Selective dimerization/oligomerization of α-olefins by cobalt bis(imino)pyridine catalysts stabilized by trifluoromethyl substituents: Group 9 metal catalysts with productivities matching those of iron systems

Article abstract of DOI:10.1021/om049297q

A series of bis(imino)pyridine ligands containing aryl substituents with o-trifluoromethyl units, along with their cobalt and iron complexes, has been prepared and characterized. The crystal structures of several complexes are reported. Both the cobalt and iron complexes, when activated by methylaluminoxane cocatalysts, afford much higher olefin oligomerization/ polymerization activities than their nonfluorinated relatives. Enhanced performance is seen not only in higher peak activities but also in longer catalyst lifetimes, suggesting that the trifluoromethyl group significantly improves catalyst stability. The most impressive activity increase is observed for a cobalt catalyst combining an o-CF₃ and an o-F unit. This system is more active than its iron counterpart in ethylene polymerization, reaching > 100 000 g mmol^-1 h^-1 bar^-1. Propene, 1-butene, and 1-hexene are oligomerized by this catalyst at rates much higher than for nonfluorinated relatives. Highly linear dimers predominate in each case, the remainder being trimeric and tetrameric products. The principal product of propene dimerization is 1-hexene (60-73% of total), whereas for 1-butene and 1-hexene internal olefins are obtained (E isomers predominate). No isomerization of 1-hexene is seen under the reaction conditions. The catalysts operate by a mechanism involving (1,2)- followed by (2,1)-insertion steps. Combinations of chain growth and step growth account for the formation of linear trimers and tetramers of propene.

Full text of DOI:10.1021/om049297q

280
Organometallics 2005, 24, 280-286
Selective Dimerization/Oligomerization of r-Olefins by
Cobalt Bis(imino)pyridine Catalysts Stabilized by
Trifluoromethyl Substituents: Group 9 Metal Catalysts
with Productivities Matching Those of Iron Systems
Kilian P. Tellmann, Vernon C. Gibson,* Andrew J. P. White, and
David J. Williams
Department of Chemistry, Imperial College London, Exhibition Road, London SW7 2AZ, U.K.
Received September 9, 2004
A series of bis(imino)pyridine ligands containing aryl substituents with o-trifluoromethyl
units, along with their cobalt and iron complexes, has been prepared and characterized.
The crystal structures of several complexes are reported. Both the cobalt and iron complexes,
when activated by methylaluminoxane cocatalysts, afford much higher olefin oligomerization/
polymerization activities than their nonfluorinated relatives. Enhanced performance is seen
not only in higher peak activities but also in longer catalyst lifetimes, suggesting that the
trifluoromethyl group significantly improves catalyst stability. The most impressive activity
increase is observed for a cobalt catalyst combining an o-CF3 and an o-F unit. This system
is more active than its iron counterpart in ethylene polymerization, reaching >100 000 g
mmol-1 h-1 bar-1. Propene, 1-butene, and 1-hexene are oligomerized by this catalyst at rates
much higher than for nonfluorinated relatives. Highly linear dimers predominate in each
case, the remainder being trimeric and tetrameric products. The principal product of propene
dimerization is 1-hexene (60-73% of total), whereas for 1-butene and 1-hexene internal
olefins are obtained (E isomers predominate). No isomerization of 1-hexene is seen under
the reaction conditions. The catalysts operate by a mechanism involving (1,2)- followed by
(2,1)-insertion steps. Combinations of chain growth and step growth account for the formation
of linear trimers and tetramers of propene.
Introduction
ligands. Here, we show not only that judicious fluorina-
tion of the imino aryl substituents results in unprec-
edented activities for cobalt catalysts, activities that in
some cases surpass those observed for the most active
iron catalysts, but also that these catalysts dimerize
propene, 1-butene, and 1-hexene with remarkably high
selectivities. These findings represent significant im-
provements in terms of activity and selectivity over
previous reports for nonfluorinated as well as other
fluorinated cobalt bis(imino)pyridine complexes.6-8 In-
deed, some examples of the latter have been reported
as being inactive for ethylene polymerization.8
The discovery of highly active iron- and cobalt-based
olefin polymerization catalysts in the late 1990s has led
to much interest in the chemistry of transition-metal
complexes bearing terdentate bis(imino)pyridine lig-
ands.1-4 Many studies have reported the effects of
ligand substitution patterns on activity and selectivity.5
However, in all cases it has generally been found that
the iron-based catalysts are typically at least an order
of magnitude more active than their cobalt relatives.
The difference in activity is even more pronounced for
oligomerization systems, which bear sterically less
demanding substituents on the bis(imino)pyridine
Complex Synthesis and Characterization
The series of bis(imino)pyridine ligands 1a-c con-
taining an o-trifluoromethyl substituent were prepared
by simple Schiff-base condensation procedures (Scheme
1).9 1a-c are readily complexed to cobaltous and ferrous
bromides to afford complexes 2a-c and 4a-c in high
purity;10,11 the known compounds 3 and 5 were prepared
in an analogous fashion as benchmarks for the catalytic
studies.2
* To whom correspondence should be addressed. E-mail: v.gibson@
imperial.ac.uk.
(1) (a) Britovsek, G. J. P.; Gibson, V. C.; Kimberley, B. S.; Solan, G.
A.; White, A. J. P.; Williams, D. J. Chem. Commun. 1998, 849. (b)
Small, B. L.; Brookhart, M.; Bennett, A. M. A. J. Am. Chem. Soc. 1998,
120, 4049. (c) Britovsek, G. J. P.; Bruce, M.; Gibson, V. C.; Kimberley,
B. S.; Maddox, P. J.; Mastroianni, S.; McTavish, S. J.; Redshaw, C.;
Solan, G. A.; Stro¨mberg, S.; White, A. J. P.; Williams, D. J. J. Am.
Chem. Soc. 1999, 121, 8728.
(2) (a) Small, B. L.; Brookhart, M. J. Am. Chem. Soc. 1998, 120,
7143. (b) Britovsek, G. J. P.; Mastroianni, S.; Solan, G. A.; Baugh, S.
P. D.; Redshaw, C.; Gibson, V. C.; White, A. J. P.; Williams, D. J.;
Elsegood, M. R. J. Chem. Eur. J. 2000, 6, 2221.
(6) Bennett, A. M. A. (DuPont) World Patent 51550, 1999.
(7) Small, B. L. Organometallics 2003, 22, 3178.
(8) (a) Chen, Y.; Qian, C.; Sun, J. Organometallics 2003, 22, 1231.
(b) Chen, Y.; Chen, R.; Qian, C.; Dong, X.; Sun, J. Organometallics
2003, 22, 4312.
(3) Reardon, D.; Conan, F.; Gambarotta, S.; Yap, G.; Wang, Q. J.
Am. Chem. Soc. 1999, 121, 9318.
(4) Esteruelas, M. A.; Lo´pez, A. M.; Me´ndez, L.; Oliva´n, M.; On˜ate,
E. Organometallics 2003, 22, 395.
(5) See: Gibson, V. C.; Spitzmesser, S. K. Chem. Rev. 2003, 103,
283 for leading references.
(9) Ligand 1c was independently prepared and recently published
by Esteruelas et al.4
10.1021/om049297q CCC: $30.25 © 2005 American Chemical Society
Publication on Web 12/21/2004

Cobalt Bis(imino)pyridine Catalysts
Organometallics, Vol. 24, No. 2, 2005 281
Scheme 1a
a Reagents and conditions: (i) Dean-Stark, benzene, cat.
H2SO4; (ii) CoCl2, CoBr2‚6H2O, FeCl2 or FeBr2 in THF.
The magnetic moments of the resulting paramagnetic
complexes were determined using an Evans balance to
be 3.9-4.2 µB for cobalt complexes 2a-c and 4.8-5.2
µB for iron complexes 4a-c. These values are in good
agreement with the spin-only values expected for high-
spin d7 (Co2+) and d6 (Fe2+) centers. Crystals of 2a-c
and 4a,b suitable for X-ray analysis were obtained by
recrystallization from saturated dichloromethane or
acetone solutions layered with hydrocarbon solvent. In
the case of 2c, crystals solvated with both dichlo-
romethane (2c‚0.5CH2Cl2) and acetone (2c‚0.5Me2CO)
were obtained. The structures of the corresponding
cobalt and iron complexes (2a and 4a, 2b and 4b) are
isomorphous, and the two solvates of 2c have unit cells
very similar to each other. The molecular structures of
2c‚0.5CH2Cl2 and 4b are shown in Figure 1, whereas
those of 2a,b, 2c‚0.5Me2CO, and 4a are provided in the
Supporting Information.
The complexes 2a-c and 4a,b all have very similar
structures with distorted-square-based-pyramidal geom-
etries at the metal centers and approximate Cs sym-
metry overall. It is notable that the aryl ring substitu-
ents within each complex adopt a mutually syn orien-
tation with the CF3 groups pointing away from the
apical ligand. This arrangement is also seen in other
bis(imino)pyridine complexes with mono-ortho-substi-
tuted N-aryl rings.1b,2b,12,13 Furthermore, as is typical
in complexes of this kind, the aryl rings are oriented
approximately orthogonal with respect to the ligand
backbone to minimize intra- and interligand repulsions.
Figure 1. Thermal ellipsoid plots of 2c (top, 30% prob-
ability ellipsoids) and 4b (bottom, 50% probability el-
lipsoids). Hydrogen atoms and cocrystallized solvent mol-
ecules are omitted for clarity.
For the two complexes shown in Figure 1, the coordi-
nating atoms of the basal plane (the nitrogen atoms and
the basal bromine) are coplanar to within 0.10 Å (2c)
and 0.04 Å (4b), with the metal centers residing 0.49 Å
(2c, M ) Co) and 0.44 Å (4b, M ) Fe) above this plane.
As expected, the M-N(py) distances (2.052(4)-2.122-
(5) Å) are noticeably shorter than those to the imino
nitrogen atoms (2.142(5)-2.212(4) Å), and with the
Fe-N bonds being generally longer than their Co-N
counterparts. The double-bond character of the imino
linkages has been retained in each case (CdN in the
range 1.255(9)-1.293(8) Å), and the py-imino bonds
C2-C7 and C6-C9 (between 1.469(10) and 1.498(9) Å)
show no evidence of delocalization between these π
systems. Further structural data for these and the other
structures are included in the Supporting Information.
Polymerization and Oligomerization of
Ethylene
Upon activation by a methylaluminoxane cocatalyst
(MAO or coMAO14), all of the fluorinated precatalysts
gave highly active ethylene polymerization catalysts
(Table 1), with activities typically in the range 6000-
17 000 g mmol-1 h-1 bar-1 over 60 min runs. For the
fluorinated cobalt catalysts 2a-c (8000-15 000 g mmol-1
h-1 bar-1) this represents a marked contrast to the
benchmark system 3, which typically shows activites of
13-26 g mmol-1 h-1 bar-1 (cf. entry 1, Table 1 and ref
3b) if activated in the absence of ethylene and up to
about 100 g mmol-1 h-1 bar-1 if activated in the
(10) The complexes were at times already of analytical purity at
this point. Satisfactory elemental analyses have been obtained for all
complexes by recrystallization from dichloromethane/hexane.
(11) Complex 2c has recently been claimed, but no characterization
data were reported. Pelascini, F.; Peruch, F.; Lutz, P.; Wesolek, M.;
Kress, J. Macromol. Rapid. Commun. 2003, 24, 768.
(12) Drew, M. G. B.; Hollis, S. Acta Crystallogr. 1978, B34, 2853.
(13) Gibson, V. C.; McTavish, S.; Redshaw, C.; Solan, G. A.; White,
A. J. P.; Williams, D. J. Dalton 2003, 221.
(14) MAO is made by partial, extremely careful hydrolysis of AlMe3.
For coMAO, a 90:10 mixture of AlMe3 and AliBu3 is hydrolyzed.

282 Organometallics, Vol. 24, No. 2, 2005
Table 1. Ethylene Polymerization Resultsa
product yieldb
Tellmann et al.
precat.
run no. (amt (µmol))
C2H4 press run time
total
solid
chain end activityf (g mmol-1
d
cocat. (Al:M)
(bar)
(min)
mass (g) share (%)
Rc/Mn
ratioe
h-1 bar-1
)
1g
2
3 (20.0)
MAO (400)
4.0
4.0
4.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
60
60
60
60
60
60
60
60
60
10
2
1.05
8.62
7.94
4.18
4.48
5.24
4.42
6.22
7.66
4.50
2.15
0
0
0.66/-
0.57/-
0.55/-
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
1.0:1
13
108
3 (20.0)
MAO (400)
3
3 (20.0)
coMAO (400)
MAO (2000)
coMAO (2000)
MAO (2000)
coMAO (2000)
MAO (2000)
coMAO (2000)
coMAO (2000)
coMAO (2000)
0
99
4
2a (0.50)
2a (0.50)
2b (0.50)
2b (0.50)
2c (0.50)
2c (0.50)
2c (0.50)
2c (0.50)
13
12
7
0.73/1600
0.73/1900
0.74/1200
0.74/940
-/2600
8 360
8 960
10 500
8 840
12 400
15 300
54 000
129 000
5
6
7
7
8
95
95
95
96
9
-/2400
10
11
-/2200
-/2400
12
13
14
15
16
17
18
19
5 (10.0)
MAO (100)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
60
60
60
60
60
60
60
60
22.72
27.38
3.16
7.70
5.98
8.66
3.43
3.61
64
15
0.77/520
0.70/470
-/2700
-/2800
-/2700
-/2100
-/4300
-/3900
3.1:1
2.1:1
2.3:1
1.8:1
1.7:1
1.9:1
5.1:1
8.3:1
2 270
2 740
5 (10.0)
coMAO (100)
MAO (2000)
coMAO (2000)
MAO (2000)
coMAO (2000)
MAO (2000)
coMAO (2000)
4a (0.50)
4a (0.50)
4b (0.50)
4b (0.50)
4c (0.50)
4c (0.50)
94
6 320
97
15 400
12 000
17 300
6 860
96
96
>99
>99
7 220
a General conditions: a 0.5 L glass autoclave filled with a suspension of precatalyst in toluene (200 cm3) at ambient temperature is
saturated with ethylene and subsequently activated with either MAO or coMAO. b Oligomers were characterized and quantified by GC,
using 2,2,4,4,6,8,8-heptamethylnonane as internal standard; the solid share is expressed in mass percent. c R value calculated from the
molar ratio of C16 to C14.
d Determined by 1H NMR for solid product in d10-p-xylene at 110 °C. e Ratio of saturated (methyl) to unsaturated
(vinyl) chain ends. f Quoted per hour for all entries, including shorter runs (#10 and #11), and rounded to three significant figures for
larger numbers. g Precatalyst activated with MAO in toluene prior to ethylene exposure under an inert atmosphere.
presence of ethylene (entry 2). This latter, improved
method of activation was employed in subsequent runs
for all precatalysts15 (the effect of in situ activation is
negligible for the iron-based catalysts). It is therefore
clear that introduction of an o-trifluoromethyl group
boosts the activities of cobalt catalysts by 2 orders of
magnitude, whereas the activities of iron catalysts are
increased by up to a factor of 6.
In the cobalt series, catalyst activity increases in the
series 2a < 2b < 2c, with values of 9600, 10 500, and
15 300 g mmol-1 h-1 bar-1, respectively. The overall
trend in activity implies that the addition of extra
fluorine atoms to the ligand framework further en-
hances monomer turnover rates in addition to the
already beneficial impact of the trifluoromethyl group.
In this respect, an o-F appears to have a more potent
influence than a p-F group, although this effect may be
more than purely electronic in nature. In terms of the
cocatalyst, which also serves as a poison scavenger,
there are no major differences between MAO and
coMAO: the latter is somewhat better than the former
for precatalysts 2a,c (entry 5 vs 4, 9 vs 8), whereas the
opposite appears to be the case for precatalyst 2b (entry
6 vs 7).
achieve higher turnover numbers than those activated
with MAO, an effect that appears more pronounced for
the more open systems 4a,b.
Significantly higher activity figures may be obtained
for shorter runs, given the greater catalyst activity in
the early stages of the run (see below), and particularly
if the activity figures are extrapolated to 60 min.
Relative to a 60 min ethylene polymerization run with
catalyst 2c/coMAO (entry 9), activity figures are boosted
to 54 000 and 129 000 g mmol-1 h-1 bar-1 (entries 10
and 11) when reducing the run time to 10 and 2 min.
The latter figure corresponds to an activity of 2.2 × 106
g (g of Co)-1 h-1, the equivalent of a turnover fre-
quency in excess of 76 000 min-1 at 1.0 bar of ethylene
pressure. To the best of our knowledge, this currently
represents the most active group 9 olefin polymeri-
zation catalyst. Indeed, ligand system c is the first
example where a cobalt bis(imino)pyridine catalyst is
more active for C-C bond formation than its iron
counterpart.
Examination of the kinetic profiles (see Supporting
Information) for ethylene oligomerization/polymeriza-
tion catalysts derived from 2b, 3 and 4b, 5 clearly
illustrate how the fluorinated catalysts 2b/MAO and
4b/MAO outperform their respective benchmarks 3
and 5 in terms of both stability and peak activity.
The lifetimes of the fluorinated derivatives exceed those
of the benchmark systems by a factor of approximately
4, whereas, in terms of peak activity, 2b/MAO and
4b/MAO may be estimated to consume about 50% more
ethylene than either 3/MAO or 5/MAO. The kinetic
profiles also shed light on the higher activity of the
o-CF3-containing catalysts at shorter reaction times, as
described for 2c above, since ethylene uptake peaks
around 5 min into the run, with catalyst death ensuing
after about 30 min under the conditions employed. In
view of the fact that catalyst activity diminishes over
time and that Mn does not vary over time (entries 9-11),
For the iron catalysts, the productivity order is 4c <
4a < 4b, with activities of 7220, 15 400, and 17 300 g
mmol-1 h-1 bar-1, respectively. For this metal, the
effects of ligand and cocatalyst are quite different from
those of the cobalt catalysts. Placement of a fluorine
atom in the second ortho position of the phenyl ring
actually reduces catalyst activity, whereas a p-F in-
creases it. Also, precatalysts activated with coMAO
(15) Activation of the cobalt complex bearing 2,6-difluorophenyl
groups attached to the imine moiety (R1 ) R2 ) F, R3 ) H in Scheme
1)sindependently prepared and recently reported by Qian and co-
workers8saccording to this protocol showed it to be active for ethylene
oligomerization. After treatment with MAO, its activity was deter-
mined as 4 g mmol-1 h-1 bar-1 in a 60 min run at 1.0 bar of ethylene
pressure (R ) 0.71).

Cobalt Bis(imino)pyridine Catalysts
Organometallics, Vol. 24, No. 2, 2005 283
Table 2. Oligomerization of r-Olefins by
it is clear that 2c/coMAO does not represent a highly
controlled system for ethylene polymerization, unlike
the effect of fluorinated ligands on some early-transi-
tion-metal systems reported recently.16,17
Precatalyst 2ca
activity
dimer
product share mmol-1
no. R-olefin feed (°C) mass (g)b (%)c h-1
g
g
As has been shown for other ethylene oligo- and
polymerization catalysts, the steric influence exerted by
the groups placed in the ortho positions of the N-aryl
groups has, among other contributing factors, a pro-
found influence on the rate of chain termination relative
to chain propagation and, hence, molecular weight of
the generated poly(ethylene)s. This is also the case for
the complexes reported here. For cobalt complexes 2a-c
relative to 3, replacement of the o-CH3 group with an
o-CF3 group leads to products with increased molecular
weights. For precatalysts 2a and 2b, oligomers following
a Schulz-Flory distribution of higher R-value than
benchmark 3 represent the main product in combination
with some solid PE of low molecular weight. In ac-
cordance with expectations, the introduction of an
additional ortho substituent in precatalyst 2c further
raises the product molecular weight, the principal
product now being solid poly(ethylene) with an Mn value
of 2400 Da. The polymers produced from all these
catalysts are linear PEs containing both saturated and
unsaturated end groups in a 1:1 ratio and, therefore,
simply represent R-olefins of elevated molecular weight.
The equimolar amounts of methyl and vinyl end groups
in the polymer are consistent with chain termination
arising by â-H transfer exclusively.
Similarly, for the iron complexes 4a-c, introduction
of the trifluoromethyl substituent increases product
moleceular weights to the extent that the vast majority
of the product is a solidsas opposed to the liquid
oligomers made by 5sand consist of linear poly-
(ethylene)s with Mn values ranging from 2100 to 4300
Da. The greater quantity of saturated chain ends to
unsaturated chain ends is readily accounted for by chain
transfer to aluminum as a second termination chain
mechanism (in addition to â-H transfer). This catalyst
behavior is also observed in nonfluorinated systems,
where it has been studied in some detail.1c
run
θ
(g of
Co)-1 h-1
20d C3H6 (1.0 bar) -20
21 C3H6 (1.0 bar)
22 C3H6 (1.0 bar) -20
0.077
0.369
2.796
0.131
0.371
0.160
0.544
0.894
1.225
88
81
72
92
90
92
90
86
83
15
261
2 510
19 300
887
2 510
1 090
3 690
6 070
8 310
0
148
1140
52
23 C4H8 (10 g)
24 C4H8 (10 g)
25 C6H12 (10 g)
26 C6H12 (10 g)
27 C6H12 (neat)
28 C6H12 (neat)
0
-20
0
-20
0
-20
148
64
217
358
490
a General conditions: a Schlenk tube containing a suspension
of precatalyst 2c (0.0050 mmol) in either 100 cm3 of toluene (runs
1-7) or 50.0 cm3 of 1-hexene (runs 8 and 9) was cooled to the
desired temperature, exposed to the monomer (runs 1-7), and
subsequently activated with MAO (Al:Co ) 1000:1). Runs were
terminated after 30 min by addition of an excess of 1 M HCl (aq).
b Products quantified by GC, using 2,2,4,4,6,8,8-heptamethyl-
nonane as internal standard. c Dimer share of total product mass.
d Performance of precatalyst 3 at 0.020 mmol loading under
otherwise identical conditions.
The results of propene, 1-butene, and 1-hexene oli-
gomerization runs are collected in Table 2. These
substrates are oligomerized by 2c/MAO with high
activities: up to 19 000, 2510, and 8310 g (g of Co)-1
h-1, respectively,19 over 30 min runs at -20 °C, despite
the reduced temperatures employed to dissolve/con-
dense adequate amounts of the gaseous monomers. The
performance of 2c relative to 3 under identical condi-
tions mirrors earlier observations, in that the fluori-
nated catalyst is about 2 orders of magnitude more
active than its nonfluorinated analogue. Regardless of
R-olefin feedstock, linear dimers are the principal
products in every case (72-92%), whereas the trimers
are always branched in the case of products derived
from 1-butene and 1-hexene, and partly so for products
derived from propene.
For propene, the outstanding activity increases rela-
tive to the benchmark system 3/MAO (entries 20 vs 22)
are accompanied by significantly higher product selec-
tivities, with 91% of the C6 fraction being 1-hexene for
2c/MAO (benchmark 3, 65%). At 0 °C, the 1-hexene
selectivity within the C6 fraction is slightly lower (87%)
for the fluorinated catalyst, but given the lower con-
centration of tri- and tetramers, a larger proportion of
the overall product fraction is 1-hexene (72%) relative
to the run at -20 °C (66%) as detailed in Table 3. The
only other products within the C6 fraction are 2-hexenes,
where the E isomer prevails over the Z isomer by about
2:1 (Figure 2, Table 3). None of the thermodynamically
most stable dimer (3-hexene) is observed by either GC
or NMR, consistent with 1- and 2-hexenes arising by a
(1,2)- followed by a (2,1)-propene insertion. The absence
of 3-hexene confirms that 1-hexene is not isomerized by
the catalyst. Indeed, examination of the residual feed-
stock from 1-hexene oligomerization runs 25 and 26
revealed it to be 99.6-99.7% 1-hexene, with e0.4%
representing the thermodynamically more stable E
Oligomerization of Propene, 1-Butene, and
1-Hexene
In an initial screen, precatalysts 2a,c (Co), as well as
precatalysts 4a,c (Fe), were tested for their activity in
propene oligomerization at 0 °C and 1.0 bar, following
activation with MAO (Al:M ) 1000). After 30 min, the
runs were terminated by addition of aqueous hydro-
chloric acid and worked up. Surprisingly, no product
was formed from either of the iron catalysts. This result
stands in contrast to the observed capability of non-
fluorinated iron catalysts to either polymerize or oligo-
merize propene, when more or approximately equally
bulky, but electron-releasing, aryl substituents are em-
ployed.18 The cobalt derivatives 2a,c, however, formed
oligomers. The superior activity of the catalyst derived
from 2c/MAO led us to investigate this system in more
detail.
(16) (a) Tian, J.; Hustad, P.; Coates, G. W. J. Am. Chem. Soc. 2001,
123, 5134. (b) Saito, J.; Mitani, M.; Mohri, J.; Yoshida, Y.; Matsui, S.;
Ishii, S.; Kojoh, S.; Kashiwa, N.; Fujita, T. Angew. Chem., Int. Ed. 2001,
40, 2918.
(17) Reinartz, S.; Mason, A. F.; Lobkovsky, E. B.; Coates, G. W.
Organometallics 2003, 22, 2542.
(18) Small, B. L.; Brookhart, M. Macromolecules 1999, 32, 2120.
(19) The activities for R-olefin dimerization/oligomerization are
expressed in g (g of Co)-1 h-1 to aid comparison with other systems in
the literature, where this is a commonly encountered unit. For ease of
reference with the ethylene polymerization data, catalyst activities are
also quoted in units of g mmol-1 h-1 bar-1 in Table 2.

284 Organometallics, Vol. 24, No. 2, 2005
Tellmann et al.
analysis by GC and NMR revealed the dimers to be
almost exclusively linear (98%-99%), internal olefins.
The 1-hexene trimers, however, are branched without
exception, as in the case of 1-butene. All linear 1-hexene
dimers are internal olefins with the double bond not
located in any of the 1-, 2-, and 6-positions of the chain,
as determined by NMR, leaving the isomers of 3-, 4-,
and 5-dodecene as possible reaction products. From a
mechanistic perspective (see Discussion) the 3-alkene
may be ruled out, to leave 4- and 5-dodecene as the
predicted 1-hexene dimers to be formed by action of the
catalyst 2c/MAO. In terms of double-bond geometries,
a significant bias exists toward E isomers (E/Z > 3),
based on the relative peak heights in the 13C{1H} NMR.
Figure 2. GC trace of propylene dimers made by the
MAO-activated cobalt complex 2c (91% 1-hexene, run #22,
Table 3).
isomers of 2- and 3-hexene. This implies that the oligo-
merization products, as well as the monomer feedstock,
are not prone to isomerization by the fluorinated cobalt
catalyst. This aspect of the system is particularly note-
worthy, since many dimerization catalysts also isomer-
ize substantial amounts of the kinetic product to the
thermodynamic product. This frequently represents an
undesired side reaction, given the higher economic and
environmental value of terminal (and/or linear) olefins
with respect to internal (and/or branched) olefins as
precursors to important chemical derivatives.20
Higher propene oligomers (n > 2) are only observed
up to n ) 3 for the benchmark system (run 20, Table
3), whereas with the catalyst 2c/MAO, tetramers are
always detected, albeit in small quantities. Both the C9
and C12 fractions consist entirely of internal olefins. In
the case of 2c/MAO, both fractions contain some linear
products, which become increasingly significant at lower
reaction temperatures and constitute the majority (56%)
of the trimer fraction at -20 °C (run 22, Table 3). The
benchmark system, in contrast, produces a far smaller
proportion (8 mol %) of trimeric product under these con-
ditions, 94% of which are branched (run 20, Table 3).
By variation of the standard run time from 10 to 60 min
(runs 22 and 29-31), we were able to establish that the
higher propene oligomers (n > 2) are formed in larger
proportion in the later stages of a run and that the
linearity of the higher oligomers (cf. C9 and C12
fractions) increases with longer run times. This issue,
as well as the occurrence of linear trimers and tetra-
mers of propene, will be revisited in the mechanistic
discussion.
For 1-butene, analysis of the products by GC and
NMR indicated the dimers to be highly linear olefins,
consisting of almost equimolar amounts of 2- and
3-octenes, in a 47:53 ratio for the experiment carried
out at 0 °C and a 46:54 ratio for the experiment per-
formed at -20 °C. In each case, the E isomer predomi-
nates, as determined quantitatively by GC for the
2-octenes (supported by 13C{1H} NMR), or qualitatively
for the 3-octenes (by 13C{1H} NMR). Relative to the
results obtained for propene oligomerization, the cata-
lyst activities are lower and trimers are also less easily
formed. Both of these factors may be rationalized in
terms of the additional steric requirements for insertion
of a butene molecule relative to propene. Furthermore,
catalyst regioselectivity is also diminished.
Discussion
The presence of an o-trifluoromethyl group in the aryl
substituents of cobalt and iron bis(imino)pyridine cata-
lysts clearly gives rise to greater olefin oligomerization/
polymerization activities than for their nonfluorinated
relatives. Enhanced performance is seen not only in
higher peak activities, presumably a consequence of the
enhanced electrophilicity of the metal centers in the
fluorinated derivatives, but also in longer catalyst
lifetimes, suggesting that the trifluoromethyl group
significantly improves catalyst stability. The most im-
pressive activity increases are observed for the cobalt
catalyst 2c, which contains one o-CF3 and an o-F unit.
This system is even more active than its iron counter-
part in ethylene polymerization, reaching >100 000 g
mmol-1 h-1 bar-1
.
Propene, 1-butene, and 1-hexene are oligomerized by
2c/MAO with activities much higher than for their
nonfluorinated relatives. Highly linear dimers are mostly
obtained in each case, the remainder being trimeric and
tetrameric products. The principal product of propene
dimerization is 1-hexene (60-73% of total), whereas for
1-butene and 1-hexene internal olefins are obtained (E
isomers predominate). Furthermore, the cobalt catalyst
2c/MAO effectively does not isomerize 1-hexene under
the reaction conditions. These findings compare favor-
ably, in terms of monomer isomerization, with a recent
report7 by Small on cobalt bis(imino)pyridine catalysts
stabilized by electron-rich ligands. In addition, the
benchmark system 3/MAO is outperformed in terms of
turnover rates and dimer selectivity. Under the condi-
tions employed by Small,7 however, the linearity of the
propene tri- and tetramers generated by 3/MAO is much
greater (>94%) than is found by us (C9 fraction: 6.3%)
in toluene at -20 °C (Table 3). The higher oligomers (n
> 2) also account for a larger share of the total product
weight than observed by us, presumably due to much
higher monomer concentrations (liquid monomer versus
gaseous/partially dissolved monomer).
Mechanism of r-Olefin Oligomerization
The mechanism governing the formation of the oli-
gomeric products must satisfy the following observa-
tions: (a) all R-olefin dimers are linear; (b) 3-hexenes
are not observed in the propene dimer fraction; (c)
The catalyst generated from 2c/MAO oligomerizes
1-hexene with activities ranging from 1090 g (g of Co)-1
h-1 at 0 °C and relatively modest monomer concentra-
tion to 8310 g (g of Co)-1 h-1 at -20 °C. The product
(20) Cornils, B., Herrmann, W. A., Eds. Applied Homogeneous
Catalysis with Organometallic Compounds; Wiley-VCH: Weinheim,
Germany, 2002.

Cobalt Bis(imino)pyridine Catalysts
Organometallics, Vol. 24, No. 2, 2005 285
Table 3. Analysis of Propene Oligomers Formed by Precatalyst 2c: Effect of Temperature and Run Timea
product distribn (C6 fraction)b
molar ratioc
run time
(min)
activity (g
1-hexene (E)-2-hexene (Z)-2-hexene
C9 linearity C12 linearity
run no. θ (°C)
(g of Co)-1 h-1
(%)
(%)
(%)
C6/C9 C9/C12
(%)
(%)
20d
21
29
30
22
31
-20
0
-20
-20
-20
-20
30
30
10
20
30
60
261
2 510
27 200
27 300
19 300
10 900
65
87
91
90
91
90
22
9
13
4
11.2:1
7.34:1
6.73:1
5.57:1
4.63:1
3.98:1
n/a
6.3
33
n/a
12
12
19
28
34
10.7:1
12.3:1
9.42:1
7.38:1
6.33:1
6
3
35
6
3
45
6
3
56
6
3
63
a Products analyzed by GC (checked against references where possible) and NMR. b Percentages expressed relative to total weight of
C6 fraction; sum may be less than unity because of rounding errors. c Molar ratios of fraction totals. d Performance of precatalyst 3 at
0.020 mmol loading under otherwise identical conditions.
Scheme 2. Proposed Mechanism for the
catalyst then passes through one product-generating
cycle and terminates the chain by â-H elimination, to
give a cobalt hydride-like species, formulated here as
“LCoH”. However, it should be noted that this cobalt
hydride complex “LCoH” must differ from the reported
neutral, diamagnetic Co(I) hydride complex,23 because
of the inability of the latter to form C-C bonds. Whether
this is due to the charge, the spin state, the oxidation
state, or the participation of a secondary agent, or even
a combination thereof, is uncertain.
Dimerization of r-Olefins
The putative cobalt hydride species “LCoH” then
undergoes two consecutive insertions, followed prefer-
entially by chain termination via â-H elimination. The
(1,2)-(2,1) order of monomer additions adequately ac-
counts for the observation of the linear olefin dimers 1-
and 2-hexene for propene and 2- and 3-octene for
1-butene, as determined by GC and NMR. Assuming
that the same mechanism is adopted for 1-hexene, the
dimerization products ought to be a mixture of 4- and
5-dodecene. This is consistent with the experimental
observations. Similar conclusions about the underlying
mechanism have been proposed for the nonfluorinated
oligomerization catalysts.7
Despite the catalyst’s pronounced preference for
R-olefin dimerization, insertion of a third monomer unit
into an R-branched cobalt alkyl must occur, albeit
slowly, to generate the branched trimers (or tetramers)
detectable by GC. The ratio of the molar sum of all the
branched higher n-mers (n > 2) to the sum of dimers
would therefore represent the underlying relative rates
of further monomer insertions into an R-branched cobalt
alkyl (chain propagation) and â-H elimination/transfer
from this species (chain termination).24 From the GC
data, values of the ratio kprop/kterm are estimated to be
ca. 1/9 for propene, 1/9 for 1-butene, and 1/11 for
1-hexene in toluene solution. The similarity of the
values for each medium supports the assumption of a
single fundamental mechanism for formation of branched
n-mers (n > 2).
1-butene and 1-hexene dimers formed by 2c/MAO are
invariably internal olefins; (d) a mixture of branched
and linear n-mers (n > 2) is observed during propene
oligomerization; (e) no 1- or 2-nonenes occur among the
propene trimers; (f) linearity in the higher propene
n-mers (n > 2) increases with time; (g) the n-mers (n >
2) of 1-butene and 1-hexene are always branched; (h)
product distributions do not follow a standard math-
ematical distribution (e.g. Schulz-Flory, Poisson); (i) the
degree of product isomerization is very low (cf. propene
dimers; 1-hexene feedstock).
Since no branched dimeric products are observed,
dimer formation must occur first by a (1,2)-insertion,
followed by (2,1)-insertion. Furthermore, the absence
of a distribution of oligomers indicates that product
formation for higher oligomers (n > 2) does not neces-
sarily occur via consecutive insertions alone. This con-
clusion is supported by the fact that the molar quanti-
ties of higher oligomers diminish more quickly than in
a geometric series, as illustrated for the C6/C9 and
C9/C12 mole ratios of the propene oligomers. From
these two observations it is plausible to conclude that,
following insertion of a second monomer unit in a (2,1)-
fashion, an R-branched cobalt alkyl is formed which
preferentially undergoes â-H elimination, as opposed to
further insertion of another R-olefin.
Clearly, it is impossible to form linear propene trimers
and tetramers by successive insertions, since monomer
insertion into the cobalt-hexyl species § (Scheme 2),
(21) Gibson, V. C.; Humphries, M. J.; Tellmann, K. P.; Wass, D. F.;
White, A. J. P.; Williams, D. J. Chem. Commun. 2001, 2252.
(22) 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.
(23) (a) Gibson, V. C.; Tellmann, K. P.; Humphries, M. J.; Wass, D.
F. Chem. Commun. 2002, 2316. (b) Tellmann, K. P.; Humphries, M.
J.; Rzepa, H.S Gibson, V. C. Organometallics 2004, 23, 5503.
(24) Chain termination normally occurs exclusively by â-H elimina-
tion/transfer in cobalt bis(imino)pyridine catalysts, regardless of ligand
size.
Consideration of these points leads us to propose the
mechanism for R-olefin dimerization shown in Scheme
2. Following initial treatment of the cobalt(II) dihalide
precursor with MAO, an active catalyst is obtained
according to a pathway recently outlined.21,22 This

286 Organometallics, Vol. 24, No. 2, 2005
Tellmann et al.
Scheme 3. Plausible Mechanistic Pathways for
Generation of Propene Trimers
probably greater for propene than for 1-hexene on steric
grounds (cf. relative oligomerization activities of C3H6
vs C6H12).
By direct analogy, the linear propene tetramers must
be formed by dimerization of the 1-hexene, formed by
initial dimerization of the monomer. Indeed, the fact
that 1-hexene always constitutes the principal reaction
product in the oligomerization of propene with 2c/MAO
implies that it is consumed significantly more slowly
(for formation of linear tri- and tetramers) than it is
produced (by simple dimerization). It is therefore pos-
sible to conclude that propene inserts far more readily
into a cobalt 1-propyl species than 1-hexene.
Summary and Conclusion
Exceptionally high productivities have been achieved
in ethylene oligomerization/polymerization reactions
using bis(imino)pyridine complexes of iron and cobalt
containing trifluoromethyl groups. These judiciously
fluorinated catalysts show greater activities than their
nonfluorinated relatives as a consequence of both higher
peak activities and longer catalyst lifetimes, suggesting
that the trifluoromethyl group not only increases the
electrophilicity of the metal centers but also significantly
improves catalyst stability. The most impressive activity
increases are observed for the cobalt catalyst 2c, which
contains one o-CF3 and an o-F unit. This system is even
more active than its iron counterpart in ethylene
whether in (1,2)- or (2,1)-fashion, must lead to a
branched product (Scheme 3, path 1).25 By logical
consequence, these olefins have to be produced by
another mechanism that has to operate, by necessity,
in stepwise fashion. This view is also supported by the
increasing degree of linearity in the propene trimer and
tetramer fractions, the longer the run time.
polymerization, reaching >100 000 g mmol-1 h-1 bar-1
.
The fluorinated catalysts are also shown to dimerize
propene, 1-butene, and 1-hexene with high selectivities
via a mechanism dominated by (1,2)- followed by (2,1)-
insertion steps. Combinations of chain growth and step
growth account for the formation of linear trimers and
tetramers of propene. Significantly, no isomerization of
terminal olefins to internal olefins is seen.
Two reaction sequences that afford linear propene
trimers involve the codimerization of propene and
1-hexene, following the same regioselectivity as estab-
lished for homodimerization, and differ only by the order
of addition (cf. Scheme 3). In path 2, propene is inserted
first in a (1,2)-fashion, to give a cobalt 5-nonyl species,
whereas by path 3, propene is inserted (2,1) after
1-hexene, to afford a cobalt 2-nonyl species. Following
chain termination by â-H elimination, these organoco-
balt complexes yield 4-/5-nonene (path 2) or 1-/2-nonene
(path 3). Of the two possible products, only the former
is observed by GC and NMR, and hence the mechanistic
pathway 3 may be ruled out. This preferred order of
addition is consistent with the higher relative concen-
tration of propene at any stage during the reaction
(being present in excess); furthermore, the ease of
insertion into a putative cobalt hydride like species is
Acknowledgment. We thank BP Chemicals Ltd for
funding and Crompton Corp. for generous donation of
MAO and coMAO. We also thank Jane Boyle and Dick
Sheppard for their helpful assistance in NMR spectros-
copy.
Supporting Information Available: Kinetic profiles
showing monomer uptake for ethylene oligomerization/poly-
merizations, experimental procedures for the preparation of
ligands and complexes, as well as the methods employed for
olefin oligomerization/polymerization, and solid-state struc-
tures and crystallographic details for complexes 2a-c and
4a,b. This material is available free of charge via the Internet
at http://pubs.acs.org.
(25) At least in the absence of chain walking, a phenomenon only
observed in certain group 10 complexes, but for which no evidence is
seen in 2c/MAO or closely related group 9 catalysts. For leading
references concerning chain walking in group 10 metal complexes,
see: Ittel, S. D.; Johnson, L. K.; Brookhart, M. Chem. Rev. 2000, 100,
1169 and references therein.
OM049297Q