Half-sandwich group 4 metal siloxy and silsesquioxane complexes: Soluble model systems for silica-grafted olefin polymerization catalysts

Article abstract of DOI:10.1021/om9904495

The cuboctameric hydroxysilsesquioxane (C-C₅H₉)₇Si₈O₁₂(OH) (2), obtained after hydrolysis of(C-C₅H₉)₇Si₈O₁₂Cl (1), and triphenylsilanol have been applied as model supports for silicagrafted olefin polymerization catalysts. The ligands were introduced on group 4 metals by either chloride metathesis or protonolysis. Treatment of Cp″MCl₃ (M = Ti, Zr; Cp″ = 1,3-C₅H₃(SiMe₃)₂) with silsesquioxane and siloxylithium or -thallium salts, [(C-C₅H₉)₇Si₈O₁₃]M′ (M′ = Tl (3), Li (4), Li-TMEDA (5)) or Ph₃SiOTl gave either the dichloride complexes Cp″-[(C-C₅H₉)₇Si₈O ₁₃]MCl₂ (M = Ti (6a), Zr (7a)) and Cp″[Ph₃SiO]TiCl₂ (8a) or the monochloride species Cp″[(c-C₅H₉)₇Si₈O ₁₃]₂MCl (M = Ti (6b), Zr (Tb)) and Cp″[Ph₃SiO]₂MCl (M = Ti (8b), Zr (9)). Similarly, protonolysis of Cp″MR₃ with the silanols 2 and Ph₃SiOH yielded the corresponding silsesquioxane bis(alkyl) complexes Cp″[(c-C₅H₉)₇Si₈O ₁₃]TiR₂ (R = CH₂Ph (10a), Me (10b)) and triphenylsiloxy bis(alkyl) compounds Cp″[Ph₃SiO]MR₂ (M = Ti, R = CH₂Ph (11a), Me (11b); M = Zr, R - CH₂Ph (12a)) and the monobenzyl complex Cp″[Ph₃SiO]₂ZrCH₂-Ph (12b). When activated with MAO, not only the dichloride complexes (6a, 7a, 8a) but also the monochlorides (6b, 7b, 8b, 9) yield active ethylene polymerization catalysts. The observation that even complexes containing a tridentate silsesquioxane ligand, [(C-C₅H₉)₇Si₈O ₁₂]-MCp″ (M = Ti (13), Zr (14)), form active ethylene polymerization catalysts when activated with MAO indicates that silsesquioxane and siloxy ligands are easily substituted by MAO. The silsesquioxane and siloxy bis(alkyl) complexes (10, 11, 12a) form active olefin polymerization catalysts when activated with B(C₆F₅)₃, which leaves the M-O bond unaffected. Although the different cone angles of (C-C₅H₉)₇Si₈O₁₃ (155°) and Ph₃SiO (132°) suggest otherwise, the effective steric congestion around the metal center of (C-C₅H₉)₇Si₈O₁₃- and Ph₃SiO-stabilized complexes was found to be reasonably comparable. The electronic differences between (C-C₅H₉)₇Si₈O₁₂(OH) (2) and Ph₃SiOH are more pronounced. pK_a measurements and DFT calculations indicate that 2 is notably more Br?nsted acidic and electron withdrawing than Ph₃SiOH.

Full text of DOI:10.1021/om9904495

Organometallics 1999, 18, 5447-5459
5447
Articles
Ha lf-Sa n d w ich Gr ou p 4 Meta l Siloxy a n d Silsesqu ioxa n e
Com p lexes: Solu ble Mod el System s for Silica -Gr a fted
Olefin P olym er iza tion Ca ta lysts
Robbert Duchateau,*,† Ulrich Cremer,† Roelant J . Harmsen,† Said I. Mohamud,†
Hendrikus C. L. Abbenhuis,† Rutger A. van Santen,† Auke Meetsma,‡
Sven K.-H. Thiele,§ Maurits F. H. van Tol,§ and Mirko Kranenburg§
Schuit Institute of Catalysis, Eindhoven University of Technology, P.O. Box 513,
5600 MB Eindhoven, The Netherlands, University of Groningen, Nijenborgh 4,
9747 AG Groningen, The Netherlands, and DSM Research BV, P.O. Box 18,
6160 MD Geleen, The Netherlands
Received J une 10, 1999
The cuboctameric hydroxysilsesquioxane (c-C5H9)7Si8O12(OH) (2), obtained after hydrolysis
of (c-C5H9)7Si8O12Cl (1), and triphenylsilanol have been applied as model supports for silica-
grafted olefin polymerization catalysts. The ligands were introduced on group 4 metals by
either chloride metathesis or protonolysis. Treatment of Cp′′MCl3 (M ) Ti, Zr; Cp′′ ) 1,3-
C5H3(SiMe3)2) with silsesquioxane and siloxylithium or -thallium salts, [(c-C5H9)7Si8O13]M′
(M′ ) Tl (3), Li (4), Li‚TMEDA (5)) or Ph3SiOTl gave either the dichloride complexes Cp′′-
[(c-C5H9)7Si8O13]MCl2 (M ) Ti (6a ), Zr (7a )) and Cp′′[Ph3SiO]TiCl2 (8a ) or the monochloride
species Cp′′[(c-C5H9)7Si8O13]2MCl (M ) Ti (6b), Zr (7b)) and Cp′′[Ph3SiO]2MCl (M ) Ti (8b),
Zr (9)). Similarly, protonolysis of Cp′′MR3 with the silanols 2 and Ph3SiOH yielded the
corresponding silsesquioxane bis(alkyl) complexes Cp′′[(c-C5H9)7Si8O13]TiR2 (R ) CH2Ph (10a ),
Me (10b)) and triphenylsiloxy bis(alkyl) compounds Cp′′[Ph3SiO]MR2 (M ) Ti, R ) CH2Ph
(11a ), Me (11b); M ) Zr, R ) CH2Ph (12a )) and the monobenzyl complex Cp′′[Ph3SiO]2ZrCH2-
Ph (12b). When activated with MAO, not only the dichloride complexes (6a , 7a , 8a ) but also
the monochlorides (6b, 7b, 8b, 9) yield active ethylene polymerization catalysts. The obser-
vation that even complexes containing a tridentate silsesquioxane ligand, [(c-C5H9)7Si8O12]-
MCp′′ (M ) Ti (13), Zr (14)), form active ethylene polymerization catalysts when activated
with MAO indicates that silsesquioxane and siloxy ligands are easily substituted by MAO.
The silsesquioxane and siloxy bis(alkyl) complexes (10, 11, 12a ) form active olefin polym-
erization catalysts when activated with B(C6F5)3, which leaves the M-O bond unaffected.
Although the different cone angles of (c-C5H9)7Si8O13 (155°) and Ph3SiO (132°) suggest
otherwise, the effective steric congestion around the metal center of (c-C5H9)7Si8O13- and
Ph3SiO-stabilized complexes was found to be reasonably comparable. The electronic
differences between (c-C5H9)7Si8O12(OH) (2) and Ph3SiOH are more pronounced. pKa
measurements and DFT calculations indicate that 2 is notably more Brønsted acidic and
electron withdrawing than Ph3SiOH.
In tr od u ction
with narrow molecular weight distributions. Addition-
ally, immobilization of the catalyst can reduce the for-
mation of dormant sites, which, in principle, makes it
possible to decrease the necessary amount of cocata-
lyst.3a,h-j The most common method to immobilize
Heterogenization of well-defined homogeneous olefin
polymerization catalysts offers opportunities to combine
the advantages of both supported and homogeneous cat-
alysts.1 In the ideal case, morphology control by replica-
tion leads to uniform polymer particles with narrow
particle size distribution and high bulk density,2 whereas
the single-site nature of the catalyst provides polymers
(1) (a) Zakharov, V. A.; Yermakov, Y. I. Catal. Rev.-Sci. Eng. 1979,
19, 67. (b) Collette, J . W.; Tullock, C. W. U.S. Pat. Appl. 4,298,722,
1981 (c) Gahleitner, M.; Ledwinka, H.; Hafner, N.; Neissl, W. Proceed-
ings of Sixth International Business Forum on Specialty Polyolefins,
SPO ‘96, 1996, p 283. (d) Marsden, C. E. Plast., Rubber Compos.
Process. Appl. 1994, 21, 193. (e) Catalytic Polymerization of Olefins,
Studies in Surface Science and Catalysis; Keii, T., Soga, K., Eds.;
Elsevier: New York, 1986. (f) Transition Metals and Organometallics
as Catalysts for Olefin Polymerization; Kaminsky, W., Sinn, H., Eds.;
Springer-Verlag: New York, 1988. (g) Olabisi, O.; Atiqulla, M.;
Kaminsky, W. Makromol. Chem. Phys. 1997, C37(3), 519 and refer-
ences therein.
* To whom correspondence should be addressed. E-mail:
R.Duchateau@tue.nl. Fax: +31-40-2455054.
† Eindhoven University of Technology.
‡ University of Groningen.
§ DSM Research BV.
10.1021/om9904495 CCC: $18.00 © 1999 American Chemical Society
Publication on Web 11/19/1999

5448 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
homogeneous olefin polymerization catalysts consists of
physisorption of metallocenes on a support that is pre-
treated with methylalumoxane (MAO) or other cocata-
lysts.1e-g,3 Whereas this method results in heteroge-
neous catalysts, leaching remains one of the major prob-
lems. Improvement of this and alternative immobiliza-
tion techniques, such as catalyst tethering4 and graft-
ing,1a-d,5 remains therefore a topic of great importance.
amido, cyclopentadienyl),7g,h,8 which are active olefin
epoxidation catalysts and mimic titanium sites in
heterogeneous epoxidation catalysts such as TS-1 and
titanium grafted on MCM-41.9
As part of a project on immobilization of homogeneous
olefin polymerization catalysts, we were interested in
modeling grafting of group 4 metal complexes onto
partly dehydroxylated silica. With regard to olefin
polymerization, silica-grafted species of the type [SiO2]-
O-M(Cp)X2 (X ) chloride, alkyl), containing two chlo-
ride or alkyl substituents, are of interest.1g,10 Prepara-
tion of such heterogeneous catalysts can be simplified
considerably using synthetic strategies developed for
homogeneous model complexes. Hence, we were looking
for simple yet realistic models for isolated silanol func-
tionalities as found in partially dehydroxylated silica.
Here we report the use of triphenylsilanol, Ph3SiOH,
and the cuboctameric hydroxysilsesquioxane (c-C5H9)7-
Si8O12(OH) as homogeneous supports to mimic grafting
of group 4 metal half-sandwich complexes onto partly
dehydroxylated silica. The olefin polymerization activity
of the metallasiloxy and metallasilsesquioxane com-
plexes is studied, as well as the stability of the com-
plexes against leaching. Finally, to get an insight into
the differences between Ph3SiOH and (c-C5H9)7Si8O12-
(OH) and to assess the suitability of these ligands as
models for silica surface silanol sites, the steric and
electronic differences between Ph3SiOH and (c-C5H9)7-
Si8O12(OH) are investigated. Parts of this work have
been communicated.11
To get more insight, at a molecular level, into pro-
cesses taking place at silica surfaces and to improve the
immobilization methodologies, suitable homogeneous
model systems for silica surface silanol sites are of great
importance. The major advantage of such homogeneous
model systems is that their structural properties and
reactivity can be studied in detail using a wide range
of powerful techniques such as multinuclear solution
NMR spectroscopy. Several siloxy- and silsesquioxane-
stabilized group 4 metal complexes have been used as
models for silica-grafted systems.6,7 Recently, the im-
portance of homogeneous model systems for heteroge-
neous catalysis has clearly been demonstrated by the
development of titanasilsesquioxane complexes [R7-
Si7O12]TiX (R ) c-C5H9, c-C6H11; X ) alkyl, alkoxide,
(2) For example see: Hungenberg, K. H.; Kerth, J .; Langhauser, F.;
Marczinke, B.; Schlund, R. In Ziegler Catalysts; Fink, G., Mu¨lhaupt,
R., Brintzinger, H. H., Eds.; Springer-Verlag: New York, 1995; Chapter
20.
(3) (a) Chien, J . C. W.; He, D. J . Polym. Sci., Part A: Polym. Chem.
1991, 1603. (b) Van Looveren, L. K.; Geysen, D. F. M. C.; Vercruysse,
K. A. L.; Wouters, B. H. J .; Grobet, P. J .; J acobs, P. A. Angew. Chem.
1998, 110, 540. (c) Kaminsky, W.; Renner, F. Makromol. Chem., Rapid
Commun. 1993, 14, 239. (d) Ko, Y. S.; Han, T. K.; Park, J . W.; Woo, S.
I. Macromol. Rapid Commun. 1996, 17, 749. (e) Kaminaka, M.; Soga,
K. Makromol. Chem., Rapid Commun. 1991, 12, 367. (f) Soga, K.;
Kaminsky, W. Makromol. Chem., Rapid Commun. 1992, 13, 221. (g)
Soga, K.; Kaminsky, W. Makromol. Chem. 1993, 194, 1745. (h) Chien,
J . C. W. Top. Catal. 1999, 7, 23 and references therein. (i) Welborn,
H. C. J r. U.S Pat. Appl. 4,897,455, 1990. (j) Canich, J .-A. M.; Licciardi,
G. P. U.S. Pat. Appl. 5,057,475, 1991.
(4) (a) Iiskola, E. I.; Timonen, S.; Pakkanen, T. T.; Ha¨rkki, O.;
Lehmus, P.; Seppa¨la¨, J . V. Macromolecules 1997, 30, 2853. (b) Arai,
T.; Ban, H. T.; Uozumi, T.; Soga, K. J . Polym. Sci., Part A: Polym.
Chem. 1998, 36, 421. (c) Arai, T.; Ban, H. T.; Uozumi, T.; Soga, K.
Macromol. Chem. Phys. 1997, 198, 237.
(5) (a) Ballard, D. G. H. Adv. Catal. 1973, 23, 263. (b) Schwartz, J .;
Ward, M. D. J . Mol. Catal. 1980, 8, 465. (c) King, S. A.; Schwartz, J .
Inorg. Chem. 1991, 30, 3771. (d) Scott, S. L.; Basset, J .-M.; Niccolai,
G. P.; Santini, C. C.; Candy, J .-P.; Lecuyer, C.; Quignard, F.; Choplin,
A. New J . Chem. 1994, 18, 115. (e) Vidal, V.; The´olier, A.; Thivolle-
Cazat, J .; Basset, J .-M.; Corker, J . J . Am. Chem. Soc. 1996, 118, 4595.
(f) Quignard, F.; Le´cuyer, C.; Choplin, A.; Basset, J .-M. J . Chem. Soc.,
Dalton Trans. 1994, 1153. (g) Ajjou, J . A.; Scott, S. Organometallics
1997, 16, 86. (h) Bade, O. M.; Blom, R.; Ystenes, M. Organometallics
1998, 17, 2524. (i) Theopold, K. H. Eur. J . Inorg. Chem. 1998, 15 and
references therein.
(6) For example see: (a) Bennet, J . L.; Wolczanski, P. T. J . Am.
Chem. Soc. 1997, 119, 10696 and references therein. (b) Murugavel,
R.; Chandrasekhar, V.; Roesky, H. W. Acc. Chem. Res. 1996, 29, 183.
(c) Winkhofer, N.; Voigt, A.; Dorn, H.; Roesky, H. W.; Steiner, A.;
Stalke, D.; Reller, A. Angew. Chem. 1994, 106, 1414. (d) Voigt, A.;
Murugavel, R.; Montero, M. L.; Wessel, H.; Liu, F.-Q.; Roesky, H. W.;
Uso´n, I.; Albers, T.; Parisini, E. Angew. Chem. 1997, 109, 1020. (e)
Murugavel, R.; Voigt, A.; Walawalkar, M. G.; Roesky, H. W. Chem.
Rev. 1996, 96, 2205.
Resu lts a n d Discu ssion
Cu bocta m er ic Hyd r oxysilsesqu ioxa n e a n d An -
ion ic Der iva tives. Recently, we reported the synthesis
of the hydroxysilsesquioxane (c-C5H9)7Si8O12(OH).11
Treating (c-C5H9)7Si7O9(OH)3 with 1 equiv of SiCl4 in
the presence of NEt3 yielded the corner-capped chlo-
rosilsesquioxane (c-C5H9)7Si8O12Cl (1; eq 1),12 which,
after subsequent hydrolysis in a mixture of THF and
water (2:1), afforded the monosilanol (c-C5H9)7Si8O12-
(OH) (2; eq 1) as a white microcrystalline material. The
hydrolysis is surprisingly slow, and 2 days of refluxing
is required to complete the reaction.
Hydroxysilsesquioxane 2 is an interesting homoge-
neous model system for isolated silanol sites in partially
dehydroxylated silica. Its high symmetry (C3v) signifi-
cantly facilitates NMR studies, whereas the cubic sil-
sesquioxane framework of 2 is stable toward degrada-
tion by strong nucleophilic reagents.13 Like all silses-
quioxane silanols, compound 2 reacts cleanly with 1
equiv of TlOEt in toluene to afford thin white needles
(8) (a) Abbenhuis, H. C. L.; Krijnen, S.; van Santen, R. A. Chem.
Commun. 1997, 331. (b) Crocker, M.; Herold, R. H. M. PCT Pat. Appl.,
WO97/24344. (c) Maschmeyer, T.; Klunduk, M. C.; Martin, C. M.;
Shephard, D. S.; Thomas, J . M.; J ohnson, B. F. G. Chem. Commun.
1997, 1847.
(7) (a) Feher, F. J .; Blanski, R. L. J . Chem. Soc., Chem. Commun.
1990, 1614. (b) Feher, F. J .; Blanski, R. L. J . Am. Chem. Soc. 1992,
114, 5886. (c) Liu, J .-C. Chem. Commun. 1996, 1109. (d) Feher, F. J .;
Blanski, R. L. Makromol. Chem., Macromol. Symp. 1993, 66, 95. (e)
Feher, F. J .; Tajima, T. L. J . Am. Chem. Soc. 1994, 116, 2145. (f) Feher,
F. J .; Walzer, J . F. Inorg. Chem. 1991, 30, 1689. (g) Crocker, M.; Herold,
R. H. M.; Orpen, A. G. Chem. Commun. 1997, 2411. (h) Krijnen, S.;
Abbenhuis, H. C. L.; Hanssen, R. W. J . M.; van Hooff, J . H. C.; van
Santen, R. A. Angew. Chem., Int. Ed. 1998, 37, 356. (i) Duchateau, R.
Abbenhuis, H. C. L.; van Santen, R. A.; Meetsma, A.; Thiele, S. K.-H.;
van Tol, M. F. H. Organometallics 1998, 17, 5663.
(9) Corma, A.; Esteve, P.; Mat´ınez, A.; Valencia, S. J . Catal. 1995,
152, 18 and references therein.
(10) For example see: (a) Park, J . R.; Shiono, T.; Soga, K. Macro-
molecules 1992, 25, 521. (b) Soga, K.; Lee, D. H. Makromol. Chem.
1992, 193, 1687. (c) Soga, K.; Koide, R.; Uozumi, T. Makromol. Chem.,
Rapid Commun. 1993, 14, 511. (d) Kratochvila, J .; Shiono, T.; Soga,
K. Makromol. Chem., Rapid Commun. 1993, 14, 85. (e) Uozumi, T.;
Toneri, T.; Soga, K. Macromol. Rapid Commun. 1997, 18, 9. (f) Xu, J .;
Zhao, J .; Fan, Z.; Feng, L. Macromol. Rapid Commun. 1997, 18, 875.
(g) Malinge, J . Proceedings of Seventh International Business Forum
on Specialty Polyolefins, SPO ‘97; 1997, p 123.

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5449
6b can directly be synthesized by salt metathesis of
Cp′′TiCl3 with 2 equiv of 4 in hexane (Scheme 1).
Despite its high solubility in organic solvents, which
prevents crystallization, the high selectivity of this
reaction allows isolation of analytically pure 6b as
lemon yellow foam after evaporation of the solvent.
Chloride substitution of Cp′′ZrCl3 with 1 equiv of 3
is less selective than for titanium and yields a 2:1
mixture of Cp′′[(c-C5H9)7Si8O13]ZrCl2 (7a ) and Cp′′[(c-
C5H9)7Si8O13]2ZrCl (7b; Scheme 1). As for titanium,
attempts to improve the selectivity of the reaction by
varying the reaction conditions or silsesquioxane re-
agents failed. The similar solubilities of 7a and 7b
prevented their separation, and as a result, complex 7a
could not be purified. On the other hand, analytically
pure 7b can be obtained by reacting Cp′′ZrCl3 with 2
equiv of 3 in toluene (Scheme 1). Attempts to substitute
the last chloride in the bis(silsesquioxane) complexes
6b and 7b with another [(c-C5H9)7Si8O13] group failed,
exclusively yielding ill-defined product mixtures. Prob-
ably steric reasons prevent clean substitution of the final
chloride in these complexes.
of the corresponding thallium salt [(c-C5H9)7Si8O13]Tl
(3; eq 1) after crystallization from hexane. In accordance
with the structurally characterized {[(c-C6H11)7Si7O9-
(OSiMe3)2O]Tl}2,13b the thallium silsesquioxane 3 is
probably also dimeric. Similarly, deprotonation of 2 by
an equimolar amount of n-BuLi in hexane solution
selectively affords {[(c-C5H9)7Si8O13]Li}n (4; eq 1). The
extremely high hexane solubility of 4 suggests an
oligomeric structure in apolar solvents. Treatment of 4
with TMEDA results in splitting of this oligomeric
structure and gives the corresponding Lewis base ad-
duct [(c-C5H9)7Si8O13]Li‚TMEDA (5; eq 1), which crys-
tallizes as thin needles from hexane.
The selectivity of the reactions of Cp′′MCl3 (M ) Ti,
Zr) with Ph3SiOTl is similar to that observed with the
silsesquioxane salts 3-5. For example, the reaction of
Cp′′TiCl3 with Ph3SiOTl provides analytically pure Cp′′-
[Ph3SiO]TiCl2 (8a ; Scheme 1) after repeated recrystal-
lization from hexane to remove small amounts of
Cp′′TiCl3 and Cp′′[Ph3SiO]2TiCl (8b). On the other hand,
the equimolar reaction between Ph3SiOTl and Cp′′ZrCl3
yields a 1:1 mixture of the bis(siloxy) complex Cp′′[Ph3-
SiO]2ZrCl (9) and starting material, Cp′′ZrCl3, indicat-
ing that for zirconium the formation of the bis(siloxy)
complex is clearly favored. Like 6b/7b, the bis(triphen-
ylsiloxy)titanium and -zirconium compounds Cp′′[Ph3-
SiO]2MCl (M ) Ti (8b), Zr (9)) are selectively formed
by treatment of Cp′′MCl3 with 2 equiv of Ph3SiOTl in
toluene (Scheme 1). Attempts to substitute the final
chloride in 8b and 9 invariably failed.
In CDCl3, all cuboctameric silsesquioxanes described
above display simple yet particularly informative 13C
and 29Si NMR spectra showing well-resolved sets of CH-
methine and Si resonances, consistent with the C3v
symmetry of the silsesquioxane ligand. Not unexpect-
1
edly, the H NMR spectra are much less informative,
due to the large number of similar CH2 and CH protons
present.
The NMR data of the half-sandwich titanium and
zirconium silsesquioxane complexes correspond well
with each other. For example, the 13C{1H} NMR spectra
of the mono(silsesquioxane) complexes Cp′′[(c-C5H9)7-
Si8O13]MCl2 (M ) Ti (6a ), Zr (7a )) display a single
resonance for all seven methine (CH) carbons. In
contrast, the 13C{1H} NMR spectra of the bis(silsesqui-
oxane) complexes Cp′′[(c-C5H9)7Si8O13]2MCl (M ) Ti
(6b), Zr (7b)) show a well-resolved set of methine CH
resonances (1:3:3 ratio), in agreement with the C3v
symmetry of the silsesquioxane ligand. 29Si NMR spec-
troscopy proves to be a very useful tool for the charac-
terization of silsesquioxane complexes, as the differences
in chemical shift are large enough to distinguish be-
tween very similar complexes such as 6 and 7, and the
integrated relative intensities are in good agreement
with the expected stoichiometries. For the metallasil-
sesquioxane complexes 6 and 7, the 29Si NMR spectra
display four resonances in a 2:3:4:1 ratio for 6a /7a
and a 1:3:4:1 ratio for 6b/7b, respectively: the Cp-
SiMe3 resonances are typically found around δ -7 ppm.
The alkyl-substituted silsesquioxane silicon atoms,
(O)3SiC5H9, show two resonances in a 3:4 ratio (δ -65
and -68 ppm). Three of the seven silicon atoms,
probably the ortho Si, are clearly distinct from the other
Ha lf-Sa n d w ich Gr ou p 4 Meta l Ch lor id e Silses-
qu ioxa n e a n d Tr ip h en ylsiloxy Com p lexes. We first
investigated halogen displacement routes to mono-
(silsesquioxane) and mono(triphenylsiloxy) chloride com-
plexes Cp′′[(c-C5H9)7Si8O13]MCl2 and Cp′′[Ph3SiO]MCl2
(Cp′′ ) 1,3-C5H3(SiMe3)2), respectively.14 The reaction
of Cp′′TiCl3 with an equimolar amount of [(c-C5H9)7-
Si8O13]Li (4) in hexane provides Cp′′[(c-C5H9)7Si8O13]-
TiCl2 (6a ) as a bright yellow microcrystalline material
after repeated recrystallization from hexane (Scheme
1). Independent of the silsesquioxane salt (3-5), solvent
(hexane, toluene, ether, THF), and reaction temperature
(-30 °C to reflux) used, small amounts (∼10%) of Cp′′-
[(c-C5H9)7Si8O13]2TiCl (6b) were also formed. Compound
(11) Duchateau, R.; Abbenhuis, H. C. L.; van Santen, R. A.; Thiele,
S. K.-H.; van Tol, M. F. H. Organometallics 1998, 17, 5222.
(12) Feher, F. J .; Weller, K. J .; Schwab, J . J . Organometallics 1995,
14, 2009.
(13) Several silsesquioxanes have found to be sensitive to nucleo-
philic attack. For example see: (a) Feher, F. J .; Budzichowski, T. A.;
Rahimian, K.; Ziller, J . W. J . Am. Chem. Soc. 1992, 114, 3859. (b)
Feher, F. J .; Rahimian, K.; Budzichowski, T. A.; Ziller, J . W. Organo-
metallics 1995, 14, 3920. (c) Day, V. W.; Klemperer, W. G.; Mainz, V.
V.; Millar, D. M. J . Am. Chem. Soc. 1985, 107, 8262.
(14) Introduction of bulky trimethylsilyl substituents on the cyclo-
pentadienyl ligands significantly improved the selectivity of the
reactions.

5450 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
Sch em e 1
four, resulting in two resonances with observed relative
intensities of 3:4. Finally, the ipso Si atom resonances
appear around δ -110 ppm, characteristic for silicon
surrounded by four oxygen atoms. For the corresponding
triphenylsiloxy complexes 8 and 9 the 29Si NMR spectra
show two resonances in a 2:1 (8a , δ -6.3, -9.2 ppm) or
1:1 (8b, δ -7.3, -12.6 ppm; 9, δ -8.5, -16.5 ppm) ratio
for the Cp-SiMe3 and Ph3SiO silicon atoms, respec-
tively.
X-r a y Str u ctu r e
Deter m in a tion
of Cp ′′-
[P h 3SiO]2Zr Cl (9). The poor crystallinity of the sil-
sesquioxane complexes 6 and 7 prevented their single-
crystal structure determinations.15 The crystallinity of
the corresponding triphenylsiloxy complexes 8 and 9 is
much better, and slow crystallization of a saturated
hexane solution at room temperature afforded Cp′′[Ph3-
SiO]2ZrCl (9) as large colorless needles suitable for an
X-ray diffraction study. A perspective view of the
molecular structure of 9 is shown in Figure 1. Crystal
data are collected in Table 1. Conformational disorder
was observed for the two cyclopentadienyl trimethylsilyl
substituents. One of the two conformations of both
SiMe3 substituents is shown. The structure consists of
a three-legged piano-stool configuration in which the
F igu r e 1. ORTEP drawing of Cp′′[Ph3SiO]2ZrCl (9). El-
lipsoids are scaled to enclose 40% of the electron density.
Selected bond distances (Å): Zr-Cl, 2.392(2); Zr-O(1),
1.921(4); Zr-O(2), 1.929(5); Zr-Cg, 2.217(4). Bond angles
(deg): Zr-O(1)-Si(3), 175.3(3); Zr-O(2)-Si(4), 159.3(2);
Cl-Zr-O(1), 103.32(14); Cl-Zr-O(2), 102.82(13); O(1)-
Zr-O(2), 102.17(19); Cl-Zr-Cg, 111.35(10); O(1)-Zr--g,
119.05(18); O(2)-Zr-Cg, 116.15(15).
tetrahedrally surrounded zirconium atom is bonded to
two siloxy groups, one chloride and one cyclopentadienyl
ligand. The Zr-O bond distances (Zr-O(1) ) 1.921(4)
Å; Zr-O(2) ) 1.929(5) Å) in 9 are short compared to
those in zirconasilsesquioxane complexes [(c-C6H11)7-
(15) Poor crystallinity is
a general characteristic of silsesqui-
oxane compounds: (a) Feher, F. J . J . Am. Chem. Soc. 1986, 108,
3850. (b) Feher, F. J .; Budzichowski, T. A.; Ziller, J . W. Inorg. Chem.
1992, 31, 5100. (c) Buys, I. E.; Hambley, T. W.; Houlton, D. J .;
Maschmeyer, T.; Masters, A. F.; Smith, A. K. J . Mol. Catal. 1994, 86,
309.

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5451
Ta ble 1. Deta ils of th e X-r a y Str u ctu r e Deter m in a tion of Cp ′′[P h 3SiO]2Zr Cl (9) a n d Cp ′′[P h 3SiO]Ti(CH2P h )2
(11a )
9
11a
formula
fw
C
47H51ClO2Si4Zr
C43H50OSi3Ti
714.99
886.94
cryst syst
space group, No.
a, Å
b, Å
c, Å
trigonal
R3h (H, obv.), 148
48.71(1)
triclinic
P1h, 2
11.461(1)
12.709(1)
14.702(1)
83.503(5)
70.018(5)
84.972(5)
1996.9(3)
1.189
10.998(2)
R, deg
â, deg
γ, deg
V, Å3
22 599(8)
1.173
18
D
Z
calcd, g cm-3
2
F(000), electrons
µ(Mo ΚR), cm-1
cryst size, mm
8316
4.0
760
3.4
0.45 × 0.50 × 0.55
0.25 × 0.30 × 0.38
T, K
130
130
θ range, deg: min, max
1.45, 26.0
0.710 73
graphite
∆ω ) 0.90 + 0.34 tan θ
h, 0-60; k, -59 to 0; l, 0-11
1.48, 27.0
0.710 73
graphite
∆ω ) 0.90 + 0.34 tan θ
h, -13 to +14; k, -16 to 16; l, 0-18
λ(Mo ΚR), Å
monochromator
ω/ 2θ scan, deg
index ranges
total no. of data
no. of unique data
no. of data with criterion Fo g 4.0σ(Fo)
10 145
9318
7567
0.058
0.038
9318
574
9035
8689
6626
0.0612
0.0408
8688
633
2
2
Rint ) ∑[|(Fo - Fo2(mean)|)/∑[Fo
]
2
Rσ ) ∑σ(Fo2)/∑[Fo
]
2
no. of rflns with Fo g 0
no. of refined params
final agreement factors
2
Rw(F2) for Fo > 0a
0.2354
0.1099, 217.3
0.0785
1.163
<0.001
0.1063
0.0622, 0.1390
0.0428
1.021
<0.001
weighting scheme: a, bb
R(F) for Fo > 4.0σ(Fo)c
GOF ) Sd
max (shift/σ) final cycle
a
2
b
Rw(F2) ) [∑[w(Fo - Fc2)2]/∑[w(Fo2)2]]1/2
.
w ) 1/[σ2(Fo2) + (aP)2 + bP] and P ) [max(Fo2,0) + 2Fc2]/3. c R(F) ) ∑(||Fo| - |Fc||)/∑|Fo|.
GOF ) S ) [∑[w(Fo - Fc2)2]/(n/p)]1/2; n ) number of reflections, and p ) number of parameters refined.
d
2
Si7O12]ZrCp* (1.958 Å),15a [(c-C5H9)7Si7O11(OSiMe3)]2Zr‚
2THF (2.00 Å average),7i and {[(c-C5H9)7Si7O12]ZrCH2-
Ph}2 (σ-bonded: 1.96 Å average)7i but compare well
with the Zr-O distances in (Ph3SiO)2ZrCl2‚DME (1.91-
(1) Å) and electron-deficient zirconium alkoxide com-
plexes.16 The combination of short Zr-O distances
(1.925 Å average) and nearly linear Zr-O-Si bond
angles (167.3° average) in 9 provide a clear manifesta-
tion of strong oxygen pπ-dπ donation of the siloxy
ligand to the zirconium center. To reduce the steric
congestion within the complex, the two bulky trimeth-
ylsilyl substituents of the cyclopentadienyl ligand oc-
cupy the least crowded position between the two siloxy
groups, whereas the phenyl groups of the siloxy ligands
are twisted like propeller blades with an average torsion
angle of 47°.
Ha lf-Sa n d w ich Gr ou p 4 Meta l Silsesqu ioxa n e
a n d Tr ip h en ylsiloxy Alk yl Com p lexes. In accor-
dance with the grafting of group 4 metal alkyls onto
partially dehydroxylated silica5 and the alcoholysis with
ROH (e.g. ROH ) (Me3C)3COH, 2,6-(Me3C)2-C6H3OH,
2,2′-CH2(C6H2-4-Me-6-(CMe3)OH)2),16 protonolysis reac-
tions of Ph3SiOH and (c-C5H9)7Si8O12(OH) (2) were
carried out.17 Reactions of Cp′′TiR3 with 2 proceed
selectively, providing the corresponding mono(silses-
quioxane) bis(alkyl) complexes Cp′′[(c-C5H9)7Si8O13]TiR2
(R ) CH2Ph (10a ), Me (10b); Scheme 1). Attempts to
introduce a second silsesquioxane failed, as both alkyls
10a ,b refuse to react with another 1 equiv of silsesqui-
oxane 2. Heating toluene solutions of 10a ,b exclusively
resulted in thermolysis of the bis(alkyl) complexes. The
reactivity of Ph3SiOH toward Cp′′TiR3 corresponds
surprisingly well to that of (c-C5H9)7Si8O12(OH) (2). The
triphenylsiloxy analogues of 10a ,b, Cp′′[Ph3SiO]TiR2 (R
) CH2Ph (11a ), Me (11b) Scheme 1), are selectively
formed by treatment of CpTiR3 (R ) CH2Ph, Me) with
equimolar amounts of Ph3SiOH. Like 10a ,b, the tita-
nasiloxy complexes 11a ,b refuse to react with a second
equivalent of Ph3SiOH.
For zirconium, the protonolysis reaction between
Cp′′ZrR3 (R ) Me, CH2Ph) and silsesquioxane 2 is much
less selective than for titanium. Instead of the expected
product, Cp′′[(c-C5H9)7Si8O13]ZrR2, treatment of Cp′′ZrR3
with 2 invariably yields complex product mixtures.
Multiple SiMe3 and silsesquioxane resonances in the
NMR spectra of these mixtures indicate the presence
of several products of the type Cp′′[(c-C5H9)7Si8O13]x-
ZrR3-x. The formation of these product mixtures indi-
cates that the rate of protonolysis for the inter-
mediate silsesquioxane metal benzyl complexes is at
least of the same order as for Cp′′ZrR3. In contrast with
the nonselective reactions described above, the proto-
nolysis reaction between Cp′′Zr(CH2Ph)3 and Ph3SiOH
proceeds very selectively with either 1 or 2 equiv of
(16) (a) Babaian, E. A.; Hrcir, D. C.; Bott, S. G.; Atwood, J . L. Inorg.
Chem. 1986, 25, 4818. (b) Latesky, S. L.; McMullen, A. K.; Niccolai,
G. P.; Rothwell, I. P.; Huffman, J . C. Organometallics 1985, 4, 902. (c)
Lubben, T. V.; Wolczanski, P. T.; Van Duyne, G. D. Organometallics
1984, 3, 977.
(17) Alkylation (MeLi, MeMgCl, Me2Zn, PhCH2MgCl) of the chloride
complexes 6-9 did not afford well-defined products.

5452 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
triphenylsilanol, affording Cp′′[Ph3SiO]Zr(CH2Ph)2
(12a ) or Cp′′[Ph3SiO]2ZrCH2Ph (12b, Scheme 1), re-
spectively.
The NMR (1H, 13C, 29Si) spectra of the titanasilses-
quioxane alkyl complexes 10a ,b show roughly the same
features as observed for the corresponding titanasil-
sesquioxane dichloro compound 6a . For example, only
one resonance is observed for all seven methine CH
carbon atoms of the silsesquioxane ligand and the 29Si
NMR spectra of 10a ,b display four resonances in a 2:3:
4:1 ratio. The chemical shifts of the methyl resonances
1
of 10b and 11b (10b, H δ 1.01 ppm, 13C δ 56.56 ppm,
1
1J C-H ) 123 Hz; 11b, H δ 1.10 ppm, 13C δ 55.80 ppm,
1J C-H ) 128 Hz) are virtually identical. Likewise, the
chemical shifts for the methylene resonances of 10a and
11a are also very similar. For both dibenzyl complexes
the methylene protons (TiCHH′) are diastereotopic,
giving an AB spin system in the 1H NMR spectrum (10a ,
F igu r e 2. ORTEP drawing of Cp′′[Ph3SiO]Ti(CH2Ph)2
(11a ). Ellipsoids are scaled to enclose 40% of the electron
density. Selected bond distances (Å): Ti-O(1), 1.8055(15);
Ti-C(30), 2.140(2); Ti-C(37), 2.145(2); Ti-Cg, 2.0644(11).
Bond angles (deg): Ti-O(1)-Si(3), 174.51(10); Ti-C(30)-
C(31), 113.54(14); Ti-C(37)-C(38), 118.49(15); O(1)-Ti-
C(30), 102.37(8); O(1)-Ti-C(37), 101.58(8); C(30)-Ti-
C(37), 102.23(8); O(1)-Ti-Cg, 126.64(5); C(30)-Ti-Cg,
109.14(8); C(37)-Ti-Cg, 112.02(8).
2
1
1H δ 3.48, δ 2.16 ppm, J H-H ) 10 Hz; 11a , H δ 3.16,
2
δ 2.22 ppm, J H-H ) 9.9 Hz), while the methylene
(TiCH2) carbons appear as triplets (13C: 10a , δ 86.83
1
1
ppm, J C-H ) 127 Hz; 11a , δ 86.1 ppm, J C-H ) 125
Hz). These values are comparable with those found for
structurally comparable cyclopentadienyl-alkoxo com-
plexes such as [η5:η1-C5Me4-2-C6H3(4-Me)O]Ti(CH2Ph)2
(1H, δ 2.50, δ 2.32 ppm, J H-H ) 10.2 Hz; 13C, δ 83.85
2
[η5:η1-C5R4(CH2)3O]TiCl2 (R ) H, 1.762(2) Å;20a R ) Me,
1.767(1) Å),20b and C5Me5[2,6-(i-Pr)2C6H3O]TiCl2 (1.772-
(3) Å)20c and the titanium alkoxide complexes (2,6-
Ph2C6H3O)2TiC4H8 (average 1.806(1) Å)20d and 2,2′-(4-
OMe,6-t-BuC6H2O)2Ti(CH2Ph)2 (average 1.810(8) Å)20e
and siloxy complexes [(Me3C)3SiO]2Ti(THF)NSiCMe3
(1.824(4) Å), [(Me3C)3SiO]3TiNH2 (1.815 Å),6a and Ph3-
SiOTi(OCH2CH2)3N (1.834 Å).21 Similar to what was
observed for 9, the partial Ti-O double-bond character
and the nearly linear Ti-O(1)-Si(3) angle of 174.51-
(10)° reflect additional oxygen π-donation to the metal.
The two benzyl ligands in 11a are inequivalent, but can
both be regarded as being η1-bound to titanium, which
is in agreement with the solution 1H and 13C NMR
spectra of 11a .
Olefin P olym er iza tion a n d Ca ta lyst Sta bility.
Olefin polymerization experiments of the described
metallasilsesquioxane and metallasiloxane chloro and
alkyl complexes have been carried out. Besides polym-
erization activity studies, possible “leaching” of the
complexes was also investigated.
With MAO as the cocatalyst, not only do the dichloride
complexes Cp′′[(c-C5H9)7Si8O13]MCl2 (M ) Ti (6a ), Zr
(7a )) and Cp′′[Ph3SiO]TiCl2 (8a ) generate active ethyl-
ene polymerization catalysts but also the corresponding
monochlorides Cp′′[(c-C5H9)7Si8O13]2MCl (M ) Ti (6b),
Zr (7b)) and Cp′′[Ph3SiO]2MCl (M ) Ti (8b), Zr (9))
polymerize ethylene.22 Obviously, the monochloride
ppm, 1J C-H ) 127.5 Hz)18a and [η5:η1-C5Me4(CH2)3O]Ti-
(CH2Ph)2 (1H, δ 2.22, δ 1.92 ppm, 2J H-H ) 10.3 Hz; 13C,
δ 78.35 ppm, 1J C-H ) 121.2 Hz).18b The normal chemical
1
shift and J C-H value of the TiCH2 group and the
absence of a high-field shift of benzyl ortho phenyl
protons in 10a and 11a excludes significant η2-bonding
of the benzyl group.19 In accordance with the titanium
dibenzyl complexes 10a and 11a , the zirconium monosi-
loxy dibenzyl complex Cp′′[Ph3SiO]Zr(CH2Ph)2 (12a )
shows an AB spin pattern for its diastereotopic meth-
2
ylene CHH′ protons (1H: δ 2.50, δ 2.10 ppm, J H-H
)
11.0 Hz) and a triplet for the methylene carbon (13C: δ
63.1 ppm, J C-H ) 124 Hz). On the other hand, the
1
methylene protons of the corresponding monobenzyl
complex Cp′′[Ph3SiO]2ZrCH2Ph (12b) are identical and
appear as a singlet (δ 2.84 ppm) in the 1H and as a
1
triplet (δ 59.3 ppm, J C-H ) 119 Hz) in the 13C NMR
spectrum.
X-r a y Str u ctu r e Deter m in a tion of Cp ′′[P h 3SiO]-
Ti(CH2P h )2 (11a ). The solid-state structure of 11a as
derived from single-crystal X-ray diffraction is shown
in Figure 2. The geometry around titanium is slightly
distorted tetrahedral, with an O(1)-Ti-Cp (centroid)
angle of 126.64°. Clearly, this large angle results from
the steric hindrance between the cyclopentadienyl and
the siloxy ligand. The Ti-O(1) bond length in 11a
(1.8055(15) Å) is comparable to those reported for
titanium cyclopentadienyl-alkoxide complexes, such as
[η5:η1-C5Me4-2-C6H3(4-Me)O]Ti(CH2Ph)2 (1.851(7) Å),18a
(20) (a) Trouve´, G.; Laske, D. A.; Meetsma, A.; Teuben, J . H. J .
Organomet. Chem. 1996, 511, 255. (b) Fandos, R.; Meetsma, A.;
Teuben, J . H. Organometallics 1991, 10, 59. (c) Nomura, K.; Naga,
N.; Miki, M.; Yanagi, K.; Imai, A. Organometallics 1998, 17, 2154. (d)
Thorn, M. G.; Hill, J . E.; Waratuke, S. A.; J ohnson, E. S.; Fanwick, P.
E.; Rothwell, I. P. J . Am. Chem. Soc. 1997, 119, 8630. (e) van der
Linden, A.; Schaverien, C. J .; Meijboom, N.; Ganter, C.; Orpen, A. G.
J . Am. Chem. Soc. 1995, 117, 3008.
(21) Menge, W. M. P. B.; Verkade, J . G. Inorg. Chem. 1991, 30,
4628.
(22) Conditions: 20 mmol of MAO; 20 µmol of catalyst; 5 atm of
ethylene, 80 °C, 7 min. Observed activities range between 0.4 and 1.2
kg of PE/ (mmol[cat] h atm), based on the polymer yield after 7 min.
(18) (a) Chen, Y.-X.; Fu, P.-F.; Stern, C.; Marks, T. J . Organome-
tallics 1997, 16, 5958. (b) Gielens, E. E. C. G.; Tiesnitsch, J . Y.; Hessen,
B.; Teuben, J . H. Organometallics 1998, 17, 1652.
(19) For example, see: (a) Bochmann, M.; Lancaster, S. J .; Hurst-
house, M. B.; Malik, K. M. A. Organometallics 1994, 13, 2235
and references therein. (b) Horton, A. D.; de With, J .; van der Linden,
A. J .; van de Weg, H. Organometallics 1996, 15, 2672. (c) Pellecchia,
C.; Immirzi, A.; Grassi, A.; Zambelli, A. Organometallics 1993, 12,
4473.

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5453
Ta ble 2. P olym er iza tion Resu lts of B(C6F 5)3
Activa ted System s
complexes can only yield active ethylene polymeriza-
tion catalysts when at least one of the silsesquioxane
or siloxy ligands is substituted. This clearly demon-
strates the poor stability of monodentate siloxy sys-
tems toward substitution by aluminum alkyls.23 To test
whether the chelate effect of silsesquioxane ligands
might improve the stability toward leaching, titanium
and zirconium precursors with tridentate silsesquioxane
ligands have been synthesized, [(c-C5H9)7Si7O12]MCp′′
(M ) Ti (13), Zr (14); eq 2).24
catalyst
activity (kg/
Mw
(amt, µmol) monomer yield (g) (mol h atm) (×10-3
)
Mw/Mn
10a (20)a
10b (20)a
11a (20)a
11b (20)a
12a (40)b
10a (20)c
11a (20)c
ethylene
ethylene
ethylene
ethylene
ethylene
1-hexene
1-hexene
10.2
9.8
26.2
17.0
0.6
880
840
2240
1460
3
260
3.3
440
360
2.9
3.7
4.5
5.8
3
4
2.9
11.0
2.0
2.1
a
Conditions: 1 mmol of B(C6F5)3, 80 °C, 7 min, 5 atm of
b
ethylene. Conditions: 1 mmol B(C6F5)3, 80 °C, 1 h, 5 atm of
ethylene. c Conditions: 0.2 mmol of B(C6F5)3, 25 °C, 16 h, neat
1-hexene.
both 10a and 11a form single-site catalysts when
activated with B(C6F5)3.25 The somewhat broader mo-
lecular weight distributions of the polyethylenes formed
can be explained by partial thermolysis of the catalysts
under the conditions applied (80 °C). The zirconium
dibenzyl complex 12a forms a very poor catalyst when
activated with B(C6F5)3. Ethylene was polymerized only
slowly, while 1-hexene was not polymerized at all. To
test whether B(C6F5)3 is capable of substituting silses-
quioxane or triphenylsiloxy ligands, 19F NMR studies
were carried out on solutions of Cp′′[(c-C5H9)7Si8O13]2-
TiCl (6b) and Cp′′[Ph3SiO]2TiCl (8b) with B(C6F5)3. As
expected, the 19F NMR spectra exclusively showed
unreacted B(C6F5)3.
St er ic a n d E lect r on ic P r op er t ies of (c-C5H 9)7-
Si8O12(OH) a n d P h 3SiOH. Although simple silanols
have been assumed to be less suitable models for silica
than silsesquioxanes,7f,26 the reactivity of Ph3SiOX and
(c-C5H9)7Si8O13X (X ) H, Li/Tl) toward Cp′′MX3 (M )
Ti, Zr; X ) Cl, alkyl) shows some remarkable similari-
ties, though some distinct differences could be observed
as well. To get insight into the steric and electronic
differences between Ph3SiOH and (c-C5H9)7Si8O12(OH),
and to assess their suitability as models for silica surface
silanol sites, the steric and electronic properties of both
ligands were investigated in more detail using molecular
modeling, DFT calculations, and pKa measurements.
To determine the steric bulk of both (c-C5H9)7Si8O12-
(OH) (2) and Ph3SiOH, CPK models (Figure 3) have
been built using molecular modeling (force field). Both
ligands are C3 symmetric, which makes the comparison
of cone angles more reliable. On the basis of the cone
angles of (c-C5H9)7Si8O13 (155°) and Ph3SiO (132°),27 it
is clear that silsesquioxane 2 is considerably larger than
triphenylsilanol. However, describing the steric bulk of
a system on the basis of cone angles alone is often not
sufficient. For example, in complexes containing sils-
esquioxane 2 the cyclopentyl substituents are located
The fact that even solutions of 13 and 14 treated with
MAO give active ethylene polymerization catalysts22
excludes any positive influence of the chelate effect of
the silsesquioxane ligands on the stability toward leach-
ing. Worth mentioning in this respect is a study by
Feher et al., who demonstrated that with equimolar
amounts of aluminum alkyls two of the three V-O
bonds in the vanadasilsesquioxane [(c-C6H11)7Si7O12]Vd
O are selectively split.7b,d Hence, the high affinity of
aluminum for silsesquioxane oxo functionalities makes
aluminum alkyls unsuitable as cocatalysts for these
systems, when used in excess. The facile activation of
6b/7b and 13/14 also gives rise to considerable doubt
about the stability of the M-O bonds in silica-grafted
mono(cyclopentadienyl) systems,1g,10 in the presence of
excess of aluminum alkyls. This is emphasized by Soga
et al.,10d who showed that significant amounts of tita-
nium dissolve in the solution in which the silica-
supported (C5Me5)TiCl3 is treated with MAO.
When activated with B(C6F5)3, the titanium bis(alkyl)
complexes 10 and 11 form active olefin polymerization
catalysts (Table 2). Ethylene was polymerized rapidly,
and 1-hexene was slowly transformed into atactic poly-
1-hexene (Table 2). The narrow molecular weight dis-
tributions of the poly-1-hexenes formed indicate that
(25) 19F NMR spectroscopy (toluene-d8, 25 °C) showed that treat-
ment of 10a with B(C6F5)3 affords a contact ion pair with [PhCH2B-
(C6F5)3]- 19F NMR (δ): -53.0, (Fo), -83.80 (Fp), -87.80 (Fm) ppm; ∆δ-
(
(Fm-Fp) ) 4.0 ppm), whereas the reaction of 11a with B(C6F5)3 yields
a solvent-separated ion pair (19F NMR (δ): -53.0 (Fo), -87.1 (Fp), -89.7
(Fm) ppm; ∆δ(Fm-Fp) ) 2.6 ppm). See also ref 19b. The 1H NMR spectra
(toluene-d8) were not very informative, as broad signals were observed
at both -45 and +25 °C.
(23) This is not very surprising, since CpTi(OBu)3/MAO and Cp*Ti-
(OCH2CH2)3N/MAO are known styrene polymerization catalysts,
indicating that Ti-O bonds can be split by MAO: (a) Chien, J . C. W.;
Salajka, Z. J . Polym. Sci., Part A: Polym. Chem. 1991, 29, 1253. (b)
Kim, Y.; Hong, E.; Lee, M. H.; Kim, J .; Han, Y.; Do, Y. Organometallics
1999, 18, 36.
(24) (a) Feher, F. J . J . Am. Chem. Soc. 1986, 108, 3850. (b) Buys, I.
E.; Hambley, T. W.; Houlton, D. J .; Maschmeyer, T.; Masters, A. F.;
Smith, A. K. J . Mol. Catal. 1994, 86, 309.
(26) Feher, F. J .; Newman, D. A.; Walzer, J . F. J . Am. Chem. Soc.
1989, 111, 1741.
(27) As input for the calculations, force field optimized structures
of the ligands were used, to which a Ti atom was attached at 1.85 Å,
with a Si-O-Ti angle of 180°. The cone angles were calculated using
a modified version of Steric: Taverner, B. C. J . Comput. Chem. 1996,
14, 1612.

5454 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
F igu r e 3. CPK models for (c-C5H9)7Si8O13 and Ph3SiO
siloxy fragments.
approximately 1 Å further away from the metal center
than the phenyl groups in the corresponding triphenyl-
siloxy compounds (Figure 3). As a result, the effective
steric congestion around the metal centers in both
systems is less different than assumed on the basis of
cone angles alone. This assumption is supported by the
similar selectivities of the chloride metathesis reactions
of [(c-C5H9)7Si8O13]Li and Ph3SiOTl with Cp′′MCl3 (M
) Ti, Zr; vide supra) and the lack of both 10a and 11a
to react with another 1 equiv of (c-C5H9)7Si8O12(OH) and
Ph3SiOH, respectively.
Feher et al. have reported that, unlike usual sila-
nols, silsesquioxanes are strongly electron-withdrawing
ligands with an electrophilicity resembling that of cor-
responding silica surface silanol sites. The electron-with-
drawing capacity of the silsesquioxane in the Schrock-
type alkylidene complex [(c-C6H11)7Si7O11(OSiMe3)]Mo-
(CHCMe2Ph)(NAr) resembles that of the two (CF3)2-
(CH3)CO ligands in [(CF3)2(CH3)CO]Mo(CHCMe2Ph)-
(NAr).7d However, on the basis of Hammett correlations,
the electron-withdrawing tendency of the [Si8O12] frame-
work in aryl-substituted silsesquioxanes was even found
to approach that of a trifluoromethyl group.28 To study
the electronic character of Ph3SiOH and (c-C5H9)7Si8O12-
(OH), more direct methods to determine the electronic
properties of these ligands were sought.
As there is a clear correlation between the electron-
withdrawing tendency of the R substituent and the
acidity of ROH, the Brønsted acidities of Ph3SiOH and
(c-C5H9)7Si8O12(OH) (2) were determined by pKa mea-
surements.29 Subsequently, these pKa values were
compared with the calculated deprotonation energies of
the silanols. The pKa measurements were performed in
THF using the overlapping indicator method,29b-f and
the deprotonation energies are the result of DFT
calculations for the silanols at 0 K under vacuum.30
As can be seen from Figure 4, Ph3SiOH is significantly
less Brønsted acidic than (c-C5H9)7Si8O12(OH) (2). In-
teresting to note is the difference in acidity between (c-
F igu r e 4. pKa values and calculated deprotonation ener-
gies.
C5H9)7Si8O12(OH) (2) and the tris(silanol) (c-C5H9)7-
Si7O9(OH)3 (vide infra), which resembles the difference
in acidity of isolated silanols and silanol nests in zeolitic
materials.31 The pKa values are in good agreement with
the calculated deprotonation energies for H7Si7O9(OH)3,
H7Si8O12(OH), and Ph3SiOH, respectively.32 The depro-
tonation energy for H7Si8O12(OH) is significantly higher
than for CF3OH (∆E ) 47.9 kJ mol-1). Hence, on the
basis of these calculations, the electron-withdrawing
capacity of the [Si8O12] framework is less pronounced
than is found on the basis of Hammett correlations28
and is intermediate between those of the Ph3Si and the
CF3 substituent. Surprisingly, the calculated deproto-
nation energy of the tris(silanol) H7Si7O9(OH)3 was
found to be even lower than that of CF3OH. DFT
calculations show that the high acidity of (c-C5H9)7Si7O9-
(OH)3 is the result of effective stabilization of the charge
in the conjugated base [(c-C5H9)7Si7O10(OH)2]- by means
of intramolecular hydrogen bridges.30 The strong acidity
of (c-C5H9)7Si7O9(OH)3 and corresponding high stability
of the monoanion [(c-C5H9)7Si7O10(OH)2]- explain the
fact that silylation of the first hydroxyl group in (c-
C5H9)7Si7O9(OH)3 was found to be approximately 3
orders of magnitude faster than subsequent silylation
of the second and third silanols.33
As a second approach to estimate the electron-
withdrawing tendency of silanol ligands, the protonation
energy of L(L′)TiH2 was calculated. Strong electron-
withdrawing ligands L and L′ will hamper hydride
abstraction from L(L′)TiH2, as these ligands are less
able to stabilize the corresponding cation L(L′)TiH+. We
used this model reaction to investigate how alkyl
abstraction of a titanium bis(alkyl) complex by e.g.
B(C6F5)3 is influenced by the electron-withdrawing
capacity of the ancillary ligands.34
As can be seen from Figure 5, the energy required to
remove a hydride from complexes containing CF3O
ancillary ligands is substantially higher than for the
(28) Feher, F. J .; Budzichowski, T. A. J . Organomet. Chem. 1989,
379, 33.
(29) (a) Cookson, R. F. Chem. Rev. 1974, 74, 5. (b) Matthews, W. S.;
Bares, J . E.; Bartmess, J . E.; Bordwell, F. G.; Cornforth, F. J .; Drucker,
G. E.; Margolin, Z.; McCallum, R. J .; McCollum, G. J .; Vanier, N. R.
J . Am. Chem. Soc. 1975, 97, 7006. (c) Olmstread, W. N.; Bordwell, F.
G. J . Org. Chem. 1980, 45, 3299. (d) Bordwell, F. G.; McCallum, R. J .;
Olmstead, W. N. J . Org. Chem. 1984, 49, 1424. (e) Kaufmann, M. J .;
Streitwieser, A., J r. J . Am. Chem. Soc. 1987, 109, 6092. (f) Bordwell,
F. G. Acc. Chem. Res. 1988, 21, 456 and references therein.
(30) A detailed discussion about pKa values and calculated depro-
tonation energies of silsesquioxanes will be reported elsewhere: Harm-
sen, R. J .; Duchateau, R.; Krijnen, S.; van Santen, R. A. Manuscript
in preparation.
(31) For example, see: Kramer, G. J .; van Santen, R. A.; Emeis, C.
A.; Nowak, A. K. Nature 1993, 363, 529 and references therein.
(32) To reduce DFT calculation time, the silsesquioxane cyclopentyl
substituents were replaced by hydrogens.
(33) (a) Feher, F. J .; Newman, D. A. J . Am. Chem. Soc. 1990, 112,
1931. (b) Feher, F. J .; Phillips, S. H.; Ziller, J . W. Chem. Commun.
1997, 829.
(34) The cationic titanium species [(C5Me5)Ti(OC6F5)2]+ was found
to effectively compete with B(C6F5)3 for the alkyl group: Tremblay, T.
L.; Ewart, S. W.; Sarsfield, M. J .; Baird, M. C. Chem. Commun. 1997,
831.

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5455
polymerization of olefins and serves as a well-defined
model for corresponding silica-grafted surface com-
plexes. Due to their high affinity for the siloxy/silses-
quioxane oxo functionalities, aluminum alkyls are, when
used in excess, unsuitable cocatalysts/scavengers for
siloxy and silsesquioxane complexes and probably also
for the corresponding silica-grafted systems. On the
other hand, triphenylsiloxy- and silsesquioxane-stabi-
lized bis(alkyl) complexes (10a ,b, 11a ,b, 12a ) proved to
be active olefin polymerization catalysts when activated
with B(C6F5)3. This gives rise to promising perspectives
for their silica-supported analogues when activated with
boron-based cocatalysts.
The reactivities of silsesquioxane- and triphenylsilyl-
stabilized group 4 metal complexes proved to be quite
similar. Although the different cone angles of (c-C5H9)7-
Si8O13 and Ph3SiO suggest otherwise, the effective steric
congestion around the metal center of (c-C5H9)7Si8O13-
and Ph3SiO-stabilized complexes was found to be rea-
sonably comparable. Silsesquioxane (c-C5H9)7Si8O12(OH)
(2) is notably more Brønsted acidic than Ph3SiOH. DFT
calculations also show that silsesquioxane 2 is more
electron withdrawing than Ph3SiO. However, 2 is
substantially less electron-withdrawing than a CF3
substituent, as was found for the [Si8O12] framework
in (XC6H4)8Si8O12.28 Although silsesquioxane 2 is prob-
ably a slightly better homogeneous model for electro-
philic, isolated silica surface silanols compared to the
less Brønsted basic, more electron rich Ph3SiOH, the
differences in electronic properties between 2 and Ph3-
SiOH are quite small. Much larger differences are
observed when comparing Ph3SiOH with the tris-
(silanol) (c-C5H9)7Si7O9(OH)3, which is significantly
more Brønsted acidic than both Ph3SiOH and (c-C5H9)7-
Si8O12(OH). This is in line with the higher acidity of
silanol nests compared to isolated silanols in silicates
and zeolites.
F igu r e 5. Protonation energies of L2TiH2 and Cp(L)TiH2
for different ligands L.
corresponding compounds stabilized by H7Si8O13 or Ph3-
SiO ligands, which indicates that CF3O is more electron-
withdrawing than both siloxides, H7Si8O13 and Ph3SiO.
Protonolysis of [H7Si8O13]2TiH2 and Cp[H7Si8O13]TiH2
is also more difficult than for [Ph3SiO]2TiH2 and Cp[Ph3-
SiO]TiH2, respectively, indicating that the silsesquiox-
ane H7Si8O13 ligand is less electron donating than the
Ph3SiO ligand. However, the energy differences are
notably smaller in comparison with the CF3O-containing
systems. The outcome of these calculations corresponds
well with the lower deprotonation energy and pKa value
of H7Si8O12(OH) compared to Ph3SiOH.
Exp er im en ta l Section
Gen er a l Con sid er a tion s. All manipulations were per-
formed under an argon atmosphere using glovebox (Braun MB-
150 GI) and Schlenk techniques. Solvents were distilled from
Na (toluene), K (THF), Na/K alloy (ether, hexanes), or CaH2
(CH2Cl2) and stored under argon. NMR solvents were dried
over Na/K alloy (benzene-d6) or 4 Å molecular sieves (CDCl3).
NMR spectra were recorded on Varian Gemini 300 (1H, 300
MHz; 13C, 75.4 MHz) and Bruker AC400 (29Si, 79.5 MHz)
spectrometers. Chemical shifts are reported in ppm and
referenced to residual solvent resonances (1H, 13C NMR) or
external TMS (29Si NMR) or CF3CO2H (19F NMR). The IR
spectrum of 2 was recorded on a Hitachi 270-30 spectropho-
tometer. Elemental analyses were carried out at the Analytical
Departments of the University of Groningen (The Nether-
lands). To reduce the often-observed silicon carbide formation,
V2O5 was added to improve the combustion. Titanium and
zirconium analyses were performed using complexometry. The
mass spectrum (FAB, matrix m-nitrobenzyl alcohol) of 2 was
measured on a Finnigan MAT-90 high-resolution doublefo-
cusing mass spectrometer at the Analytical Department of the
University of Twente (Twente, The Netherlands). The molec-
ular weights and molecular weight distributions were mea-
sured at 135 °C by gel permeation chromatography (GPC)
(GPC210, Polymer Labs) at the University of Groningen, using
1,2,4-trichlorobenzene as solvent. (c-C5H9)7Si7O9(OH)3,26 (c-
In an attempt to confirm these calculated energy
differences, dibenzyl complexes Cp[(c-C5H9)7Si8O13]Ti-
(CH2Ph)2 (10a ) and Cp′′[Ph3SiO]Ti(CH2Ph)2 (11a ) were
treated with B(C6F5)3. The 19F NMR spectra showed
distinct resonances of both B(C6F5)3 and [PhCH2B-
(C6F5)3]-.25 For an equimolar mixture of 10a + B(C6F5)3,
the B(C6F5)3:[PhCH2B(C6F5)3]- ratio was found to be 1:1.
With a B(C6F5)3:[PhCH2B(C6F5)3]- ratio of 0.2:1 for an
equimolar mixture of 11a + B(C6F5)3, benzyl abstraction
is more facile for the triphenylsilyl system than for the
corresponding silsesquioxane system. Clearly, both
Lewis acids [Cp[(c-C5H9)7Si8O13]TiCH2Ph]+ and [Cp′′-
[Ph3SiO]TiCH2Ph]+ effectively compete with B(C6F5)3
for possession of the benzyl group.34 Although steric
factors cannot be excluded, these results are in agree-
ment with the calculated energy difference for the
hydride abstraction of Cp[H7Si8O13]TiH2 and Cp[Ph3-
SiO]TiH2 and indicate that the triphenylsilyl ligand is
less electron withdrawing than the silsesquioxane.
Con clu d in g Rem a r k s
A new class of group 4 metal compounds has been
developed, which provides catalyst precursors for the

5456 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
C5H9)7Si7O9(OSiMe3)(OH)2,35 Ph3SiOTl,13b and Cp′′MX3 (M )
Ti, Zr; X ) Cl, CH2Ph, Me)36 were prepared by following
literature procedures. Ph3SiOH (Aldrich) was used as pur-
chased.
(CDCl3, δ): 2.38 (s, 4H, CH2-TMEDA), 2.32 (s, 12H, CH3-
TMEDA), 1.75 (m, 14H, CH2-C5H9), 1.52 (m, 42H, CH2-C5H9),
0.96 (m, 7H, CH-C5H9). 13C NMR (CDCl3, δ): 56.86 (t, CH2-
TMEDA, 1J C-H ) 133 Hz), 45.86 (q, CH3-TMEDA, 1J C-H ) 137
Hz). 13C{1H} NMR (CDCl3, δ): 27.74, 27.34, 27.28, 27,18, 27.07
(CH2-C5H9), 22.71, 22.41 (CH-C5H9). 29Si NMR (toluene, δ):
-66.75, -68.35, -103.67 (4:3:1). Anal. Calcd for C41H79LiN2O13-
Si8: C, 47.36; H, 7.66. Found: C, 47.00; H, 7.74.
Cp ′′[(c-C5H 9)7Si8O13]TiCl2 (6a ). BuLi (3.4 mL, 2.5 M in
hexanes, 8.5 mmol) was added to a hexane solution of 2 (7.7
g, 8.4 mmol). Subsequently, Cp′′TiCl3 (3.07 g, 8.44 mmol) was
added to the in situ prepared solution of [(c-C5H9)7Si8O13]Li
and the mixture was stirred overnight at room temperature.
Filtration, concentration, and cooling to -30 °C yielded a
yellow powder consistent with a 9:1 mixture of 6a and 6b, as
determined by 1H NMR analysis. Repeated recrystallization
from hexane yielded analytically pure 6a (5.3 g, 4.3 mmol,
51%). 1H NMR (CDCl3, δ): 7.10 (s, 1H, C5H3(SiMe3)2), 7.02 (s,
2H, C5H3(SiMe3)2), 1.63 (m, 14H, CH2-C5H9), 1.37 (m, 42H,
CH2-C5H9), 0.87 (m, 7H, CH-C5H9), 0.21 (s, 18H, Si(CH3)3). 13C-
{1H} NMR (CDCl3, δ): 141.42 (C5H3(SiMe3)2), 132.97 (C5H3-
(SiMe3)2), 129.81 (C5H3(SiMe3)2), 27.31, 27.24, 27.00 (CH2-
C5H9), 22.12 (CH-C5H9), -0.59 (Si(CH3)3). 29Si NMR (THF, δ):
-5.63 (SiMe3), -65.92, -66.66, -111.76 (2:3:4:1). Anal. Calcd
for C46H84Cl2O13Si10Ti: C, 44.38; H, 6.80; Ti, 3.85. Found: C,
44.20; H, 6.87; Ti, 3.85.
(c-C5H9)7Si8O12Cl (1). To a suspension of (c-C5H9)7Si7(OH)3
(6.96 g, 7.95 mmol) in ether (300 mL) and NEt3 (3.3 mL, 23.8
mmol) was added SiCl4 (0.9 mL, 7.86 mmol) at room temper-
ature. The mixture was refluxed overnight and then filtered
to remove Et3N(H)Cl. The salt was washed with ether (2 × 30
mL), and evaporation of the filtrate yielded crude 1 as a white
powder (6.85 g, 7.31 mmol, 92%). Cooling of a hot saturated
toluene/acetonitrile solution afforded 1 as microcrystals. 1H
NMR (CDCl3, δ): 1.72 (m, 14H, CH2-C5H9), 1.55 (m, 42H, CH2-
C5H9), 1.05 (m, 7H, CH-C5H9). 13C{1H} NMR (CDCl3, δ): 27.23,
27.18, 26.99, 26.90 (CH2-C5H9), 22.12, 22.05, 21.85 (CH-C5H9,
1:3:3). 29Si NMR (CH2Cl2, δ): -66.28, -66.85, -66.98, -90.20
(3:3:1:1). Anal. Calcd for C35H63ClO12Si8: C, 44.91; H, 6.78.
Found: C, 44.85; H, 6.95.
(c-C5H9)7Si8O12(OH) (2). A suspension of (c-C5H9)7Si8O12
-
Cl (1; 2.11 g, 2.25 mmol) in THF/H2O (2:1, 50 mL) was refluxed
for 40 h. Evaporation of the volatiles afforded crude 2 as a
white solid. To remove the lattice THF, the product was
dissolved in toluene (5 mL) and subsequently dried under
vacuum. Thorough drying and recrystallization from a hot
toluene/acetonitrile mixture gave 2 as a white microcrystalline
material (1.65 g, 1.80 mmol, 80%). IR (Nujol, cm-1): 3650 (br,
OH). 1H NMR (CDCl3, δ): 2.93 (s broad, 1H, OH), 1.79 (m,
14H, CH2-C5H9), 1.55 (m, 42H, CH2-C5H9), 1.05 (m, 7H, CH-
C5H9). 13C{1H} NMR (CDCl3, δ): 27.25, 26.98, 26.92 (CH2-
C5H9), 22.17, 22.12, 22.05 (CH-C5H9, 1:3:3). 29Si NMR (CH2Cl2,
δ): -63,48, -64.18, -97.75 (4:3:1). A satisfactory combustion
analysis simply could not be obtained. One possible explana-
tion is the formation of refractory silicon carbides. Anal. Calcd
for C35H64O13Si8: C, 45.81; H, 7.03. Found: C, 44.87; H, 7.09.
Mass (FAB-nitrobenzyl alcohol): 917 (M+), 847 (M - H -
C5H9), 777 (M - H - 2 × C5H9), 709 (M - H - 3 × C5H9).
[(c-C5H9)7Si8O13]Tl (3). To a solution of 2 (3.5 g, 3.8 mmol)
in toluene (25 mL) was added TlOEt (0.27 mL, 3.8 mmol) at
room temperature. The mixture was left overnight, after which
the volatiles were removed under vacuum, leaving crude 3 as
a white foam. Recrystallization at -30 °C from hexane yielded
Cp ′′[(c-C5H9)7Si8O13]2TiCl (6b). A solution of 4 (4.2 g, 4.6
mmol) in hexane (50 mL) was treated with Cp′′TiCl3 (0.79 g,
2.17 mmol), and the resulting suspension was stirred for 2 days
at room temperature. The salt was removed by centrifuge, and
the solvent was evaporated, yielding 6b (3.6 g, 1.7 mmol, 78%)
as a bright yellow foam. 1H NMR (CDCl3, δ): 7.23 (s, 1H, C5H3-
(SiMe3)2), 7.04 (s, 2H, C5H3(SiMe3)2), 1.83 (m, 28H, CH2-C5H9),
1.55 (m, 84H, CH2-C5H9), 1.04 (m, 14H, CH-C5H9), 0.38 (s, 18H,
Si(CH3)3). 13C{1H} NMR (CDCl3, δ): 136.44 (C5H3(SiMe3)2),
131.51 (C5H3(SiMe3)2), 127.85(C5H3(SiMe3)2), 27.31, 27.27,
27.09, 27.04 (CH2-C5H9), 22.26, 22.21, 22.11 (1:3:3, CH-C5H9),
-0.49 (Si(CH3)3). 29Si NMR (toluene, δ): -6.54 (SiMe3), -65.99,
-66.54, -111.25 (1:3:4:1). Anal. Calcd for C81H147ClO26Si18Ti:
C, 45.76; H, 6.97; Ti, 2.25. Found: C, 45.36; H, 7.07; Ti, 2.19.
Cp ′′[(c-C5H9)7Si8O13]Zr Cl2 (7a ). To a toluene (30 mL)
solution of Cp′′ZrCl3 (0.70 g, 1.72 mmol) was added 3 (1.90 g,
1.70 mmol) at room temperature. The white suspension that
formed was stirred overnight. After the volatiles were removed
under vacuum, the crude product was extracted (30 mL) with
1
3 as thin colorless needles (2.2 g, 1.96 mmol, 52%). H NMR
(CDCl3, δ): 1.78 (m, 14H, CH2-C5H9), 1.54 (m, 42H, CH2-C5H9),
1.00 (m, 7H, CH-C5H9). 13C{1H} NMR (CDCl3, δ): 27.36, 27.32,
27.21, 27.07 (CH2-C5H9), 22.33, 22.30, 22.27 (CH-C5H9, 3:1:3).
29Si NMR (CH2Cl2, δ): -67.06, -67.31, -101.23 (4:3:1). Anal.
Calcd for C35H63O13Si8Tl: C, 37.50; H, 5.67. Found: C, 37.64;
H, 5.65.
{[(c-C5H9)7Si8O13]Li}n (4). At room temperature, n-BuLi
(0.9 mL, 2.5 M in hexanes, 2.25 mmol) was added to a solution
of 2 (2.08 g, 2.27 mmol) in hexanes (50 mL). After the mixture
was stirred for 0.5 h, the solvent was evaporated, leaving 4 as
a white foam (1.87 g, 2.0 mmol, 89%). Due to the extreme
solubility of 4 in common organic solvents, purification by
means of recrystallization proved to be impossible. 1H NMR
(CDCl3, δ): 1.78 (m, 14H, CH2-C5H9), 1.50 (m, 42H, CH2-C5H9),
0.94 (m, 7H, CH-C5H9). 13C{1H} NMR (CDCl3, δ): 27.37, 27.29,
27.01 (CH2-C5H9), 22.26, 22.21, 22.09 (CH-C5H9).
1
hexane. Evaporation of the solvent yielded a white foam. H
NMR spectroscopy revealed that a 2:1 mixture of Cp′′[(c-C5H9)7-
Si8O13]ZrCl2 (7a) and Cp′′[(c-C5H9)7Si8O13]2ZrCl (7b) was formed.
Due to their extreme solubility, separation of these products
proved to be impossible. Cp′′[(c-C5H9)7Si8O13]ZrCl2 (7a ): 1H
NMR (CDCl3, δ) 7.14 (t, 1H, C5H3(SiMe3)2, 4J H-H ) 2 Hz), 7.03
(d, 2H, C5H3(SiMe3)2, 4J H-H ) 2 Hz), 1.80 (m, 14H, CH2-C5H9),
1.56 (m, 42H, CH2-C5H9), 1.18 (m, 7H, CH-C5H9), 0.39 (s, 18H,
C5H3(Si(CH3)3)2); 13C{1H} NMR (CDCl3, δ) 135.24 (C5H3-
(SiMe3)2), 130.09 (C5H3(SiMe3)2), 126.17 (C5H3(SiMe3)2), 27.32
(CH2-C5H9), 27.26 (CH2-C5H9), 27.01 (CH2-C5H9), 22.13 (CH-
C5H9), -0.49 (C5H3(Si(CH3)3)2); 29Si NMR (THF, δ) -7.36,
-66.68, -66.84, -111.40 (2:3:4:1).
Cp ′′[(c-C5H9)7Si8O13]2Zr Cl (7b). Using the same proce-
dures as for 7a , starting with Cp′′ZrCl3 (0.77 g, 1.89 mmol)
and 3 (4.24 g, 3.78 mmol), 7b was obtained as an off-white
foam (2.90 g, 1.34 mmol, 71%). 1H NMR (CDCl3, δ): 7.13 (t,
[(c-C5H9)7Si8O13]Li‚TMEDA (5). A room-temperature solu-
tion of 2 (4.38 g, 4.77 mmol) in ether (30 mL) was treated with
BuLi (1.9 mL, 4.75 mmol, 2.5 M in hexanes). A few minutes
after TMEDA (1.5 mL, 10 mmol) was added, needle-shaped
crystals of 5 started to form. Subsequent cooling to 4 °C yielded
5 as colorless crystals (1.72 g, 1.65 mmol, 35%). 1H NMR
4
4
C5H3(SiMe3)2, J H-H ) 2 Hz), 6.95 (d, C5H3(SiMe3)2, J H-H ) 2
Hz), 1.80 (m, 14H, CH2-C5H9), 1.56 (m, 42H, CH2-C5H9), 0.95
(m, 7H, CH-C5H9), 0.36 (s, 18H, C5H3(Si(CH3)3)2). 13C{1H} NMR
(CDCl3, δ): 132.15 (C5H3(SiMe3)2), 124.80 (C5H3(SiMe3)2),
27.32, 27.27, 27.04 (CH2-C5H9), 22.25, 22.19, 22.12 (1:3:3, CH-
C5H9), -0.44 (C5H3(Si(CH3)3)2). 29Si NMR (toluene, δ): -7.85,
(35) Abbenhuis, H. C. L.; Burrows, A. D.; Kooijman, H.; Lutz, M.;
Palmer, M. T.; van Santen, R. A.; Spek, A. L. Chem. Commun. 1998,
2627.
(36) (a) Winter, C. H.; Zhou, X.-X.; Dobbs, D. A.; Heeg, M. J .
Oganometallics 1991, 10, 210. (b) Lancaster, S. J .; Robinson, O. B.;
Bochmann, M.; Coles, S. J .; Hursthouse, M. B. Organometallics 1995,
14, 2456.
-65.88, -66.42, -109.74 (1:3:4:1). Anal. Calcd for C81H147
-
ClO26Si18Zr: C, 44.85; H, 6.83; Zr, 4.21. Found: C, 44.47; H,
6.93; Zr, 4.09.

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5457
1
Cp ′′[P h 3SiO]TiCl2 (8a ). At room temperature, a solution
of Cp′′TiCl3 (4.9 g, 13.6 mmol) in toluene (35 mL) was added
to a toluene (20 mL) solution of Ph3SiOTl (6.5 g, 13.6 mmol).
Immediately a white salt precipitated. After the mixture was
stirred for 24 h, the toluene was evaporated and the residue
was dried thoroughly. The crude product was dissolved in
hexane (15 mL), and after filtration the clear orange to yellow
filtrate was concentrated until crystallization started. Re-
peated crystallization (2×) at -30 °C yielded 8a (6.0 g, 9.9
mmol, 73%) as an orange microcrystalline solid. 1H NMR
(CDCl3, δ): 7.80 (d, 6H, C6H5), 7.48 (m, 9H, C6H5), 7.23 (s, 1H,
C5H3(SiMe3)2), 6.84 (s, 2H, C5H3(SiMe3)2), 0.24 (s, 18H, C5H3-
(Si(CH3)3)2). 13C NMR (CDCl3, δ): 141.8 (s, ipso-C5H3(SiMe3)2),
136.3 (d, C6H5, 1J C-H ) 159 Hz), 135.0 (s, ipso-C6H5), 134.1 (d,
126.16 (d, C6H5, J C-H ) 157 Hz), 123.03 (d, C5H3(SiMe3)2,
1J C-H ) 158 Hz), 86.82 (t, CH2Ph, J C-H ) 127 Hz), 27.90 (t,
1
1
1
CH2-C5H9, J C-H ) 130 Hz), 27.50 (t, CH2-C5H9, J C-H ) 130
1
Hz), 22.78 (d, CH-C5H9, J C-H ) 119 Hz), 0.14 (q, Si(CH3)3,
1J C-H ) 120 Hz). 29Si NMR (toluene, δ): -7.98, -66.22, -66.70,
-111.53 (2:3:4:1). Anal. Calcd for C60H98O13Si10Ti: C, 53.14;
H, 7.28; Ti, 3.53. Found: C, 52.81; H, 7.19; Ti, 3.47.
Cp ′′[(c-C5H9)7Si8O13]TiMe2 (10b). Using the same proce-
dure as for 10a , starting with Cp′′TiMe3 (1.04 g, 3.44 mmol)
and 2 (3.09 g, 3.37 mmol), afforded 10b (2.10 g, 1.74 mmol,
51%) as a yellow crystalline material. 1H NMR (benzene-d6,
δ): 7.17 (s, 1H, C5H3(SiMe3)2), 6.63 (s, 2H, C5H3(SiMe3)2), 1.68
(m, 56 H, CH2-C5H9), 1.30 (m, 7H, CH-C5H9), 1.01 (s, 6H, CH3),
0.24 (18H, Si(CH3)3). 13C NMR (benzene-d6, δ): 131.16 (s, C5H3-
(SiMe3)2), 124.71 (d, C5H3(SiMe3)2, 1J C-H ) 163 Hz), 123.22 (d,
C5H3(SiMe3)2, 1J C-H ) 164 Hz), 56.56 (q, CH3, 1J C-H ) 123 Hz),
27.94, 27.85, 27.50 (t, CH2-C5H9, 1J C-H ) 130 Hz), 22.88, 22.77
(d, CH-C5H9, 1J C-H ) 122 Hz), -0.23 (q, Si(CH3)3, 1J C-H ) 120
Hz). 29Si NMR (toluene, δ): -8.35, -66.26, -66.65, -111.06
(2:3:4:1). Anal. Calcd for C48H90O13Si10Ti: C, 47.88; H, 7.53;
Ti, 3.98. Found: C, 47.63; H, 7.49; Ti, 3.67.
Cp ′′[P h 3SiO]Ti(CH2P h )2 (11a ). At -80 °C, a toluene (20
mL) solution of Ph3SiOH (1.2 g, 4.3 mmol) was added to a
stirred solution of Cp′′Ti(CH2Ph)3 (2.2 g, 4.1 mmol) in toluene
(10 mL). After the addition, the mixture was slowly warmed
to room temperature and the toluene was removed under
vacuum, leaving a red oily residue. After the crude product
was stripped with hexane (2 × 10 mL), it was dissolved in
hexane (50 mL) and filtered hot. Slow cooling to 4 °C for
crystallization afforded 11a as large block-shaped crystals (2.0
g, 2.8 mmol, 68%). Concentration and cooling to -30 °C of the
mother liquor yielded a second crop of 11a as small red crystals
1
1
C5H3(SiMe3)2, J C-H ) 172 Hz), 131.0 (d, C6H5, J C-H ) 160
1
Hz), 129.6 (d, C5H3(SiMe3)2, J C-H ) 174 Hz), 128.6 (d, C6H5,
1J C-H ) 160 Hz), 0.2 (q, C5H3(Si(CH3)3)2, 1J C-H ) 119 Hz). 29Si
NMR (toluene, δ): -6.3 (C5H3(SiMe3)2), -9.2 (Ph3SiO) (2:1).
Anal. Calcd for C29H36Cl2OSi3Ti: C,57.70; H, 6.01. Found: C,
57.49; H, 6.04.
Cp ′′[P h 3SiO]2TiCl (8b). The same procedures as for 8a
were used. Starting from Cp′′TiCl3 (1.0 g, 2.7 mmol) and Ph3-
SiOTl (2.5 g, 5.2 mmol), gave 8b (1.6 g, 1.9 mmol, 70%) as
large rod-shaped yellow crystals. 1H NMR (CDCl3, δ): 7.68 (d,
12H, C6H5), 7.42 (m, 6H, C6H5), 7.31 (m, 12H, C6H5), 6.70 (s,
1H, C5H3(SiMe3)2), 6.58 (s, 2H, C5H3(SiMe3)2), 0.03 (s, 18H,
C5H3(Si(CH3)3)2). 13C NMR (CDCl3, δ): 137.0 (s, ipso-C5H3-
(SiMe3)2), 136.4 (d, C6H5, 1J C-H ) 159 Hz), 136.1 (s, ipso-C6H5),
130.8 (d, C6H5, 1J C-H ) 151 Hz), 130.9 (d, C5H3(SiMe3)2, 1J C-H
1
) 170 Hz), 128.3 (d, C6H5, J C-H ) 159 Hz), 0.3 (q, C5H3(Si-
1
(CH3)3)2, J C-H ) 121 Hz). 29Si NMR (toluene, δ): -7.3 (C5H3-
(SiMe3)2), -12.6 (Ph3SiO) (1:1). Anal. Calcd for C47H51ClO2-
Si4Ti(C6H14)0.5: C, 67.73; H, 6.59. Found: C, 67.78; H, 6.63.
1
(0.2 g, 0.3 mmol, 7%). H NMR (benzene-d6, δ): 7.59 (m, 6H,
3
C6H5), 7.07 (m, 9H, C6H5), 6.92 (t, 4H, C6H5CH2, J H-H ) 7.4
Cp ′′[P h 3SiO]2Zr Cl (9). To a toluene (30 mL) solution of Ph3-
SiOTl (2.0 g, 4.2 mmol) was added solid Cp′′ZrCl3 (0.8 g, 2.0
mmol) at room temperature. After the mixture was stirred for
24 h, the toluene was removed under vacuum and the residue
was dried thoroughly. The crude product was dissolved in
hexane (15 mL), filtered, and concentrated until crystallization
started. Slow crystallization at room temperature afforded 9
(0.8 g, 0.9 mmol 45%) as long white needles, suitable for an
X-ray crystal structure. 1H NMR (CDCl3, δ): 7.57 (d, 12H,
C6H5, 3J H-H ) 7.3 Hz), 7.35 (t, 6H, C6H5, 3J H-H ) 7.4 Hz), 7.25
(t, 12H, C6H5, 3J H-H ) 7.5 Hz), 6.64 (t, 1H, C5H3(SiMe3)2, 4J H-H
3
Hz), 6.73 (t, 6H, C6H5CH2, J H-H ) 7.4 Hz), 6.69 (t, 1H, C5H3-
4
4
(SiMe3)2, J H-H ) 1.8 Hz), 6.15 (d, 2H, C5H3(SiMe3)2, J H-H
)
1.8 Hz), 3.16 (d, 2H, CH2Ph, 2J H-H ) 9.9 Hz), 2.22 (d, 2H, CH2-
Ph, 2J ) 9.9 Hz), 0.40 (s, 18H, C5H3(Si(CH3)3)2). 13C NMR
(benzene-d6, δ): 149.9 (s, ipso-C6H5), 136.4 (s, ipso-C6H5), 136.4
1
(d, C6H5, J C-H ) 156 Hz), 133.7 (s, ipso-C5H3(SiMe3)2), 131.0
(d, C5H3(SiMe3)2, 1J C-H ) 170 Hz), 130.3 (d, C6H5, 1J C-H ) 166
1
1
Hz), 128.6 (d, C6H5, J C-H ) 158 Hz), 128.2 (d, C6H5, J C-H
157 Hz), 127.3 (d, C6H5, J C-H ) 145 Hz), 126.0 (d, C5H3-
(SiMe3)2, J C-H ) 171 Hz), 122.9 (d, C6H5, J C-H ) 163 Hz),
86.1 (t, CH2Ph, 1J C-H ) 125 Hz), 0.3 (q, C5H3(Si(CH3)3)2, 1J C-H
) 120 Hz). 29Si NMR (toluene, δ): -8.1 (C5H3(SiMe3)2), -13.8
(Ph3SiO) (2:1). Anal. Calcd for C43H50OSi3Ti: C, 72.23; H, 7.05.
Found: C, 72.26; H, 6.83.
)
1
1
1
4
) 1.9 Hz), 6.56 (d, 2H, C5H3(SiMe3)2, J H-H ) 1.9 Hz), 0.00 (s,
18H, C5H3(Si(CH3)3)2). 13C NMR (CDCl3, δ): 135.7 (s, ipso-
1
C6H5), 135.3 (d, C6H5, J C-H ) 158 Hz), 132.4 (s, ipso-C5H3-
1
(SiMe3)2), 129.8 (d, C6H5, J C-H ) 159 Hz), 128.1 (d, C5H3(Si-
1
1
(CH3)3)2, J C-H ) 168 Hz), 127.9 (d, C6H5, J C-H ) 159 Hz),
Cp ′′[P h 3SiO]Ti(CH3)2 (11b). The same procedure as for
11a , staring with Cp′′TiMe3 (1.1 g, 3.6 mmol) and Ph3SiOH
(1.0 g, 3.6 mmol), afforded 11b (1.7 g, 2.9 mmol, 81%) as a
yellow solid after removal of the hexane and thorough drying.
1H NMR (benzene-d6, δ): 8.15 (m, 6H, C6H5), 7.45 (m, 9H,
C6H5), 6.78 (s, 1H, C5H3(SiMe3)2), 6.72 (s, 2H, C5H3(SiMe3)2),
1
125.1 (d, C5H3(SiMe3)2, J C-H ) 170 Hz), -0.4 (q, C5H3(Si-
1
(CH3)3)2, J C-H ) 119 Hz). 29Si NMR (toluene, δ): -8.5 (C5H3-
(SiMe3)2), -16.5 (Ph3SiO) (1:1). The crystalline product gradu-
ally lost hexane from the lattice, preventing a reliable elemental
analysis.
1.10 (s, 6H, CH3), 0.24 (s, 18H, C5H3(Si(CH3)3)2). 13C NMR
(benzene-d6, δ): 136.7 (s, ipso-C6H5), 135.8 (d, C6H5, J C-H
158 Hz), 131.1 (s, ipso-C5H3(SiMe3)2), 130.0 (d, C6H5, J C-H
Cp ′′[(c-C5H9)7Si8O13]Ti(CH2P h )2 (10a ). To a hexane (25
mL) solution of Cp′′Ti(CH2Ph)3 (1.21 g, 2.28 mmol) was added
solid 2 (2.10 g, 2.29 mmol) at 0 °C, and the mixture was
warmed to room temperature. After the mixture was stirred
for 2 h, traces of insoluble impurities were filtered off and the
dark red solution was concentrated to approximately 5 mL.
Slow crystallization at -30 °C afforded 10a as a red micro-
crystalline material (2.00 g, 1.48 mmol, 64%). 1H NMR
(benzene-d6, δ): 7.36 (m, 4H, m-C6H5), 7.28 (d, 4H, o-C6H5,
1
)
)
1
1
160 Hz), 128.0 (d, C6H5, J C-H ) 157 Hz), 124.3 (d, C5H3-
1
1
(SiMe3)2, J ) 170 Hz), 122.9 (d, C5H3(SiMe3)2, J ) 170 Hz),
55.8 (q, CH3, J C-H ) 128 Hz), 0.3 (q, C5H3(Si(CH3)3)2, 1J )
1
116 Hz). 29Si NMR (toluene, δ): -8.5 (C5H3(SiMe3)2), -16.4
(Ph3SiO) (2:1). Anal. Calcd for C31H42OSi3Ti: C, 66.15; H, 7.52.
Found: C, 65.54; H, 7.63.
3
3J H-H ) 7 Hz), 7.00 (t, 2H, p-C6H5, J H-H ) 7 Hz), 6.89 (s, 1H,
Cp ′′[P h 3SiO]Zr (CH2P h )2 (12a ). At -80 °C, solid Ph3SiOH
(1.2 g, 4.3 mmol) was added to a stirred solution of Cp′′Zr-
(CH2Ph)3 (2.44 g, 4.25 mmol) in toluene (50 mL). The mixture
was slowly warmed to room temperature, after which the
solvent was removed under vacuum. To remove traces of
toluene, the sticky solid was dissolved in hexane (10 mL) and
subsequently dried under vacuum. This procedure was re-
peated (2×) until a nonsticky solid was obtained. Crystalliza-
C5H3(SiMe3)2), 6.25 (s, 2H, C5H3(SiMe3)2), 3.48 (d, 2H, CH2-
2
2
Ph, J H-H ) 10 Hz), 2.16 (d, 2H, CH2Ph, J H-H ) 10 Hz), 1.97
(m, 14H, CH2-C5H9), 1.75 (m, 28H, CH2-C5H9), 1.58 (m, 14H,
CH2-C5H9), 1.26 (m, 7H, CH-C5H9), 0.37 (s, 18H, Si(CH3)3). 13
C
NMR (benzene-d6, δ): 149.04 (s, ipso-C6H5), 133.62 (s, C5H3-
(SiMe3)2), 129.55 (d, C5H3(SiMe3)2, 1J C-H ) 159 Hz), 128.53 (d,
1
1
C6H5, J C-H ) 158 Hz), 127.75 (d, C6H5, J C-H ) 158 Hz),

5458 Organometallics, Vol. 18, No. 26, 1999
Duchateau et al.
tion from hexane (20 mL) at -30 °C yielded 12a as yellow
crystals (1.8 g, 2.37 mmol, 56%). 1H NMR (benzene-d6, δ): 7.77
(m, 6H, C6H5), 7.30 (m, 9H, C6H5), 7.09 (m, 4H, C6H5CH2), 6.97
(m, 6H, C6H5CH2), 6.90 (s, 1H, C5H3(SiMe3)2), 6.45 (s, 2H, C5H3-
) 168 Hz), 27.78 (t, CH2-C5H9, 1J C-H ) 131 Hz), 27.34 (t, CH2-
1
1
C5H9, J C-H ) 130 Hz), 27.11 (t, CH2-C5H9, J C-H ) 129 Hz),
22.78 (d, CH-C5H9, 1J C-H ) 119 Hz), 22.41 (d, CH-C5H9, 1J C-H
) 117 Hz), 22.34 (d, CH-C5H9, 1J C-H ) 117 Hz), -0.34 (q, C5H3-
2
1
(SiMe3)2), 2.50 (d, 2H, CH2Ph, J H-H ) 11.0 Hz), 2.10 (d, 2H,
(Si(CH3)3)2, J C-H ) 119 Hz). 29Si NMR (toluene, δ): -9.17
2
CH2Ph, J H-H ) 11.0 Hz), 0.20 (s, 18H, C5H3(Si(CH3)3)2). 13C
(SiMe3), -64.32, -66.53, -68.29 (1:3:1:3). Anal. Calcd for
C46H84O12Si9Zr: C, 47.10; H, 7.22; Zr, 7.78. Found: C, 46.85;
H, 7.25; Zr, 7.74.
NMR (benzene-d6, δ): 145.4 (s, ipso-C6H5), 136.8 (s, ipso-C6H5),
1
1
136.2 (d, C6H5, J C-H ) 158 Hz), 130.4 (d, C6H5, J C-H ) 159
1
Hz), 130.2 (s, ipso-C5H3(SiMe3)2), 129.7 (d, C6H5, J C-H ) 158
X-r a y Diffr a ction Stu d ies of 9 a n d 11a . Suitable crystals
were measured at 130 K with graphite-monochromated Mo
KR radiation on an Enraf-Nonius CAD-4F diffractometer
equipped with a low-temperature unit.37 Precise lattice pa-
rameters and their standard deviations and orientation matrix
were derived from the angular settings of 22 reflections (9,
18.24° < θ < 20.08°; 11a , 16.52° < θ < 20.34°) in four
alternative settings.38 The unit cell was identified as rhom-
bohedral (space group R3h (H, obv.)) for 9 and triclinic (space
group P1h) for 11a .39 Intensity data were corrected for Lorentz
and polarization effects and scale variation and reduced to
1
1
Hz), 128.3 (d, C6H5, J C-H ) 157 Hz), 128.2 (d, C6H5, J C-H
)
1
145 Hz), 127.4 (d, C5H3(SiMe3)2, J C-H ) 171 Hz), 123.4 (d,
1
1
C6H5, J C-H ) 156 Hz), 123.2 (d, C5H3(SiMe3)2, J C-H ) 177
1
Hz), 63.1 (t, CH2Ph, J C-H ) 124 Hz), 0.5 (q, C5H3(Si(CH3)3)2,
1J C-H ) 119 Hz). 29Si NMR (toluene, δ): -9.0 (C5H3(SiMe3)2),
-17.4 (Ph3SiO) (2:1). Anal. Calcd. for C43H50OSi3Zr: C, 68.10;
H, 6.65. Found: C, 67.81; H, 6.73.
Cp ′′[P h 3SiO]2Zr (CH2P h ) (12b). Using the same procedure
as for 12a and starting with Ph3SiOH (0.9 g, 3.3 mmol) and
Cp′′Zr(CH2Ph)3 (0.9 g, 1.6 mmol) yielded 12b (0.8 g, 0.8 mmol,
52%) as an off-white crystalline solid. 1H NMR (benzene-d6,
2
40
Fo
.
The structures were solved by Patterson methods, and
3
δ): 7.78 (d, 12H, C6H5, J H-H ) 6.8 Hz), 7.21 (m, 18H, C6H5,
extension of the model was accomplished by direct methods
applied to difference structure factors using the program
DIRDIF.41 The positional and anisotropic thermal displace-
ment parameters for the non-hydrogen atoms were refined on
F2 with full-matrix least-squares procedures minimizing the
3J H-H ) 6.8 Hz), 7.04-6.83 (m, 5H, C6H5), 6.79 (t, 1H, C5H3-
4
4
(SiMe3)2, J H-H ) 1.9 Hz), 6.74 (d, 2H, C5H3(SiMe3)2, J H-H
)
1.9 Hz), 2.84 (s, 2H, CH2Ph), 0.15 (s, 18H, C5H3(Si(CH3)3)2).
13C NMR (benzene-d6, δ): 148.3 (s, ipso-C6H5), 137.0 (s, ipso-
1
1
function Q ) ∑h[w(|Fo2 - kFc |)2], where w ) 1/[σ2(Fo2) + (aP)2
2
C6H5), 136.2 (d, C6H5, J C-H ) 158 Hz), 130.3 (d, C6H5, J C-H
) 159 Hz), 130.0 (s, ipso-C5H3(SiMe3)2), 129.5 (d, CH2Ph, 1J -H
2
+ bP] and P ) [max(Fo2,0) + 2Fc ]/3. Fo and Fc are the observed
1
) 158 Hz), 128.4 (d, C6H5, J C-H ) 163 Hz), 128.2 (d, C6H5,
and calculated structure factor amplitudes, respectively; a and
b were refined. Reflections were stated observed if they
satisfied the F2 > 0 criterion of observability. For 9, refinement
was complicated by distortion of the cyclopentadienyl tri-
methylsilyl groups. For both Si1-C6‚‚‚C8 and Si2-C9‚‚‚C11,
two positions were found with major site occupancy factors of
0.53(7) and 0.55(3), respectively. Probably due to the presence
of the highly distorted trimethylsilyl groups and to a large
extent disappeared solvent (hexane) molecules from the holes
in the lattice, peaks not exceeding 1.5 e/Å3 in the last residual
Fourier maps were found, but their geometry could not be
rationalized. Neutral atom scattering factors and anomalous
dispersion corrections were taken from ref 42. All calculations
were carried out on a HP9000/735 computer at the University
of Groningen with the program packages SHELXL,43 PLA-
TON,44 and ORTEP.45
1J C-H ) 145 Hz), 126.9 (d, C5H3(SiMe3)2, 1J C-H ) 168 Hz), 124.7
(d, C5H3(SiMe3)2, 1J C-H ) 170 Hz), 122.4 (d, C6H5, 1J C-H ) 161
1
Hz), 59.3 (t, CH2Ph, J C-H ) 119 Hz), 0.4 (q, C5H3(Si(CH3)3)2,
1J C-H ) 119 Hz). 29Si NMR (toluene, δ): -9.1 (C5H3(SiMe3)2),
-17.6 (Ph3SiO) (1:1). Anal. Calcd for C54H58O2Si2Zr: C, 68.81;
H, 6.20. Found: C, 68.08; H, 6.26.
[(c-C5H9)7Si7O12]TiCp ′′ (13). To a deep yellow solution of
Cp′′TiCl3 (1.64 g, 4.51 mmol) in CH2Cl2 (40 mL) was added
(c-C5H9)7Si7O9(OH)3 (3.98 g, 4.55 mmol). The color of the
suspension changed to light yellow immediately following the
addition of pyridine (1.5 mL, 18.6 mmol). After the mixture
was warmed to reflux for 5 min, the solvent was evaporated
and the product was extracted with hexanes (30 mL). The
filtrate was evaporated to dryness, and the product was
washed with acetone (4 × 10 mL). Recrystallization of the
crude product from a toluene/acetonitrile mixture at room
temperature yielded 13 as pale yellow crystals (2.67 g, 2.36
p Ka Mea su r em en ts. The pKa measurements were carried
out using the overlapping indicator method as described by
Bordwell et al.29 9-Cyanofluorene (pKa(DMSO) ) 8.3) and
9-(methoxycarbonyl)fluorene (pKa(DMSO) ) 10.4) were the
indicators used. Lithium salts of the indicators were prepared,
isolated, and recrystallized before use. The pKa values of the
indicators are determined in DMSO.29 Due to the very low
solubility of silsesquioxanes in DMSO, THF was used as the
1
mmol, 52%). H NMR (CDCl3, δ): 6.87 (s, 1H, C5H3(SiMe3)2),
6.79 (s, 2H, C5H3(SiMe3)2), 1.79 (m, 14H, CH2-C5H9), 1.56 (m,
42H, CH2-C5H9), 0.98 (m, 7H, CH-C5H9), 0.36 (s, 18H, C5H3-
(Si(CH3)3)2). 13C NMR (CDCl3, δ): 132.91 (d, CH-C5H3(SiMe3)2,
1J C-H ) 170 Hz), 130.28 (s, Cipso-C5H3(SiMe3)2), 125.45 (d, CH-
1
1
C5H3(SiMe3)2, J C-H ) 167 Hz), 27.89 (t, CH2-C5H9, J C-H
)
1
130 Hz), 27.36 (t, CH2-C5H9, J C-H ) 131 Hz), 27.07 (t, CH2-
1
1
C5H9, J C-H ) 129 Hz), 22.80 (d, CH-C5H9, J C-H ) 118 Hz),
22.29 (d, CH-C5H9, 1J C-H ) 117 Hz), -0.27 (q, C5H3(Si(CH3)3)2,
1J C-H ) 120 Hz). 29Si{1H} NMR (toluene, δ): -8.17 (SiMe3),
(37) Van Bolhuis, F. J . Appl. Crystallogr. 1971, 4, 263.
(38) De Boer, J . L.; Duisenberg, A. J . M. Acta Crystallogr. 1984,
A40, C410.
(39) No higher metric lattice symmetry or extra metric symmetry
elements were found: (a) Spek, A. L. J . Appl. Crystallogr. 1988, 21,
578. (b) Le Page, Y. J . Appl. Crystallogr. 1987, 20, 264. (c) Le Page, Y.
J . Appl. Crystallogr. 1988, 21, 983.
-64.43, -66.60, -67.64 (2:3:1:3). Anal. Calcd for C46H84O12
-
Si9Ti: C, 48.90; H, 7.49; Ti, 4.24. Found: C, 48.87; H, 7.55;
Ti, 4.20.
[(c-C5H9)7Si7O12]Zr Cp ′′ (14). To a frozen (liquid N2) solu-
tion of Cp′′ZrMe3 (0.56 g, 1.62 mmol) in toluene (20 mL) was
added (c-C5H9)7Si7O9(OH)3 (1.42 g, 1.62 mmol). The mixture
was slowly warmed to room temperature. The resulting pale
yellow solution was stirred for an additional 1 h, after which
the solvent was evaporated, leaving a crystalline solid. Re-
crystallization from CH2Cl2 at -30 °C yielded 14 as colorless
crystals (1.17 g, 1.00 mmol, 62%). 1H NMR (CDCl3, δ): 6.86
(t, 1H, C5H3(SiMe3)2, 4J H-H ) 2 Hz), 6.82 (d, 2H, C5H3(SiMe3)2,
4J H-H ) 2 Hz), 1.76 (m, 14H, CH2-C5H9), 1.54 (m, 42H, CH2-
(40) Spek, A. L. HELENA Program for Data Reduction; University
of Utrecht, Utrecht, The Netherlands, 1993.
(41) Beurskens, P. T.; Beurskens, G.; Bosman, W. P.; de Gelder, R.;
Garc´ıa-Granda, S.; Gould, R. O.; Israe¨l, R.; Smits, J . M. M. The
DIRDIF-97 Program System; Crystallography Laboratory, University
of Nijmegen, Nijmegen, The Netherlands, 1992.
(42) International Tables of Crystallography; Wilson, A. J . C., Ed.;
Kluwer Academic: Dordrecht, The Netherlands, 1992; Vol. C.
(43) Sheldrick, G. M. SHELX-97: Program for the Solution and
Refinement of Crystal Structures; University of Go¨ttingen, Go¨ttingen,
Germany, 1997.
(44) Spek, A. L. PLATON: Program for the Automated Analysis of
Molecular Geometry; University of Utrecht, Utrecht, The Netherlands,
1996.
C5H9), 0.96 (m, 7H, CH-C5H9), 0.36 (18H, C5H3(Si(CH3)3)2). 13
C
NMR (CDCl3, δ): 129.94 (s, Cipso-C5H3(SiMe3)2), 128.82 (d,
(45) J ohnson, C. K. ORTEPII; Report ORNL-5138; Oak Ridge
National Laboratory, Oak Ridge, TN, 1976.
C5H3(SiMe3)2, 1J C-H ) 170 Hz), 123.34 (d, C5H3(SiMe3)2, 1J C-H

Group 4 Metal Siloxy and Silsesquioxane Complexes
Organometallics, Vol. 18, No. 26, 1999 5459
solvent instead. pKa values can differ dramatically, depending
on the solvent used (e.g. PhOH: pKa(H2O) ) 10.0, pKa(DMSO)
) 18.0),29d ion pairing, and chelate formation.29e Hence, the
reported pKa's are no absolute values but are solely qualitative
values relative to the (arbitrary) indicator pKa values.
DF T Com p u ta tion a l Stu d ies. Density functional theory
(DFT) forms the basis of our calculations, as implemented by
the Amsterdam density functional (ADF) code.46 The exchange
correlation functionals in the local-density approximation47
were augmented by generalized gradient approximations to
the exchange48 and correlation.49 All the corrections were
applied self-consistently. We choose a double-ú set of Slater
type basis functions with polarization function for carbon,
oxygen, and fluorine with frozen core of the 1s electrons. The
same basiss, but with no frozen core and a 2p core, was used
for hydrogen and silicon, respectively. The basis for titanium
was triple-ú with polarization function and a 2p frozen core.
A quasi Newtonian approach50 to geometry optimization is
combined with the direct inversion in the iterative subspace
method (DIIS).51 Full geometry optimization has been used
in all calculations in this work. The convergence criteria used
in these calculations are 10-3 hartree for the energy, 5 × 10-3
hartree/Å for the gradients, and 5 × 10-3 Å for the Cartesian
displacement. ADF code makes use of the natural symmetry
properties of the different clusters. The silsesquioxanes were
approximated by substituting the large cyclopentadienyl groups
on the silicon atoms by hydrogens.
was brought into the reactor. The contents were heated with
stirring to the required temperature. Additionally, a constant
ethylene pressure was maintained. A toluene (10 mL) solution
of the cocatalyst (MAO or B(C6F5)3) was syringed into a 100
mL catalyst-dosing vessel containing PMH (25 mL). Subse-
quently, the solution was transferred into the reactor. After
15 min, the desired amount of catalyst precursor was intro-
duced into the reactor in the same way, after which the
catalyst-dosing vessel was washed automatically with PMH
(2 × 50 mL). The polymerization reactions were carried out
under isothermal conditions and constant ethene pressure. At
the end of the reaction, the reaction mixture was collected and
quenched with 20 mL of methanol. Antioxidant (Irganox 1076)
was added to stabilize the polymer. The polymer was dried
under vacuum at 70 °C for 24 h and analyzed by GPC.
1-Hexen e P olym er iza tion . In a glovebox, 10a and 11a (20
µmol) and B(C6F5)3 (0.2 mmol) were dissolved in neat 1-hexene
(10 mL). The homogeneous solution was stirred for 16 h at 25
°C. The polymer obtained was washed with ethanol and H2O
to remove the catalyst. Thorough drying yielded the polymer
1
as a viscous, sticky oil. The polymer was identified by H and
13C NMR and characterized by GPC analysis.
Ack n ow led gm en t. This work was financially sup-
ported by DSM Research BV, which is gratefully ac-
knowledged. We also thank Dr. S. J . R. Creve (DSM
Research) for calculating the cone angles. R.J .H. thanks
the German Science Foundation (DFG) and the National
Computing Facilities Foundation (NCF) for financial
support.
Eth ylen e P olym er iza tion . Ethylene polymerization ex-
periments were carried out in a 1.3 L stainless steel reactor
equipped with a stationary steel stirrer. In a typical experi-
ment, under nitrogen PMH (pentamethylheptane, 600 mL)
(46) ADF 2.3.0; Theoretical Chemistry, Vrije Universiteit Amster-
dam, Amsterdam, The Netherlands: (a) Baerends, E. J .; Ellis, D. E.;
Ros, P. Chem. Phys. 1973, 2, 41. (b) te Velde, G.; Baerends, E. J . J .
Comput. Phys. 1992, 99, 84.
(47) Vosko, S. H.; Wilk, L.; Nusair, N. Can. J . Phys. 1980, 58, 1200.
(48) Becke, A. D. Phys. Rev. A 1988, 38, 3098.
(49) Perdew, J . P. Phys. Rev. B 1986, 33, 8822.
(50) Fan, L.; Ziegler, T. J . Chem. Phys. 1991, 95, 7401.
(51) Fisher, T. H.; Almlo¨f, J . J . Chem. Phys. 1992, 96, 9768.
Su p p or tin g In for m a tion Ava ila ble: Tables of atomic
coordinates, thermal displacement parameters, bond lengths,
bond angles, and torsion angles for 9 and 11a . This material
is available free of charge via the Internet at http://pubs.acs.org.
OM9904495