Trovitch, R. J.; Lobkovsky, E.;
Bouwkamp, M. W.; Chirik, P. J. Organometallics 2008, 27, 6264−6278.
(8) (a) Doherty, J. C.; Ballem, K. H. D.; Patrick, B. O.; Smith, K. M.
Organometallics 2004, 23, 1487−1489. (b) MacLeod, K. C.; Conway, J.
L.; Tang, L. M.; Smith, J. J.; Corcoran, L. D.; Ballem, K. H. D.; Patrick,
B. O.; Smith, K. M. Organometallics 2009, 28, 6798−6806.
(9) (a) Halpern, J.; Maher, J. P. J. Am. Chem. Soc. 1965, 87, 5361−
5366. (b) Chock, P. B.; Halpern, J. J. Am. Chem. Soc. 1969, 91, 582−&.
(c) Schneider, P. W.; Phelan, P. F.; Halpern, J. J. Am. Chem. Soc. 1969,
91, 77−. (d) Marzilli, L. G.; Marzilli, P. A.; Halpern, J. J. Am. Chem.
Soc. 1971, 93, 1374−. (e) Halpern, J.; Phelan, P. F. J. Am. Chem. Soc.
1972, 94, 1881−1886.
Computational Details. All geometries were optimized with
Turbomole43 using the TZVP basis set,44 the b3-lyp functional,45 and
the unrestricted DFT formalism in combination with an external opti-
mizer (PQS OPTIMIZE).46 The low-spin state was found to be the
lowest in energy for most species studied; square-planar Co(I)
complexes preferred a broken-symmetry Sz = 0 solution.20 Vibrational
analyses were done to confirm the nature of all stationary points and to
calculate thermal corrections (enthalpy and entropy, gas phase, 298 K,
1 bar) using the standard formulas of statistical thermodynamics.
Improved single-point energies were obtained using the TZVPP basis
set47 at TZVP geometries and combined with TZVP-level thermal
corrections to generate the final free energies.
(10) Ogoshi, H.; Setsune, J.; Yoshida, Z. J. Am. Chem. Soc. 1977, 99,
3869−3870.
(11) Zhu, D.; Budzelaar, P. H. M. Organometallics 2010, 29, 5759−
5761.
(12) Smith, A. L.; Hardcastle, K. I.; Soper, J. D. J. Am. Chem. Soc.
2010, 132, 14358−14360.
(13) Enright, D.; Gambarotta, S.; Yap, G. P. A.; Budzelaar, P. H. M.
Angew. Chem., Int. Ed. 2002, 41, 3873−3876.
(14) Zhu, D.; Thapa, I.; Korobkov, I.; Gambarotta, S.; Budzelaar, P.
H. Inorg. Chem. 2011, 50, 9879−9887.
(15) (a) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30,
4999−5009. (b) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30,
4269−4283. (c) Cheung, C. W.; Chan, K. S. Organometallics 2011, 30,
1768−1771.
ASSOCIATED CONTENT
* Supporting Information
■
S
(16) Bowman, A. C.; Milsmann, C.; Atienza, C. C. H.; Lobkovsky, E.;
Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2010, 132, 1676−1684.
(17) (a) 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,
151−159. (b) Tellmann, K. F.; Humphries, M. J.; Rzepa, H. S.;
Gibson, V. C. Organometallics 2004, 23, 5503−5513.
CIF file containing X-ray structure data for the structures
determined in this work; total energies and xyz coordinate files
for all optimized geometries. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Peter_Budzelaar@umanitoba.ca.
■
(18) PhCHCHPh and PhCH2CH2Ph detected by GC/MS after
hydrolysis.
(19) (a) Erker, G.; Fromberg, W.; Atwood, J. L.; Hunter, W. E.
Angew. Chem., Int. Ed. Engl. 1984, 23, 68−69. (b) Kaplan, A. W.; Polse,
J. L.; Ball, G. E.; Andersen, R. A.; Bergman, R. G. J. Am. Chem. Soc.
1998, 120, 11649−11662. (c) Uhl, W.; Matar, M. Z. Naturforsch. Sect.
B: J. Chem. Sci. 2004, 59, 1214−1222. (d) Temprado, M.;
McDonough, J. E.; Mendiratta, A.; Tsai, Y. C.; Fortman, G. C.;
Cummins, C. C.; Rybak-Akimova, E. V.; Hoff, C. D. Inorg. Chem.
2008, 47, 9380−9389.
(20) Knijnenburg, Q.; Hetterscheid, D.; Kooistra, T. M.; Budzelaar,
P. H. M. Eur. J. Inorg. Chem. 2004, 1204−1211.
(21) The (3)Co(N2) used for this reaction was prepared via Na/Hg
reduction of (3)CoCl2.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Mr. Mark Cooper (University of Manitoba) for help
with X-ray structure determinations and Prof. Frank
Hawthorne for the use of their single-crystal X-ray diffrac-
tometer. This work was supported by NSERC, CFI, and MRIF
grants (to P.H.M.B.) and a University of Manitoba Graduate
Fellowship and Manitoba Graduate Scholarship (to D.Z.).
3970
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971

Organometallics
Article
(22) Note that since LCoAr and LCoX are formed in a ratio <1:1,
“yields” of LCoX calculated from eq 1 may exceeed 100%.
(23) In our original communication11 this product was incorrectly
identified as (1)Co(2,4,6-tBu3)C6H2.
(24) (a) Childress, B. C.; Rice, A. C.; Shevlin, P. B. J. Org. Chem.
1974, 39, 3056−3058. (b) Brunton, G.; Griller, D.; Barclay, L. R. C.;
Ingold, K. U. J. Am. Chem. Soc. 1976, 98, 6803−6811.
(25) Bowry, V. W.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1991,
113, 5687−5698.
(26) By contrast, 1-hexene was detected in the reaction between (3)
Co(N2) and 6-bromohexene. For details, refer to the Experimental
Section.
(27) However, in that case one would expect a concentration
dependence of the product ratio, which we did not observe. Further
work is probably needed here.
(28) Bart, S. C.; Chlopek, K.; Bill, E.; Bouwkamp, M. W.; Lobkovsky,
E.; Neese, F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem. Soc. 2006, 128,
13901−13912.
(29) In the addition of most substrates other than vinyl acetate, the
alkyl group is lost after the addition, forming a formally Fe(I) product,
which in reality contains Fe(II) and a ligand radical anion.
(30) Reliable assignments of the paramagnetically shifted resonances
were not possible; hence we do not know whether there is more than
one paramagnetic component.
(31) Knijnenburg, Q.; Gambarotta, S.; Budzelaar, P. H. M. Dalton
Trans. 2006, 5442−5448.
(32) Reardon, D.; Conan, F.; Gambarotta, S.; Yap, G.; Wang, Q. Y. J.
Am. Chem. Soc. 1999, 121, 9318−9325.
(33) Arrowsmith, M.; Hill, M. S.; Kociok-Kohn, G. Organometallics
2010, 29, 4203−4206.
(34) (a) Cahiez, G.; Moyeux, A. Chem. Rev. 2010, 110, 1435−1462.
(b) Amatore, M.; Gosmini, C.; Perichon, J. Eur. J. Org. Chem. 2005,
989−992.
(35) Fan, R. Q.; Zhu, D. S.; Mu, Y.; Li, G. H.; Yang, Y. L.; Su, Q.;
Feng, S. H. Eur. J. Inorg. Chem. 2004, 4891−4897.
(36) Schmidt, R.; Welch, M. B.; Knudsen, R. D.; Gottfried, S.; Alt, H.
G. J. Mol. Catal. A: Chem. 2004, 222, 9−15.
(37) Zhu, D.; Janssen, F. F. B. J.; Budzelaar, P. H. M. Organometallics
2010, 29, 1897−1908.
(38) Kooistra, T. M.; Knijnenburg, Q.; Smits, J. M. M.; Horton, A.
D.; Budzelaar, P. H. M.; Gal, A. W. Angew. Chem., Int. Ed. 2001, 40,
4719−+.
(39) SMART program suite, 6; Bruker AXS Inc: Madison, WI.
(40) Sheldrick, G. M. SADABS; University of Gottingen: Germany:
̈
1996.
(41) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr.
2008, 64, 112−122.
(42) Spek, A. L. PLATON. A Multipurpose Crystallographic Tool;
Utrecht University: Utrecht, The Netherlands, 2003.
(43) (a) Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346−354.
(b) Ahlrichs, R.; Bar, M.; Baron, H.-P.; Bauernschmitt, R.; Bocker, S.;
̈
̈
Ehrig, M.; Eichkorn, K.; Elliott, S.; Furche, F.; Haase, F.; Haser, M.;
̈
Hattig, C.; Horn, H.; Huber, C.; Huniar, U.; Kattannek, M.; Kohn, A.;
Kolmel, C.; Kollwitz, M.; May, K.; Ochsenfeld, C.; Ohm, H.; Schafer,
A.; Schneider, U.; Treutler, O.; Tsereteli, K.; Unterreiner, B.; Von
Arnim, M.; Weigend, F.; Weis, P.; Weiss, H. Turbomole, 5; Theoretical
Chemistry Group, University of Karlsruhe, 2002.
̈
̈
̈
̈
(44) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,
5829−5835.
(45) (a) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37,
785−789. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(c) Becke, A. D. J. Chem. Phys. 1993, 98, 1372−1377.
(46) (a) Baker, J. J. Comput. Chem. 1986, 7, 385−395. (b) Baker, J.
PQS, 2.4; Parallel Quantum Solutions: Fayetteville, AR, 2001.
(47) Weigend, F.; Furche, F.; Ahlrichs, R. J. Chem. Phys. 2003, 119,
12753−12762.
3971
dx.doi.org/10.1021/om300182c | Organometallics 2012, 31, 3958−3971