Hydrosilylation of allyl alcohol with [HSiMe2OSiO1.5]8: Octa(3-hydroxypropyldimethylsiloxy)octasilsesquioxane and its octamethacrylate derivative as potential precursors to hybrid nanocomposites

Article abstract of DOI:10.1021/ja000318r

Octakis(3-hydroxypropyldimethylsiloxy)octasilsesquioxane (OHPS) can be synthesized by direct hydrosilylation of allyl alcohol with octakis(dimethylsiloxy)octasilsesquioxane, (HSiMe₂O)Si₈O₁₂ (Q₈M₈(H)), using platinum divinyltetramethyldisiloxane [Pt(dvs)] as catalyst. Surprisingly, C-hydrosilylation occurs in preference to O-silylation. Hydrosilylation of trimethylsiloxy-2-propene with Q₈M₈(H), followed by desilylation also gives pure OHPS. OHPS reacts with methacryloyl chloride to give octakis(3-methacryloxypropyldimethylsiloxy)octasilsesquioxane (OMPS), a thermal and UV/vis curable precursor to organic/inorganic nanocomposites. Direct hydrosilylation of Q₈M₈(H) with 2-allyloxyethanol also proceeds primarily via C- rather than O-silylation. In contrast, compounds such as 1,3,5,7-tetramethylcyclotetrasiloxane (D₄(H)), 1,1,3,3-tetramethyldisiloxane (TMDS) or terminal Si - H functionalized poly(dimethylsiloxane) (PDMS-H, MW = 400), give significant amounts of O-silylation along with C-silylation. Initial catalyst concentration studies suggest that the catalytic cycle requires the intermediacy of Pt cluster complexes in contrast to recent studies on the mechanism of hydrosilylation which suggest monometallic complex catalysis.

Full text of DOI:10.1021/ja000318r

J. Am. Chem. Soc. 2000, 122, 6979-6988
6979
Hydrosilylation of Allyl Alcohol with [HSiMe2OSiO1.5]8:
Octa(3-hydroxypropyldimethylsiloxy)octasilsesquioxane and Its
Octamethacrylate Derivative as Potential Precursors to Hybrid
Nanocomposites
Chunxin Zhang and Richard M. Laine*
Contribution from the Departments of Chemistry, and Materials Science and Engineering, and
the Macromolecular Science and Engineering Center, UniVersity of Michigan,
Ann Arbor, Michigan 48109-2136
ReceiVed January 28, 2000
Abstract: Octakis(3-hydroxypropyldimethylsiloxy)octasilsesquioxane (OHPS) can be synthesized by direct
hydrosilylation of allyl alcohol with octakis(dimethylsiloxy)octasilsesquioxane, (HSiMe2O)Si8O12 (Q8M8H),
using platinum divinyltetramethyldisiloxane [Pt(dvs)] as catalyst. Surprisingly, C-hydrosilylation occurs in
preference to O-silylation. Hydrosilylation of trimethylsiloxy-2-propene with Q8M8H, followed by desilylation
also gives pure OHPS. OHPS reacts with methacryloyl chloride to give octakis(3-methacryloxypropyl-
dimethylsiloxy)octasilsesquioxane (OMPS), a thermal and UV/vis curable precursor to organic/inorganic
nanocomposites. Direct hydrosilylation of Q8M8H with 2-allyloxyethanol also proceeds primarily via C- rather
than O-silylation. In contrast, compounds such as 1,3,5,7-tetramethylcyclotetrasiloxane (D4H), 1,1,3,3-
tetramethyldisiloxane (TMDS) or terminal Si-H functionalized poly(dimethylsiloxane) (PDMS-H, MW )
400), give significant amounts of O-silylation along with C-silylation. Initial catalyst concentration studies
suggest that the catalytic cycle requires the intermediacy of Pt cluster complexes in contrast to recent studies
on the mechanism of hydrosilylation which suggest monometallic complex catalysis.
I. Introduction
polymerize or copolymerize3 with other functionalized cubes
to provide nanocomposites whose length scales and interfacial
interactions are exactly defined. Note that all of the silicon atoms
in cubes are at the “surface” of these reinforcing particles.
Nanocomposite materials, because of the length scales
involved, incorporate extensive interfacial interactions that can
result in nonlinear changes in composite properties.1,2 These
changes can be so great, that the rule-of-mixtures approximation
commonly used to estimate macroscopic composite properties
fails. Thus, nanocomposites offer potential access to completely
new classes of materials with unique properties. Unfortunately
one of the more difficult problems with nanocomposites is in
developing synthetic and processing approaches that precisely
define the volume and shape of the individual phases and their
periodicity.2 Such efforts are needed to generate nanocomposite
materials that permit one to probe how small changes in
nanostructure affect macroscopic properties.
For these reasons, we have sought to develop methods of
preparing organic/inorganic hybrid nanocomposites based on
cubic silsesquioxanes.3 Cubic silsesquioxanes, R8Si8O12, consist
of a rigid, crystalline silica-like core that is perfectly defined
spatially (0.5-0.7 nm) and that can be linked covalently to eight
R groups.4 By designing the functionality of the R group, it is
possible to create octafunctional macromonomers that will self-
As detailed elsewhere,3b we seek to develop polymerizable
cubes with good-to-excellent control of the lengths and archi-
tectures of the organic segments that tether two points on two
cubes. Because the structure of the organic phase between the
rigid, hard particles can be varied systematically, the potential
exists to carefully probe mechanical, photonic and/or electronic
properties to establish structure-processing-property relation-
ships.3 Such relationships are expected to provide fundamental
principles for the selective design of nanocomposites for specific
applications.
(1) (a) Novak, B. M. AdV. Mater. 1993, 5(6), 422. (b) Loy, D. A.; Shea,
K. J. Chem. ReV. 1995, 95, 1431. (c) Ajayan, P. M. Chem. ReV. 1999, 99,
1787. (d) http://itri.loyola.edu/nano/final/.
We report here novel syntheses of octahydroxy functionalized
cubes that offer potential access to diverse polyester nanocom-
posites and related polymers.9 These compounds also offer
potential utility for developing new ligands for inorganic and
(2) Laine, R. M.; Asuncion, M.; Baliat, S.; Dias Filho, N. L.; Harcup,
J.; Sutorik, A. C.; Viculis, L.; Yee, A. F.; Zhang, C.; Zhu, Q. In Organic/
Inorganic Hybrid Materials; Klein, L., De Guire, M., Lorraine, F., Mark,
J., Eds.; MRS Symposium Series, Vol. 576, Dec. 1999; pp 3-15.
(3) (a) Sellinger, A.; Laine, R. M. Macromolecules 1996, 29, 2327. (b)
Zhang, C.; Laine, R. M. J. Organomet. Chem. 1996, 521, 199. (c) Sellinger,
A.; Laine, R. M. Chem. Mater. 1996, 8, 1592. (d) Zhang, C.; Babonneau,
F.; Bonhomme, C.; Laine, R. M.; Soles, C.; Hristov, H. A.; Yee, A. F. J.
Am. Chem. Soc. 1998, 120, 0(33), 8380.
(4) For reviews: (a) Voronkov, M. G.; Lavrent’yev, V. I. Top. Curr.
Chem. 1982, 102, 199-236. (b) Baney, R. H.; Itoh, M.; Sakakibara, A.;
Suzuki, T. Chem. ReV. 1995, 95, 1409. (c) Provatas, A.; Matisons, J. G.
Trends Polym. Sci. 1997, 5(10), 327.
10.1021/ja000318r CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/28/2000

6980 J. Am. Chem. Soc., Vol. 122, No. 29, 2000
Zhang and Laine
organometallic chemistry,10a,b and as cores for hyperbranched
or dendritic materials.10c,d The current work emphasizes a
polymethacrylate precursor to organic/inorganic nanocompos-
ites.
Iron supplied by Perkin-Elmer. Measurements were performed under
a continuous flow of nitrogen or synthetic air (25 mL/min), at 20 °C/
min.
II.B.5. Differential Scanning Calorimetry. The calorimetry was
performed on a Perkin-Elmer DSC-7 differential scanning calorimetry
(Perkin-Elmer Co., Norwalk, CT). The instrument was calibrated with
indium supplied by Perkin-Elmer. Measurements were performed under
a continuous flow of nitrogen (25 mL/min). Samples (1-2 mg) were
typically equilibrated at 30 °C, ramped to the desired temperature (∼20
°C lower than its decomposition temperature), and then allowed to cool
back to 30 °C at 10 °C/min. Three heating-cooling cycles were
recorded for each sample unless otherwise noted.
II. Experimental Section
II.A. Materials. Solvents for hydrosilylation were purified under
N2 as follows: CH2Cl2 was distilled from P2O5, toluene from benzo-
phenone/Na, THF from CaH2. Allyl alcohol and methacryloyl chloride
H
were distilled before use. Q8M8 was synthesized following modified
literature procedures.6 Platinum divinyltetramethyldisiloxane complex,
Pt(dvs) was obtained from PCR Co. and diluted to a 2 mM solution in
distilled toluene before use. Dicyclopentadiene Pt complex, Pt(dcp),
was synthesized following literature procedures and used as a 2 mM
solution in toluene.11 Hexachloroplatinic acid, H2PtCl6 was obtained
from Aldrich and used as a 2 mM solution in isopropyl alcohol. 1,1,3,3-
Tetramethyldisiloxane (TMDS) and Si-H terminated poly(dimethyl-
siloxane) (PDMS-H, MW ) 400) were obtained from Hu¨ls America
Inc. 1,3,5,7-tetramethylcyclotetrasiloxane (D4H) was obtained from
Gelest Inc. Other reagents and solvents were obtained from standard
vendors and used as received.
II.B.6. Gel Permeation Chromatography. Chromatographic analy-
ses were performed on a Waters GPC system, using a Waters 410 RI
detector, Waters Styragel columns (7.8 × 300, HR 0.5, 1, 3, 4), and a
PL-DCU data capture unit from Polymer Laboratory. The system was
calibrated using polystyrene standards obtained from Polymer Labora-
tory. THF was used as the eluent, at a flow rate of 1.0 mL/min.
II.C. Synthesis of Octakis(3-hydroxypropyldimethylsiloxy)octa-
H
silsesquioxane (OHPS). II.C.1. Reaction of Q8M8 with Allyl
Alcohol. Q8M8H (5.00 g, 4.91 mmol) was placed in a 100 mL Schlenk
flask equipped with a reflux condenser and a magnetic stirrer. Toluene
(25 mL) was added to dissolve the cube, followed by addition of allyl
alcohol (3.34 mL, 49.1 mmol). The reaction flask was evacuated and
refilled with N2 three times. Pt(dvs) (2 mM solution, 1.2 mL) was added
via a syringe. The reaction began to reflux immediately due to the
exothermic hydrosilylation and was complete in ∼30 min. Two layers
were formed. Triphenylphosphine, PPh3 (∼10 mg), was added to
deactivate the Pt(dvs) catalyst, and the mixture was stirred for ∼1 h.
The bottom layer containing the product, OHPS, was recovered.
Residual allyl alcohol and toluene were removed first under N2 stream,
and then under vacuum. OHPS is a white solid (6.2 g, 86%), soluble
in methanol, THF, acetone, CH2Cl2 etc. Selected characterization
data: FTIR (KBr, cm-1) υ O-H: 3339s; υ C-H: 2959s, 2931s,
II.B. Techniques. B.1. NMR analyses. All 1H- and 13C NMR
analyses were done in CDCl3 and recorded on a Varian INOVA 400
1
spectrometer. H NMR spectra were collected at 400.0 MHz using a
6000 Hz spectral width, a relaxation delay of 3.5 s, a pulse width of
38°, 30k data points, and CHCl3 (7.259 ppm) as internal reference.
13C{1H } NMR spectra were obtained at 100.6 MHz using a 25 000
Hz spectral width, a relaxation delay of 1.5 s, a pulse width of 40°,
75k data points, and CDCl3 (77.23 ppm) as internal reference.
All 29Si NMR spectra were recorded on a Bruker 360 spectrometer
operated at 71.5 MHz using a 32000 Hz spectral width, a pulse angle
of 90°, a relaxation delay of 10.0 s, 32k data points, and TMS as
external reference.
II.B.2. GC-Mass Spectral Analyses. Analyses were conducted on
a Finningan model 4021 quadrupole GC/mass spectrometer by electron
ionization. Data were recorded and analyzed using the software package
provided by Galwin.
1
2875m; δ CH3: 1258m; υ Si-C: 1173m; υ Si-O: 1082s. H NMR
(400 MHz, CDCl3, CHCl3 ref) 3.76 (s, 8H, OH), 3.54 (t, 16H, CH2-
OH), 1.59 (p, 16H, CH2-CH2-CH2), 0.58 (t, 16H, CH2-Si), 0.13 (s,
48H, Si(CH3)2). 13C NMR (100 MHz, CDCl3, CDCl3 ref) 65.2 (CH2-
OH), 26.3 (CH2-CH2-CH2), 13.6 (CH2-Si), -0.2 (Si(CH3)2). 29Si
NMR (71.5 MHz, CH2Cl2, TMS ref) 13.9 (-OSi(CH3)2CH2), -108.3
(SiO4). TGA (ceramic yield in air) 51.0% (64.8% theoretical, see results
and discussion). DSC: mp 84 °C. Elemental analysis found (%) C:
32.35, H: 6.98; Calculated (%) C: 32.40, H: 7.07, Si: 30.31, O: 30.22.
GPC: Mn 1.72 × 103, Mw 1.80 × 103, PDI 1.04 (FW of OHPS 1482).
II.B.3. Fourier Transform Infrared Spectra. Spectra were recorded
on a Nicolet 5DXB FT-IR 300 spectrometer. Random cuttings of
crystalline, optical-grade KBr from International Crystal Laboratories
were used to prepare samples. About 600 mg of KBr was ground in a
mortar with a pestle, and enough solid sample was ground with KBr
to make a 1 wt % mixture for making KBr pellets. Liquid samples
were cast on salt plates. After the sample was loaded, the sample
chamber was purged with nitrogen for a minimum of 10 min before
data collection. A minimum of 32 scans was collected for each sample
NMR data in methanol: 1H NMR (400 MHz, CD3OD, CD3OD ref,
taken with a fresh sample) 4.78 (s, ∼2H, OHs), 3.47 (t, ∼1.4H, CH2-
OH), 3.44 (t, ∼0.6H, CH2-OSi), 1.55 (m, 2H, CH2-CH2-CH2), 0.59
(m, 2H, CH2-Si), 0.13 (s, ∼4H, Si(CH3)2), 0.05 (s, ∼2H, Si(CH3)2).
1H NMR (400 MHz, CD3OD, CD3OD ref, taken after 2 days, see Figure
8) 4.78 (s, ∼2H, OHs), 3.44 (t, 2H, CH2-OSi), 1.50 (p, 2H, CH2-
CH2-CH2), 0.55 (t, 2H, CH2-Si), 0.05 (s, ∼6H, Si(CH3)2). 13C NMR
(100 MHz, CD3OD, CDCl3 ref, almost no change with time) 65.6
(CH2-O), 27.1 (CH2-CH2-CH2), 14.5 (CH2-Si), -0.02 (Si(CH3)2),
-2.6 (Si(CH3)2). 29Si NMR (71.5 MHz, CD3OD, TMS ref) 13.9
(-OSi(CH3)2CH2), -100.4 (see Figure 7 in results and discussion),
-108.3 (SiO4).
at a resolution of (4 cm-1
.
II.B.4. Thermal Gravimetric Analyses. Analyses were performed
on a Perkin-Elmer TGA-7 thermogravimetric analyzer (Perkin-Elmer
Co., Norwalk, CT). The instrument was calibrated with Alumel and
(5) (a) Agaskar, P. A. Inorg. Chem. 1991, 30, 2707. (b) Olsson, K. ArkiV.
Kemi. 1958, 13(5), 367. (c) Voronkov, M. G.; Martynova, R. G.; Mirskov,
R. G.; Belyi, V. I. Zh. Obsh. Khim. 1979, 49(7), 1522.
(6) (a) Hasegawa, I. J. Sol-Gel Sci. Technol. 1993, 1, 57. (b) Hasegawa,
I.; Motojima, D. J. Organomet. Chem. 1992, 441, 373.
(7) Gravel, M.-C.; Zhang, C.; Laine, R. M. Appl. Organomet. Chem.
1998, 13, 329-36
(8) (a) Kreuzer, F.-H.; Maurer, R.; Spes, R. Makromol. Chem. Macromol.
Symp. 1991, 50, 215. (b) Sellinger, A.; Laine, R. M.; Chu, V.; Viney, C. J.
Polym. Sci.: Part A: Polym. Chem. 1994, 32, 3069. (c) Mehl, G. H.;
Goodby, J. W. Angew. Chem., Int. Ed. Engl. 1996, 35(22), 2641. (d) Zhang,
C.; Laine, R. M. Manuscript in preparation.
(9) Majoros, I.; Marsalko, T. M.; Kennedy, J. P. Polymer Bull. 1997,
38, 15,.
(10) (a) Feher, F. J.; Budzichowski, T. A. Polyhedron 1995, 14(22), 3239.
(b) Feher, F. J.; Tajima, R. L. J. Am. Chem. Soc. 1994, 116, 2145. (c)
Hong, B.; Thomas, T. P. S.; Murte, H. J.; Lebrun, M. M. Inorg. Chem.
1997, 36, 6146. (d) Feher, F. J. Wyndham, K. D. Chem. Commun. 1998,
323.
A GC-mass spectrum was taken when the above 1H NMR sample
in CD3OD was first dried, and then redissolved in CDCl3 (m/z, %) P+
116 (8.1%), P-1 115 (0.2%), P-15 101 (100%), P-17 99 (24.3%), P-28
88 (38.5%). (see Results and Discussion).
II.C.2. Octakis[(3-trimethylsiloxypropyl)dimethylsiloxy]silses-
quioxane (OSPS) from Allyloxytrimethylsilane. Q8M8H (5.00 g, 4.91
mmol) was placed in a 100 mL Schlenk flask equipped with a reflux
condenser and a magnetic stirrer. Toluene (25 mL) was added to
dissolve the cube, followed by allyloxytrimethylsilane (8.28 mL, 49.1
mmol). The reaction flask was degassed and refilled with N2 three times.
Pt(dvs) (0.1 mL, 2 mM solution) was added via a syringe. The reaction
was then heated to ∼65 °C/∼5 h and 100 °C/4 h to completion, giving
octakis[(3-trimethylsiloxypropyl)dimethylsiloxy]silsesquioxane (OSPS).
Triphenylphosphine, PPh3 (∼2 mg), was added to deactivate the Pt-
(11) Laub, R. J.; Finkelmann, H.; Apfel, M. A.; Janini, G. M.; Lu¨hmann,
B. H.; Price, A.; Roberts, W. L.; Shaw, T. J.; Smith, C. A. Anal. Chem.
1985, 57, 651.

H
Hydrosilylation of Allyl Alcohol with O8M8
J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6981
(dvs) catalyst. Excess allyloxytrimethylsilane and toluene were evapo-
rated using N2 stream and then vacuum, to give analytically pure OSPS
as a white solid (8.30 g, 82%). Selected characterization data: IR (KBr,
cm-1) υ C-H 2960s, 2936, 2876m; δ CH3: 1252s; υ Si-C: 1186m;
υ Si-O: 1096s. 1H NMR (400 MHz, CDCl3, CHCl3 ref) 3.51 (t, 16H,
CH2-O), 1.54 (p, 16H, CH2-CH2-CH2), 0.55 (t, 16H, CH2-Si), 0.12
(s, 48H, Si(CH3)2), 0.08 (s, 72H, Si(CH3)3). 13C NMR (100 MHz,
CDCl3, CDCl3 ref) 65.5 (CH2-O), 26.4 (CH2-CH2-CH2), 13.6 (CH2-
Si), -0.17 (Si(CH3)2), -0.22 (Si(CH3)3). 29Si NMR (71.5 MHz, CH3-
OH, TMS ref) 13.3 (OSi(CH3)2CH2), 16.5 (OSi(CH3)3), -108.9 (SiO4).
TGA (ceramic yield in air) 46.2% (69.9% theoretical, see results and
discussion). DSC: mp 35 °C. Elemental analysis found (%) C: 37.42,
H: 8.06; Calculated (%) C: 37.31, H: 8.22, Si: 32.72, O: 21.75.
GPC: Mn 1.99 × 103, Mw 2.03 × 103, PDI 1.02 (FW of OSPS 2060).
II.C.3. OHPS from OSPS. The OSPS (2 g) obtained above was
combined with 10 mL of anhydrous methanol, and stirred at room
temperature for 4 h. All of the trimethylsilyl groups were removed as
determined by NMR analysis. OHPS was isolated by removing the
entire methanol and the methoxytrimethylsilane under vacuum. The
resultant white powder is identical to the OHPS obtained via the direct
hydrosilylation route based on NMR, GPC, and EA.
groups disappeared within 30 min. The products were obtained as white
solids in ∼80% yield, after excess allyl alcohol and toluene were
removed. GPC analyses consist of a sharp peak at ∼1600 Da, and a
small peak at ∼3000 Da corresponding to dimers. The polydispersity
indices (PDI) of these products ranged from 1.02 to 1.04. No
1
O-silylation was detected by H NMR analysis. When Pt (dvs)] ) 20
mM, the reaction required heating to 65 °C for g12 h to go to
completion. The product was a waxy solid after solvent removal. About
1
3% O-silylation was observed by H NMR analysis. Approximately
two-thirds of the product was monomer at ∼1600, and one-third higher-
molecular weight species based on the GPC trace: Mn 2.11 × 103, Mw
2.78 × 103, PDI 1.32. When [Pt] ) 10 mM, the reaction had to be
refluxed for g12 h to force completion. The product is a viscous liquid.
About 7% O-silylation was observed. Approximately one-third of the
product was monomer at ∼1600, and two-third higher-molecular weight
species: Mn 1.40 × 104, Mw 6.84 × 104, PDI 17.2.
II.D.4. Direct Hydrosilylation of Other Unsaturated Alcohols.
Two additional alcohols, propargyl alcohol and 2-allyloxyethanol were
H
reacted with Q8M8 to explore the applicability of the direct hydro-
H
silylation method. Concentrations of Q8M8 and alcohols were kept
the same as above, and [Pt] ) 100 mM. Toluene was used as the
solvent.
II.D. Kinetic Studies on Direct Hydrosilylation of Allyl Alcohol
with Q8M8H. II.D.1. Catalyst Selectivity Studies. Three common
catalysts for hydrosilylation, Pt(dvs), Pt(dcp), and H2PtCl6 were tested.
In all reactions, concentrations of reactants and catalyst were kept
constant, that is, [Si-H] ) 1.6 M, [allyl alcohol] ) 2.0 M, [Pt] ) 100
mM. Toluene was used as solvent.
H
II.D.5. Reaction of Propargyl Alcohol. Q8M8 (0.500 g, 0.491
mmol), propargyl alcohol (0.29 mL, 4.9 mmol), Pt(dvs) (0.015 mL, 2
mM), toluene (2.5 mL). The reaction required heating at 60 °C for 5
h to complete. On cooling, two layers formed. The top layer was mainly
toluene, and the product recovered from the bottom layer (0.610 g,
1
Q8M8H (0.500 g, 0.491 mmol) was placed in a 25 mL Schlenk flask
equipped with a reflux condenser and a magnetic stirrer. Toluene (2.5
mL) was added to dissolve the cube, followed by allyl alcohol (0.334
mL, 4.91 mmol). The reaction flask was carefully degassed and refilled
with N2 three times. Pt catalyst solution (2 mM, 0.l5 mL) was added.
When Pt(dvs) was used, the reaction was so exothermic that it began
to reflux, and NMR analysis showed that all the Si-H disappeared in
∼30 min. Toluene and the excess allyl alcohol were evaporated using
an N2 stream and then vacuum. The product was a white solid (0.580
g, 77% of theoretical). NMR analysis indicates that the majority of the
product results from C-silylation. GPC analysis showed a sharp peak:
Mn 1.67 × 103, Mw 1.71 × 103, PDI 1.02 (calculated FW 1482). When
Pt(dcp) and H2PtCl6 were used, the reactions had to be heated at 85 °C
for 3 h before all the Si-H reacted, as determined by 1H NMR analysis.
The resultant products were both viscous liquids. With Pt(dcp), ∼86%
C-silylation and ∼14% O-silylation occurred based on NMR analysis,
and the resultant product had a broad weight distribution: Mn 3.25 ×
103, Mw 8.49 × 103, PDI 2.61. With H2PtCl6, ∼14% C-silylation and
∼86% O-silylation occurred based on NMR analysis. GPC analysis:
Mn 3.04 × 103, Mw 2.59 × 104, PDI 8.53.
85%) was analyzed as follows. H NMR (400 MHz, CDCl3, CHCl3
ref) 6.31 (dt, 3.6H, CHdCHSi, â-cis), 5.95 (dt, 3.6H, CHdCHSi, â-cis),
5.84 (m, 4.4H, CHdCHSi, â-trans), 5.50 (m, 4.4H, CHdCHSi,
â-trans), 4.29 (m, 8.8H, trans-CHdCHCH2O), 4.17 (dd, 7.2H, cis-
CHdCHCH2O), 2.71 (s, 8H, OH), 0.28 [s, 26H, trans-CHdCH-Si-
(CH3)2], 0.23 [s, 22H, cis-CHdCH-Si(CH3)2]. As shown above,
-OSi(CH3)2-CHdCHCH2OH and -OSi(CH3)2-CHdCHCH2OSi-
(CH3)2- structures are not differentiable from NMR analysis. However,
GPC analysis gave a relatively broad distribution with higher molecular
species (Mn 2.61 × 103, Mw 3.96 × 103, PDI 1.52), which suggests
simultaneous C- and O-silylation. Thus no further characterization was
pursued.
H
II.D.6. Reaction of 2-Allyloxyethanol. Q8M8 (0.500 g, 0.491
mmol), 2-allyloxyethanol (0.52 mL, 4.9 mmol), Pt(dvs) (0.015 mL, 2
mM), toluene (2.5 mL). The reaction was completed ∼30 min after
Pt(dvs) was added per NMR analysis. Two layers formed on cooling,
the top layer was mainly toluene, and bottom layer contained the
reaction product. The bottom layer was collected, and the residual
2-allyloxyethanol and toluene were removed first under N2 stream, and
then under vacuum to provide a white solid (0.790, 87%). On the basis
of the characterization data, the product was found to be octakis[2-(2-
hydroxylethoxy)ethyldimethylsiloxy]octasilsesquioxane. 1H NMR (400
MHz, CDCl3, CHCl3 ref) 3.71 (t, 16H, HO-CH2-CH2-O), 3.53 (t,
16H, HO-CH2-CH2-O), 3.45 (t, 16H, OCH2CH2CH2Si), 2.70 (s, 8H,
OH), 1.66 (m, 16H, OCH2CH2CH2Si), 0.61 (t, 16H, OCH2CH2CH2-
Si), 0.15 [s, 48H, Si(CH3)2]. 13C NMR (100 MHz, CDCl3, CDCl3 ref)
73.9 (OCH2CH2OH), 72.1 (CH2OCH2CH2OH), 65.2 (CH2-OH), 23.2
(CH2-CH2-CH2), 13.7 (CH2-Si), -0.3 (Si(CH3)2). 29Si NMR (71.5
MHz, CH2Cl2, TMS ref) 13.6 [OSi(CH3)2CH2], -108.5 (SiO4). GPC
analysis: Mn 1.91 × 103, Mw 2.01 × 103, PDI 1.05.
II.D.2. Solvent Effects. Three additional solventssheptane, dichloro-
methane, and THF were also tested. Pt(dvs) was used as the catalyst.
Concentrations of Q8M8H, allyl alcohol, and Pt(dvs) were the same as
above, and the same procedure was followed. When heptane was used,
the reaction proceeded in the same way as in toluene. When CH2Cl2
was used, the reaction had to be refluxed for 2 h to force completion.
NMR analysis indicated ∼81% C-silylation and ∼19% O-silylation.
GPC analysis gave: Mn 2.37 × 103, Mw 4.01 × 104, PDI 1.69. When
THF was used, the reaction required 8 h of reflux to force completion.
NMR analysis showed ∼87% C-silylation and ∼13% O-silylation. GPC
analysis gave: Mn 2.10 × 103, Mw 2.57 × 104, PDI 1.22.
II.E. Direct Hydrosilylation of Allyl Alcohol with Other Si-H
Compounds. Three additional Si-H compounds, tetramethylcyclox-
tetrasiloxane (D4H), tetramethyldisiloxane (TMDS) and Si-H capped
poly(dimethylsiloxane) (PDMS-H, MW 400) were reacted with allyl
alcohol. The concentration of Si-H and the allyl alcohol were kept
the same as above, with [Pt] ) 100 mM. Toluene was used as the
solvent.
II.E.1. Reaction of D4H. D4H (1 mL, 4.12 mmol), allyl alcohol (1.40
mL, 20.6 mmol), toluene (10 mL), Pt(dvs) (0.50 mL, 2 mM). Violent
bubbling was observed when Pt(dvs) was added. The reaction product
was a cross-linked gel after stirring at room-temperature overnight, and
was not further characterized.
II.D.3. Pt(dvs) Concentration Studies. On the basis of the above
results, Pt(dvs) was chosen as the catalyst, and toluene the solvent.
Different catalyst concentrations ranging from 10 to 100 mM were
H
tested. A typical procedure was as follows: Q8M8 (0.500 g, 0.491
mmol) was placed in a 25 mL Schlenk flask equipped with a reflux
condenser and a magnetic stirrer. Toluene (2.5 mL) was added via a
syringe to dissolve the cube, followed by allyl alcohol (0.334 mL, 4.91
mmol). The reaction flask was carefully degassed and refilled with N2
three times. Pt(dvs) was added via a syringe {2 mM solution, amount
varied from 0.15, 0.12, 0.10, 0.075, 0.030, to 0.015 mL for different
reactions, corresponding to [Pt(dvs)] ) 100, 80, 67, 50, 20, and 10
1
mM}. Reaction was followed by H NMR until complete. When Pt-
II.E.2. Reaction of TMDS. TMDS (1 mL, 5.65 mmol), allyl alcohol
(0.96 mL, 14.1 mmol), toluene (6 mL), Pt(dvs) (0.35 mL, 2 mM).
(dvs)] g 50 mM, the reaction was extremely exothermic and Si-H

6982 J. Am. Chem. Soc., Vol. 122, No. 29, 2000
Zhang and Laine
Violent bubbling was also observed after Pt(dvs) was added. The Si-H
peak disappeared after ∼1 h. The product was a clear, viscous liquid.
NMR analysis indicated approximately ∼75% C-silylation and ∼25%
O-silylation. GPC analysis: Mn 7.08 × 102, Mw 7.84 × 102, PDI 1.11.
II.E.3. Reaction of PDMS-H. PDMS-H (2 g, 5.00 mmol), allyl
alcohol (0.85 mL, 12.5 mmol), toluene (6 mL), Pt(dvs) (0.35 mL, 2
mM). The reaction was exothermic, with no significant bubbling. All
of the Si-H disappeared in ∼30 min. NMR analysis indicated ∼16%
O-silylation, ∼84% C-silylation. GPC analysis: Mn 1.02 × 103, Mw
1.09 × 103, PDI 1.07.
2-allyloxyethyl-methacrylate cube requires hydrosilylation tem-
peratures of 80-90 °C, leading to premature polymerization.12
To avoid the problems associated with the above syntheses,
we sought a simple method of introducing OH functionality
via hydrosilylation. Such a route, if identified, would provide a
general approach to polyester nanocomposites including the
desired methacrylates. Much to our surprise, we discovered that
H
direct hydrosilylation of allyl alcohol with Q8M8 can lead to
exclusive C-silylation to give octakis(3-hydroxypropyl-dimeth-
ylsiloxy)octasilsesquioxane (OHPS). This contrasts with many
literature reports wherein O-silylation competes so successfully
with C-silylation that protective groups must be used to avoid
O-silylation:13,14
II.F. Synthesis of Octakis(methacryloxypropyldimethylsiloxy)octa-
silsesquioxane (OMPS). OHPS cube (2.57 g, 1.73 mmol) was weighed
into a 50 mL Schlenk flask equipped with an addition funnel, a reflux
condenser, and a magnetic stir bar. CH2Cl2 (20 mL) was added to
dissolve the OHPS. Triethylamine (1.98 mL, 14.2 mmol) was added.
The mixture was cooled in an ice bath and stirred for 15 min, then
methacryloyl chloride (1.50 mL, 15.2 mmol) was added dropwise
through the addition funnel. A light pink precipitate of triethylammo-
nium chloride formed on adding the methacryloyl chloride. After all
of the methacryloyl chloride was added, the reaction was further stirred
at room temperature for 2 h. To isolate the methacrylate product, the
reaction mixture was first filtered, and then the filtrate was transferred
to a 100 mL separatory funnel and washed with DI water three times,
5% NaHCO3 solution twice, 0.5 N NaOH twice, and last brine once.
The solution was dried over Na2SO4. Solvent was removed, to give
OMPS as clear, yellowish, slightly viscous oil (7.2 g, 75%). Selected
characterization data: IR (cm-1) υ C-H: 2959s, 2930s, 2896m; υ
CdO: 1719s; υ CdC: 1638w; δ CH3: 1254m; υ Si-C: 1164s; υ
H
Thus, if Q8M8 is reacted with neat allyl alcohol, using
conditions discussed below, then the product recovered is the
octahydroxypropyl product of reaction 4. However, if low
concentrations of catalyst or allyl alcohol are used then
considerable O-silylation results.
To further understand the reaction process and briefly probe
the catalytic cycle, simple kinetic studies were conducted. The
effects of changes in reaction conditions, including types of
solvents, catalyst precursors, and precursor concentrations on
C- vs O-silylation were identified. The goal was first to optimize
the process and then explore the scope of the reaction using
different Si-H compounds and unsaturated alcohols.
1
Si-O: 1088s. H NMR (400 MHz, CDCl3, CHCl3 ref) 6.08 [s, 8H,
CH2dC(CH3)(CdO)], 5.54 [s, 8H, CH2dC(CH3)(CdO)], 4.08 (t, 16H,
CH2-O), 1.93 (s, 24H, CH2dC(CH3)(CdO)), 1.69 (p, 16H, CH2-
CH2-CH2), 0.62 (t, 16H, CH2-Si), 0.15 (s, 48H, Si(CH3)2). 13C NMR
(100 MHz, CDCl3, CDCl3 ref) 167.4 (CdO), 136.6 (CH2dCH-), 125.3
(CH2dCH-), 67.0 (CH2-O), 26.4 (CH2-CH2-CH2), 18.4 (CH2dCH-
(CH3)-), 13.7 (CH2-Si), -0.33 (Si(CH3)2). 29Si NMR (71.5 MHz, CH3-
OH, TMS ref) 14.4 (OSi(CH3)2CH2), -107.8 (SiO4). TGA (ceramic
yield in air) 45.8% (47.4% theoretical). DSC: (broad exothermal onset
at 176 °C in first heating cycle, which was absent from subsequent
heating cycles). Elemental analysis found (%) C: 42.29, H: 6.65;
Calculated (%) C: 42.66, H: 6.76, Si: 28.41, O: 22.17. GPC
analysis: Mn 2.00 × 103, Mw 2.08 × 103, PDI 1.06 (FW of OMPS
2027).
III.A. Kinetic Studies of Direct Hydrosilylation of Allyl
1
Alcohol. The reaction kinetics were followed using H NMR
III. Results and Discussion
and GPC, to determine the percent C- vs O-silylation, and the
molecular weight distributions of the products, respectively. If
the reaction gives selective C- or O-silylation, the respective
OHPS or octakis(allyloxy-dimethylsiloxy)octasilsesquioxane
formed would be monodisperse with MW ≈ 1500. However,
if both C- and O-silylation occur with similar reaction rates,
polymeric species should form as demonstrated when 4 equiva-
lents of allyl alcohol are used, resulting in a cross-linked,
transparent gel. In this case, both C- and O-silylation occur as
allyl alcohol serves as a bifunctional cross-linker (Figure 1).
Our initial work on polymerizable R8Si8O12 cubes targeted
low viscosity, light curable, single-phase nanocomposite precur-
sors as novel dental restoratives.3 To be compatible with current
restoratives, initial efforts were made to synthesize methacrylate
cubes by reacting H8Si8O12 (T8H) or Q8M8 with propargyl
H
methacrylate and 2-allyloxyethyl methacrylate, reactions 1 and
2.3a,c Unfortunately, propargyl methacrylate derived mac-
romonomers contain internal double bonds that react further with
unreacted Si-H groups leading to untimely polymerization. The
Figure 1. Cross-link produced by simultaneous C- and O-silylation
of allyl alcohol.
Note that GPC can only be used to estimate the molecular
weights and distributions of cubes, or their polymers, because
the linear polystyrene standards have limited relevance to cubes
(12) Sellinger, A. Ph.D. Dissertation, University of Michigan, 1997.
(13) (a) Greber, V. G.; Jager, S. Makromol. Chem. 1962, 55, 96. (b)
Ghose, B. N. J. Organomet. Chem. 1979, 164, 11. (c) Braun, F.; Willner,
L.; Hess, M.; Kosfeld, R. J. Organomet. Chem. 1987, 332, 63. (d) Braun,
F.; Willner, L.; Hess, M.; Kosfeld, R. J. Organomet. Chem. 1989, 366, 53.
(14) ComprehensiVe Handbook on Hydrosilylation; Marciniec, B., Ed.;
Pergamon Press: Elmsford, NY, 1992; Chapter 3.

H
Hydrosilylation of Allyl Alcohol with O8M8
J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6983
Figure 2. 1H NMR spectra of products obtained from hydrosilylation of allyl alcohol with Q8M8H using (a) H2PtCl6, (b) Pt(dcp) (a broad hump for
OH at ∼3.2 ppm does not show up in this reduced spectrum), and (c) Pt(dvs) as catalyst at 100 mM.
or 3-D structures.15 However, these standards provide a
convenient way to estimate molecular weight, especially when
comparing two similar systems.
The detailed results obtained with different catalyst precursors,
solvents, Pt concentrations, and different Si-H compounds and
unsaturated alcohols are presented below.
Catalyst Selectivity. The utility of three types of Pt catalyst
precursors, hexachloroplatinic acid (H2PtCl6), platinum dicy-
clopentadienyl dichloride [Pt(dcp)], and platinum divinyltet-
ramethyldisiloxane [Pt(dvs)] were assessed for reaction 4. The
amount of added catalyst was kept at [Pt] ) 100 mM to be
consistent with studies described below. Note that for simplicity,
the catalyst concentration is treated as the amount of catalyst
precursor added (see experimental).
Under a set of standard reaction conditions, for reaction 4,
1H NMR analysis indicates that Pt(dvs) provides the cleanest
C-silylation, while Pt(dcp) gave ∼14% O-silylation in addition
Figure 3. GPC of products obtained from hydrosilylation of allyl
alcohol with Q8M8H using different Pt catalyst precursors at 100 mM.
to ∼86% C-silylation. In contrast, H2PtCl6 gave ∼14% C-
silylation and ∼86% O-silylation, with both allyl alcohol and
the 2-propanol used as the catalyst solvent. Figures 2 and 3
were followed by NMR and GPC, as above. When [Pt(dvs)] )
10 mM and 20 mM, the reaction required heating at reflux (110
show 1H NMR spectra and GPC traces of the products obtained
using different catalyst precursors. Pt(dvs) appears to be the
most effective C-silylation catalyst precursor for this reaction,
and was used for the remaining studies. We have not exhaus-
tively assessed all possible catalyst precursors.
III.A.1. Solvent Effects. The influence of solvent on regi-
oselectivity was tested, under standard conditions, using Pt(dvs),
[Pt] ) 100 mM. Heptane, CH2Cl2, and THF were compared
with toluene. Reaction 4 proceeds much the same in both
heptane and toluene providing selective C-silylation. In CH2-
Cl2 and THF, 19% and 13% O-silylation were observed,
respectively. Toluene was chosen for the remaining studies
because it provides the cleanest C-silylation.
°C) for g 12 h to go to completion giving either a viscous liquid
or a waxy solid product. Both GPC traces show two major
components, a sharp peak at ∼1600, and a higher-molecular
weight peak with a broad distribution, indicating a mixture of
different molecular weight species. For [Pt(dvs)] ) 100 mM,
the reaction goes to completion in e30 min at room temperature
to give a white, solid product. The GPC trace shows only a
minor high molecular weight component, in addition to the sharp
peak at ∼1600 (Figure 4).
Products from the above reactions were also analyzed using
NMR. At lower [Pt(dvs)], for example, 10 and 20 mM, about
7 and 3% O-silylation were observed, respectively. However,
no CH2dCH-CH2-O-Si- species were detectable, only
-Si-CH2-CH2-CH2-O-Si- linkages as shown in Figure
1. This suggests that CH2dCH-CH2-O-Si- is more suscep-
tible to C-silylation than free allyl alcohol. At higher Pt(dvs)
III.A.2. Catalyst Concentration Studies. The effects of [Pt-
(dvs)] on regioselectivity in reaction 4 were probed. Reactions
(15) Sperling, L. H. In Introduction to Physical Polymer Science; John
Wiley & Sons: 1986; Chapter 3.

6984 J. Am. Chem. Soc., Vol. 122, No. 29, 2000
Zhang and Laine
Figure 4. GPC traces of OHPS obtained from OSPS and products obtained from direct hydrosilylation of allyl alcohol using different concentrations
of Pt(dvs).
Scheme 1. Chalk-Harrod Mechanism of C-Silylation
Scheme 2. Silane Activation in O-Silylation
(PPh3)2]SbF6 (S)solvent) catalyzed O-silylation also indicate
that the reaction is first order in catalyst concentration, and zero
order in silane and alkene.20
On the basis of these observations, when the concentrations
of the allyl alcohol and the cube {i.e., [Si-H]} are kept constant,
as above, the ratio of C- to O-silylated product should be a
constant, independent of changes in [catalyst]. This contradicts
the above observations, where almost no O-silylation was
obtained at [Pt(dvs)] g 50 mM, and ∼3 and 7% O-silylation
occurred when [Pt] ) 20 mM and 10 mM, respectively.
Furthermore, much more forcing conditions were required at
the lower catalyst concentrations. These results suggest a change
in the nature of the active catalyst species.
Pt(dvs), or Karstedt’s catalyst, is obtained by treating H2-
PtCl6 with divinyltetramethyldisiloxane to form a Pt(0) species
as suggested by the X-ray structure of [Pt{(I-CH2dCHMe2-
Si)2O}(P-t-Bu3)], obtained by adding P-t-Bu3 to a Pt(dvs)
solution.21,22 The principle component of Karstedt’s catalyst is
suggested to be Pt(dvs)2 in the same report.22 However, dinuclear
Pt2(H2CdCHSiMe2OSiMe2CHdCH2)3, that is, Pt2(dvs)3 was
isolated by removing excess divinyltetramethyldisiloxane (dvs)
from a Pt(dvs) solution and characterized by X-ray crystal
studies.23 In solution however, Pt2(dvs)3 is in equilibrium with
Pt(dvs)2 (Figure 5), and the concentration of Pt(dvs)2 increases
with increasing concentrations of dvs.24
concentrations, for example, [Pt(dvs)] g 50 mM, no O-silylation
is detected by NMR and the product is essentially identical to
the OHPS obtained via protection-deprotection (reactions 8 and
9).
III.A.3. Comments on the Catalysis Mechanism. The above
results allow us to make several observations concerning the
type of catalysis mechanism that is operative for both C-
silylation and O-silylation.
C-silylation has been studied extensively since the 1950s.14
The widely accepted Chalk-Harrod mechanism for alkene
hydrosilylation is illustrated in Scheme 1.16 Kinetic studies on
homogeneous H2PtCl6 catalyzed C-silylation indicate that the
process is first order in [molecular Pt catalyst], and zero order
in both [silane] and [alkene].17
Si-H alcoholysis resulting in O-silylation has also been
studied.18 Most of the mechanistic studies suggest transition
metal activation of Si-H bonds, either through oxidative
addition or formation of a σ complex, which favors attack of
the alcohol on Si (Scheme 2).19 Kinetic studies on [IrH2S2-
(19) (a) Sommer, L. H.; Lyons, J. E. J. Am. Chem. Soc. 1969, 91(25),
7061. (b) Barber, D. E.; Lu, Z.; Richardson, T.; Crabtree, R. H. Inorg. Chem.
1992, 31, 4709.
(20) Luo, X.-L.; Crabtree, R. H. J. Am. Chem. Soc. 1989, 111, 2527.
(21) (a) Willing, D. N. U.S. Patent 3,419,593, 1968. (b) Karstedt, B. D.
U.S. Patent 3775452, 1973.
(16) (a) Harrod, J. F.; Chalk, A. J. J. Am. Chem. Soc. 1964, 86, 1776.
(b) Harrod, J. F.; Chalk, A. J. ibid 1965, 87, 16. (c) Harrod, J. F.; Chalk,
A. J. ibid 1966, 88, 3491.
(17) (a) Reikhsfel’d, V. O.; Astrakhanov, M. I. Zh. Obshch. Khim. 1970,
40(3), 699. (b) Reikhsfel’d, V.; Koroleva, G. N. ibid. 1966, 36(8), 1474.
(18) For a review see: Lukevics, E.; Dzintara, M. J. Organomet. Chem.
1985, 295, 265.
(22) Chandra, G.; Lo, P. Y.; Hitchcock, P. B.; Lappert, M. F. Organo-
metallics 1987, 5, 191.

H
Hydrosilylation of Allyl Alcohol with O8M8
J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6985
where cluster catalysis becomes dominant. Although on careful
reading of the Lewis et al. paper, it appears that the presence
of polynuclear species as intermediates and/or transition states
in the catalytic cycle cannot be ruled out.
The solvent effect studies suggest that nonpolar solvents, that
is, heptane or toluene, favor C-silylation, whereas polar solvents,
that is, THF and CH2Cl2, gave 10-20% O-silylation. Electron-
donating solvents,20,28 for example, THF, may displace the
bridging olefins of di- or polynuclear Pt complexes to produce
mononuclear complexes, thus promoting O-silylation, whereas
nonpolar solvents would favor C-silylation by preserving the
di- or polynuclear species. The influence of CH2Cl2 may be
understood similarly, considering the possibility that it acts as
a ligand as found for MeI.29
Still another important aspect of this work is the fact that
C-silylation can be effected almost exclusively. Although most
O-silylation catalysts catalyze C-silylation, catalysts regiospe-
cific selective for O-silylation in the presence of CdC and
CdO moieties are known.20,30 However, efforts to develop
catalysts that exhibit selective C-silylation can be expected to
be frustrated, because of the strong preference for Si-O bond
(bond strength ∼120 kcal/mol) over Si-C bond formation (bond
strength ∼90 kcal/mol).31 For example, ion-exchange resin
immobilized H2PtCl6 catalyzed reaction of allyl alcohol with
triorganosilanes, gave a 3:2 ratio of C-:O-silylated products, the
highest ever reported.31b Another report describes selective
C-silylation of propargyl alcohol using platinum catalysts with
bulky norbornene and tri(tert-butyl)phosphine ligands.32 How-
ever, aside from the present study, there are no literature reports
on selective C-silylation of carbon-carbon double bonds in the
presence of OH groups. Note that hydrogermylation of propynol
derivatives leads to a high degree of C-germylation via a
π-complex transition state. This process may be selective
C-germylation but insufficient details are available to verify this
fact.33
Figure 5. Equilibrium between Pt2(dvs)3 and Pt(dvs)2.
The above-described studies document equilibria between
complexes of different nuclearity in the Pt(dvs) precursor
solution. In an attempt to identify the nature of the active species
in Pt(dvs) catalysis systems, Lewis et al. proposed the inter-
mediacy of colloidal platinum.25 However, recent studies suggest
that simple hydrosilylation involves a catalytic cycle wherein
the active species is monometallic.26 Hence our results appear
to be quite unusual.
Our brief examination of the effects of variations in reaction
conditions on products and reaction rates reveal some unusual
effects that deserve further discussion vis a` vis the possibility
of cluster catalysis. Rosenberg and Laine27 recently delineated
five criteria for polynuclear metal complex catalysis including:
(1) increase in turnover frequency (TF) with increases in catalyst
concentration; (2) product selectivities that are different from
known mononuclear catalyzed reactions; (3) enhanced catalyst
reactivity with the addition of a second metal; (4) enhanced
reactivity upon modification of conditions to favor metal-metal
bond formation; and (5) catalytic asymmetric induction.
In the above concentration studies, when [Pt] ) 100 and 10
mM, the reaction times were 0.5 h (RT) and g12 h (refluxing
toluene), respectively. Based on these results, the turnover
frequency {TF, [product]‚[catalyst]-1‚time-1} at [Pt] 100 mM
is g2.4 times that of the TF at [Pt] 10 mM not counting the
∼90 °C difference in reaction temperatures. This increase in
TF with increasing Pt concentration, along with the unusual
regioselectivity (C- over O-silylation) which differs from known
mononuclear Pt complex catalysis13,14 (eq 3), implies that the
direct hydrosilylation of allyl alcohol in this study involves di-
or polynuclear Pt complexes.
Given that higher Pt(dvs) concentrations favor C-silylation
under conditions that favor formation of di- or polynuclear
species, it appears that di- or polynuclear Pt complexes favor
C-silylation, while a mononuclear Pt catalyst leads to O-
silylation. The solvent effects, while preliminary, are also in
accord with di- or polynuclear Pt complex catalyzed C-silylation
of allyl alcohol.
As noted above, the Lewis et al. paper suggests that
C-silylation using Karstedt’s catalyst proceeds via mononuclear
species.26 While they do observe formation of polynuclear
species, they suggest that these polynuclear species represent a
“resting state,” not involved in the catalytic cycle. However,
the considerable structural differences between our reactants and
those used by Lewis et al. perhaps give rise to a special case
The best explanation for the above behavior is that a cluster
catalyst intermediate forms wherein OH groups are shielded
from reaction with silyl groups at the metal center but alkenes
are not. The exact form this intermediate takes is beyond
reasonable speculation based on the evidence currently available.
However, the reason for its formation might be because the
silsesquioxane, Q8M8H, is polyfunctional and may induce
formation of polynuclear catalytic intermediates at its “surface.”
A similar explanation might account for the very highly selective
and extensive hydrosilylative cross-linking achieved when
H
vinyl 3(d)
Q8M8 is reacted with Q8M8
.
If true, this change in
catalytic mechanism might be classified as arising because of
the presence of a nanostructured material. The unique role that
H
Q8M8 appears to play may also account for the discrepancy
between Lewis’ recent studies26 and our results.
A reviewer has suggested that the above results might be
explained in terms of h3 coordination of the allyl alcohol to
(28) Speier, J. L. AdVances in Organometallic Chemistry; Academic
Press: London, 1979; Vol. 17, p 407.
(29) (a) Crabtree, R. H.; Faller, J. W.; Mellea, M. F.; Quirk, J. M.
Organometallics 1982, 1, 1361. (b) Burk, M. J.; Segmuller, B.; Crabtree,
R. H. Organometallics 1987, 6, 2241.
(23) Hitchcock, P. B.; Lappert, M. F.; Warhurst, N. J. W. Angew. Chem.,
Int. Engl. 1991, 30(4), 438.
(24) Lappert, M. F.; Scott, F. P. A J. Organomet. Chem. 1995, 492, C11.
(25) (a) Lewis, L. N.; Lewis, N. J. Am. Chem. Soc. 1986, 108, 7228. (b)
Lewis, L. N.; Uriarte, R. J. Organometallics 1990, 9, 621. (c) Lewis, L. N.
J. Am. Chem. Soc. 1990, 112, 5998.
(26) Stein, J.; Lewis, L. N.; Gao, Y.; Scott, R. A. J. Am. Chem. Soc.
1999, 121, 3693.
(27) Rosenberg, E.; Laine, R. M. In Catalysis by Di- and Polynuclear
Metal Cluster Complexes; Adams, R. D., Cotton, F. A., Eds.; Wiley-VCH:
New York, 1998; Chapter 1.
(30) (a) Stille, J. K.; Fox, D. B. J. Am. Chem. Soc. 1970, 92, 2(5), 1274.
(b) Sommer, L. H.; Lyons, J. E. J. Am. Chem. Soc. 1969, 91(25), 7061. (c)
Doyle, M. P.; High, K. G.; Bagheri, V.; Pieters, R. J.; Lewis, P. J.; Pearson,
M. M. J. Org. Chem. 1990, 50, 6082.
(31) (a) Lukevits, E. A.; Voronkov, M. G. J. Heterocycl. Comp. 1965,
1(2), 116. (b) Zaslavskaya, T. N.; Reikhsfel’d, V. O.; Filippov, N. A.;
Shornik, N. A. Zh. Obshch. Khim. 1980, 50, 2001.
(32) Murphy, P. J.; Spencer, J. L.; Proctor, G. Tetrahedron Lett. 1990,
31(7), 1051.
(33) Esteruelas, M. A.; Martin, M.; Oro, L. A. Organometallics 1999,
18, 2267.

6986 J. Am. Chem. Soc., Vol. 122, No. 29, 2000
Zhang and Laine
H
Figure 6. Hydrosilylation of Q8M8 with 2-allyloxyethanol.
metal combined with both steric effects and selectivity during
reductive elimination, rather than invoking a bimetallic/cluster
catalyst explanation. While we consider these possibilities above,
the change in selectivity favoring C-silylation on going to higher
catalyst concentrations does argue for the intermediacy of
polynuclear species as we have discussed elsewhere.27 The
above results are not definitive only suggestive. Clearly, more
detailed studies must be done to confirm these statements.
III.B. Direct Hydrosilylation of Other Unsaturated Alco-
hols and Using other Si-H Compounds. III.B.1. Direct
Hydrosilylation of Allyl Alcohol with other Si-H Com-
pounds. Once the hydrosilylation of allyl alcohol was optimized,
we sought to extend this methodology to other Si-H compounds
H
including D4H, TMDS, and PDMS-H. The reaction of D4
resulted in a gel, suggesting significant simultaneous O- and
C-silylation. Reactions of TMDS and PDMS-H with allyl
alcohol gave C- and O-silylation products as the molecular
weights were higher than expected and/or exhibited a broad
H
distribution. It appears that Q8M8 is a very unique substrate
for this type of direct hydrosilylationsas suggested above.
Again, the influence of Si-H substrates on this reaction needs
further study.
III.B.2. Direct Hydrosilylation of Other Alcohols. The
reactions of Q8M8H with propargyl alcohol and 2-allyloxyethanol
were also examined. Reaction with propargyl alcohol resulted
in a product with a broad molecular weight distribution,
suggesting cross-linking through both C- and O-silylation.
However, when 2-allyloxyethanol was used, the reaction gave
exclusive C-silylation, resulting in OH functionalized cube, with
a longer tether between the OH and the cube than in OHPS
(Figure 6). The success of this reaction makes it possible to
attach methacrylates with different spacer lengths to cubes,
hence tailor the microstructure of the resultant nanocomposites
after polymerization.
III.C. Characterization of OHPS. OHPS characterization
data including FTIR, 1H and 13C NMR, and elemental analysis
(EA) are all those expected. 29Si NMR in CH2Cl2 gives Si(OSi)4
at -108.4 ppm, and Si(OSi)C3 at ∼13.6 ppm, in accord with
that expected based on the work of Sellinger et al.3(a),3(c)
However, 29Si NMR data in methanol showed an unusual, small
peak around -100.4 ppm, besides the corresponding peaks for
Si(OSi)4 and Si(OSi)C3 (Figure 7). This small peak was absent
from both OSPS and OMPS, whichever solvent was used.
Furthermore, when the desilylation of OSPS to OHPS was
followed by 29Si NMR in methanol, the small peak at -100.4
ppm showed up gradually along with the desilylation product.
The appearance of this peak is directly related to the CH2CH2-
CH2OH structure and also depends on the nature of the solvent.
Hasegawa and co-workers report that the (SiO)3SiOH struc-
ture formed from incomplete silylation of octaanion species
Si8O208- appears characteristically at -99.5 to -100.5 ppm,6,34
bracketing the -100.4 ppm peak seen above. This coincidence
Figure 7. 29Si NMR of OHPS in CH2Cl2 (a), OHPS in methanol (b),
OSPS in methanol (c), and OMPS in methanol (d).
Scheme 3. Backbiting in OHPS
led us to tentatively attribute the peak at -100.4 ppm to
(SiO)3SiOH species. This in turn suggested a back-biting process
per Scheme 3.
To prove the above process occurs, we sought to detect the
formation of 2,2-dimethyl-1-oxa-2-silacyclopentane in our
system. The literature indicates that this cyclic compound forms
via dehydration of 1,3-bis(hydroxypropyl)tetramethyldisiloxane
on heating, presumably by backbiting of OH on Si (Scheme
4).35 Pure 2,2-dimethyl-1-oxa-2-silacyclopentane easily poly-
merizes to HO[(CH3)2Si(CH2)3O]nH (Scheme 4).36
The following set of experiments was designed to detect the
formation of 2,2-dimethyl-1-oxa-2-silacyclopentane from OHPS.
1H NMR spectra of OHPS/CD3OD and OHPS/CDCl3 solutions
were compared. Interestingly, while the Si(CH3)2 peak on the
cube is a singlet in CDCl3, it splits into two peaks in CD3OD,
a broad one at 0.12 ppm and a sharp one at 0.05 (Figure 8).
The broad peak disappears after 2 days, coincident with increase
of the sharp one.
(35) Speier, J. L.; David, M. P.; Eynon, B. A. J. Org. Chem. 1960, 26,
1637.
(34) Hasegawa, I. Trends Organomet. Chem. 1994, 1, 131.
(36) Gu, T.-Y. Y.; Weber, W. P. J. Am. Chem. Soc. 1980, 102, 1641.

H
Hydrosilylation of Allyl Alcohol with O8M8
J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6987
Figure 8. 1H NMR of OHPS in CDCl3 (a) and CD3OD for 10 min (b) and 2 days (c).
Scheme 4. Formation and Polymerization of
Scheme 5. Transsilylation for OSPS and
2,2-Dimethyl-1-oxa-2-silacyclopentane37
2,2-Dimethyl-1-oxa-2-silacyclopentane35
When CD3OD was removed from the sample, the residue
left was partially soluble. The soluble component was analyzed
using gas chromotography-mass spectral (GC-MS) analysis.
GC-MS showed one major peak beside the solvent peak, with
a mass spectrum that corresponds to that reported for 2,2-
dimethyl-1-oxa-2-silacyclopentane (m/z) parent ion P 116 (13%),
P-1 (1.7%), P-15 (100%), P-17 (16.9%), P-28 (22.4%).36 Thus,
the soluble part of the residue from the NMR sample was 2,2-
dimethyl-1-oxa-2-silacyclopentane. The insoluble component
most likely consists of polymers formed by condensation of
the second product of backbiting, Si8O12(OH)8 (T3OH), as shown
in eq 5.
While the OHPS is very stable as a dry powder, it turns into
viscous oil in methanol solution after a few days, and GPC
analysis indicates a significant amount of high molecular weight
species. This phenomenon can also be explained by the
backbiting process discussed above. As a dry powder, the OH
groups on OHPS are likely immobilized by H-bonding between
OHs, and thus are unable to backbite. On dissolution in
methanol, H-bonding with methanol competes with that between
the propanol groups and provides the propanol groups with the
mobility essential for backbiting to occur.
The same observation was made in ethanol. However, in
aprotic solvents, such as THF, CH3CN, or acetone, no backbiting
1
was observed by H NMR at room temperature.
OHPS melts at ∼84 °C and decomposes (by TGA) at ∼225
°C. The TGA ceramic yield (to SiO2, not shown) in air (51.0%)
is much lower than expected (64.8%), probably due to a loss
of some of the 2,2-dimethyl-1-oxa-2-silacyclopentane (bp 97-
100 °C),36 which occurs when the OH groups obtain sufficient
mobility in the melt. If all the side chains were lost via
backbiting, the expected ceramic yield of SiO2 would be 43.2%.
III.D. Hydrosilylation of Protected Allyl Alcohol. When
allyloxytrimethylsilane was used instead of allyl alcohol, the
reaction goes exclusively via C-silylation as expected. The
isolated and purified OSPS is a white solid. Deprotection by
stirring in methanol gives OHPS as a white powder after
workup.37 Its NMR, TGA, and elemental analyses are virtually
the same as those for OHPS obtained via direct hydrosilylation.
Considering the backbiting of OH groups on OHPS as shown
above, methanol is not a good solvent for the desilylation of
Returning to the 1H NMR of OHPS in CD3OD (Figure 8), it
appears that the broad Si(CH3)2 peak likely results when some
of the OSiCH2CH2CH2OH side chains cyclize to coincidentally
generate Si8O12(OH)x[OSi(CH3)2CH2CH2CH2OH]8-x, while the
sharp Si(CH3)2 peak is from 2,2-dimethyl-1-oxa-2-silacyclo-
pentane. Spectrum (c) in Figure 8 correlates well with that
reported for 2,2-dimethyl-1-oxa-2-silacyclopentane.35,36
To sum up, the formation of 2,2-dimethyl-1-oxa-2-silacyclo-
pentane in OHPS in methanol was firmly established. Thus the
backbiting mechanism proposed in Scheme 3 most likely occurs.
Likewise, the small peak at ∼-100.4 ppm in the 29Si NMR
(Figure 7) is most likely due to the presence of Si(OSi)3OH
components on cubes Si8O12(OH)x[OSi(CH3)2CH2CH2CH2OH]8-x
,
formed as a result of backbiting.

6988 J. Am. Chem. Soc., Vol. 122, No. 29, 2000
Zhang and Laine
Scheme 6
OSPS. However, backbiting is not a problem in this study, as
the OHPS obtained from OSPS is virtually the same as that
from the direct hydrosilylation. Furthermore, we find that almost
all the OHPS in 1H NMR sample (∼5 mg/mL) tube converts to
2,2-dimethyl-1-oxa-2-silacyclopentane and Si8O12(OH)8 (Figure
8), while only a small portion in a 29Si NMR sample (∼300
mg/mL) undergoes backbiting as revealed by the small peak at
-100.4 ppm (Figure 7). On the basis of these observations, the
proposed backbiting in Scheme 3 is likely an equilibrium
reaction. Indeed, 2,2-dimethyl-1-oxa-2-silacyclopentane is known
to ring-open polymerize,35,36 which supports this possibility. As
a result, at low concentration of OHPS (e.g., 1H NMR sample),
the equilibrium shifts toward the right, while at higher concen-
tration (e.g., 29Si NMR sample), the resultant silanolcube and
2,2-dimethyl-1-oxa-2-silacyclopentane push the reaction more
toward the left, and backbiting is not very evident.
The OSPS TGA ceramic yield in air (SiO2, not shown),
46.2%, is lower than the expected value, 69.9%, calculated based
on the assumption that all the Si is oxidized to SiO2. This
discrepancy can be explained by the loss of Si through a
transsilylation per eq 6 (Scheme 5), similar to that responsible
for the transsilylation of 2,2-dimethyl-1-oxa-2-silacyclopentane
(eq 7).38 According to eq 6, one-third of the Si might be lost as
2,2-dimethyl-1-oxa-2-silacyclopentane, instead of being oxidized
to SiO2. As a result, the expected ceramic yield to SiO2 would
be 46.6%. The found ceramic yield of 46.2% (by TGA) is close
to the expected value, and seems to confirm the transsilylation
process shown in eq 6.
Preliminary polymerization studies were conducted using
(1R)-(-)-camphorquinone (0.1 wt %) as a photoinitiator, and
(N,N-dimethylamino)ethyl methacrylate (0.13 wt %) as an
accelerator. The liquid OMPS cured readily under blue light (λ
) 460 nm) within one minute to give hard, transparent and
insoluble composites. We will report the mechanical properties
of nanocomposites prepared from OMPS and copolymers with
other methacrylates soon (Scheme 6).40
IV. Conclusions
(1) We report here the first example of regioselective
hydrosilylation of allyl alcohol at the CdC. Hydrosilylation of
allyl alcohol with Q8M8H was discovered to provide C-silylation
at CdC, which provides one-step access to OHPS, and hence
an efficient route to OMPS. The success of the reaction with
2-allyloxyethanol suggests a broad spectrum of OH function-
alized cubes can be synthesized using this facile methodology,
and thus provides access to a wide variety of macromonomers
incorporating ester functionality on the periphery of octasil-
sesquioxane cores, such as the octamethacrylatesilsesquioxane.
In another words, OHPS provides a basis for the development
of polyester nanocomposite precursors.
(2) Even though the catalytic Pt species are not identified in
this study, a change of nuclearity of the Pt complex is suggested.
Further kinetic and mechanistic studies are needed to exploit
this type of selective C- over O-silylation.
(3) Back-biting of OH on Si in OHPS and the simultaneous
formation of 2,2-dimethyl-1-oxa-2-silacyclopentane were identi-
fied, which correctly explains the 29Si NMR data, TGA ceramic
yield of OHPS, and its instability in protic solutions.
(4) Octakis(3-methacryloxypropyldimethylsiloxy)octasilses-
quioxane (OMPS) is easily prepared directly from OHPS.
Previous unwanted polymerization that accompanies hydrosi-
lylation of allyl or propargyl methacrylates was eliminated.
Preliminary polymerization studies show that OMPS readily
polymerizes under UV light (λ ) 460 nm) to give a cross-linked
nanocomposite.
III.E. OMPS Prepared from OHPS. OHPS prepared either
via direct silylation or via the silylated allyl alcohol route was
used to prepare OMPS (eqs 8 and 9). The final product, OMPS,
was isolated as a slightly viscous liquid at room temperature
(viscosity ≈ 100 centipoise).39 Unlike the OHPS, OMPS is
stable in solution and its GPC determined molecular weight
distribution (PDI ) 1.06) is close to that of OHPS (PDI ) 1.04).
This suggests that no significant oligomerization of methacry-
lates, or decomposition of the cube occurs during the preparation
of OMPS.
DSC of the OMPS shows a broad exotherm with an onset at
∼176 °C on the first heating cycle, which disappears in
subsequent heating and cooling traces. This exotherm is certainly
associated with the polymerization of the methacrylate groups.
Acknowledgment. We thank NIDR for the generous support
through Grant No. DE09839-03. We thank Professor John R.
Wiseman and Professor M. David Curtis for helpful discussions,
and Dr. Valerie A. Lee and Dr. Steven Rasmussen for help with
the UV-curing study. We would like to thank the reviewers for
their comments, suggestions, and proofreading.
(37) Knoth, W. H. Jr.; Lindsey, R. V. Jr. J. Am. Chem. Soc. 1958, 80,
4106.
(38) (a) Marchand, A.; Gerval P.; Massol, M.; Barrau, J. J. Organomet.
Chem. 1974, 74, 209. (b) Massol, M.; Barrau, J.; Satge, J.; Bouyssieres, B.
ibid 1974, 80, 47.
JA000318R
(40) Cost, R. O. R.; Dias Filho, N. L.; Choi, J.; Laine, R. M. Work in
progress.
(39) Private communication from Dr. Walter Zimbeck.