Molecular understanding of alumina supported single-site catalysts by a combination of experiment and theory

Article abstract of DOI:10.1021/ja0616736

The nature and structure of grafted organometallic complexes on γ-alumina are studied from a combination of experimental data (mass balance analysis, IR, NMR) and density functional theory calculations. The chemisorptive interactions of two complexes are analyzed and compared. The reaction of [Zr(CH₂Su)₄] with alumina dehydroxylated at 500 °C gives {[(Al_sO)₂Zr(CH₂tBu)] ⁺[(tBuCH₂)(Al_s)]^-}, a bisgrafted cationic complex as major surface species. The DFT calculations show that the reaction with surface hydroxyls is very exothermic and that alkyl transfer on Al atoms is favored. In contrast, [W(≡CtBu)(CH₂tBu)₃] reacts with an alumina treated under identical conditions to give selectively a monografted neutral surface complex, [(Al_sO)W(≡CtBu)(CH ₂tBu)₂], This was inferred by the evolution of 1 equiv of tBuCH₃ per grafted W and the presence of remaining hydroxyls. The calculations show that the reaction of [W(≡CtBu)(CH₂tBu) ₃] with surface hydroxyls is in fact less exothermic and has a considerably higher activation barrier than the one of the Zr complex. Additionally, the transfer of an alkyl ligand onto an adjacent Al center is disfavored, and hence cationic species are not formed. Some ligands of this monoaluminoxy surface complex interact with remaining surface hydroxyls, which explains the complexity of the experimental NMR and IR data.

Full text of DOI:10.1021/ja0616736

Published on Web 06/22/2006
Molecular Understanding of Alumina Supported Single-Site
Catalysts by a Combination of Experiment and Theory
Je´roˆme Joubert,† Franc¸oise Delbecq,† Philippe Sautet,*,† Erwan Le Roux,‡
Mostafa Taoufik,‡ Chloe´ Thieuleux,‡ Fre´de´ric Blanc,‡ Christophe Cope´ret,‡
Jean Thivolle-Cazat,‡ and Jean-Marie Basset*,‡
Contribution from the Laboratoire de Chimie (UMR 5182 CNRS/ENS), Ecole Normale
Supe´rieure de Lyon, 46 Alle´e d’Italie, F-69364 Lyon Cedex 07, France, and the Laboratoire de
Chimie Organome´tallique de Surface (UMR 9986 CNRS/ESCPE Lyon), ESCPE Lyon,
F-308-43 BouleVard du 11 NoVembre 1918, F-69616 Villeurbanne Cedex, France
Received March 10, 2006; E-mail: Philippe.Sautet@ens-lyon.fr; basset@cpe.fr
Abstract: The nature and structure of grafted organometallic complexes on γ-alumina are studied from a
combination of experimental data (mass balance analysis, IR, NMR) and density functional theory
calculations. The chemisorptive interactions of two complexes are analyzed and compared. The reaction
of [Zr(CH2tBu)4] with alumina dehydroxylated at 500 °C gives {[(AlsO)2Zr(CH2tBu)]+[(tBuCH2)(Als)]-}, a
bisgrafted cationic complex as major surface species. The DFT calculations show that the reaction
with surface hydroxyls is very exothermic and that alkyl transfer on Al atoms is favored. In contrast,
[W(tCtBu)(CH2tBu)3] reacts with an alumina treated under identical conditions to give selectively a
monografted neutral surface complex, [(AlsO)W(tCtBu)(CH2tBu)2]. This was inferred by the evolution of 1
equiv of tBuCH3 per grafted W and the presence of remaining hydroxyls. The calculations show that the
reaction of [W(tCtBu)(CH2tBu)3] with surface hydroxyls is in fact less exothermic and has a considerably
higher activation barrier than the one of the Zr complex. Additionally, the transfer of an alkyl ligand onto an
adjacent Al center is disfavored, and hence cationic species are not formed. Some ligands of this
monoaluminoxy surface complex interact with remaining surface hydroxyls, which explains the complexity
of the experimental NMR and IR data.
Introduction
characterization of such a heterogeneous catalyst is indeed very
difficult. This molecular understanding of the structure of surface
The need for a sustainable and cleaner chemistry drives an
important demand for highly efficient catalysts, which should
be simultaneously active, selective for the desired product, and
easily recycled. One successful direction toward this aim is to
graft well-defined organometallic complexes on a large surface
area oxide support, used as a solid ligand. This has been inspired
by the success of homogeneous catalysis, which relies on the
structure-reactivity relationship as a method to develop better
catalytic systems, and which typically enters directly a catalytic
cycle through the preparation of well-defined reaction interme-
diates.1 This approach has already brought considerable success
for olefin polymerization, hydrogenation, or metathesis.2,3
complexes is critically needed if one wants to undertake a
rational development of heterogeneous catalysts with a control
on reaction selectivity. One first important question is the nature
of the surface organometallic complex, in terms of molecular
ligands remaining on the complex and number and type of bonds
established with the oxide support (solid ligand). The cationic
or neutral character of the surface metal species is also a key
aspect for its catalytic properties, not easily accessible from
experiment alone. Finally, one wants to understand which sites
of the oxide surface are most appropriate for the grafting process.
Do we really form a single well-defined surface/metal/ligands
complex with a unique reactivity or can several species
simultaneously be formed?
However, in heterogeneous catalysis, a fundamental under-
standing of the chemisorptive interaction and reaction of the
metal-organic complex with the oxide surface is still lacking
although Surface Organometallic Chemistry is designed to
produce well-defined surface metallic species. The structural
The choice of the oxide support plays a critical role. After
years of research, we have shown that silica could be used as
a tunable solid ligand, and the chemistry is now understood at
a molecular level. Typically one can generate either monosiloxy,
bissiloxy, and even in some cases trissiloxy surface complexes.
These systems have been characterized by several methods
(elemental analysis, IR, NMR, and EXAFS), and they have been
used as catalyst precursors in several reactions (hydrogenation,
polymerization, depolymerization, olefin metathesis, oxidation,
and alkane metathesis). Typically, the siloxy substituent of the
† Laboratoire de Chimie, ENS Lyon.
‡ Laboratoire de Chimie Organome´tallique de Surface, ESCPE Lyon.
(1) Cope´ret, C.; Chabanas, M.; Petroff Saint-Arroman, R.; Basset, J.-M. Angew.
Chem., Int. Ed. 2003, 42, 156.
(2) Cornils, B.; Herrmann, W. A. Applied Homogeneous Catalysis with
Organometallic Compounds, 2nd ed.; Wiley-VCH: 2002; Vol. 1.
(3) Cornils, B.; Herrmann, W. A. Applied Homogeneous Catalysis with
Organometallic Compounds, 2nd ed.; Wiley-VCH: 2002; Vol. 2.
9
10.1021/ja0616736 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 9157-9169
9157

A R T I C L E S
Joubert et al.
surface activates the metal center by generating more electro-
philic metal centers,1,4-6 but in some instances, it is necessary
to use other supports such as alumina to obtain a gain in
reactivity. Several hydrogenation, metathesis, and polymeriza-
tion supported catalysts have been prepared, for which the
alumina support is critical to obtain catalytic activity. For
example, chemisorption of [Zr(CH2tBu)4] on alumina provides
an efficient olefin polymerization catalyst, while when supported
on silica it is inactive.7-9 Alumina has a richer chemistry than
silica since it combines a variety of hydroxyl groups, with
several modes of coordination, with a Lewis acid character on
the aluminum. Various types of aluminum atoms are potentially
present on the surface, from tri- to pentacoordinated, originating
from the tetrahedral or octahedral cations of bulk γ-alumina.10,11
The role of the aluminum Lewis centers for the chemisorptive
interaction of the metal-organic complex and more generally
the specificity of alumina for this process is still unclear. In
some cases the formation of partially or fully cationic species
has been proposed based on NMR spectroscopy.7-9,12-21 Most
recently, we have found that well-defined W carbyne complexes
are inactive catalyst precursors for alkane metathesis when
supported on silica, while they become active when supported
on alumina.22
Whether in olefin polymerization or in alkane metathesis, the
development of better catalysts is still a challenge, which
requires a better understanding of the structure of active sites
on this type of support. Therefore, we have decided to study
the reactivity of alumina toward two organometallic reagents:
[Zr(CH2tBu)4] (1) and [W(tCtBu)(CH2tBu)3] (2). Our approach
combines a characterization of the reaction products (mass
balance analysis), a spectroscopic study of the surface species
(NMR and in situ IR), and a quantum chemical exploration of
reaction pathways and stable intermediates for the grafting of
the metal-organic species on alumina. For an optimal consis-
tency of the approach, the modeling part includes the simulation
of spectroscopic NMR and IR data. It is based on a model of a
partially hydrated γ-alumina surface which has been previously
developed and validated.11 Our specific aim is to compare the
Figure 1. Monitoring the reaction of [Zr(CH2tBu)4] and Al2O3-(500) (100
mg) by IR spectroscopy: (a) (i) (Al2O3-(500)) and (ii) 1/Al2O3. (b) Zoom in
the ν(OH) region [4000-3400 cm-1] of (i) (Al2O3-(500)) and (ii) 1/Al2O3.
(c) Subtracted spectrum.
grafting mechanism for these two complexes on alumina, and
a special focus is put on the existence (or non existence) of
cationic metallic species and on the unique (or non unique)
character of the formed surface entity.
Results and Discussions
1. Analytical Results on the Reaction of [Zr(CH2tBu)4]
(1) with Alumina Partially Dehydroxylated at 500 °C. 1.1.
In Situ IR Study. Monitoring by IR the reaction of an excess
of 1, [Zr(CH2tBu)4], on an alumina disk partially dehydroxylated
at 500 °C (Al2O3-(500)), shows that all the ν(OH) vibrations are
affected: the bands at 3795, 3776, and 3730 cm-1 (vide infra
for assignment) have mostly disappeared (70%), whereas the
bands at 3695 cm-1 are partially consumed (Figure 1). This is
consistent with the existence of OH groups not accessible to 1.
Moreover, the characteristic bands in the 3000-2700 cm-1 and
1500-1300 cm-1 regions associated with the ν(CH) and δ(CH)
of the perhydrocarbyl ligands around Zr have also appeared.
1.2. Mass Balance Analyses. After 1 is contacted at room
temperature with γ-Al2O3-(500), prepared from Boehmite, in
pentane for 2 h, a solid [1/Al2O3]B is obtained along with 1.8
equiv of tBuCH3/grafted Zr. After this solid was washed and
dried, elemental analysis shows the presence of Zr to the extent
of 2.8%wt, which corresponds to 0.31 mmol of Zr/g. As 1.8
equiv of tBuCH3 evolved per grafted Zr, 1.8 OH are therefore
consumed, and thereby 0.56 mmol of OH/g of alumina, in
agreement with the disappearance of a large fraction (50%) of
the surface hydroxyls (1.1 mmol of OH/g in the initial
γ-Al2O3-(500)). Moreover, 11 C/Zr are found by elemental
analysis, which is consistent with the quantity of tBuCH3
evolved upon grafting. Similarly, using a γ-alumina from
Degussa, 2.0 equiv of tBuCH3 are formed per grafted Zr, about
50% of the OH are consumed, and the resulting complex
contains around 11 carbons/Zr. Overall, the data are consistent
with the formation of [(AlsO)2Zr(CH2tBu)2]DA, 3, as the major
(4) Ballard, D. G. H. AdV. Catal. 1973, 23, 263.
(5) Cope´ret, C. New. J. Chem. 2004, 28, 1.
(6) Solans-Monfort, X.; Clot, E.; Coperet, C.; Eisenstein, O. J. Am. Chem.
Soc. 2005, 127, 14015.
(7) Tullock, C. W.; Tebbe, F. N.; Mulhaupt, R.; Ovenall, D. W.; Setterquist,
R. A.; Ittel, S. D. J. Polym. Sci., A 1989, 27, 3063.
(8) Collette, J. W.; Tullock, C. W.; MacDonald, R. N.; Buck, W. H.; Su, A.
C. L.; Harrell, J. R.; Mulhaupt, R.; Anderson, B. C. Macromolecules 1989,
22, 3851.
(9) Tullock, C. W.; Mulhaupt, R.; Ittel, S. D. Makromol. Chem., Rapid
Commun. 1989, 10, 19.
(10) Digne, M.; Sautet, P.; Raybaud, P.; Toulhoat, H.; Artacho, E. J. Phys. Chem.
B 2002, 106, 5155.
(11) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. J. Catal. 2002,
211, 1.
(12) Burwell, R. L., Jr. J. Catal. 1984, 86, 301.
(13) Ballard, D. G. H. Coord. Polym. 1975, 223.
(14) Ballard, D. G. H. J. Polym. Sci., Polym. Chem. Ed. 1975, 13, 2191.
(15) Toscano, P. J.; Marks, T. J. J. Am. Chem. Soc. 1985, 107, 653.
(16) Dahmen, K. H.; Hedden, D.; Burwell, R. L., Jr.; Marks, T. J. Langmuir
1988, 4, 1212.
(17) Gillespie, R. D.; Burwell, R. L., Jr.; Marks, T. J. Langmuir 1990, 6, 1465.
(18) Marks, T. J. Acc. Chem. Res. 1992, 25, 57.
(19) Ahn, H.; Marks, T. J. J. Am. Chem. Soc. 2002, 124, 7103.
(20) Buffon, R.; Leconte, M.; Choplin, A.; Basset, J.-M. J. Chem. Soc., Dalton
Trans. 1994, 1723.
(21) Jezequel, M.; Dufaud, V.; Ruiz-Garcia, M. J.; Carrillo-Hermosilla, F.;
Neugebauer, U.; Niccolai, G. P.; Lefebvre, F.; Bayard, F.; Corker, J.; Fiddy,
S.; Evans, J.; Broyer, J.-P.; Malinge, J.; Basset, J.-M. J. Am. Chem. Soc.
2001, 123, 3520.
(22) Le Roux, E.; Taoufik, M.; Coperet, C.; de Mallmann, A.; Thivolle-Cazat,
J.; Basset, J.-M.; Maunders, B. M.; Sunley, G. J. Angew. Chem., Int. Ed.
2005, 44, 6755.
9
9158 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
Scheme 1. Proposed Structures for the Reaction of 1 with
Al2O3-(500) (x ) 1, 2, or 3)
surface species, which contains 10 C/Zr and for which 2 equiv
of tBuCH3 are expected upon grafting (Scheme 1).
1.3. Solid-State NMR Spectroscopy. While no specific
1
feature is observed on the H NMR spectrum (a single broad
peak centered at 1 ppm, Figure S1), the 13C CP MAS solid state
NMR spectrum of the unlabeled complex presents an intense
peak at 32 ppm along with a shoulder at 29 ppm and two very
broad peaks of low intensity at 84 and 26 ppm (Figure 2a).
When the carbons directly attached to Zr in 1 are selectively
13C labeled (33%), the two broad peaks at 26 and 84 ppm
increase in intensity (Figure 2b), which is consistent with their
attribution to methylene groups of tBuCH2 fragments. Moreover,
the signal at 84 ppm is very broad (∆ν1/2 > 4 kHz) and displays
a shoulder at 99 ppm. The detailed attribution of NMR spectra
will be presented later in the theoretical section 3.5.
Figure 3. Monitoring the reaction of [W(tCtBu)(CH2tBu)3] and Al2O3-(500)
(100 mg) by IR spectroscopy: (a) (i) (Al2O3-(500)) and (ii) 2/Al2O3. (b) Zoom
in the ν(OH) region [4000-3400 cm-1] of (i) (Al2O3-(500)) and (ii) 2/Al2O3.
(c) Subtracted spectrum.
2. Analytical Results on the Reaction of [W(tCtBu)-
(CH2tBu)3] with Alumina Partially Dehydroxylated at 500
°C. 2.1. In Situ IR Study. Monitoring by IR the sublimation
of an excess of 2, [W(tCtBu)(CH2tBu)3], onto an alumina disk
partially dehydroxylated at 500 °C (γ-Al2O3-(500)) shows that
the ν(OH) vibrations at 3795, 3776, and 3730 cm-1 disappeared
(Figure 3), while some OHs appear as a large band centered at
3650 cm-1. This is very different from what was observed
during the grafting of [Zr(CH2tBu)4] (1) (vide supra, disappear-
ance of a large proportion of the OH bands and absence of bands
shifted at lower frequencies). Moreover, the characteristic bands
in the 3000-2700 cm-1 and 1500-1300 cm-1 regions associ-
ated with the ν(CH) and δ(CH) of the perhydrocarbyl ligands
around W also have appeared.
2.2. Mass Balance Analyses. The reaction of a mechanical
mixture of 2 and γ-Al2O3-(500) (prepared from Boehmite) at 66
°C for 4 h generates 0.9 equiv of tBuCH3/grafted W, and the
corresponding solid is washed 3 times with pentane before
drying under vacuum. Elemental analysis shows the presence
of W to the extent of 3.8 wt %, which corresponds to 0.20 mmol
of W/g. As 1 equiv of tBuCH3 evolved per grafted W, 0.20
mmol of OH/g of alumina has been consumed, in agreement
with the disappearance of a small fraction of the surface
hydroxyls as evidenced by IR spectroscopy (consumption of
18% of 1.1 mmol of OH/g in the initial Al2O3-(500)). Moreover,
an average of 14.7 C/W is found on the resulting solid by both
elemental analysis and hydrogenolysis. Overall, the data are
consistent with the formation of the surface complex [(AlsO)W-
(tCtBu)(CH2tBu)2]DA, 4 (Scheme 2), which contains 15C/W
and for which 1 tBuCH3 is expected during grafting.
Figure 2. 13C CP solid-state NMR spectrum of [Zr(CH2tBu)4/Al2O3-(500)
]
2.3. Solid-State NMR Spectroscopy. The 1H and 13C solid-
state NMR spectra of 2/Al2O3-(500) display, respectively, a single
peak centered at 1.0 ppm (Figure S2) and two peaks at 85 and
32 ppm (Figure 4a). When the compound 2 13C labeled on the
carbon R to the metal was used to prepare 2/Al2O3-(500), the
recorded under MAS of 10 kHz. The simulated and deconvoluted spectra
are shown below the experimental one. (a) 1/Al2O3-(500). The number of
scans was 56 000, and the recycle delay was set to 1 s. A CP step of 2 ms
was used. (b) 1*/Al2O3-(500) (33% labeled 13C on CH2). The number of
scans was 4096, and the recycle delay was set to 2 s. A CP step of 2 ms
was used.
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9159

A R T I C L E S
Joubert et al.
Scheme 2. Proposed Structures for the Reaction of 2 with
Al2O3-(500) Based on Chemical Analyses
13C CP/MAS NMR spectrum displays six major peaks at 318,
103, 95, 85, 52, and 32 ppm (Figure 4b, see deconvolution).
By comparison with the 13C NMR spectrum of the correspond-
ing silica supported W complex,24 it was clear that the carbyne
ligand was left intact during grafting as evidenced by the peak
at 318 and 52 ppm assigned, respectively, to the carbynic carbon
(tCtBu) and the quaternary carbon of the tBu directly attached
to this carbon (tCCMe3). Nonetheless, several species are
possibly present as evidenced by the broad peak at 318 ppm
and several peaks around 90 ppm (85, 95, and 103 ppm) and
30 ppm (32 and a shoulder at 29 ppm) attributed, respectively,
to several types of methylene carbon (CH2tBu) and several tBu
groups. Based on what was proposed on alumina supported
group 4 transition metal complexes,18,21 we proposed earlier the
presence of several surface species22 such as neutral and possibly
cationic surface species 4a-c (vide infra for further comments
and assignments).
3. Computional Study on the Reaction of Alumina with
ZrR4 (R ) Me and CH2tBu) and W(CR′)(R) (R′ ) R ) Me
and R′ ) tBu and R ) CH2tBu). 3.1. Model Surface. Since
γ-Al2O3 crystallites mainly expose (110) facets (between 70 and
83%), a periodic slab with (110) orientation was considered.
Moreover the other surface present on the crystallite, (100),
would not show any hydroxyl group at the considered pretreat-
ment temperature and, hence, would not be reactive for
grafting.11 Ab initio atomistic thermodynamic calculations
showed that in the experimental pretreatment conditions (500
°C) a coverage of 8.5 OH per nm2 is present on the (110)
surface,11 which corresponds to three adsorbed H2O molecules
per surface unit cell (5.6 OH per nm2 on γ-alumina). This OH
density is slightly overestimated compared to experimental
measurements (4 OH/nm2).25 On this most stable surface
structure, several OH adsorption sites are present (see Figure
5a), and this termination reproduces reasonably well the diversity
in the IR spectra of the real surface. It is hence a good sample
to probe the reactivity of various types of OH group for grafting
organometallic complexes.
Figure 4. 13C CP solid-state NMR spectrum of [W(tCtBu)(CH2tBu)3/
Al2O3-(500)] recorded under MAS of 10 kHz. The simulated and deconvo-
luted spectra are shown below the experimental one. (a) 2/Al2O3-(500). The
number of scans was 55 000, and the recycle delay was set to 1 s. A CP
step of 5 ms was used. (b) 2*/Al2O3-(500) (30% labeled 13C on CH2). The
number of scans was 55 000, and the recycle delay was set to 1 s. A CP
step of 5 ms was used.
orientation. Moreover, it has been proven that there is a
nonhomogeneous repartition of hydroxyl groups on the surface,26
which can explain both the overestimation of the hydroxyl
density and the lack of some frequencies in this periodic model.
A first intrinsic measure of the reactivity of a hydroxyl group
can be performed through the evaluation of its Lewis acidity,
i.e., the capability of the proton to interact with a Lewis base
without breaking the O-H bond. An example of such an
interaction is shown in Figure 5c in the case of pyridine as a
Lewis base. The adsorption energy of pyridine on the hydroxyl
group is characteristic of its acidity: the higher the absolute
value of adsorption energy, the more acidic the hydroxyl group.
These adsorption energies are reported in Table 1 together with
OH stretching frequencies prior to pyridine adsorption.
A clear trend emerges: the lower the Al atom coordination,
the higher the vibration frequency for the OH coordinated to
this Al atom. Hydrogen bonds affect drastically the vibrational
properties of hydroxyl groups with a marked decrease of the
stretching frequency. The adsorption strength of pyridine clearly
correlates with the force constant of the hydroxyl group, the
one with the largest Lewis acidity (i.e., the one which leads to
the strongest pyridine adsorption) corresponding to the OH
bonded to the tetrahedral Al (HO-µ1-AlIV). Conversely, the OH
sites corresponding to a low OH frequency (µ3 or hydrogen
bonded) do not interact with pyridine (the molecule moves to
It must be noticed that the complete description of the IR
spectrum of realistic samples is not possible with a single
(23) Nedez, C.; Theolier, A.; Lefebvre, F.; Choplin, A.; Basset, J. M.; Joly, J.
F. J. Am. Chem. Soc. 1993, 115, 722.
(24) Le Roux, E.; Taoufik, M.; Chabanas, M.; Alcor, D.; Baudouin, A.; Coperet,
C.; Thivolle-Cazat, J.; Basset, J.-M.; Lesage, A.; Hediger, S.; Emsley, L.
Organometallics 2005, 24, 4274.
(26) Metivier, R.; Leray, I.; Roy-Auberger, M.; Zanier-Szydlowski, N.; Valeur,
(25) Zhang, W.; Sun, M.; Prins, R. J. Phys. Chem. B 2002, 106, 11805.
B. New J. Chem. 2002, 26, 411.
9
9160 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
Figure 5. (a) Trihydrated γ-Al2O3 model surface. The roman numbers in index of Al is the coordination number. (b) Simplified representation of the same
surface that will be used below for clarity. (c) Adsorption of pyridine on OH-µ1-AlIV group.
an alumina surface can provide several species. First, the d0
Table 1. Adsorption Energies (Eads) of Pyridine on OH Groups
and Stretching Frequencies (νstr) of the Corresponding OH Groups
metal center can coordinate to the oxygen atom of the surface
Prior to Adsorption
hydroxyl, leading to a pentacoordinated metal center as pre-
1
1
OH group
E
ads (kJ mol-
)
ν )
str (cm-
sented in Figure 7a. In the case of Zr(CH3)4, such an intermedi-
HO-µ1-AlIV
-40
-31
-26 (free OH)
3787
3698
3490 (free OH)
ate on the HO-µ1-AlIV is slightly more stable (by 7 kJ.mol-1
)
HO-µ2-(AlV,AlV)
HO2-µ1-AlV
than the starting complex isolated from the surface, while it is
isoenergetic for the other hydroxyl sites. These weakly bound
precursors are probably not stable intermediates, if entropic
effects are taken into account. Nonetheless, they hint to the
grafting mechanism (vide infra), leading to the monoaluminoxy
surface complex, (AlsO)Zr(CH3)3 (5m), with the production of
one CH4 molecule (Figure 7b).
The grafting reaction is largely exoenergetic for the dif-
ferent hydroxyl sites (Table 2) and favored by 14-15 kJ/mol
on HO-µ1-AlIV compared to HO-µ2-(AlV,AlV) and H2O-µ1-AlV.
Note that the higher the hydroxyl group stretching frequency
is, the more exoenergetic the grafting is, as observed for the
coordination of pyridine. Therefore, the OH frequency can
indeed be used as a probe for the capacity of an oxygen atom
to react with a Lewis acid (a H atom or a metal center).
The grafting step on HO-µ1-AlIV provides 5m AlsOZr(CH3)3
for which the Zr-O distance (1.99 Å) compares well with
experimental data.21,31 The Al-O distance is not affected by
the replacement of H by Zr(CH3)3, as a Lewis acid is replaced
by another one.
We have subsequently studied the reaction of the surface
complex 5m, AlIVOZr(CH3)3, obtained by the interaction with
the most reactive hydroxyl, with adjacent Lewis acid sites and
surface hydroxyls. The transfer of one methyl ligand onto an
adjacent Al Lewis acid site can generate a partially cationic
complex 6m AlsOZr(CH3)2(µ-CH3)Als, where the µ-CH3 ligand
lies between Zr and a AlV atom (Figure 8a). Such a structure is
98 kJ mol-1 more stable than the monoaluminoxy neutral Zr
complex 5m, AlsOZr(CH3)3. In 6m AlsOZr(CH3)2(µ-CH3)Als,
the Zr complex is further stabilized by an oxygen atom of a
1243-1325 (O-H- - -O)
HO-µ3-(AlV,AlVI,AlVI
HO-µ2-(AlIV,AlV)
)
a
a
2521
1528 (O-H- - -O)
a No adsorption on this site.
another site during geometry optimization). This suggests that
they are probably not good candidates for grafting an organo-
metallic complex so that they have not been considered later.
Upon pyridine adsorption, the calculated OH stretching
frequency decreases by 600 cm-1 (800 cm-1 experimentally).27
The calculated value is underestimated, but the trend is
reproduced. Note that the frequencies are sensitive to weak
interaction, which makes them difficult to calculate accurately
by DFT.
3.2. Model Complexes. The complexes Zr(CH2tBu)4 (1) and
W(≡CtBu)(CH2tBu)3 (2) are too large for a systematic descrip-
tion of the grafting reaction path by quantum modeling. There-
fore, the neopentyl and the neopentylidyne ligands have been,
respectively, replaced by methyl and ethylidyne ligands (Figure
6). Nonetheless, to check the influence of the real ligands, a
selected set of structures has been recalculated using the com-
plete ligand structure. No structural data is available for these
complexes, but the calculated bond lengths are in good agree-
ment with those reported for similar molecular complexes.28-30
3.3. Thermodynamics of Grafting of Zr(CH3)4 (1m) on
Alumina. Grafting of a tetrahedral organometallic complex on
(27) Pichat, P.; Mathieu, M. V.; Imelik, B. Bull. Soc. Chim. Fr. 1969, 2611.
(28) Churchill, M. R.; Youngs, W. J. Inorg. Chem. 1979, 18, 2454.
(29) Orpen, A. G.; Brammer, L.; Allen, F. H.; Kennard, O.; Watson, D. G.;
Taylor, R. J. Chem. Soc., Dalton Trans. 1989, S1.
(30) Schrock, R. R. Chem. ReV. 2002, 102, 145.
(31) Howard, W. A.; Parkin, G. Polyhedron 1993, 12, 1253.
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9161

A R T I C L E S
Joubert et al.
Figure 6. Optimized geometries of the model complexes: (a) ZrMe4 (1m)
and (b) W(tCMe)Me3 (2m).
Figure 8. (a) The partially cationic monoaluminoxy Zr complex 6m,
(AlsO)Zr(CH3)2(µ-CH3)(Als). (b) The bisaluminoxy Zr surface complex
3am, (AlsO)2Zr(CH3)2.
Figure 7. (a) Precursor state for grafting of Zr(CH3)4 on OH-µ1-AlIV. (b)
Monoaluminoxy Zr species grafted on O-AlIV, AlIVO- Zr(CH3)3 (5m).
Table 2. Calculated Reaction Energy for the Grafting of Zr(CH3)4
on Different γ-Al2O3 Sites
1
surface site
E (kJ mol-
)
HO-µ1-AlIV
-201
-186
-187
HO-µ2-(AlV,AlV)
HO2-µ1-AlV
Figure 9. (a) The surface complex 3bm, (AlsO)2Zr(CH3)(µ-CH3)(Als). (b)
Trisaluminoxy Zr surface complex, (AlsO)3ZrCH3 (7m).
oxygen atom between two Al and the Zr atoms being negatively
charged. The trisaluminoxy surface complex, (AlsO)3Zr(CH3)
7m, is slightly more stable than the partially cationic Zr(CH3)-
(µ-CH3) complex 3bm; however thermodynamics is not suf-
ficient to understand the reactivity, and therefore pathways and
activation barriers need to be addressed.
3.4. Kinetic Aspects for the Grafting of Zr(CH3)4 on
Alumina. Since Zr(CH3)4 has only metal-carbon bonds and a
d0 configuration, only σ-bond metathesis between Zr-C and
O-H has been studied for the Zr-O formation mechanism.
The energetic pathway for the three successive cleavages of
the Zr-C bonds by surface hydroxyls resulting in the elimi-
nation of alkane molecules is reported in Scheme 3 including
the possible rearrangement of surface species by reaction
with adjacent Lewis acid sites. The structures of transition states
are reported in Figure 10 with selected geometric data in
Table 3.
The OH site that has been chosen to test the kinetic profile
is on the tetrahedral Al site, HO-µ1-AlIV, because it has the
largest interaction with pyridine and thereby it corresponds to
the case where the proton is the most able to interact with a
Lewis base (pyridine or an alkyl ligand) and probably the most
reactive one. Additionally, the oxygen of this site has the greatest
capability to interact with a Lewis acid (see the OH stretching
frequency), and thus the formation of the metal-oxygen bond
should also be favored. However, considering the large exo-
thermicity of the reaction and the small difference in the stability
of 5m for the different hydroxyl sites, it is clear that the grafting
reaction will also take place on the other sites (Table 2).
The first step (grafting 1m f 5m) is associated with the
largest activation barrier (∆rEq ) 37 kJ.mol-1). The subsequent
reactions are all easier except for the transformation of 3bm
(AlsO)2Zr(CH3)(µ-CH3)(Als) into 7m (AlsO)3Zr(CH3), either
surface aluminoxane bridge, to compensate the loss of electrons
from the shared ligand between Zr and Al.
Starting again from the monoaluminoxy surface complex 5m,
(AlIVO)Zr(CH3)3, a second elimination of methane can occur
by reaction with an adjacent AlsOH, namely a HO-µ1-AlV which
leads to a neutral bisaluminoxy species 3am, (AlsO)2Zr(CH3)2
(Figure 8b). This reaction is exoenergetic: ∆rE ) -154 kJ
mol-1. The energy is less favorable than for the first grafting
step, probably because of geometric constraints on the surface.
In fact, the OH groups are not placed in an ideal position to
allow a perfect tetrahedral environment around the Zr atom.
The bisaluminoxy surface complex, (AlsO)2Zr(CH3)2 3am,
or the partially cationic surface complex, (AlsO)Zr(CH3)2-
(µ-CH3)(Als) 6m, can further react with adjacent Lewis acid
sites or an adjacent AlOH, respectively, to form a partially
cationic bisaluminoxy surface complex 3bm (AlsO)2Zr(CH3)-
(µ-CH3)(Als) (Figure 9a). Both reactions are exoenergetic: ∆rE
) -80 kJ mol-1 and -136 kJ mol-1, respectively. Like in the
partially cationic (AlsO)Zr(CH3)2(µ-CH3)(Als) complex 6m, the
coordination sphere of Zr is completed by an oxygen atom from
an adjacent aluminoxane bridge.
Further reaction with a surface AlsOH, namely a HO-µ2-
(AlV,AlV), to give a third equivalent of methane, whether starting
from (AlsO)2Zr(CH3)2 3am or from the partially cationic surface
complex, (AlsO)2Zr(CH3)(µ-CH3)(Als) 3bm, is still possible, but
the geometric constraint is even more important, which probably
explains the lower energy gain: ∆rE ) -90 kJ mol-1 (Figure
9b). In this case, the structure 7m, (AlsO)3Zr(CH3), presents
two quasi covalent Zr-O bonds (2.00 Å) and a slightly longer
one (2.17 Å); the latter corresponds to a two-electron donation
by an oxygen atom from the surface. Furthermore, there is a
very long Zr-O bond (2.51 Å) of electrostatic character, the
9
9162 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
a
Scheme 3. Energetic Pathway for Three Successive Methane Eliminations Starting from 1m + Al2O3
a Reactions with alkane elimination proceed horizontally, while surface isomerizations are indicated vertically with bold arrows.
through a direct pathway (which involves a σ-bond metathesis
with the µ-CH3 ligand) for which the activation barrier is >
150 kJ mol-1 or through a two-step reaction, which involves
the reformation of 3am (AlsO)2Zr(CH3)2 followed by the
reaction of a third OH group (Ea > 92 kJ mol-1). The activation
energy for the rearrangement of 5m (AlsO)Zr(CH3)3 into 6m
(AlsO)Zr(CH3)2(µ-CH3)Als via reaction with an adjacent alu-
minum is 15 kJ mol-1. The geometry (Figure 10b) shows a
very early transition state with a very low imaginary frequency
along the reaction coordinate (61 cm-1). Similarly, the activation
energy for the rearrangement of 3am (AlsO)2Zr(CH3)2 into 3bm
(AlsO)2Zr(CH3)(µ-CH3)(Als) is lower than 5 kJ mol-1. There-
fore, 3bm (AlsO)2Zr(CH3)(µ-CH3)(Als) is likely to be obtained
more rapidly than 7m (AlsO)3Zr(CH3). This probably explains
why the third Zr-C bond cleavage leading to a third equivalent
of alkane is not observed experimentally even if it is slightly
more favored on a thermodynamic basis.
ligands are better electron donors, which prevents the Zr-C
bond from being lengthened too much by the steric effects.
Hence, the energy of the transition state is not raised too much
by the increase of the Zr-C bond length when changing methyl
by neopentyl ligands. In any case, this study shows that the
replacement of neopentyl ligands by methyl ligands in ZrR4
does not dramatically affect the description of the elimination
process.
In contrast, changing methyl for neopentyl ligands greatly
affects the steps 5m f 6m and 3am f 3bm. In fact, while 5m
f 6m is exoenergetic (-98 kJ mol-1), the corresponding
reaction with neopentyl ligands is endoenergetic (+38 kJ mol-1
)
and in this case the neopentyl ligand cannot bridge Zr and Al
like µ-CH3 in 6m (AlsO)Zr(CH3)2(µ-CH3)Als: it is coordinated
only on Al thus forming a cationic Zr complex.
The same trend is observed for 3am f 3bm, but the reaction
remains exoenergetic (-23 kJ mol-1 for neopentyl vs -80 kJ
mol-1 for methyl), and the neopentyl ligand is not a bridging
ligand such as the case in 3bm (AlsO)2Zr(CH3)(µ-CH3)(Als):
there is a clear charge separation generating {[(AlsO)2Zr-
(CH2tBu)+][Als(CH2tBu)-]} (3c, Figure 11).
It can be concluded from the kinetic study that 3bm is the
most probable model structure, obtained through 5m and 3am
or through 5m and 6m in an equiprobable way. The third
elimination yielding 7m is excluded from a kinetic argu-
ment. While considering neopentyl ligands, the reaction path
through the equivalent of 6m is excluded, and the reaction goes
through the equivalent of 5m and 3am giving {[(AlsO)2Zr-
(CH2tBu)+][Als(CH2tBu)-]} (3c) as the most probable structure.
3.5. Comparison and Interpretation of Analytical and
Computational Data for Zr. The proposed structure 3c is in
agreement with the mass balance analysis. To further refine the
comparison between theory and experiment, we have calculated
the 13C NMR chemical shifts of the major surface species and
compared them with experimental data (Table 4). Calculated
NMR chemical shifts for [Zr(CH2tBu)4] (1) are relatively close
to the experimental data, especially for the carbon directly
attached to Zr, which shows that the chosen basis set (IGLO-II
combined with LANL2DZ) describes quite well the electronic
properties of the Zr complexes.32 For the proposed surface
In summary, the calculations show that the reaction of a
Zr-CH3 bond with surface AlsOH is always exoenergetic and
that the activation barriers are quite low (12-37 kJ mol-1
)
except for a µ-CH3 ligand. Moreover, they also show that the
reaction with adjacent aluminum Lewis acid centers is probably
faster than that with AlsOH (Ea <14 kJ mol-1), yielding partially
cationic species. Therefore, the calculations would be consistent
with the formation of 3bm (AlsO)2Zr(CH3)(µ-CH3)(Als) with
the release of 2 equiv of alkane in the gas phase as observed
experimentally (vide supra).
To check the influence of the size of the ligand (replacement
of CH2tBu by methyl ligands) the activation barrier of the second
Zr-C bond cleavage was evaluated with the full ligand. Note
that the study with the neopentyl ligands requires quadrupling
the size of the surface unit cell, which renders the calculation
very CPU intensive, and therefore only certain reaction steps
have been considered and only the activation barriers of the
second elimination step have been evaluated. The activation
barrier for the reaction of (AlsO)Zr(CH2tBu)3 (5m′) with an
adjacent AlsOH to form (AlsO)2Zr(CH2tBu)2 (3am′) and tBuCH3
is 26 kJ mol-1, which is comparable to that obtained for the
methyl system, i.e., 20 kJ mol-1. Nonetheless, the use of the
neopentyl ligands has two influences. First, the bond distances
in the four-membered ring transition state are larger, probably
because of greater steric repulsions (Table 3). Second, these
(32) Ziegler, T.; Autschbach, J. Chem. ReV. 2005, 105, 2695.
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9163

A R T I C L E S
Joubert et al.
Figure 11. {[(AlsO)2Zr(CH2tBu)+][Als(CH2tBu)-]} (3c) structure optimized
on the model surface with the complete neopentyl ligands.
Table 4. Comparison of Calculated vs Experimental Carbon
Chemical Shift for 1 and 3
δ
(ppm)
CH2tBu
CH2CMe3
CH3
calculated for 1
98
102.6
42
35.4
37
34.7
experimental for 1a
calculated for 3c
ZrCH2tBu
AlVICH2tBu 90
AlIVCH2tBu 25
86
40
49
36
40-41
36-46
33-41
experimental for 1/Al2O3
99, 84, 26c
b
32, 29
a Recorded in C6D6. b Not observed. c Broad (see Figure 2 for details).
Table 5. Calculated Reaction Energy for the Grafting of
W(CCH3)(CH3)3 on Different γ-Al2O3 Sites
1
surface site
E (kJ mol-
)
HO-µ1-AlIV
-150
-139
-130
HO-µ2-(AlV,AlV)
HO2-µ1-AlV
and 90 ppm, respectively, which can be associated to the broad
peak around 84 ppm and the shoulder at 99 ppm in the spectrum.
For the Zr(CH2tBu) unit, the calculated upfield chemical shift
compared to 1 is typical of the replacement of a carbon by a
more electronegative atom (oxygen) in the coordination sphere
of Zr.33 Structure 3c however does not account for the intense
methylene peak at 26 ppm in the spectrum. In this geometry,
the neopentyl ligand is bound to an octahedral Al atom. As
surface aluminum atoms have several coordination geometries,
we have also calculated a surface structure for which the
neopentyl is borne by a tetrahedral AlIV center. In this case, the
calculated chemical shift for the methylene carbon is 25 ppm,
in very good agreement with the data. This shows that neutral
Zr neopentyl intermediates react with both AlV and AlIII centers
at the surface forming two types of [Als(CH2tBu)-] species.
The analysis of experimental data in light of the model, which
reproduces well the different features of the system, leads to
the identification of the most probable surface structure for the
grafted Zr complex. While initial grafting does not probably
occur selectively on a tetrahedral Al site because of the high
exothermicity of the process (along with the small difference
of reaction energy, -186 to -201 kJ/mol) and the low activation
barriers (12-37 kJ/mol), all the data are in agreement with the
formation, as a major species, of a cationic bisaluminoxy mono-
neopentyl zirconium complex bound to an aluminoxane bridge
and in proximity of a neopentyl aluminate, the Al being either
penta- (3c) or tetracoordinated.
Figure 10. (a) Transition state structures for successive alkane eliminations
in Zr(CH3)4 grafting. (b) Transition state structure for the rearrangement of
5m AlsOZr(CH3)3 to 6m AlsOZr(CH3)2(µ-CH3)Als.
Table 3. Distances (in Å) in the Four-membered Ring Transition
States for Successive Alkane Eliminations for the Grafting of
Zr(CH3)4 on Alumina
reaction with a second AlsOH
with a third
grafting step
AlsOH
methyl ligands
on the neutral
complex
methyl ligands
on the cationic
complex
methyl
ligands
TS1m
neopentyl
ligands
TS5m
methyl
ligands
TS3am
TS5m
TS6m
-5m
-
3am
-3bm
′-3am
′
-7m
Zr-O
Zr-C
O-H
C-H
2.19
2.47
1.18
1.54
2.29
2.39
1.06
1.73
2.27
2.49
1.11
1.55
2.37
3.30
2.41
1.24
1.43
2.45
1.04
1.80
species {[(AlsO)2Zr(CH2tBu)]+[(Als)(tBuCH2)]-}, our analysis
will be centered on the methylene carbon atoms which are direct
neighbors of the metal Zr and Al atoms, and for which
contributions have been highlighted in the experiment by 13C
labeling (Figure 2). In structure 3c, the calculated chemical shift
for the methylene carbon atoms attached to Zr and AlVI are 86
(33) Berger, S.; Bock, W.; Frenking, G.; Jonas, V.; Mueller, F. J. Am. Chem.
Soc. 1995, 117, 3820.
9
9164 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
Scheme 4. Two Possible Pathways for the Initial Grafting Step
Figure 12. (a) Monoaluminoxy W surface complex on tetrahedral
aluminoxy site, (AlIVO)W(CH3)2(tCCH3), 4am. (b) Bisaluminoxy W
surface complex 8m, (AlsO)2W(tCCH3)(CH3) (remaining WC distances
are not significantly modified by the reaction with the second AlsOH
group).
After grafting, it is experimentally observed that some
hydroxyl groups remain unreacted. Therefore, the stretching
frequencies of these remaining hydroxyls have been calculated
for the proposed final structure 3bm (AlsO)2Zr(CH3)(µ-CH3)-
(Als) and show no significant change compared to their original
values as observed experimentally (see Figure 1).
R-H-abstraction thus releasing an alkane molecule in the gas
phase and restoring the WtC bond (Scheme 4).
While the addition of an AlsOH onto the carbyne requires
going through a rather high activation barrier, i.e., ∆rEq ) 126
kJ mol-1, the direct electrophilic cleavage of the WsC bond
by AlsOH gives a much lower activation energy (81 kJ mol-1),
which shows that grafting probably occurs via the latter pathway.
The activation energy associated with the reaction of 4am
(AlIVO)W(CH3)2(tCCH3) with an adjacent hydroxyl group is
only slightly lower (70 kJ mol-1) than that of the grafting step
(81 kJ mol-1), but remains high (See Scheme 5 for the reaction
path and Figure 13 for the corresponding TS). Note that,
experimentally, alkylidyne complexes are in general not par-
ticipating in the grafting process,35 in contrast to alkyl and
alkylidene ligands.35-38 The comparison of the distances in
Tables 3 and 6 shows that TS2m-4am is later than TS1m-5m which
is reflected by the higher barrier.
As in the case of Zr, a test with the large neopentyl ligands
on a quadruple unit cell has been performed. In contrast to Zr,
the energy barrier for the second elimination increases substan-
tially from 70 to 97 kJ mol-1. In the case of W, the influence
of realistic bulky ligands clearly disfavors the formation of the
bisaluminoxy surface complex in contrast to Zr where the
activation energy is little affected by changing the methyl by a
neopentyl ligand. This probably explains why the second release
of tBuCH3 is not observed experimentally in contrast to what
is observed during the grafting of the Zr complex, for which 2
equiv of tBuCH3 are released. All these activation barriers are
high compared to those obtained for Zr, and they are probably
associated with the presence of the carbyne ligand, which is
not prone to geometric deformation during the necessary
formation of a pentacoordinated structure in the transition state.
This is in agreement with the experimental results since the
complete gratfing process is slower (greater experimental time
and grafting temperature) for W than for Zr.
3.6. Thermodynamics of Grafting of W(tCMe)(CH3)3 on
Alumina. For the W(tCCH3)(CH3)3 complex, the first step of
the grafting process is quite similar to that for Zr(CH3)4. The
major difference is that the precursor states are less stable than
the complex isolated from the surface (by 3 to 7 kJ mol-1
depending on the AlsOH site). Generally speaking, the tungsten-
carbyne complex is more difficult to distort than the Zr complex
because the formation of a pentacoordinated tungsten-carbyne
complex by addition of a σ-donor ligand on W(tCCH3)(CH3)3
weakens the tungsten-carbon triple bond, hence giving less
stable precursor states (vide infra). As already noticed in the
case of Zr(CH3)4, this weakly bound precursor is probably not
a stable intermediate, if entropic effects are taken into account.
The stability of the monoaluminoxy W surface complexes
as a function of the surface AlsOH site follows the same trend
as that calculated for Zr complexes (Table 5): HO-µ1-AlIV
>
HO-µ2-(AlV,AlV) > HO2-µ1-AlV. The structure of the most
stable monoaluminoxy complex is reported in Figure 12. The
W-O distance reported in Figure 12 is in good agreement with
X-ray diffraction data obtained for alkoxy tungsten complexes.34
Further reaction of 4am (AlIVO)W(tCCH3)(CH3)2 with an
adjacent AlsOH, leading to (AlsO)2W(tCCH3)(CH3) (8m) and
a second equivalent of methane, is thermodynamically favorable
(-90 kJ mol-1), but less than for the first grafting step (-150
kJ mol-1). Moreover and in contrast to Zr, all calculated cationic
complexes, with a shift of a methyl group on a neighboring Al
center, are less stable (g 49 kJ mol-1) than 4am (AlIVO)-
W(CH3)2(tCCH3). This already infers that the formation of
4b-c (Scheme 2), both cationic species, is probably unlikely
in contrast to what we proposed earlier.22
3.7. Kinetic Aspects of W(CH3)3(CCH3) Grafting. Grafting
of W(CH3)3(tCCH3) on alumina can take place via two reaction
pathways. The first one goes through a σ-bond metathesis
between WsCH3 and OsH to form the OsW and H3CsH
bonds. The second one results from the addition of the hydroxyl
group onto the carbyne to give a carbene, which undergoes an
(35) Chabanas, M.; Baudouin, A.; Cope´ret, C.; Basset, J.-M.; Lukens, W.;
Lesage, A.; Hediger, S.; Emsley, L. J. Am. Chem. Soc. 2003, 125, 492.
(36) Dufaud, V.; Niccolai, G. P.; Thivolle-Cazat, J.; Basset, J.-M. J. Am. Chem.
Soc. 1995, 117, 4288.
(37) Chabanas, M.; Quadrelli, E. A.; Fenet, B.; Cope´ret, C.; Thivolle-Cazat, J.;
Basset, J.-M.; Lesage, A.; Emsley, L. Angew. Chem., Int. Ed. 2001, 40,
4493.
(38) Le Roux, E.; Chabanas, M.; Baudouin, A.; de Mallmann, A.; Cope´ret, C.;
Quadrelli, E. A.; Thivolle-Cazat, J.; Basset, J.-M.; Lukens, W.; Lesage,
A.; Emsley, L.; Sunley, G. J. J. Am. Chem. Soc. 2004, 126, 13391.
(34) Akiyama, M.; Chisholm, M. H.; Cotton, F. A.; Extine, M. W.; Haitko, D.
A.; Little, D.; Fanwick, P. E. Inorg. Chem. 1979, 18, 2266.
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9165

A R T I C L E S
Joubert et al.
Scheme 5. Energetic Pathway for the Grafting of W(tCCH3)(CH3)3 on Alumina
Table 6. Geometric Parameters (in Å) in Four-membered Rings of
Transition States in Successive Alkane Eliminations in
W(CH3)3(CCH3) Grafting
An orbital interaction diagram of the two situations (Figure
14), representing the starting complex and the pentacoordinated
fragment, has been built to understand the behavior of 2 toward
grafting. The surface hydroxyl is represented by a molecule of
water for the pentacoordinated fragment. While the interaction
between the px orbital on the carbyne fragment and the dxz orbital
on the metallic fragment is approximately the same for a
W(CH3)3 and a W(CH3)3(OH2) fragment, the interaction of the
py orbital on the carbyne fragment and the dyz orbital on the
metallic fragment is less favorable in the case of W(CH3)3(OH2)
than W(CH3)3. The coordination of a fifth ligand to W(CCH3)-
(CH3)3 therefore disrupts the π-bond system of the tungsten-
carbon triple bond, thus destabilizing the complex.
3.8. Comparison of Analytical and Computational Data
for W. From kinetic and thermodynamic considerations, 4am
seems to represent a good model for the grafted complex. This
statement is further argued by a comparison between experi-
mental and computational data, especially geometric features
and IR and NMR data. First, the analogue of 4am with real
ligands (4d) has geometric features which agree very well with
experimental EXAFS data (see Figure 15).22
grafting step
reaction with a second AlsOH
methyl ligands
methyl ligands
TS4am
neopentyl
ligands
TS2m
-4am
-8m
W-O
W-C
O-H
C-H
2.29
2.33
1.25
1.43
2.23
2.33
1.27
1.42
2.29
2.29
1.27
1.42
hydroxyl and the perhydrocarbyl ligands. For 4am in which
the OH is interacting with the π system of the carbyne, a large
shift is obtained (∆ν ) 70 cm-1, 3630 cm-1), while, for 4am′,
a conformer of 4am in which the OH interacts with the methyl
ligand, a smaller shift is obtained (∆ν ) 12 cm-1, 3688 cm-1).
Such a short contact between the carbyne and a surface hydroxyl
group is specific of structure 4am (AlIVO)W(tCCH3)(CH3)2
(H‚‚‚C distance of 2.78 Å) and does not appear in the case of
8m (AlsO)2W(tCCH3)(CH3) (H‚‚‚C distance of 3.40 Å). The
calculated frequency shifts reproduce well the experimental data
(see Figure 3) where a broad signal at 3650 cm-1 appears upon
reaction of the tungsten complex with alumina, clearly indicating
a perturbation of the remaining surface OH groups by the surface
complex (such an interaction is not observed experimentally or
by calculation in the case of zirconium).
To further evaluate the model, vibrational characteristics of
hydroxyls interacting with 4am (AlIVO)W(CH3)2(tCCH3) have
been calculated. It appears that the stretching frequency of the
remaining µ2-OH is shifted compared to the bare surface, and
the shift is associated with an extra interaction between this
The 13C NMR spectrum of the W surface complex (Figure
4) displays six major peaks and was initially assigned to the
presence of several species on the surfaces.22 The presence of
significant interactions between the hydrocarbon ligands and
the surface OH groups, already seen in the νOH frequencies,
opens a new interpretation where the multiple peaks are pro-
duced by a single surface W species, of 4a type, but presenting
several types of interactions between hydrocarbon ligands and
surface OH groups. Experimentally, we observe a broadening
of the carbynic signal and several types of methylene carbons:
three peaks at 103, 90, and 85 ppm. We have therefore
calculated the carbon chemical shifts of two conformers of the
monoaluminoxy surface complex (AlsO)W(tCtBu)(CH2tBu)2
(Table 7): one showing an interaction of the carbyne with a
vicinal hydroxyl group (4d, corresponding to 4am for the model
Figure 13. Transitions states structures TS2m-4am and TS4am-8m for
successive alkane eliminations during the grafting of W(CH3)3(tCCH3)
on alumina (for geometric data, see Table 6).
9
9166 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
Figure 14. Orbital interaction diagram between CCH3 and W(CH3)3 and W(CH3)3(OH2).
Scheme 6. Possible Interactions between a Surface Hydroxyl and
the Ligands of (AlsO)W(tCtBu)(CH2tBu)2
the carbynic carbon being shifted by only 5 ppm. This shows
that the interaction with hydroxyls has a great influence on the
chemical shift of the methylene carbon, which probably explains
the upfield signal experimentally observed. Moreover, in the
case of 4d, for which the carbyne ligand interacts with adjacent
hydroxyl groups, a large upfield shift of the carbynic carbon
(∆δ ) -29) is calculated, while little influence is obtained for
the methylene carbon (∆δ ) -4). Both types of interactions
could explain the broadening of the carbynic signal. Thus, by
comparing calculated and experimental data, we can propose
that grafting 2 on alumina generates a monoaluminoxy surface
species with several conformations, which interacts with residual
surface hydroxyls as evidenced by IR spectroscopy, and which
gives rise to a complex NMR spectrum. Such conformers of a
single species allow a complete interpretation of the data.
Moreover the combination of total energy calculations and
experimental and calculated NMR data clearly demonstrates the
absence of cationic species in the case of the grafting of the W
carbyne complex.
Figure 15. Calculated and experimental (in italic letters) geometric features
of the grafted tungsten complex 4d.
Table 7. Comparison of Calculated vs Experimental Carbon
Chemical Shifts for 2 and 4a
δ
(ppm)
tCtBu
CCMe3
CH2CMe3
CH2tBu
CH3
calculated for 2
341
58
52
41
32
89
103
32-37
34
experimental data for 2a 316
calculated for 4d
calculated for4e
312
55
38-39
85
31-39
(-29) (-3) (-3/-2) (-4)
336
56
38-39
70 (-19)b 31-39
(-5)
(-2) (-3/-2) 85 (-4)
experimental data for
318(2)c 52(0)c
d
103 (0)c
32,29
2/Al2O3
90 (-8)c
85 (-19)c
a The value in parentheses corresponds to ∆δ between the 13C calculated
chemical shift for 2 and 4i (i ) d or e). b Recorded in C6D6. c This calculated
chemical shift corresponds to the one of the methylenes of the neopentyl
ligand in interaction with a surface hydroxyl. d Broad peaks (see Figure 4
for details). e Not observed.
Conclusion
system) and the other having an interaction of one of its
neopentyl ligands with a vicinal hydroxyl group (4e, corre-
sponding to 4am′ for the model system) (Scheme 6).
A combination of physical chemical techniques and of
quantum simulations has allowed a novel and complete char-
acterization of the nature and structure of alumina single site
catalysts. So far, silica supported organometallic complexes have
been quite successfully characterized only from experimental
physical chemical techniques, which have been the foundation
of Surface Organometallic Chemistry (SOMC). As shown
above, in the case of alumina, the surface chemistry can be so
diverse that it has been necessary to include theoretical modeling
as an essential tool of characterization in addition to the usual
techniques of SOMC. The synergy between these approaches
First of all, the calculated 13C chemical shifts for the carbons
far from the metal centers are very accurate and do not vary
significantly between complex 2 and the various surface
complexes 4 (Table 7) as observed experimentally. Second, in
the case of 4e, the calculated chemical shift of the methylene
carbon of the neopentyl ligand in interaction with the hydroxyl
is shifted to 70 ppm (∆δ ) -19 ppm compared to the isolated
complex 2), while that of the remaining free neopentyl ligand
is shifted upfield by only 4 ppm at 85 ppm (∆δ ) -4 ppm),
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9167

A R T I C L E S
Joubert et al.
has been illustrated by the understanding at a molecular level
of the difference of reactivity of [ZrR4] and [W(tCR′)R3]
complexes with alumina.
on alumina are active polymerization catalysts, while their
corresponding silica supported analogues, neutral, are inactive.7-9
The combination of theory and physical chemistry characteriza-
tions hence appears as a promising tool for the understanding,
at a molecular scale, of well-defined surface catalytic species,
enabling the rational design of highly efficient catalysts.
The first step of the reaction of ZrR4 with an alumina partially
dehydroxylated at 500 °C is an electrophilic cleavage of a
ZrsC bond preferentially on HOsAlIV (∆rE ) -201 kJ/mol)
through σ-bond metathesis with a low activation barrier (Ea )
37 kJ/mol). This reaction gives [(AlsO)ZrR3] along with one
alkane molecule. In the case of W(tCR′)R3, the first step is
also an electrophilic cleavage of the WsR bond preferentially
on HOsAlIV (∆rE ) -150 kJ/mol) to form [(AlsO)W(tCR′)-
R2], but this reaction is associated with a much higher activation
barrier (Ea ) 81 kJ/mol). The lower reactivity of the W complex
is due to the carbyne ligand, which hinders the necessary
deformation of the complex for the approach of the AlsOH
toward the WsC bond. Although preferred on HOsAlIV, the
grafting reaction is also possible on other accessible OH groups
such as HOs(Alv,Alv), especially for the case of the Zr complex
where the reaction barrier is low.
Computational Methods and Systems
The calculations were performed on the framework of density
functional theory (DFT) using a periodic description of the system as
implemented in the VASP code.39,40 The generalized gradient ap-
proximation was used in the formulation of Perdew and Wang PW91.41
Atomic cores were described with the projected augmented wave
method (PAW) which is equivalent to an all electron frozen core
approach.42,43 The one electron wave functions are developed on a basis
set of plane waves. With the selected PAW potentials, a cutoff energy
of 275 eV is adequate and yields a converged total energy.
Brillouin zone integration was converged with a 331 k-point mesh
generated by the Monkhorst-Pack algorithm.44 Vibrational frequencies
were calculated in the harmonic approximation by a numerical
evaluation of the Hessian matrix. An anharmonicity term of 80 cm-1
,
For Zr, the favored second step is the reaction of [(AlsO)-
ZrR3] with an adjacent hydroxyl to give a neutral complex,
[(AlsO)2ZrR2], and the evolution of a second molecule of alkane.
This second step is still exoenergetic (-154 kJ/mol), its barrier
is lower than that of the first step and is not affected by the
size of the ligands (R ) Me or CH2tBu). This surface complex
further reacts with the alumina surface in an almost barrierless
process through the transfer of one of its alkyl ligands onto an
adjacent Als Lewis center, giving a cationic surface complex
{[(AlsO)2ZrR]+[(Als)R-]}. The alkyl transfer to give [(AlIV)R-]
calculated previously on hydroxyl groups on boehmite, has been applied
a posteriori.45
Chemical shifts have been evaluated with the GIAO method46
implemented in the Gaussian03 code47 at a DFT/B3LYP level.48-51 The
IGLO-II basis set52 has been used for carbon and hydrogen. For other
atoms, the Hay and Wadt effective core potential53-55 has been used
with the adapted LANL2DZ basis set. The chemical shift calculations
have been performed on selected clusters. The interested reader will
find more details in the Supporting Information in which a validation
of computational parameters on well-defined molecules is also proposed.
and [(AlVI)R-] has been confirmed by the combined use of 13
C
Experimental Section
CP-MAS solid-state NMR and chemical shift calculations.
Moreover, this complete reaction pathway involves the forma-
tion of two alkane molecules per grafted Zr as observed
experimentally.
General Procedure. All experiments were conducted under
strict inert atmosphere or vacuum conditions using standard Schlenk
techniques. Solvents were purified and dried according to standard
procedures. tBuCH2MgCl was prepared from tBuCH2Cl (98%, Aldrich)
and Mg turnings (99%, Lancaster). [(1-13C 33%) tBu13CH2MgCl]
was prepared according to the literature procedure (It was pre-
pared from a mixture of 1:2 [(1-13C 99%) tBu13CH2MgCl] and un-
labeled tBuCH2MgCl. [(1-13C 99%) tBu13CH2Cl] was prepared from
(1-13C 99%) H13CONMe2 (Cambridge Isotope Laboratories) and
unlabeled tBuLi followed by reduction to the alcohol, chloration with
a Vilsmeier reagent, and finally conversion to the Grignard with Mg).35
[Zr(CH2tBu)4] (1) and [(13C 33%) Zr(CH2tBu)4] (1*) were prepared
by alkylation of ZrCl4 (Clariant, used as received) with tBuCH2MgCl
and [(1-13C 33%) tBu13CH2MgCl], respectively, and were sublimed prior
The situation is completely different for the W carbyne
complex. After the first step, the [(AlsO)W(tCR′)R2] species
is found to be nonreactive. Further reaction with surface
hydroxyls involves a high reaction barrier (∼100 kJ/mol) for
the real CH2tBu ligand. This is in agreement with the evolution
of a single alkane molecule in the reactive adsorption process.
Moreover, transfer of an alkyl ligand to an adjacent Al center
does not occur either as it involves the formation of unstable
cationic species (∆rE ) +50 kJ/mol). This is also confirmed
by the absence of typical 13C NMR signals from surface alkyl
aluminum species. Nonetheless, this neutral complex shows
weak lateral interactions with the remaining adjacent surface
hydroxyls as evidenced by a combination of IR spectroscopy
and frequency calculations. It is noteworthy that the alumina
supported W species, [(AlsO)W(tCR′)R2], is in principle well-
defined but that its 13C NMR spectrum is complicated simply
because this well-defined species interacts through its ligands
(alkyl and carbyne) with different remaining hydroxyls present
on the surface of alumina, thus providing different conformers,
whose calculated NMR spectra are consistent with experimental
data. This reflects the complexity of γ-alumina surfaces. In
contrast, the zirconium complex, which has reacted with its
surrounding hydroxyl groups and an adjacent aluminum, gives
a fairly well-defined complex as evidenced by a rather simple
NMR spectrum. It is a true cationic alkyl zirconium species,
and this explains quite well why zirconium alkyls supported
(39) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15.
(40) Kresse, G.; Furthmueller, J. Phys. ReV. B: Condens. Matter 1996, 54,
11169.
(41) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M.
R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B: Condens. Matter 1992, 46,
6671.
(42) Bloechl, P. E.; Forst, C. J.; Schimpl, J. Bull. Mater. Sci. 2003, 26, 33.
(43) Bloechl, P. E. Phys. ReV. B: Condens. Matter 1994, 50, 17953.
(44) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B: Condens. Matter 1976, 13,
5188.
(45) Raybaud, P.; Digne, M.; Iftimie, R.; Wellens, W.; Euzen, P.; Toulhoat, H.
J. Catal. 2001, 201, 236.
(46) Ditchfield, R. J. Chem. Phys. 1972, 56, 5688.
(47) Pople, J. A. et al. Gaussian 98; Gaussian Inc.: Pittsburgh, PA, 1998.
(48) Becke, A. D. Phys. ReV. A: At., Mol., Opt. Phys. 1988, 38, 3098.
(49) Dickson, R. M.; Becke, A. D. J. Chem. Phys. 1993, 99, 3898.
(50) Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
(51) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B: Condens. Matter 1988, 37,
785.
(52) Kutzelnigg, W.; Fleischer, U.; Schindler, M. In NMR basic principles and
progress; Springer: Berlin, 1990; Vol. 23, p 165.
(53) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270.
(54) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(55) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284.
9
9168 J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006

Alumina Supported Single-Site Catalysts
A R T I C L E S
to use.56 [W(tCtBu)(CH2tBu)3] (2) and [(13C 30%) W(tCtBu)-
(CH2tBu)3] (2*) were prepared with tBuCH2MgCl and [(1-13C
30%) tBu13CH2MgCl], respectively, and were sublimed prior to use.57
[W(tCtBu)(CH2tBu)3/Al2O3-(500)] was synthesized according to the
litterature procedure.22 Elemental analysis was performed at the CNRS
Central Analysis Service of Solaize (Zr, W) and at the University of
Bourgogne (C).
Gas-phase analysis was performed on a Hewlett-Packard 5890 series
II gas chromatograph equipped with a flame ionization detector and a
KCl/Al2O3 on a fused silica column (50 m × 0.32 mm).
Infrared spectra were recorded on a Nicolet 550-FT by using an
infrared cell equipped with CaF2 windows, allowing in situ studies.
Typically 16 scans were accumulated for each spectrum (resolution, 2
cm-1).
Representative Procedure. A mixture of 1 (69 mg, 0.18 mmol) and
γ-Al2O3-(500) from Degussa (320 mg, 0.19 mmol AlOH) in pentane (8
mL) was stirred at 25 °C for 2 h. After filtration, the white solid was
washed 3 times with pentane, and all volatiles were condensed into
another reactor of known volume (> 6 L) in order to quantify tBuCH3
released during grafting. The resulting light yellow powder was dried
thoroughly under vaccum (1.34 Pa) to yield [1/γ-Al2O3-(500)]. Analysis
by gas chromatography indicated the formation of 0.30 mmol of
1
neopentane during grafting (2.2 tBuCH3/Zr). H MAS NMR: δ ) 1.0
ppm. 13C CP MAS NMR δ ) 84 (weak), 32 and 26 ppm. Elemental
analysis: 1.3%wt Zr, 2.3%wt C, 13 ( 2 C/Zr.
Preparation of [Zr(CH2tBu)4/γ-Al2O3-(500)]B by Impregnation of
[Zr(CH2tBu)4] onto γ-Al2O3-(500) Prepared from Boehmite (Large
Scale). The same procedure as described above was used: A mixture
of 1 (161 mg, 0.42 mmol) and γ-Al2O3-(500) prepared from Boehmite
as described above (590 mg, 0.65 mmol AlOH) in pentane (10 mL)
was stirred at 25 °C for 2 h. Analysis by gas chromatography indicated
the formation of 0.32 mmol of neopentane during grafting (1.8 tBuCH3/
Zr). 1H MAS NMR: δH ) 1.0 ppm. 13C CP MAS NMR ) 84 (weak),
32 and 26 ppm. Elemental analysis: 2.8%wt Zr, 4.1%wt C, 11 ( 2 C/Zr.
All solid-state NMR spectra were recorded under MAS (νR ) 10
kHz) on a Bru¨ker Avance 500 spectrometer equipped with a standard
4-mm double-bearing probe head and operating at 500.13 and 125.73
1
MHz for H and13C, respectively. The freshly prepared samples were
immediately introduced in the 4-mm zirconia rotor in a glovebox and
tightly closed. Compressed air was used for both bearing and driving
the rotors. Chemical shifts are reported in ppm downfield from SiMe4
((0.1 and 1 ppm for 1H and 13C NMR spectra, respectively). The typical
cross-polarization sequence was used for the 13C CP MAS NMR
spectra: 90° proton pulse, cross-polarization step to carbon spins, and
detection of the carbon magnetization under proton decoupling TPPM-
15.58 For the CP step, a ramp radio frequency (rf) field centered at νCP
) 60 kHz was applied on protons, while the carbon rf field was matched
to obtain optimal signal. The contact time for CP was set to 2 ms. An
exponential line broadening of 80 Hz was applied before Fourier
transform. All other details are given in the caption of figures.
Preparation of Partially Dehydroxylated Alumina at 500 °C,
Al2O3-(500). R-Boehmite monohydrate from Johnson Matthey (200
m2/g) or γ-Al2O3-(500) from Degussa C Aerosil (100 m2/g) was calcined
at 500 °C under N2/O2 flow overnight and then partially dehydroxylated
at 500 °C under a high vacuum (1.34 Pa) for 12 h to give a white
solid. The R-boehmite monohydrate from Johnson Matthey gave
γ-Al2O3-(500) having a specific surface area of 162 m2/g and containing
ca. 4.0 OH/nm2 (1.1 mmol/g); 27Al CP MAS NMR: δAl ) 4 (3 Al,
Oh), 62 ppm (1 Al, Td). γ-Al2O3-(500) from Degussa C Aerosil gave a
γ-Al2O3-(500) having a specific surface area of 90 m2/g and containing
ca. 4.0 OH/nm2 (0.6 mmol/g). 27Al CP MAS NMR: δAl ) 4 (3 Al,
Oh), 62 ppm (1 Al, Td).
Grafting of Molecular Complexes on Alumina Monitored by IR
Spectroscopy. Grafting of 1 (IR). Representative Procedure. Alu-
mina (100 mg) was pressed into an 18-mm self-supporting disk, adjusted
in sample holder, and put into a glass reactor equipped with CaF2
windows. The support were prepared as described above, and 1 was
then sublimed under high vacuum (1.34 Pa) at 80 °C onto the alumina
disk. The solid was then heated at 65 °C for 2 h before the excess of
complexes were removed by reverse sublimation at 80 °C and
condensed into a tube cooled by liquid N2, which was then sealed off
using a torch. An IR spectrum was recorded at each step.
Preparation of [(13C 33%) Zr(CH2tBu)4/γ-Al2O3-(500)
] by Im-
D
pregnation of [(13C 33%) Zr(CH2tBu)4] onto γ-Al2O3-(500) (Large
Scale). [1*/γ-Al2O3-(500)] was prepared using the procedure described
above using 1* (20 mg, 53 µmol) and γ-Al2O3-(500) (109 mg, 65
µmolAlOH). 1H MAS NMR δ: 1.0 ppm. 13C CP MAS NMR δ: 99(*),
84(*), 32, 26(*).
Preparation of [W(tCtBu)(CH2tBu)3/Al2O3-(500)]B by Impregna-
tion of [W(tCtBu)(CH2tBu)3] onto Al2O3-(500) (Large Scale). An
excess of 2 (311 mg, 0.66 mmol) and γ-Al2O3-(500) prepared from
Boehmite (1.8 g, 2 mmol OH) were stirred at 66 °C for 4 h. All volatile
compounds were condensed into another reactor (of known volume)
in order to quantify tBuCH3 evolved during grafting. Analysis by gas
chromatography indicated the formation of 0.21 ( 0.05 mmol of
tBuCH3 (0.9 ( 0.1 CH3tBu/W). Pentane (10 mL) was introduced into
the reactor by distillation, and the solid was washed 3 times. After
evaporation of the solvent, the resulting light brown powder was dried
1
under a vacuum to yield [2/γ-Al2O3-(500)]. H MAS NMR: δH ) 1.0
ppm. 13C CP MAS NMR ) 85 (weak) and 32 ppm. Elemental analysis:
3.8%wt W, 3.6%wt C, 14 ( 2 C/W.
Preparation of [(13C 30%) W(tCtBu)(CH2tBu)3/Al2O3-(500)] by
Impregnation of (13C 30%) [W(tCtBu)(CH2tBu)3] onto Al2O3-(500)
(Large Scale). [2*/γ-Al2O3-(500)] was prepared using the procedure
1
described above using 2* and γ-Al2O3-(500). H MAS NMR: δH 1.0
ppm. 13C CP MAS NMR: 318(*), 103(*), 95(*), 85(*), 52, 32, 29.
Elemental analysis: 4.7%wt W, 4.6%wt C, 15 ( 2 C/W.
Treatment under H2 of the Solids. The solid was heated at 150
°C in the presence of a large excess of anhydrous H2 (77 330 Pa).
After 15 h, the gaseous product was quantified by GC.
Acknowledgment. We would like to thank BP Chemicals
and the CNRS for financial support. F.B. is grateful to the
Ministry of Research and Education for a predoctoral fellowship.
We thank the Centre Informatique National de l’Enseignement
Supe´rieur (CINES) at Montpellier and the Poˆle Scientifique de
Mode´lisation Nume´rique (PSMN) at ENS-Lyon for CPU time.
Supporting Information Available: More results on 1H NMR
spectroscopy and chemical shift modeling. This material is
available free of charge via the Internet at http://pubs.acs.org.
Grafting of 2 (IR). The same procedure was used for 2 by
sublimation of 1 in place of 2.
Preparation of [Zr(CH2tBu)4/γ-Al2O3-(500)]D by Impregnation of
[Zr(CH2tBu)4] onto γ-Al2O3-(500) from Degussa (Large Scale).
(56) King, S. A.; Schwartz, J. Inorg. Chem. 1991, 30, 3771.
(57) Clark, D. N.; Schrock, R. R. J. Am. Chem. Soc. 1978, 100, 6774-6776.
(58) Hediger, S.; Meier, B. H.; Kurur, N. D.; Bodenhausen, G.; Ernst, R. R.
Chem. Phys. Lett. 1994, 223, 283.
JA0616736
9
J. AM. CHEM. SOC. VOL. 128, NO. 28, 2006 9169