Comparative Dimerization of 1-Butene mith a Variety of Metal Catalysts, and the Investigation of a New Catalyst for C-H Bond Activation

Article abstract of DOI:10.1002/chem.200304945

Catalytic dimerization of 1-butene by a variety of catalysts is carried out, and the products are analyzed by gas chromatography and mass spectrometry. Catalysts based on cobalt and iron can produce highly linear dimers, with the cobalt-based dimers exceeding 97% linearity. Catalysts based on vanadium and aluminum prefer to make branched dimers, which are most often methyl-heptenes in the case of vanadium and almost exclusively 2-ethyl-1-butene in the case of aluminum. The vanadium catalyst also produces substantial amounts of dienes and alkanes, suggesting a competing hydrogenation/dehydrogenation pathway that appears to involve vinyl C-H bond activation. Nickel catalysts are generally less selective than those based on iron or cobalt for making linear dimers, but they can make dimers with 60% linearity. The major by-products for the nickel systems are trisubstituted internal olefins. An important side reaction that must be considered for dimerization reactions is 1-butene isomerization to 2-butene, which makes recycling the butene difficult for a linear dimerization process. Aluminum, iron, and vanadium systems promote very little isomerization, but nickel and cobalt systems tend to isomerize the undimerized substrate heavily.

Full text of DOI:10.1002/chem.200304945

FULL PAPER
Comparative Dimerization of 1-Butene with a Variety of Metal Catalysts, and
À
the Investigation of a New Catalyst for C H Bond Activation
Brooke L. Small*[a] and Roland Schmidt[b]
Abstract: Catalytic dimerization of 1-
butene by a variety of catalysts is car-
ried out, and the products are analyzed
by gas chromatography and mass spec-
trometry. Catalysts based on cobalt and
iron can produce highly linear dimers,
with the cobalt-based dimers exceeding
97% linearity. Catalysts based on vana-
dium and aluminum prefer to make
branched dimers, which are most often
methyl-heptenes in the case of vanadi-
um and almost exclusively 2-ethyl-1-
butene in the case of aluminum. The
vanadium catalyst also produces sub-
stantial amounts of dienes and alkanes,
suggesting a competing hydrogenation/
dehydrogenation pathway that appears
they can make dimers with 60% linear-
ity. The major by-products for the
nickel systems are trisubstituted inter-
nal olefins. An important side reaction
that must be considered for dimeriza-
tion reactions is 1-butene isomerization
to 2-butene, which makes recycling the
butene difficult for a linear dimeriza-
tion process. Aluminum, iron, and va-
nadium systems promote very little iso-
merization, but nickel and cobalt sys-
tems tend to isomerize the undimerized
substrate heavily.
À
to involve vinyl C H bond activation.
Nickel catalysts are generally less se-
lective than those based on iron or
cobalt for making linear dimers, but
À
Keywords: C H
dimerization
activation
homogeneous
¥
¥
catalysis ¥ vanadium
Introduction
clude branched dimer formation, linear dimer formation, 1-
butene isomerization, and butene oligomerization.
The dimerization of olefins by homogeneous catalysts is an
area of academic as well as commercial importance.[1] For
example, ethylene dimerization to 1-butene is a potentially
attractive way to produce polyethylene comonomer, espe-
cially in regions where other sources of 1-butene are
scarce.[2] Dimerization of propylene and 1-butene is prac-
ticed widely, with the generally branched dimers finding ap-
plication as plasticizer alcohol precursors or gasoline addi-
tives. The dimerization of 1-butene is a particularly interest-
ing area to study, since there are a number of different proc-
esses that can happen during the reaction, some of which in-
Recent reports by Wasserscheid et al.[3] and by ourselves[4]
illustrate the desirability of catalysts that promote linear di-
merization. While Keim and Wasserscheid reported that
nickel complexes with fluorinated acetylacetonate (acac) li-
gands dissolved in buffered ionic liquids catalyze linear di-
merization (TONs<3000, 50 70% linearity), we reported
that tridentate pyridine bisimine iron complexes, when acti-
vated with alumoxanes, are highly active and selective for
this process (TONsꢀ75000, 65 80% linearity). This unpre-
cedented combination of activity and selectivity prompted
an investigation of several other catalysts and a comparison
of their tendencies regarding olefin dimerization. Herein are
reported the results of this study.
[a] Dr. B. L. Small
ChevronPhillips Chemical Company LP, Kingwood Technology
Center
1862Kingwood Drive, Kingwood, TX 77339 3097 (USA)
Fax : (+1)281-359-6515
E-mail: smallbl@cpchem.com
Results and Discussion
[b] Dr. R. Schmidt
Although a number of different catalysts have been report-
ed for dimerizing olefins, the catalyst precursors utilized in
this work were chosen because they provide ample room for
comparison between well-known and newly emerging homo-
geneous systems. These six complexes are shown in
Scheme 1. Complexes 1 3, all of which may be activated by
alkyl alumoxanes, compare a series of pyridine bisimine cat-
alysts, using the recently reported iron-based system as a
useful benchmark. Diimine nickel complex 4, on which
ConocoPhillips, Bartlesville Technology Center,
83-F, BTC, Bartlesville, OK 74004 (USA)
Fax : (+1)918-662-2445
E-mail: roland.schmidt@conocophillips.com
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author. 1H NMR of vana-
dium-based 1-butene dimer; GC of vanadium-based 1-butene dimer,
1-butene trimer, and 1-hexene dimer with signals identified from MS
data; GC/MS of vanadium-based propylene dimer with signals identi-
fied.
1014
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200304945
Chem. Eur. J. 2004, 10, 1014 1020

1014 1020
longer to improve the conversion. Table 1, which presents
the dimerization data, allows a number of comparisons to
be made regarding catalyst selectivity and productivity.
In terms of activity, the iron catalyst 2 is by far the high-
est, outpacing the other systems by approximately an order
of magnitude. This trend was not surprising, especially upon
comparison of the iron catalyst to its cobalt analogue 1. Pre-
vious reports of these systems for ethylene oligomerization[9]
have shown that iron is significantly more active than
cobalt. However, the activity of cobalt catalyst 1 compared
very favorably to that of catalysts 3 5. In fact, complex 1 is
significantly more active than the nickel or vanadium sys-
tems studied.
Along with the dimerization reaction, several of the cata-
lysts promoted the side reaction of butene isomerization.
For both nickel systems and for cobalt, isomerization is
rapid, and the amount of isomerized butene tends to exceed
the amount of dimer formed. In a process where linear di-
merization is desired, isomerization is detrimental because
incorporation of internal olefins, if they are able to incorpo-
rate, will cause the formation of branched products. Cata-
lysts 2 and 6 only promote modest isomerization, with the
highly active iron complex 2 preferring dimerization to iso-
merization by a factor of >20:1. Vanadium system 3[10] does
not heavily isomerize the starting olefin either, but it pro-
motes a unique hydrogen transfer process that leads to sev-
eral unexpected products. The mechanism of catalyst 3 will
be discussed in this report.
All of the catalysts 1 6 (Table 1) are quite selective for
producing dimers preferentially to higher oligomers. As a
first approximation, assuming that the sterically hindered di-
and trisubstituted dimer products do not react further (ex-
cluding possible isomerization) once they have undergone
chain transfer from the metal, the degree of dimer forma-
tion is simply a factor of the catalysts’ relative rates of prop-
agation and chain transfer. Thus, in complexes of type 1 5,
in which decreased steric bulk of the ligands has been
shown to promote oligomerization relative to polymeriza-
tion, the propensity for forming dimers is somewhat expect-
Scheme 1. Precatalysts for the dimerization of 1-butene.
Brookhart and Svejda have disclosed variations for propyl-
ene dimerization,[5] is included to extend the comparative
study of nitrogen-based ligands to nickel. Complex 5, which
is activated with diethylaluminum chloride, resembles the
types of catalysts used in the IFP Dimersol Process,[6] and is
included as an example of a relatively non-selective catalyst.
Catalyst 6, which is known to selectively make vinylidene
(2-alkyl-1-alkene) dimers,[7] is incorporated to illustrate the
diverse products that can be made from olefin dimerization.
To test each of the catalyst 1 6 for dimerization activity
and selectivity, a 500-mL Zipperclave reactor was used, and
an inert gas head pressure of at least 100 psig was used to
drive the butene into the reactor as a liquid and to keep the
butene liquefied during reaction heating.[8] The reactions
were typically run for three hours, but catalyst 3 was run
Table 1. Results for the dimerization of 1-butene by catalysts 1 6.
Catalyst
Reaction
Yield
%
Conv.
%
Dim.
%
%
%
Dimer
Total
Dim./
isom
LD[a]
MBD[b]
DBD[c]
branch
conditions
[g]
index (BI)[d]
TON[e] ratio[f]
1/
2 5 mg1, Al:Co=500:1, 250 gC4, 308C,
3 h
97.7 39
99
97
3
trace
0.03
38000 0.7:1
MMAO[g]
2/MMAO 10 mg 2, Al:Fe=500:1, 250 gC4, 308C,
153
61
45.1 18
31.6 13
26.1 10
8270
84
30
trace
0.30
20.66
0.38
147000
>20:1
>30:1
3 h
3/MMAO 25 mg 3, Al:V=500:1, 250 gC4, 308C,
36[h]
62[h]
trace
16000
16 h
4/MMAO 25 mg 4, Al:Ni=500:1, 250 gC4, 308C,
726238
84
8300 1.1:1
3 h
5/DEAC [i] 2 5 mg5, Al:Ni=500:1, 250 gC4, 608C,
55
9
31
90
13
1
0.57
0.91
12200 <0.1:1
30 >20:1
3 h
6
16.6 g 6, 700 gC4, 1200 psig N2, 2 008C,
140
20
98
3 h
[a] LD=linear dimer. [b] MBD=mono-branched dimer; further analysis included in Table 2. [c] DBD=di-branched dimer. [d] BI=average number
of branches per molecule in the C8 product fraction. [e] TON=moles of butene dimerized or oligomerized per mole of pre-catalyst. [f] Approximate
ratio of moles of 1-butene dimerized or oligomerized to moles of butene isomerized; since this value may change with changing substrate concentration,
it is only a general expression of a catalyst system×s tendency to isomerize or dimerize a-olefins. [g] MMAO=isobutyl-modified methylalumoxane, pur-
chased from Akzo. [h] Refer to the discussion on catalyst 3 and to Table 2for more details on these numbers. [i] DEAC =diethylaluminum chloride.
1015
Chem. Eur. J. 2004, 10, 1014 1020
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

B. L. Small et al.
FULL PAPER
Table 2. Analysis of mono-branched dimers.
ed. The cobalt catalyst 1 is the most selective, producing
only traces of higher butene oligomers.
Catalyst
% Vinylidene
% Internal
% Internal
disubstituted
methyl-heptene
(2-ethyl-1-hexene)
trisubstututed
methyl-heptene
The composition of the C8 products made by catalysts 1 6
illustrates the diverse pathways for dimerizing a-olefins.
Table 1 breaks the C8 products from each reaction into
linear, monobranched, and dibranched components. The
branch index simply refers to the average number of branch-
es per C8 molecule. As is seen from the table, which was
composed using GC/MS data, the iron system produces pre-
dominantly linear octenes, but the octenes made by the
cobalt catalyst are 97% linear. We have recently reported
additional studies involving highly selective cobalt catalysts
for linear dimerization and isomerization.[11] Nickel catalysts
4 and 5 are also moderately selective for linear dimer forma-
tion, although the branch index of the bisphosphine complex
5 is inflated by its formation of dibranched products. In ad-
dition, the high levels of 2-butenes made by 4 and 5 tend to
make them poor candidates for linear dimerization, since
butene recycle or high substrate conversion (if possible)
would lead to significant increases in product branching.
Catalysts 3 and 6, on the other hand, make predominantly
branched dimers, albeit with different structures from each
other.
To more carefully analyze the branched dimers, Table 2
was constructed, and all of the monobranched dimers were
analyzed by GC/MS. While the iron and cobalt catalysts
make primarily 5-methyl-2-heptenes and 5-methyl-3-hep-
tenes, the vanadium system produces a mixture of methyl-
heptenes, vinylidene (2-ethyl-1-hexene), and methyl-heptane.
The origin of this saturated species will be discussed in
detail. The nickel systems make predominantly trisubstitut-
ed internal olefins, while the aluminum catalyst makes
almost exclusively 2-ethyl-1-hexene.
1/MMAO
2/MMAO
3/MMAO
4/MMAO
5/DEAC
6
trace
1.5
trace
0.5
trace
90.7[b]
77.8[b]
trace
>99
98.0
75.7[a]
9.3
24.3[a]
trace
trace
98.21.8
22.0
[a] Only the mono-branched, mono-olefins were included in this analysis
(products made by proposed complexes and 12 in Scheme 2).
[b] Nickel catalysts 4 and 5 are capable of isomerizing the products.
9
sumed to constitute only a negligible percentage of product
formation. The various products shown in Scheme 2can be
matched to the different dimers listed in Tables 1 and 2,
thus allowing the data in the tables to be presented in a
mechanistic format.
Upon analysis of the dimers made by vanadium catalyst 3,
the mass spectral data indicated a substantial presence of
both dienes and saturated species in the C8 products. Ap-
proximately 28 mol% of the total C8 products were identi-
fied as dienes, while about 19 mol% were identified as C8
alkanes. The fairly equal levels of these di-olefins and non-
olefins suggested that some form of hydrogen transfer was
responsible for the results, rather than only a dehydrogena-
tion step that would generate hydrogen gas. One possible
mechanistic explanation for this process is the dehydrogena-
tion of mono-olefins by the catalyst, followed by transfer of
the hydrogen to a second mono-olefin to generate a saturat-
ed species. However, several pieces of evidence made this
scenario unlikely. First, since the olefin with the highest con-
centration in solution is 1-butene, the transfer of hydrogen
would be expected to form much more butane than saturat-
ed C8 products. Analysis of the products indicated only
slightly elevated levels of butane, a result that may be better
explained by a different mechanism. Also, no dehydrogena-
tion of the heptane internal standard to heptenes was ob-
To better understand the different selectivities reported, it
is useful to examine Scheme 2, which illustrates the various
pathways available for 1-butene dimerization. Similar
schemes have been presented by Keim and Beach for ana-
lyzing butene dimers.[12] To simplify the picture somewhat,
incorporation of 2-butene has been omitted, since it is as-
served; if
a dehydrogenation
mechanism were at work, one
might surmise that C7 paraffins
and C8 mono-olefins should
both be susceptible to this pro-
cess. Furthermore, activation of
heptane by the catalyst would
not necessarily result in chain
transfer to form heptenes. A va-
nadium heptyl complex might
be able to react with butene to
produce C11 species. No C11 spe-
cies were observed in the prod-
ucts, thus indicating the relative
inertness of heptane in this
process.
Further analysis of the prod-
ucts provided the best clues re-
garding the vanadium dimeriza-
tion mechanism. The mass spec-
Scheme 2. Pathways for 1-butene dimerization by migratory insertion.
1016
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 1014 1020

Catalytic Dimerization of 1-Butene
1014 1020
tral data showed that virtually all of the saturated C8 prod-
ucts contained methyl branches, and that almost all of the
dienes were linear. This segregation of alkanes and dienes
between the branched and the linear products, respectively,
combined with the relatively equal amounts of the two
product types, suggested a mechanism whereby the creation
of a molecule of one product would eventually lead to the
creation of a molecule of the other. Upon consideration of
this mole balance constraint, the dimerization mechanism in
Scheme 3 is proposed, in which competitive mechanisms of
chain transfer can be used to rationalize the unique product
distribution. Beginning with the unobserved vanadium-hy-
bond of an incoming molecule of 1-butene. This reaction
generates an equivalent of 3-methyl-heptane and a new
complex 10 with a vanadium butenyl group.
À
The vinylic C H activation seems unlikely when one con-
[13]
À
siders that such a C H bond (ca. 465 kJmol)
is much
of an alkane
[14]
À
stronger than, for example, a C H bond
À
(ꢀ420 kJmol; 423 kJmol for ethane) or the H H bond
(436 kJmol).[15] Surprisingly, despite these bond enthalpy
À
differences, metathesis reactions involving vinylic C H and
metal carbon bonds have often been proposed[16] to explain
such phenomena as formation of polymers and oligomers
with saturated end groups. These interpretations will be dis-
cussed in the next section.
À
dride (’V H’) species shown, an initial primary (1,2) inser-
tion of 1-butene produces the vanadium-n-butyl complex 7,
which can insert a second butene with secondary (2,1) or
primary (1,2) regiochemistry to give the two vanadium octyl
complexes 8 and 9, respectively. Species 8, which according
to the product distribution represents the less likely inser-
tion product, may then undergo b-H elimination or b-H
transfer to form the linear internal olefin. Species 9, on the
other hand, undergoes a competing mechanism of chain
transfer. The catalyst may yield the product via the similar
b-H mechanism of 8 to produce 2-ethyl-1-hexene, or it can
At this juncture, an obvious question is why complexes 8
and 9 undergo different mechanisms of chain transfer. The
apparent answer is due to sterics, since 8 possesses an ethyl
branch at the a-carbon atom to the vanadium center, while
9 has an ethyl branch at the b-position. Complex 8 is appa-
rently too sterically hindered to facilitate the metathesis re-
action preferred by complex 9. Following the chain transfer
step of 9, the vanadium butenyl species 10 may undergo the
same reactions as the vanadium butyl complex; specifically
2,1 insertion of butene to form complex 11 or 1,2insertion
to produce species 12. Both 11 and 12 will then undergo
chain transfer processes identical to their saturated ana-
logues 8 and 9, only this time
À
undergo a net vinyl C H bond metathesis reaction between
the vanadium carbon bond and the vinyl carbon hydrogen
the products will be linear C8
dienes and methyl-branched C8
mono-olefins, with a very small
amount (~0.4%) of mono-
branched dienes.
Worth noting, the vanadium
butyl complex
7 prefers the
second insertion of butene to
occur with primary (1,2) regio-
chemistry, but butenyl complex
10 has an approximately equal
propensity for 1,2and 2,1 inser-
tion. These tendencies are esti-
mated by adding the respective
amounts of branched and linear
products that result from each
complex. To further investigate
why complexes 7 and 10 prefer
somewhat different modes of
insertion, we considered that
the methyl heptene made by
complex 12 could be made by a
different route; specifically by
successive 2,1 insertions fol-
lowed by b-H elimination (see
Scheme 2).
Formation
of
methyl-heptenes via this route
would produce both 5-methyl-
2-heptene and 5-methyl-3-hep-
tene, for a total of four prod-
ucts when the cis and trans iso-
mers are counted. Complex 12,
however, would only make 5-
Scheme 3. Proposed mechanism for dimerization by vanadium complex 3. All of the percentages were deter-
mined by GC/MS. The presence of vinylidene product was confirmed by H NMR spectroscopy. All of the per-
centages refer to mole percent values. The dotted bond lines on products refer to ambiguities in the location
of the bonds due to the suggested isomerization/insertion mechanism of 10. The trans isomers are depicted for
clarity, but cis isomers are likely present for all of the disubstituted internal olefins.
1
1017
Chem. Eur. J. 2004, 10, 1014 1020
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

B. L. Small et al.
FULL PAPER
methyl-3-heptene. By use of commercially available GC
standards, the only methyl-heptene identified in the product
was trans-5-methyl-2-heptene, a result that is inconsistent
with either mechanism. This odd result may suggest that
complex 10 undergoes a double bond shift upon conversion
to 11 or 12, or that our speculative mechanism is incorrect.
In addition to the questions regarding species 7 and 10,
their respective insertion products 9 and 12 apparently ex-
hibit quite different tendencies in their modes of chain
transfer. Complex 9 undergoes chain transfer with only a
If the catalytic cycle in Scheme 3 is correct, it is possible
to establish the following ’rules’ for the vanadium-catalyzed
dimerization:
1) The first insertion of an a-olefin at vanadium occurs
with 1,2regiochemistry (98 + %).
2) If the second insertion is 2,1, then b-H elimination/trans-
fer to generate linear internal olefins is the preferred
chain transfer step (99+ %).
3) If the second insertion is 1,2, then vinyl hydride transfer
(metathesis) is preferred.
4) The methylene (terminal) hydride of the incoming olefin
transfers exclusively when vinyl-hydride activation
occurs, thus giving an n-butenyl vanadium complex.
À
slight preference for the vinyl C H activation route, while
complex 12 almost exclusively prefers this mechanism. Since
9 and 12 differ only in the presence of a double bond on the
alkyl fragment of complex 12, the disparate product compo-
sitions are another interesting and yet unexplained phenom-
enon.
Another consideration of the cycle proposed in Scheme 3
involves the disposition of vanadium complex 7. If relatively
unhindered species such as 9 and 12 undergo chain transfer
via a metathesis step, complex 7 would also be expected to
display this behavior to produce n-butane. In fact, by careful
inspection of Scheme 3, the amount of butane may be ap-
proximated by subtracting the number of moles of alkanes
from the number of moles of dienes: (mol. butane)=(mol.
diene)À(mol. alkane)
Discussion
The proposed competing chain transfer pathways in
Scheme 3 raise the question of literature precedent for vana-
dium and other metals. For example, in polypropylene pro-
duced with
a Sc-based constrained geometry catalyst,
Bercaw et al.[16b] found no evidence of vinylidene chain ends
(from H-transfer to monomer or metal center); hence, it
À
From the GC data, it was possible to calculate a total
C8 + yield of 45.1 g, which included 37.9 g of C8 fractions,
4.1 g of C12 fractions, and 2.9 g of C16 fractions. The molar
ratio of C8 dienes to C8 alkanes (from GC/MS) was then as-
sumed to be constant for the C12 and C16 fractions, for which
the GC/MS data could not be deconvoluted.[17] From the
overall diene to alkane ratio, the amount of ™missing∫ n-
butane was predicted to be 2.2 g, or about 9 mol% of the
total product (n-butane, C8, C12, C16). Finally, by using the
residual n-butane and 1-butene signals and the reaction con-
version (18%), the amount of n-butane formed in the reac-
tion was estimated at 3.0 g, or about 12mol% of the total
product. These predicted and observed butane levels are in
fairly close agreement considering the limitations of the ex-
periment.
was speculated that olefin C H bond metathesis might be
an important termination process. Ziegler et al.[18] published
several computational papers that examined vinylic s-bond
metathesis as a chain transfer route for scandium and titani-
um catalysts. Specifically, their computations indicated that
such a process would generally be less favorable than the
parallel C=C bond insertion (in terms of both overall exo-
thermicity and reaction barrier). However, the computed
energetics are not prohibitive enough to rule out vinylic
À
C H metathesis as a minor process. Apparently, a well-de-
fined complex (i.e. adduct) with ethylene (or olefin) in a
head-on orientation toward the metal center is formed; it is
stabilized by an agostic interaction involving the vinylic
C H bond as well as electrostatic interaction between the
olefin and the electron deficient metal center.
À
In separate experiments, 1-hexene was dimerized
(16,000:1 hexene:V ratio) to 60% conversion, in order to
determine the amount of n-hexane formed. The n-hexane
constituted 7.5 mol% of the final product (n-hexane, C12,
C18); but as with the butene trimers, we were not able to
achieve sufficient separation of the C12 species to quantify
the amounts of dienes and alkanes, thus preventing estima-
tion of the predicted amount of n-hexane formation. Finally,
propylene was also dimerized by catalyst 3 to see if any
additional information could be gained from examining the
C6 products. As with the 1-butene and 1-hexene dimeriza-
tions, numerous alkane and diene species were identified.
However, quantitative analysis was limited by the formation
of substantial amounts of higher oligomers. In addition,
the propylene dimers do not appear to form by the same re-
giochemical (2,1 versus 1,2) and steric constraints as the
butene dimers. GC/MS data for both the propylene and
1-hexene dimerizations are included in the Supporting Infor-
mation.
Recently the groups of Gibson and Gal have indepen-
dently reported the isolation of tridentate pyridinebisimine
CoI alkyl complexes. These CoI species do not undergo
olefin insertion, and the authors speculate that the catalysts
may achieve an active CoIII state via oxidative addition of a
À
vinyl C H bond. These researchers clearly state that this ac-
tivation step is one of several possible ways to generate the
active species, if indeed the active catalyst is a CoIII spe-
cies.[19] In the case of vanadium complex 3, it is not clear if
the oxidation state of the metal in the active form is +3
(the same as the catalyst precursor) or if it is somewhat re-
duced. In any case, the described system likely consists of a
dn system with nꢀ2. The agostic interactions and electro-
À
static stabilization that are deemed to favor vinylic C H ac-
tivation in the case of the ScIII or TiIV catalysts would not be
that strong in the case of the vanadium catalyst. However,
this seemingly unlikely mode of chain transfer should not be
ruled out, particularly because there appears to be no other
good explanation for the formation of the alkane products.
1018
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 1014 1020

Catalytic Dimerization of 1-Butene
1014 1020
Furthermore, it is also noted that complex 3 possesses a sig-
nificantly different ligand set than the ScIII or TiIV catalysts.
Bearing that in mind, it is not inconceivable that the nitro-
Conclusion
The dimerization of 1-butene by a selected variety of transi-
tion metal catalysts has been used to demonstrate a number
of mechanistic pathways that are available during a-olefin
dimerization. Cobalt catalysts form almost completely linear
dimers, while iron and nickel systems can make predomi-
nantly linear products. The iron catalysts, however, are at
least an order of magnitude faster than nickel, and they do
not tend to isomerize the substrate or the product substan-
tially. Aluminum, albeit with low activity, is highly selective
for producing vinylidene dimers, and the vanadium catalyst
discussed herein dimerizes olefins by a unique mechanism
that produces significant and roughly equal amounts of
branched alkanes and linear dienes. Future work will in-
volve attempts to better understand the speculative mecha-
nism of the vanadium catalyst.
À
gen-based ligand could also be involved in the C H activa-
tion chemistry proposed for vanadium. The proposed mech-
anism of Scheme 3 should at this point only be invoked to
rationalize the product distribution, rather than to argue for
a specific mode of activation.
À
Although s-bond metathesis and vinyl C H bond activa-
tion are known for a variety of transition metal complexes,
subsequent functionalization of the activated hydrocarbon is
less common. One area where several studies have been un-
dertaken is in the field of ethylene polymerization. Specifi-
cally, it has been proposed by Reinking et al. and Kissin
et al. that Zr- and V-based polymerization catalysts can pro-
mote long-chain branching (LCB) by intramolecular or in-
termolecular s-bond metathesis reactions with polyethylene
chains.[20,21] This theory has been proposed as an alternative
to the prevailing idea that LCB arises from the reincorpora-
tion of polymer chains with unsaturated end groups into an-
other growing polymer chain. However, in systems that uti-
lize hydrogen as a chain transfer agent, the number of avail-
able terminal olefin groups can be quite small, which indi-
cates another mechanism may be responsible for LCB. Ad-
ditionally, Reinking showed that simple alkanes such an n-
heptane and cyclohexane can be activated by VCl4 on silica
with an alkylaluminum cocatalyst.[20] In the presence of hy-
drogen and ethylene, the vanadium system produces alkyl-
cyclohexanes catalytically from cyclohexane, indicative of
Experimental Section
Materials: Anhydrous THF and methanol were purchased from Aldrich
and used without further purification. Anhydrous cyclohexane was pur-
chased from Aldrich and stored over 3A molecular sieves. 1-hexene and
1-butene were obtained as commercial grades of Chevron Phillips’
Normal Alpha Olefins. 1-hexene was degassed and dried over 3A molec-
ular sieves. MMAO-3A, triisobutylaluminum, and diethylaluminum chlo-
ride were purchased from Akzo Nobel. 2,6-Diacetylpyridine, iron(ii)
chloride tetrahydrate, and all substituted anilines were purchased from
Aldrich and used without further purification. Nickel(ii) chloride, dime-
thoxyethane adduct; vanadium(iii) chloride, tris-tetrahydrofuran adduct;
bis(triphenylphosphine) nickel(ii) chloride; and cobalt(ii) chloride hexa-
hydrate were purchased from Strem.
À
C H bond activation followed by ethylene insertion. The
authors propose that this phenomenon is a model for the
LCB mechanism since it demonstrates how saturated species
can be incorporated into a polymer chain.
Catalyst preparation: Complexes 1,[9,11] 2,[9] and 4[5] were prepared by de-
scribed literature methods. Synthetic details for complex 3 are detailed
below:
Due to the unique product distribution, a catalytic cycle
involving a formal oxidative addition of butene to vanadium
catalyst 3, followed by successive steps involving migratory
insertion and reductive elimination, appears unlikely. Al-
though 1-butene could conceivably oxidatively add to a low-
valent vanadium center, followed by successive migratory
insertion and reductive elimination reactions to generate
the products shown, the absence of products such as
n-octane, coupled with the roughly equal amounts of linear
dienes and branched paraffins, argue against this mecha-
2,6-Bis[1-(2-methylphenylimino)ethyl]pyridine vanadium(iii) chloride (3):
2,6-Bis[1-(2-methylphenylimino)ethyl]pyridine (1.00 g, 2.9 mmol) and va-
nadium(iii)chloride tetrahydrofuran adduct (1.04 g, 2.8 mmol) were
added together under inert conditions to a 100 mL flask with a stirbar.
THF (25 mL) was added, and the reaction was stirred for 16 h in hot
THF (558C). Pentane was added, and the reaction was filtered and
washed with ether and pentane to give, after drying, 1.35 g (97%) of
complex 3. Analytical data for pre-catalyst complex 3: MS ([M+ÀCl]):
464/462; elemental analysis (%): calcd C 55.39, H 4.65, N 8.42; found: C
55.34, H 5.10, N 7.94.
À
nism. Vinyl C H bond metathesis coupled with b-hydride
Dimerization of 1-hexene: A dry two-necked flask with a stirbar was
charged with 1-hexene (100 mL) and pre-catalyst 3 (25 mg) under inert
atmosphere (16000:1 olefin:V ratio). The flask was transferred to a
Schlenk manifold and placed under a slow nitrogen purge. The pre-cata-
lyst was slurried in the 1-hexene by rapidly stirring the liquid. MMAO-
3A in heptane (100:1 Al:V) was then added via syringe. Samples were
taken for GC analysis after 3 h and after 24 h. The reaction had formed
about 50% of its final product after 3 h.
elimination/transfer remains the most likely pathway; since
both processes are known to occur with early metals, the
unique features of this system are the proposed steric-based
competition between the two mechanisms and the high
À
catalytic activity toward the vinyl C H bond activation
route.
Dimerization of 1-butene: A 500-mL Zipperclave reactor was heated
under vacuum at 508C for several hours. The reactor was cooled to room
temperature under an inert gas. The pre-catalyst was then added to the
reactor in a sealed NMR tube by tying the tube to the stirrer shaft, and
the reactor was resealed and placed under vacuum. A glass sample charg-
er was then attached to the injection port of the reactor. The internal
standard and the cocatalyst were added. The reactor was then quickly
sealed and charged with liquid butene. The reactor was further pressur-
ized with at least 100 psi of nitrogen or argon to keep the butene in the
liquid phase.[8] Stirring was begun, thus breaking the NMR tube and ini-
tiating the reaction. The temperature was monitored using a thermocou-
À
As a final consideration, the prevalence of vinyl C H
bond activation in the vanadium system magnifies a reaction
that may be occurring to a lesser extent in a variety of other
catalytic systems, ranging from Ziegler Natta polymeriza-
tions and oligomerizations to catalytic olefin dimerization.
As presented in this work, the presence of trace amounts of
dienes in any of these processes may possibly be attributed
À
to small amounts of chain transfer by vinyl C H bond acti-
vation to generate vinyl metal bonds.
1019
Chem. Eur. J. 2004, 10, 1014 1020
www.chemeurj.org
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

B. L. Small et al.
FULL PAPER
ple. Internal cooling and external heating were used to maintain the de-
sired reactor temperature.
[4] B. L. Small, A. J. Marcucci, Organometallics 2001, 20, 5738.
[5] S. A. Svejda, M. Brookhart, Organometallics 1999, 18, 65.
[6] a) H. Olivier-Bourbigou, J.-A. Chodorge, P. Travers, Petroleum Tech-
nology Quarterly 1999, Autumn, 141; b) Y. Chauvin, J. F. Gaillard,
D. V. Quang, J. W. Andrews, Chem. Ind. 1974, 375; c) D. Commer-
euc, Y. Chauvin, J. Gaillard, J. Lÿonard, J. W. Andrews, Hydrocar-
bon Process. 1984, 118.
Product analysis: For the butene dimerization, the reactor was slowly
vented. The aluminum cocatalysts in all of the reactions were neutralized
by pouring the liquid products into a water wash. After removal of the
cocatalysts, the products were analyzed by gas chromatography (GC). A
Hewlett Packard 6890 Series GC System with an HP-5 50 m or 30 m
column with a 0.2mm inner diameter was used for dimer as well as a-
olefin characterization. An initial temperature of 358C and a rate of
2.48CminÀ1 were used to raise the temperature to 528C, followed by a
rate of 15.08CminÀ1 to raise the temperature to 1578C. A final ramp rate
of 22.58CminÀ1 was used to reach the final temperature of 2508C. Chem-
Station from Hewlett Packard was used to analyze the collected data.
GC/MS data were obtained using an Agilent 5973 Benchtop Mass Spec-
trometer using electron impact ionization interfaced to an Agilent 6890
gas chromatograph. The GC column was a J&W Scientific DB-5mS,
60 mî0.25 mm i.d. with a 0.25 mm film thickness. After an initial time of
5.0 min, a ramp rate of 3.08CminÀ1 was used to raise the oven tempera-
ture to 3008C. The Wiley275 L computerized database was used to assign
the mass spectrum signals. Due to ambiguities in classifying all of the
[7] K.-F. Lin, G. E. Nelson, C. W. Lanier, (Ethyl) U. S. Pat. 4973788,
1990.
[8] For catalyst 6, a 2-L reactor was used. Care should be taken when
working with liquid butene at the elevated temperature described
for catalyst 6 (ꢀ1808C). The reactor should be rated for at least
1400 psig, and the reaction should be heated slowly to avoid sudden
temperature increases.
[9] a) B. L. Small, M. Brookhart, J. Am. Chem. Soc. 1998, 120, 7143;
b) G. J. P. Britovsek, S. Mastroianni, G. A. Solan, S. P. D. Baugh, C.
Redshaw, V. C. Gibson, A. J. P. White, D. J. Williams, M. R. J. Else-
good, Chem. Eur. J. 2000, 6, 12, 2221.
[10] a) D. Reardon, F. Conan, S. Gambarotta, G. Yap, Q. Wang, J. Am.
Chem. Soc. 1999, 121, 9318; b) R. Schmidt, U. Hammon, M. B.
Welch, H. G. Alt, B. E. Hauger, R. D. Knudsen (Phillips) U. S. Pat.
6458905, 2002.
1
dimer products for catalyst 3 by GC/MS, H NMR spectroscopy was used
for further characterization. In particular, NMR spectroscopy resolved a
GC/MS discrepancy, resulting in correct classification of the vinylidene
dimer made from 1-butene. At first, the GC/MS data indicated that the
vinylidene species was actually a trisubstituted dimer, which could be
formed by isomerization of a vinylidene or a methyl-heptene. However,
control experiments with the activated vanadium catalyst 3 and 2-ethyl-1-
hexene or disubstituted internal methyl-heptenes showed no isomeriza-
tion behavior. 1H NMR spectroscopy confirmed that the species in ques-
tion was indeed a vinylidene and not a trisubstituted olefin. GC data
were then used to make all of the quantitative assessments, since the
NMR spectra could not distinguish between the dimeric and trimeric
products.
[11] B. L. Small, Organometallics 2003, 22, 3178.
[12] a) W. Keim, B. Hoffmann, R. Lodewick, M. Penckert, G. Schmitt, J.
Fleischhauer, U. Meier, J. Mol. Catal. 1979, 6, 79; b) D. L. Beach,
J. E. Bozik, C.-Y. Wu, Y. V. Kissin, J. Mol. Catal. 1986, 34, 345.
[13] a) K. M. Ervin, S. Gronert, S. E. Barlow, M. K. Gilles, A. G. Harri-
son, V. M. Bierbaum, C. H. DePuy, W. C. Lineberger, G. B. Ellison,
J. Am. Chem. Soc. 1990, 112, 5750; b) J. J. Russell, S. M. Senkan,
J. A. Seetula, D. Gutman, J. Phys. Chem. 1989, 93, 5184.
[14] J. Berkowitz, G. B. Ellison, D. Gutman, J. Phys. Chem. 1994, 98,
2744.
[15] K. P. Huber, G. Herzberg, in Molecular Spectra and Molecular Struc-
ture Constants of Diatomic Molecules, Van Nostrand, New York,
1979.
The conversions and yields were determined by comparing the product
to the internal standard integrals, and by assuming equal response factors
of the standard and the products. For hexene dimerization, unreacted 1-
hexene was the internal standard, and for the butene experiments cyclo-
hexane was used.
[16] a) H. H. Brintzinger, D. Fischer, R. M¸lhaupt, B. Rieger, R. M.
Waymouth, Angew. Chem. 1995, 107, 1255 1283; Angew. Chem. Int.
Ed. Engl. 1995, 34, 1143 1170; b) B. J. Burger, M. E. Thomson,
W. D. Cotter, J. E. Bercaw, J. Am. Chem. Soc. 1990, 112, 1566;
c) M. E. Thomson, S. M. Buxter, A. R. Bulls, B. J. Burger, M. C.
Nolan, B. D. Santasiero, W. P. Schaefer, J. E. Bercaw, J. Am. Chem.
Soc. 1987, 109, 203; d) P. L. Watson in Selective Hydrocarbon Activa-
tion (Eds.: J. A. Davis, P. L. Watson, J. F. Liebman, A. Greenberg),
VCH Publishers, New York, 1990.
[17] GC/MS was used to obtain peak MW values for the C12 fraction, in-
dicating substantial amounts of C12 alkanes (MW=170) and C12
dienes (MW=166). Reliable intensities, however, could not be ob-
tained due to multiple overlapping signals.
Acknowledgement
We thank Dr. M. Jeansonne and Dr. J. B. Green for help with dimer
characterization, and A. J. Marcucci and Eric Fernandez for performing
several of the dimerization experiments. Furthermore, we thank Dr. M.
M. Johnson, Dr. P. K. Das and Dr. D. E. Lauffer for helpful mechanistic
discussions.
[18] a) T. Ziegler, E. Folga, A. Berces, J. Am. Chem. Soc. 1993, 115, 636;
b) T. K. Woo, L. Fan, T. Ziegler, Organometallics 1994, 13, 2252;
c) T. K. Woo, P. M. Margl, T. Ziegler, Organometallics 1997, 16,
3454.
[19] a) V. C. Gibson, M. J. Humphries, K. P. Tellman, D. F. Wass, A. J. P.
White, D. J. Williams, Chem. Commun. 2001, 2252; b) T. M. Koois-
tra, Q. Knijnenburg, J. M. M. Smits, A. D. Horton, P. H. M. Budze-
laar, A. W. Gal, Angew. Chem. 2001, 113, 4855; Angew. Chem. Int.
Ed. 2001, 40, 4719.
[20] M. K. Reinking, G. Orf, D. McFaddin, J. Polym. Sci. Polym. Chem.
Ed. 1998, 36, 2889.
[21] Y. V. Kissin, T. E. Nowlin, R. I. Mink, Macromolecules 1993, 26(9),
2151.
[1] a) Y. Chauvin, H. Olivier, in Applied Homogeneous Catalysis with
Organometallic Compounds, Vol. 1 (Eds.: B. Cornils, W. A. Herr-
mann), VCH, New York 1996, pp. 258 268; b) G. W. Parshall, S. D.
Ittel in Homogeneous Catalysis:the Applications and Chemistry of
Catalysis by Soluble Transition Metal Complexes 2nd ed., Wiley,
1992, pp. 72 76; c) J. Skupinska, Chem. Rev. 1991, 91, 613; d) W.
Keim, Angew. Chem. 1990, 102, 251; Angew. Chem. Int. Ed. Engl.
1990, 29, 235 244.
[2] a) A. M. Al-Jarallah, J. A. Anabtawi, M. A. B. Siddiqui, A. M.
Aitani, A. W. Al-Sa’doun, Catal. Today 1992, 14, 1; b) A.-H. Wu
(Phillips) U. S. Pat. 5414178, 1995.
[3] a) B. Ellis, W. Keim, P. Wasserscheid, Chem. Commun. 1999, 337;
b) P. Wasserscheid, M. Eichmann, Catal. Today 2001, 66, 309.
Received: March 13, 2003
Revised: October 27, 2003 [F4945]
1020
¹ 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2004, 10, 1014 1020