Mechanism of ethylene dimerization catalyzed by Ti(OR′) 4/AlR3

Article abstract of DOI:10.1021/om3008508

Ti-alkoxide-based catalysts in combination with AlEt₃ are responsible for the production of a significant proportion of the world's 1-butene supply, via the dimerization of ethylene. A metallacycle mechanism is normally presumed to operate with this system. However, despite its importance, the catalyst is not mechanistically well understood. The mechanism of dimerization has been studied through a series of C₂H ₄/C₂D₄ co-oligomerization experiments and comparison of theoretical and experimental mass spectra. The results obtained show that the textbook metallacycle mechanism is most likely not responsible for dimerization with this catalyst. Instead, an excellent fit between the theoretical and experimental mass spectra is obtained when a conventional Cossee-type mechanism (insertion/β-hydride elimination) is modeled. The formation of both the primary product 1-butene and the secondary reaction products (ethylene/1-butene co-dimers) is best explained by this mechanism.

Full text of DOI:10.1021/om3008508

Article
pubs.acs.org/Organometallics
Mechanism of Ethylene Dimerization Catalyzed by Ti(OR′)4/AlR3
James A. Suttil and David S. McGuinness*
School of Chemistry, University of Tasmania, Private Bag 75, Hobart 7001, Australia
S
* Supporting Information
ABSTRACT: Ti-alkoxide-based catalysts in combination with AlEt3 are
responsible for the production of a significant proportion of the world's 1-butene
supply, via the dimerization of ethylene. A metallacycle mechanism is normally
presumed to operate with this system. However, despite its importance, the catalyst
is not mechanistically well understood. The mechanism of dimerization has been
studied through a series of C2H4/C2D4 co-oligomerization experiments and
comparison of theoretical and experimental mass spectra. The results obtained show that the textbook metallacycle mechanism is
most likely not responsible for dimerization with this catalyst. Instead, an excellent fit between the theoretical and experimental
mass spectra is obtained when a conventional Cossee-type mechanism (insertion/β-hydride elimination) is modeled. The
formation of both the primary product 1-butene and the secondary reaction products (ethylene/1-butene co-dimers) is best
explained by this mechanism.
tetramerization have been of much interest over recent
years.13−33 In contrast, mechanistic studies of the Ti(OR′)4/
AlR3 dimerization system are rare; the highest volume and most
mature of the selective oligomerization systems seems to be the
least well understood. The high selectivity of the system has
often been attributed to a metallacyclic mechanism for ethylene
dimerization (Scheme 1).34 Such a process involves the
INTRODUCTION
■
The oligomerization of ethylene to form linear alpha olefins
(LAOs) is a field of research that continues to receive much
attention from both industry and academia. While the majority
of ethylene oligomerization is carried out using full-range
processes that are governed by either Schulz−Flory or Poisson
distributions of LAOs,1 such processes struggle to meet the
increasing market demand for short-chained LAOs such as 1-
butene, 1-hexene, and 1-octene, which are high-value co-
monomers for the production of linear low-density poly-
ethylene. This imbalance between supply and market demand
has resulted in a greater emphasis on developing and
understanding selective, or on-purpose, oligomerization tech-
nologies (ethylene dimerization, trimerization, and tetrameriza-
tion), which avoid production of excess higher LAOs.2−7
One such technology is the IFP Energies Nouvelles/Sabic
Alphabutol process for the dimerization of ethylene to 1-
butene.2 This process was developed from Ziegler and Martin’s
initial discovery that combinations of triethylaluminum and
titanium or zirconium alkoxides readily convert ethylene to 1-
butene with high selectivity.8 A similar catalyst system,
Ti(OAr)4/AlEt3, was also employed to develop the Toso 1-
butene process,9 and more recently mixed-ligand aryloxytita-
nium complexes have been investigated.10,11 Catalyst activities
approaching 1 × 106 h−1 (TOF) with 1-butene selectivity of
93% have been achieved with the optimal system, Ti(OBu)4/
AlEt3.2 Given the high activity and selectivity of the system, it is
unsurprising that there are currently 30 Alphabutol plants
operating worldwide, providing around 25% (708 000 tons per
annum) of the world’s 1-butene supply.1,12
Scheme 1. Metallacycle Mechanism for Ethylene
Dimerization
coordination of two ethylene molecules to a divalent titanium
species, whereupon these ethylene units undergo an oxidative
addition generating a titanocyclopentane, which decomposes
via a beta-hydride transfer (either in a stepwise process or by a
concerted mechanism) to yield 1-butene. This mechanism can
also be successfully invoked to explain the assortment of C6
byproducts that are generated in the process, which come about
due to co-dimerization of ethylene and 1-butene. While such a
mechanism has now become generally accepted in the
academic literature and texts,1,2,34−38 direct evidence support-
ing such a mechanism is somewhat sparse. Studies by Rothwell
on titanium aryloxide compounds have explicitly shown they
can support metallacyclic species, although 1-butene is not
generated by these complexes.39,40 Work by Whitesides and co-
Given the industrial importance of these selective oligome-
rization technologies, it is of equally high importance to
understand the mechanistic basis for this selectivity. Such an
understanding is vital for further catalyst development, in
addition to being of significant fundamental interest. For this
reason, mechanistic studies of ethylene trimerization and
Received: September 2, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
workers utilizing preformed titanacyclopentanes has previously
demonstrated their propensity to either decompose to generate
1-butene (Scheme 2, pathway a) or revert to ethylene via a
earlier literature that alkoxyltitanium/trialkylaluminum systems
could operate via a Cossee mechanism.2,36 These findings,
combined with the fact that the oligomer distribution can be
just as easily ratified by a Cossee mechanism (with a high rate
of chain transfer relative to chain propagation), suggest that this
alternative warrants further investigation.
Scheme 2. (a) Metallacycle Decomposition to Form 1-
Butene, and (b) Cβ−Cβ′ Bond Cleavage Forming Ethylene
In a 2004 investigation on an ethylene trimerization system,
Bercaw and co-workers demonstrated that the metallacyclic and
Cossee mechanisms could be distinguished on the basis of
analysis of the isotopomer distribution resulting from
trimerization of a 1:1 mixture of ethylene and perdeutero-
ethylene.22 It was shown that, as the metallacyclic mechanism
leads to no H/D scrambling, only C6H12, C6H8D4, C6H4D8,
and C6D12 isotopomers should be observed. Conversely, for a
Cossee-based mechanism H/D scrambling results from β-
hydride chain transfer, which would yield isotopomers
containing odd numbers of hydrogen and deuterium. By
employing a well-defined chromium catalyst they demonstrated
that a metallacyclic mechanism was most likely in effect. A
further study in 2007 extended this experiment and employed a
Ni-based SHOP (full-range oligomerization) catalyst; analysis
of the isotopomers in this case correlated well with the
predicted Cossee mechanism.23 Since its inception, this simple
experiment has been employed in mechanistic investigations for
various tri/tetramerization24,32,43 and oligomerization cata-
lysts.53−56 In principle this experiment should provide a
straightforward means of elucidating the mode of 1-butene
formation with Ti(OR′)4/AlR3. Indeed, a little known study
from 1982 has previously undertaken this experiment employ-
ing Ti(OBu)4/AlMe3 as the catalyst system.57 Contrary to the
expected distribution from their proposed synchronous
coupling mechanism to form 1-butene (no H/D scrambling),
a distribution consisting of H/D scrambled isotopomers was
reported, which is more consistent with a Cossee mechanism.
However, as reported below, we have found that the complex
mass spectral fragmentation pattern of 1-butene complicates
this analysis significantly. The distributions predicted in this
1982 study seem overly simplistic, and insufficient experimental
procedures are reported to account for this. As a result, there is
considerable doubt over the conclusions that can be drawn. On
the question of the mechanism of this catalyst, the evidence to
date is conflicting, and a definitive answer is overdue given the
importance of this catalyst. Herein we report a detailed study of
the mechanism of dimerization with this system, which for the
first time provides strong evidence that a metallacycle
mechanism is not operative and fully supports a Cossee
mechanism.
carbon−carbon bond cleavage (Scheme 2, pathway b),
although this work did highlight the stability of the metal-
lacyclopentane ring.41,42 Follow-up studies by Chauvin and co-
workers confirmed this result when Ti(OBu)4 was reacted with
1,4-dilithiobutane (expected to yield a titanacyclopentane
species in the first instance).34 Additionally, our research
group has recently published results showing titanium alkoxide
and aryloxide compounds can oligomerize ethylene, yielding
products consistent with an Alphabutol-type catalyst, via a
metallacyclic mechanism.43 Several other academic studies have
also implied a titanacyclic intermediate in ethylene dimeriza-
tion.44−47
Conversely, early studies involving platinum(II) metalla-
cycles have shown that five-membered platinacyclic species are
more long-lived when compared to their seven-membered
counterparts. It has been proposed that the rigidity of the five-
membered ring inhibits it from adopting the required
platinum/β-hydrogen dihedral angle needed for elimination
of 1-butene, while the seven-membered ring has the flexibility
required for facile elimination of 1-hexene.48,49 Such a trend has
been similarly demonstrated for chromium metallacyclic
species,50 has been suggested by several theoretical studies of
Ti complexes,29−31 and is also proposed to give rise to the
selectivity of numerous ethylene tri- and tetramerization
catalysts.3−6 Interestingly, ethylene dimerization catalysts
based upon other transition metals such as Ni35,36,51,52 have
been suggested in the literature to oligomerize via a Cossee−
Arlman mechanism (Scheme 3). It has also been suggested in
Scheme 3. Cossee Mechanism for Ethylene Dimerization
RESULTS AND DISCUSSION
■
Analysis of 1-Butene. Our initial work in elucidating the
mechanism was focused on the standard system, Ti(OBu)4/
AlEt3. Ethylene oligomerization with this catalyst (Table 1)
a
Table 1. Ethylene Oligomerization with Ti(OBu)4/AlR3
entry
1
co-catalyst
TON
1-butene (%)
3-methylpent-1-ene (%)
3-methylpentane (%)
1-hexene (%)
2-ethylbut-1-ene (%)
polymer (%)
AlEt3
AlEt3
AlMe3
AlEt3
696
530
399
186
78.4
54.5
75.2
94.9
7.0
4.7
7.0
0.9
0.3
2.4
0.3
0.7
1.4
2.1
0.9
0.7
12.3
8.7
0.6
27.4
1.9
b
2
3
14.7
2.8
c
4
a
b
0.15 mmol of catalyst, 3 equiv of cocatalyst, 20 mL of toluene, 55 °C, 20 bar ethylene pressure, 30 min run time. Run employed 15 equiv of
c
cocatalyst. Run performed at 0 °C and 1 bar ethylene pressure.
B
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
yielded the expected high selectivity toward 1-butene as well as
the typical ethylene/1-butene co-dimerization products 3-
methylpent-1-ene, 1-hexene, and 2-ethylbut-1-ene.2,35 In
addition to these expected products, a small amount of 3-
methylpentane was detected by GC analysis; this most likely
results from chain transfer between an alkyltitanium species and
triethylaluminum followed by hydrolysis during reactor
quenching. A small percentage of polymer was also formed,
which increases significantly with increasing cocatalyst loading
(cf. entries 1 and 2, Table 1). This effect is known,2 with excess
cocatalyst reported to yield a catalytic species that generates
high molecular weight polyethylene. For the reasons outlined
below, we also tested a modified system in which the cocatalyst
is trimethylaluminum. This led to a significant decrease in
activity (Table 1, entry 3), which again is a known effect and
has been attributed to the more stable dimeric structure of
trimethylaluminum in solution, yielding a less effective
activator.2 At the same time, a comparable selectivity profile
to that resulting from activation with triethylaluminum is
obtained. The selectivity of the catalyst to 1-butene is highest at
0 °C, although a marked decrease in activity also results.
Replication of the experiment employing a 1:1 mixture of
C2H4/C2D4 yielded isotopomers of the oligomerization
products detailed above. Careful analysis of the H/D
isotopomer distribution in the 1-butene formed, as determined
by GC-MS, was undertaken. If 1-butene is formed via a
metallacyclic mechanism, then no H/D scrambling should be
evident, and as such, the isotopomers should consist of C4H8,
C4H4D4, and C4D8 in a ratio of 1:2:1 (full details of how
predicted distributions were obtained are given in the
Supporting Information).
spectrum obviously complicate the isotopomer analysis
significantly, and their effects have been incorporated by
employing a model whereby the probability of loss of H or D
via EI ionization is proportional to the ratio of H and D in the
isotopomer. This correction is detailed in the Supporting
Information. A final factor that could potentially affect the
isotopomer distribution is a kinetic isotope effect (KIE) related
to a β-hydride transfer step. This would not lead to H/D
scrambling, but could potentially skew the distribution toward
less deuterated isotopomers. The results presented below,
however, indicate that any KIE is small to absent. This is
expected given the selectivity of this catalyst toward dimers,
which suggests that product releasing β-hydride transfer is not a
rate-determining step (if it were, further insertion leading to
longer chain products would result). Rather olefin insertion (or
metallacycle formation) is more likely to be rate limiting, which
accounts for the selectivity to dimers.
Our initial comparison of the theoretically derived
distributions to the experimental data (Figure 2) showed no
Conversely, for a Cossee mechanism involving β-hydride
transfer58 between oligomers, scrambling will occur and the
expected theoretical distribution becomes 1:1:0:1:2:1:0:1:1
across the molecular weight range 56 to 64. In order to
convert these theoretical isotopomer distributions into the
actual expected mass spectral ion (m/z) ratios, a number of
Figure 2. Experimental and predicted mass spectrum of 1-butene
produced from C2H4/C2D4 with Ti(OBu)4/AlEt3. Data normalized to
mol. ion m/z 60.
definitive support for either the Cossee or metallacycle
mechanism. It was noted that an overexpression of the
molecular ion at m/z 56 ([M]+ for completely undeuterated
1-butene) was evident compared to either of the theoretical
distributions. We subsequently found that commercial
triethylaluminum contains ca. 5% butyl groups relative to
ethyl, including the cocatalyst used in this study. As such, it
seems likely that a portion of 1-butene formed is derived from
(undeuterated) butyl group transfer to titanium, followed by
elimination of 1-butene. The evidence for chain transfer
between the catalyst and cocatalyst that is discussed above
(formation of saturated products) supports this notion. This
incorporation of an undeuterated butyl (and most likely also
ethyl) functionality into a portion of the 1-butene skews the
distribution away from either of the theoretical distributions to
the point where deconvolution becomes fruitless, as there is no
way of knowing how much incorporation has occurred.
corrections can to be applied (natural abundance of D and 13
C
1
in ethylene, H impurity in C2D4). However, in this work we
have discovered a much greater complicating factor, specific to
1-butene, which arises due to the fragmentation pattern of this
compound under mass spectral ionization (EI). The mass
spectrum of 1-butene over the range m/z 49−57 is shown
graphically in Figure 1 and reveals significant contributions due
to ions down to [M − 7]+, with the [M − 1]+ signal being
particularly abundant.59 These contributions to the mass
The most straightforward solution to the problem of chain
transfer with triethylaluminum is to use trimethylaluminum as
the cocatalyst instead. This would eliminate any incorporation
of an even-numbered alkyl functionality into 1-butene that
could result from alkylation of titanium and subsequent
ethylene insertion/β-hydride elimination; conversely the only
side-products that should result would be odd-numbered
hydrocarbons. As shown in Table 1, activation with
trimethylaluminum generates a very similar selectivity profile
to triethylaluminum under the same conditions (cf. entries 1
Figure 1. Mass spectrum of 1-butene in the region of the molecular
ion as a result of electron impact ionization. Relative intensities shown
in comparison to [M]+ = 100.
C
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
and 3), albeit at somewhat reduced activity. Interestingly, at
most only trace amounts of propene are formed, which can
again be related back to the higher stability of the
trimethylaluminum dimer, thus not favoring chain transfer
reactions. Catalysis employing a 1:1 mixture of C2H4/C2D4 in
combination with Ti(OBu)4/AlMe3, upon analysis by GC-MS,
yields an excellent fit between the experimental data and the
theoretical distribution for 1-butene formed via a Cossee
mechanism (Figure 3a). The experimental distribution differs
of the possible insertion geometries and elimination pathways
(Scheme 4) can indeed account for the major product
Scheme 4. Co-dimerization of Ethylene and 1-Butene via a
Metallacyclic Mechanism
distributions. Oxidative addition yielding an ethyl moiety at
the C1 position (Scheme 4, pathway a) has the potential to
undergo exocyclic β-hydride transfer to give 2-hexene or
endocyclic β-hydride transfer resulting in either 1-hexene or 3-
hexene. However, it has been argued that due to unfavorable
steric interactions between the titanium center and ethyl
moiety, these oligomers should be present as minor products, a
result that has been previously supported experimentally.34,45
Conversely, oxidative addition placing the ethyl moiety at the
C2 position (Scheme 4, pathway b) gives two possible
endocyclic β-hydride transfer pathways, which result in 3-
methylpent-1-ene and 2-ethylbut-1-ene.
Alternatively, a Cossee mechanism can also be invoked to
explain the product distribution. This is somewhat more
complicated, as both the order and geometry of insertion
govern the resultant product selectivity and isotopomer
distribution. Starting from a titanium hydride, ethylene
insertion yields an ethyl-titanium moiety. Insertion of 1-butene
into this moiety, with subsequent β-hydride transfer, can result
in either 2-ethylbut-1-ene (Scheme 5, pathway a) or 2-hexene
Figure 3. Experimental and predicted mass spectrum of 1-butene
produced from C2H4/C2D4 with Ti(OBu)4/AlMe3: (a) Cossee
mechanism and (b) metallacyclic mechanism.
Scheme 5. Co-dimerization of Ethylene and 1-Butene via a
Cossee Mechanism
significantly from a metallacyclic distribution (Figure 3b),
particularly for molecular ions at m/z 57, 61, and 63, where a
negligent contribution would be expected to occur.60 These
findings strongly disfavor the prevailing metallacycle mecha-
nism and are wholly consistent with a conventional Cossee
mechanism for 1-butene formation.61 The existence of an
extremely close fit between the experimental and calculated
distributions indicates that a number of approximations built
into the model are accurate. First, a kinetic isotope effect
related to 1-butene formation is small to absent. Second, the
ratio of H:D in a given isotopomer is the controlling factor in
determining the mass spectral fragmentation pattern of that
species, without needing to introduce a kinetic isotope effect for
this fragmentation process (Supporting Information). None-
theless, this latter approximation can be eliminated entirely by
considering the ethylene/1-butene co-dimerization products,
which do not suffer from the complex fragmentation pattern of
1-butene. This analysis is discussed below.
Analysis of the C6 Oligomers. The byproducts of the
Ti(OR′)4/AlR3 system have been shown to consist of
significant portions of 2-ethylbut-1-ene, 3-methylpent-1-ene,
1-hexene, and 2-hexene as well as other minor C6 constituents.
In previous reports these oligomers have been rationalized by
the co-dimerization of one unit of ethylene with one unit of 1-
butene by a metallacycle mechanism.34,35 Careful consideration
and 3-hexene (Scheme 5, pathway b) depending on the 1-
butene insertion geometry. If, however, 1-butene were to
reinsert into a titanium hydride followed by ethylene insertion
and β-hydride transfer, then 1-hexene (Scheme 5, pathway c)
and 3-methylpent-1-ene (Scheme 5, pathway d) are generated.
Given that such oligomers are predicted to form from the
same dimerization mechanism and that they do not suffer from
D
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
the same mass spectral fragmentation issue that complicates 1-
butene, isotopomer analysis was also undertaken. As previous
studies have shown, for C6 products formed via a metallacyle
mechanism only the isotopomers C6H12, C6H8D4, C6H4D8, and
C6D12 should form in a ratio of 1:3:3:1.22,23 However,
isotopomers that result from a Cossee mechanism undergo
H/D scrambling as discussed above. The distribution in this
case is also dependent on the order of insertion during co-
dimerization; insertion of 1-butene into a metal hydride
followed by ethylene insertion yields a theoretical distribution
that can be loosely termed as bimodal, whereas if ethylene
insertion occurs prior to 1-butene insertion, then a more
Gaussian-like distribution results (see Supporting Information
for derivation).
Figure 6. Experimental and predicted mass spectrum of 2-ethylbut-1-
ene produced from C2H4/C2D4 with Ti(OBu)4/AlMe3.
Isotopomer analysis of 1-hexene and 3-methylpent-1-ene,
from the above detailed experiment with Ti(OBu)4/AlMe3 and
C2H4/C2D4, revealed an excellent fit with the theoretical
distributions predicted by a Cossee mechanism involving 1-
butene insertion and subsequent ethylene insertion (Figures 4
same mechanism; however the mechanism is most likely a
Cossee mechanism rather than the favored metallacyclic
mechanism.
One point that warrants some discussion is the mode of
formation of 1-hexene. While ethylene/1-butene co-dimeriza-
tion is typically involked to account for this, 1-hexene can also
result simply from three successive insertions of ethylene
(trimerization). Both routes involve a common intermediate, a
Ti-butyl species, into which ethylene inserts and 1-hexene is
then eliminated. Interestingly, the results shown in Figure 4
reveal that 1-butene is primarily formed by reinsertion of
scrambled 1-butene. A much poorer fit to the experimental data
is obtained for the case in which 1-hexene is formed by three
consecutive ethylene insertions (see Supporting Information),
although the theoretical distributions are similar enough that a
contribution from this route is quite possible. This conclusion is
supported by an experiment in which 1-pentene was added to
the reactor; 1-heptene was formed through co-dimerization of
ethylene and pentene, in addition to the expected brached co-
dimers. The result can be rationalized by considering the rate of
β-hydride elimination from the Ti-butyl intermediate relative to
the rate of ethylene insertion into the same intermediate. When
the rate of β-hydride elimination is much faster than ethylene
insertion (as is the case here, because the catalyst is selective for
1-butene), formation of H/D scrambled 1-butene will
preferentially occur before further ethylene insertion. As such,
the 1-hexene that does form for the most part incorporates
scrambled 1-butene.
Figure 4. Experimental and predicted mass spectrum of 1-hexene
produced from C2H4/C2D4 with Ti(OBu)4/AlMe3.
In light of these results, it is also possible to speculate as to
why 2-ethylbut-1-ene is the favored co-dimerization product. It
is known that β-hydride elimination from an ethyl chain has a
higher barrier than from longer chains (propyl and higher),62
due to stabilization of the product olefin by alkyl substitution.63
Thus, if 1-butene were to insert first (paths c and d, Scheme 5),
rapid elimination of 1-butene (the reverse reaction) can be
anticipated to largely outcompete insertion of ethylene, as we
have discussed above. This is precisely the reason that the
system is selective for 1-butene, although the near absence of 2-
butenes is somewhat puzzling given the amount of 3-
methylpent-1-ene formed. If ethylene inserts first (paths a
and b, Scheme 5), then the Ti-ethyl may be more stable against
the reverse reaction, and further insertion (ethylene or 1-
butene) occurs preferentially. The dominant formation of 2-
ethylbut-1-ene over internal linear hexenes most likely results
from a favored 1-butene insertion geometry that is under steric
control. Finally, such a situation also explains why H/D
scrambling in the C2H4/C2D4 mixed monomers is not observed
Figure 5. Experimental and predicted mass spectrum of 3-methylpent-
1-ene produced from C2H4/C2D4 with Ti(OBu)4/AlMe3.
and 5). Additionally, analysis of the 2-ethylbut-1-ene formed in
the reaction gives a strong correlation with the distribution
predicted for a Cossee mechanism of ethylene insertion
followed by subsequent 1-butene insertion (Figure 6). As the
isotopomer distributions and resulting mass spectra of the co-
dimers depend on both the order and geometry of insertion,
these results also show that H/D scrambling does not occur
following oligomer formation.60 In each of these cases, a
metallacycle mechanism leads to a very poor fit to the
experimental data (see Supporting Information). These results
confirm that dimerization and co-dimerization do occur by the
E
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(as it has been previously55), because reversible insertion and β-
hydride elimination of ethylene are not occurring.
bar of ethylene pressure, while the temperature was maintained, for a
30 min run time. Upon completion, the reactor was cooled to <10 °C,
to avoid loss of any volatile butenes, before purging of the excess
ethylene pressure. Injection of a weighed nonane standard for
quantification was then introduced before careful quenching with a
10% HCl solution. Samples of the organic phase were filtered through
glass fiber filter paper to remove any particulates and analyzed by GC-
FID and where appropriate GC-MS.
C2H4/C2D4 Co-oligomerization: Triethlyaluminum. To a 100
mL round-bottom Schlenk flask was added titanium butoxide (0.15
mmol), triethylaluminum (0.45 mmol), and toluene (20 mL total
volume). The resulting mixture was cooled to 0 °C and stirred under a
replenishing flow of 1:1 ethylene/ethylene-d4 at 1 atm of pressure.
After 30 min a weighed amount of nonane was injected for
quantification, and the catalysis mixture was quenched with
approximately 10 mL of deionized water. The organic phase was
sampled, filtered through glass fiber filter paper to remove any
particulates, and analyzed by GC-FID and GC-MS.
C2H4/C2D4 Co-oligomerization: Trimethlyaluminum. In order
to generate sufficient amounts of the trace co-dimers for accurate
analysis, the co-oligomerization of C2H4/C2D4 with Ti(OBu)4/AlMe3
was carried out under more concentrated conditions and elevated
pressure. Oligomerization was undertaken in an approximately 5 mL
microreactor prepared from a 6 mm Swagelok tee piece fitted with
needle valves that was predried under full pump vacuum for 12 h. This
was charged with titanium butoxide (0.15 mmol) and trimethylalu-
minum (0.45 mmol) to give a total toluene volume of 1 mL. The
reaction mixture was heated to 55 °C and exposed to 4.5 bar of 1:1
ethylene/ethylene-d4 for a duration of 45 min. Upon completion the
reactor was cooled to 0 °C and excess gas released. The liquid phase
was diluted with approximately 5 mL of ice cold toluene, quenched
with iced deionized water, and sampled for GC-FID and GC-MS
analysis of the isotopomer distribution.
CONCLUSION
■
We have in the past questioned the generally held assumption
that alkoxytitanium dimerization catalysts operate via a
metallacycle mechanism.5 At the same time, little solid evidence
existed for a Cossee mechanism either. Herein, we have now
shown that metallacycles are not likely responsible for the
selective formation of dimers with this catalyst. The simplest
explanation for the selectivity of this system is via a
conventional linear chain insertion/β-hydride chain transfer
process (Cossee mechanism), in which the rate of chain
transfer is much faster than the rate of chain propagation. Such
a mechanism is well supported by the results presented herein.
We believe this work should, in many cases, lead to a re-
evaluation of the textbook mechanistic explanation for the
alkoxytitanium/trialkylaluminum catalyst system. Nonetheless,
a number of uncertainties and further work remain. What are
relatively simple precatalysts become much more complex upon
activation, and as such, the structure and oxidation state of the
active species are not well understood.2 Insights from
theoretical chemistry would also be valuable, but are difficult
given these uncertainties in the identity of the active species.
EXPERIMENTAL SECTION
■
General Procedures. All manipulations were performed under an
argon atmosphere using standard Schlenk techniques. Solvents were
purified by passage through an Innovative Technologies solvent
purification system and, where appropriate, were stored over a sodium
mirror. Ethylene (BOC, C.P. Grade) was purified by passage through a
column of activated molecular sieves (3 Å) and alumina. Ethylene-d4
(98%) was purchased from Cambridge Isotope Laboratories, Inc. and
used as received. Titanium(IV) butoxide (97%) and trimethylalumi-
num (97%) were purchased from Sigma-Aldrich and prepared as 0.06
and 2.0 M toluene solutions, respectively. Triethylaluminum (1.9 M in
toluene) was purchased from Sigma-Aldrich and used as received.
GC-FID and GC-MS Analysis. Oligomer quantification was carried
out by GC-FID on a Shimadzu CG-2014 chromatograph fitted with a
BP1 column (25 m × 0.32 mm internal diameter and 0.5 μm film).
The carrier gas was helium with a flow rate of 3.0 mL per minute. The
column was held at 40 °C for 2 min and then ramped to 280 at 10 °C
per minute. Quantification was achieved by comparison to a known
mass of nonane standard. Identification of oligomers was carried out
by GC-MS. Analyses were carried out on a Varian 3800 GC coupled to
a Bruker 300 triple-quadrupole mass spectrometer in single-quadru-
pole mode. The column was a Varian “Factor Four” VF-5 ms (30 m ×
0.25 mm internal diameter and 0.25 μm film). Injections of 1 μL of
diluted samples were made using a Varian CP-8400 autosampler and a
Varian 1177 split/splitless injector at 240 °C with a split ratio of 50:1.
The ion source was at 220 °C and the transfer line at 290 °C. The
carrier gas was helium at 1.2 mL/min using constant flow mode. For
the separation of 1-hexene and 2-ethylbut-1-ene the column oven was
held at 5 °C for 2 min and then ramped to 150 °C at 8 °C/min. For all
other separations the column oven was held at 25 °C for 2 min and
then ramped to 150 °C at 8 °C/min. The range from m/z 35 to 300
was scanned four times per second. 2-Ethylbut-1-ene and 3-
methylpent-1-ene were identified by comparison to injections of
standards, while other oligomers were confirmed by a combination of
ionization spectra and Kovats’ retention indices.
ASSOCIATED CONTENT
* Supporting Information
■
S
Full details of calculated mass spectra for all products, further
comparison of experimental data with a metallacycle mecha-
nism, and raw mass spectra of all products. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: david.mcguinness@utas.edu.au.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The Australian Research Council is thanked for financial
support through grant FT100100609 to D.S.M. and an
Australian Postgraduate Award to J.A.S. Noel Davies of the
Central Science Laboratory, University of Tasmania, is thanked
for GC-MS analysis. Edward Doddridge is thanked for helpful
discussion on the mathematical treatment of 1-butene
fragmentation.
REFERENCES
■
(1) Forestier
̀
e, A.; Olivier-Bourbigou, H.; Saussine, L. Oil Gas Sci.
Technol. Rev. IFP 2009, 64, 649.
(2) Al-Sa’doun, A. W. App. Catal. A: Gen. 1993, 105, 1.
(3) Dixon, J. T.; Green, M. J.; Hess, F. M.; Morgan, D. H. J.
Organomet. Chem. 2004, 689, 3641.
(4) Wass, D. F. Dalton Trans. 2007, 816.
(5) McGuinness, D. S. Chem. Rev. 2011, 111, 2321.
Ethylene Oligomerization. In a typical run, dimerization was
conducted in a 0.3 L stainless steel Parr 5500 Compact Mini Reactor
that had been preheated to 120 °C and flushed with four vacuum/
argon cycles and two ethylene purges. The reactor was then cooled to
55 °C and loaded with toluene (total volume of 20 mL), titanium
butoxide (0.15 mmol), and alkylaluminum (0.75 mmol). Stirring was
commenced, and the reaction mixture was exposed to a constant 20
(6) van Leeuwen, P. W. N. M.; Clem
́
ent, N. D.; Tschan, M. J.-L.
Coord. Chem. Rev. 2011, 255, 1499.
F
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(7) Agapie, T. Coord. Chem. Rev. 2011, 255, 861.
(42) McDermott, J. X.; Wilson, M. E.; Whitesides, G. M. J. Am.
Chem. Soc. 1976, 98, 6529.
(8) Ziegler, K.; Martin, H. U.S. Patent 2943125 1960.
(9) Yamada, S.; Ono, I. Bull. Japan Pet. Inst. 1970, 12, 160.
(10) Grasset, F.; Cazaux, J.-B.; Magna, L.; Braunstein, P.; Oliver-
Bourbigou, H. Dalton Trans. 2012, 41, 10396.
(11) Cazaux, J.-B.; Braunstein, P.; Magna, L.; Saussine, L.; Oliver-
Bourbigou, H. Eur. J. Inorg. Chem. 2009, 2942.
(43) Suttil, J. A.; McGuinness, D. S.; Pichler, M.; Gardiner, M. G.;
Morgan, D. H.; Evans, S. J. Dalton Trans. 2012, 41, 6625.
(44) You, Y.; Girolami, G. S. Organometallics 2008, 27, 3172.
(45) Datta, S.; Fischer, M. B.; Wreford, S. S. J. Organomet. Chem.
1980, 188, 353.
(46) Spencer, M. D.; Morse, P. M.; Wilson, S. R.; Girolami, G. S. J.
Am. Chem. Soc. 1993, 115, 2057.
(47) Cohen, S. A.; Auburn, P. R.; Bercaw, J. E. J. Am. Chem. Soc.
1983, 105, 1136.
(48) McDermott, J. X.; White, J. F.; Whitesides, G. M. J. Am. Chem.
Soc. 1973, 95, 4451.
(12) The Alphabutol process is licensed by Axens; see www.axens.
net.
(13) Janse van Rensburg, W.; Grove, C.; Steynberg, J. P.; Stark, K. B.;
Huyser, J. J.; Steynberg, P. J. Organometallics 2004, 23, 1207.
(14) Jabri, A.; Mason, C. B.; Sim, Y.; Gambarotta, S.; Burchell, T. J.;
Duchateau, R. Angew. Chem., Int. Ed. 2008, 47, 9717.
(15) Vidyaratne, I.; Nikiforov, G. B.; Gorelsky, S. I.; Gambarotta, S.;
Duchateau, R.; Korobkov, I. Angew. Chem., Int. Ed. 2009, 48, 6552.
(16) Skobelev, I. Y.; Panchenko, V. N.; Lyakin, O. Y.; Bryliakov, K.
P.; Zakharov, V. A.; Talsi, E. P. Organometallics 2010, 29, 2943.
(17) Budzelaar, P. H. M. Can. J. Chem. 2009, 87, 832.
(18) McGuinness, D. S.; Brown, D. B.; Tooze, R. P.; Hess, F. M.;
Dixon, J. T.; Slawin, A. M. Z. Organometallics 2006, 25, 3605.
(19) Jabri, A.; Temple, C.; Crewdson, P.; Gambarotta, S.; Korobkov,
I.; Duchateau, R. J. Am. Chem. Soc. 2006, 128, 9238.
(20) Temple, C.; Jabri, A.; Crewdson, P.; Gambarotta, S.; Korobkov,
I.; Duchateau, R. Angew. Chem. 2006, 118, 7208.
(21) Yang, Y.; Liu, Z.; Zhong, L.; Qiu, P.; Dong, Q.; Cheng, R.;
Vanderbilt, J.; Liu, B. Organometallics 2011, 30, 5297.
(22) Agapie, T.; Schofer, S. J.; Labinger, J. A.; Bercaw, J. E. J. Am.
Chem. Soc. 2004, 126, 1304.
(23) Agapie, T.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2007,
129, 14281.
(49) McDermott, J. X.; White, J. F.; Whitesides, G. M. J. Am. Chem.
Soc. 1976, 98, 6521.
(50) Emrich, R.; Heinemann, O.; Jolly, P. W.; Kruger, C.; Verhovnik,
̈
G. P. J. Organometallics 1997, 16, 1511.
(51) Chauvin, Y. Angew. Chem., Int. Ed. 2006, 45, 3741.
(52) Speiser, F.; Braunstein, P.; Saussine, L. Acc. Chem. Res. 2005, 38,
784.
(53) Tomov, A. K.; Chirinos, J. J.; Jones, D. J.; Long, R. J.; Gibson, V.
C. J. Am. Chem. Soc. 2005, 127, 10166.
(54) Tomov, A. K.; Chirinos, J. J.; Long, R. J.; Gibson, V. C.;
Elsegood, M. R. J. J. Am. Chem. Soc. 2006, 128, 7704.
(55) Tomov, A. K.; Gibson, V. C.; Britovsek, G. J. P.; Long, R. J.; van
Meurs, M.; Jones, D. J.; Tellmann, K. P.; Chirinos, J. J. Organometallics
2009, 28, 7033.
(56) McGuinness, D. S.; Suttil, J. A.; Gardiner, M. G.; Davies, N. W.
Organometallics 2008, 27, 4238.
(57) Belov, G. P.; Dzhabiev, T. S.; Kolesnikov, I. M. J. Mol. Catal.
1982, 14, 105.
(24) Overett, M. J.; Blann, K.; Bollmann, A.; Dixon, J. T.; Haasbroek,
D.; Killian, E.; Maumela, H.; McGuinness, D. S.; Morgan, D. H. J. Am.
Chem. Soc. 2005, 127, 10723.
(25) Schofer, S. J.; Day, M. W.; Henling, L. M.; Labinger, J. A.;
Bercaw, J. E. Organometallics 2006, 25, 2743.
(26) Rucklidge, A. J.; McGuinness, D. S.; Tooze, R. P.; Slawin, A. M.
Z.; Pelletier, J. D. A.; Hanton, M. J.; Webb, P. B. Organometallics 2007,
26, 2782.
(58) For a Cossee mechanism, chain termination can occur either via
a stepwise β-hydride elimination (yielding a discrete M-H intermediate
as shown in Scheme 3) or via concerted β-hydride transfer to ethylene.
Either case yields the same result and does not affect the isotopomer
distribution.
(59) Throughout this paper, the notation [M − 1]+ refers to loss of
one hydrogen or deuterium atom, rather than loss of one atomic mass
unit. Likewise, [M − 2]+ refers to [M − 2H]+, [M − 2D]+ or [M −
HD]+, and so forth for [M − 3]+, etc.
(27) Bruckner, A.; Jabor, J. K.; McConnell, A. E. C.; Webb, P. B.
̈
Organometallics 2008, 27, 3849.
(60) It was suggested by a reviewer that H/D scrambling in the 1-
butene could occur after it is formed. This possibility was discounted
by adding 1-pentene to a C2H4/C2D4 co-dimerization experiment. No
incorporation of deuterium into the 1-pentene was evident by mass
spectroscopy (see Supporting Information). It is also noted that such a
process of scrambling could not lead to the distinct mass spectral
patterns observed for the C6 co-dimers.
(28) Janse van Rensburg, W.; van den Berg, J.-A.; Steynberg, P. J.
Organometallics 2007, 26, 1000.
(29) Blok, A. N. J.; Budzelaar, P. H. M.; Gal, A. W. Organometallics
2003, 22, 2564.
(30) de Bruin, T. J. M.; Magna, L.; Raybaud, P.; Toulhoat, H.
Organometallics 2003, 22, 3404.
(61) A reviewer raised the possibility that the Ti(OBu)4/AlEt3
system could operate via a metallacycle mechanism even though the
Ti(OBu)4/AlMe3 system does not. While this possibility cannot be
completely discounted, we do not consider it a high probability, given
the very close similarities between the two systems and their product
outputs.
(62) Budzelaar, P. H. M.; Talarico, G. Struct. Bonding (Berlin) 2003,
105, 141.
(63) Approximately half of this stabilization is felt at the transition
state; see ref 62.
(31) Tobisch, S.; Ziegler, T. Organometallics 2003, 22, 5392.
(32) Suttil, J. A.; McGuinness, D. S.; Evans, S. J. Dalton Trans. 2010,
39, 5278.
(33) Do, L. H.; Labinger, J. A.; Bercaw, J. E. Organometallics,
published online 2012, dx.doi.org/10.1021/om300492r.
(34) Bre, A.; Chauvin, Y.; Commereuc, D. Nouv. J. Chim. 1986, 10,
535.
(35) Chauvin, Y.; Olivier, H. In Applied Homogenous Catalysis with
Organometallic Compounds; Cornils, B., Herrmann, W. A., Eds.; VCH:
Weinheim, 1996; Vol. 1, p 258.
(36) Al-Jarallah, A. M.; Anabtawi, J. A.; Siddiqui, M. A. B.; Aitani, A.
M.; Al-Sa’doun, A. W. Catal. Today 1992, 14, 1.
(37) Olivier-Bourbigou, H.; Forestier
Favre, F.; Hugues, F. Oil Gas Eur. Mag. 2010, 36, 97.
(38) Commereuc, D.; Chauvin, Y.; Gaillard, J.; Leonard, J.; Andrews,
J. Hydrocarb. Process. 1984, 63, 118.
̀
e, A.; Saussine, L.; Magna, L.;
́
(39) Hill, J. E.; Balaich, G. J.; Fanwick, P. E.; Rothwell, I. P.
Organometallics 1991, 10, 3428.
(40) Thorn, M. G.; Hill, J. E.; Waratuke, S. A.; Johnson, E. S.;
Fanwick, P. E.; Rothwell, I. P. J. Am. Chem. Soc. 1997, 119, 8630.
(41) McDermott, J. X.; Whitesides, G. M. J. Am. Chem. Soc. 1974, 96,
947.
G
dx.doi.org/10.1021/om3008508 | Organometallics XXXX, XXX, XXX−XXX