Cyclization phenomena in the Sol-Gel polymerization of α,ω-bis(triethoxysilyl)alkanes and incorporation of the cyclic structures into network silsesquioxane polymers

Article abstract of DOI:10.1021/ja982751v

Intramolecular cyclizations during acid-catalyzed sol-gel polymerizations of α,ω-bis(triethoxysilyl)-alkanes substantially lengthen gel times for monomers with ethylene (1), propylene (2), and butylene (3) bridging groups. These cyclization reactions were found, using mass spectrometry and ²⁹Si NMR spectroscopy, to lead preferentially to monomeric and dimeric products based on six- and seven-membered disilsesquioxane rings. 1,2-Bis(triethoxysilyl)ethane (1) reacts under acidic conditions to give a bicyclic dimer (5) that is composed of two annelated seven-membered rings. Under the same conditions, 1,3-bis(triethoxysilyl)propane (2), 1,4-bis(triethoxysilyl)butane (3), and Z-1,4-bis(triethoxysilyl)but-2-ene (10) undergo an intramolecular condensation reaction to give the six- and seven-membered cyclic disilsesquioxanes 6, 7, and 11. Subsequently, these cyclic monomers slowly react to form the tricyclic dimers 8, 9, and 12. With NaOH as polymerization catalyst, these cyclic silsesquioxanes readily reacted to afford gels that were shown by CP MAS ²⁹Si NMR and infrared spectroscopies to retain some cyclic structures. Comparison of the porosity and microstructure of xerogels prepared from the cyclic monomers 6 and 7 with those of gels prepared directly from their acyclic precursors 2 and 3 indicates that the final pore structure of the xerogels is markedly dependent on the nature of the precursor. In addition, despite the fact that the monomeric cyclic disilsesquioxane species cannot be isolated from 1-3 under basic conditions due to their rapid rate of gelation, spectroscopic techniques also detected the presence of the cyclic structures in the resulting polymeric gels.

Full text of DOI:10.1021/ja982751v

J. Am. Chem. Soc. 1999, 121, 5413-5425
5413
Cyclization Phenomena in the Sol-Gel Polymerization of
R,ω-Bis(triethoxysilyl)alkanes and Incorporation of the Cyclic
Structures into Network Silsesquioxane Polymers
Douglas A. Loy,*,† Joseph P. Carpenter,† Todd M. Alam,§ Raef Shaltout,†
Peter K. Dorhout,‡ John Greaves, James H. Small,| and Kenneth J. Shea*,
Contribution from the Catalysts Department and the Aging and Reliability of Bulk Materials Department,
Sandia National Laboratories, Albuquerque, New Mexico 87185-1407, Departments of Chemistry,
Colorado State UniVersity, Fort Collins, Colorado 80523, and UniVersity of California,
IrVine, California 92717-2025, and Polymers and Coatings Group, MST-7, Los Alamos National
Laboratory, Los Alamos, New Mexico 87545
ReceiVed August 3, 1998
Abstract: Intramolecular cyclizations during acid-catalyzed sol-gel polymerizations of R,ω-bis(triethoxysilyl)-
alkanes substantially lengthen gel times for monomers with ethylene (1), propylene (2), and butylene (3) bridging
groups. These cyclization reactions were found, using mass spectrometry and 29Si NMR spectroscopy, to lead
preferentially to monomeric and dimeric products based on six- and seven-membered disilsesquioxane rings.
1,2-Bis(triethoxysilyl)ethane (1) reacts under acidic conditions to give a bicyclic dimer (5) that is composed
of two annelated seven-membered rings. Under the same conditions, 1,3-bis(triethoxysilyl)propane (2), 1,4-
bis(triethoxysilyl)butane (3), and Z-1,4-bis(triethoxysilyl)but-2-ene (10) undergo an intramolecular condensation
reaction to give the six- and seven-membered cyclic disilsesquioxanes 6, 7, and 11. Subsequently, these cyclic
monomers slowly react to form the tricyclic dimers 8, 9, and 12. With NaOH as polymerization catalyst, these
cyclic silsesquioxanes readily reacted to afford gels that were shown by CP MAS 29Si NMR and infrared
spectroscopies to retain some cyclic structures. Comparison of the porosity and microstructure of xerogels
prepared from the cyclic monomers 6 and 7 with those of gels prepared directly from their acyclic precursors
2 and 3 indicates that the final pore structure of the xerogels is markedly dependent on the nature of the
precursor. In addition, despite the fact that the monomeric cyclic disilsesquioxane species cannot be isolated
from 1-3 under basic conditions due to their rapid rate of gelation, spectroscopic techniques also detected the
presence of the cyclic structures in the resulting polymeric gels.
Introduction
polyhedral oligomers such as the octamethoxysilsesquioxane
cube,9 which may act as one of the primary building blocks in
silica sol-gels. Similarly, the hydrolysis and condensation
chemistry of silsesquioxanes, RSiO1.5, is heavily influenced by
cyclizations. Polymerization of trialkoxysilanes often leads to
soluble polymers4,10 due to the formation of numerous cyclic
structures in lieu of branches or cross-links. With sterically
demanding substituents for R, discrete polyhedral oligosilses-
quioxanes can be isolated in high yield.11,12
While studying the synthesis of a new class of hybrid organic/
silica materials, a related yet unique cyclization process was
discovered. Organically bridged polysilsesquioxanes, prepared
from the hydrolysis and condensation of triethoxysilanes linked
by a hydrocarbon spacer, have been synthesized with a wide
variety of organic groups incorporated in the polymer.13-21 Acid-
or base-catalyzed polymerization of R,ω-bis(triethoxysilyl)-
Cyclization phenomena in the sol-gel synthesis of silica and
polysilsesquioxanes play an integral role in the polymerization
process and in the development of the network architecture. Six-
and eight-membered rings have been identified in the hydrolysis
and condensation of tetramethoxysilane and tetraethoxysilane.1-6
Larger rings, which are probably present, have been more
difficult to detect.7,8 These two-dimensional rings undergo
further condensation reactions to generate, at least in some cases,
† Catalysts Department, Sandia National Laboratories.
§ Aging and Reliability of Bulk Materials Department, Sandia National
Laboratories.
‡ Department of Chemistry, Colorado State University.
Department of Chemistry, University of California, Irvine.
| Polymers and Coatings Group, MST-7, Los Alamos National Labora-
tory.
(1) Sanchez, J.; McCormick, A. V. J. Non-Cryst. Solids 1994, 167, 289-
294.
(9) Agaskar, P. A.; Day, V. W.; Klemperer, W. G. J. Am. Chem. Soc.
(2) Brunet, F.; Cabane, B. J. Non-Cryst. Solids 1993, 163, 211.
(3) Capozzi, C. A.; Pye, L. D.; Condrate, R. A., Sr. Mater. Lett. 1992,
15, 130.
(4) Ng, L. V.; Thompson, P.; Sanchez, J.; Macosko, C. W.; McCormick,
A. V. Macromolecules 1995, 28, 6471-6476.
(5) Van Beek, J. J.; Seykens, D.; Jansen, J. B. H.; Schuiling, R. D. J.
Non-Cryst. Solids 1991, 134, 14-22.
(6) Van Beek, J. J.; Seykens, D.; Jansen, J. B. H. J. Non-Cryst. Solids
1992, 146, 111-120.
(7) Peace, B. W.; Mayhan, K. G.; Montle, J. F. Polymer 1973, 14, 420.
(8) Kelts, L. W.; Armstrong, N. J. J. Mater. Res 1989, 4, 423-433.
1987, 109, 5554.
(10) Baney, R. H.; Itoh, M.; Sakakibara, A.; Suzuki, T. Chem. ReV. 1995,
95, 1409-1430.
(11) Feher, F. J.; Budzichowski, T. A. J. Organomet. Chem 1989, 373,
153-163.
(12) Voronkov, M. G.; Lavrent’yev, V. I. Top. Curr. Chem 1982, 102,
199-236.
(13) Shea, K. J.; Loy, D. A.; Webster, O. W. Chem. Mater 1989, 1,
572-574.
(14) Shea, K. J.; Webster, O.; Loy, D. A. Mater. Res. Soc. Symp. Proc.
1990, 180, 975-980.
10.1021/ja982751v CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/25/1999

5414 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
Scheme 1. Polymerization and Cyclization Reactions of
R,ω-Bis(triethoxysilyl)alkanes
Scheme 2. Cyclization of R,ω-Bis(triethoxysilyl)alkanes
1-3, 10, and 1,3-Bis(diethoxymethylsilyl)propane 13 To
Give Cyclic Monomers 6-11, Bicyclic 5, and Tricyclics 8,
9, 12, and 15
alkanes normally proceeds to highly condensed rigid gels within
a few hours.21 However, when the alkylene bridge is ethylene
(1), propylene (2), or butylene (3), gelation under acidic
conditions requires 2-6 months. Upon closer examination, it
was discovered that these monomers were undergoing intra-
molecular cyclization reactions to form six- and seven-
membered rings (Scheme 1) that are designated as “cyclic
disilsesquioxanes”.22 Ring closure to form cycloalkanes is well
known to be governed by both the kinetic and the thermo-
dynamic demands of cyclization, generally favoring the forma-
tion of five- and six-membered rings.23 Similarly, dihalodi-
methylsilanes are known to preferentially cyclooligomerize to
eight- and ten-membered siloxane rings.1,24,25 Since cyclic
disilsesquioxanes can be considered hybrids of the hydrocarbon
and siloxane ring systems, it is not surprising that cyclizations
to six- and seven-membered rings are the most favorable in the
alkylene-bridged series.
Naturally, the fate of these cyclic disilsesquioxanes was of
interest with regard to the final molecular structure of the
resulting polymeric networks and gels. By monitoring the
hydrolysis and condensation reactions of the bridged monomers
1-3, 10, and 13 with mass spectrometry and 29Si NMR
spectroscopy, it was possible to detect not only the monomeric
cyclic disilsesquioxanes but also a variety of oligomeric products
based predominantly on six- and seven-membered rings. By
performing the reactions with substoichiometric quantities of
water, we were able to isolate and characterize the monocyclic
(6, 7, 11, 14), bicyclic (5), and tricyclic (8, 9, 12, 15)
silsesquioxanes formed during these reactions (Scheme 2).
Furthermore, we were able to demonstrate, using infrared and
29Si CP MAS NMR spectroscopies, that polymerization of the
cyclic disilsesquioxane monomers could be accomplished
without significant ring-opening to afford a new class of
polysilsesquioxane materials. The discovery of cis syndiotactic
stereochemistry about the ring junctions of the tricyclic dimers
8, 9, 12, and 15 suggests that it might be possible to prepare
ladder-like polysilsesquioxanes structures as well.
Results and Discussion
Gelation of monomers with two or more triethoxysilyl groups
has been shown to occur as quickly as a matter of minutes,
even at monomer concentrations as low as 0.2 M.16,20,21 The
ease of this type of gelation, relative to that of tetraethoxysilane
under identical conditions, has been attributed to the higher
degree of functionality and additional connectivity built into
these monomers with the organic bridging group.16 With base-
catalyzed polymerization of R,ω-bis(triethoxysilyl)alkanes, ge-
lation is rapid, regardless of the length of the bridging alkylene
group; if cyclic species are forming, they are not interfering
with the growth of polymers capable of forming gels. Under
acidic conditions, the length of the bridging group had a
remarkable effect on the gel times (Figure 1). Gelation occurred
after months for monomers 1-3 but occurred after only hours
for monomers with bridging alkylenes greater than six carbons
in length. Gelation of the latter would readily occur, even with
only 2 equiv of water. It was this disparity in gelation times
that gave us the first indication that different reaction manifolds
could be accessed on the basis of the length of the alkylene
bridging group.22 The fact that bridging alkylenes longer than
six carbons gave no evidence of soluble cyclic monomers and
oligomers in either mass spectrometric or 29Si NMR studies,
coupled with their short gel times, indicated that acyclic and
intermolecular condensation reactions predominate.
(15) Shea, K. J.; Loy, D. A.; Webster, O. W. Polym. Mater. Sci. Eng.
1990, 63, 281-285.
(16) Shea, K. J.; Loy, D. A.; Webster, O. J. Am. Chem. Soc. 1992, 114,
6700-6710.
(17) Loy, D. A.; Shea, K. J.; Russick, E. M. Mater. Res. Soc. Symp.
Proc. 1992, 271, 699-704.
(18) Corriu, R.; Moreau, J.; Thepot, P.; Wong Chi Man, M. Chem. Mater.
1992, 4, 1217-1224.
(19) Corriu, R. J. P.; Moreau, J. J. E.; Thepot, P.; Wong Chi Man, M.;
Chorro, C.; Lere-Porte, J. P.; Sauvajol, J. L. Chem. Mater 1994, 6, 640-
649.
(20) Small, J. H.; Shea, K. J.; Loy, D. A. J. Non-Cryst. Solids 1993,
160, 234-246.
(21) Oviatt, H. W., Jr.; Shea, K. J.; Small, J. H. Chem. Mater. 1993, 5,
943-950.
(22) Loy, D. A.; Carpenter, J. P.; Myers, S. A.; Assink, R. A.; Small, J.
H.; Greaves, J.; Shea, K. J. J. Am. Chem. Soc. 1996, 118, 8501-8502.
(23) Illuminati, G.; Mandolini, L. Acc. Chem. Res. 1981, 14, 95.
(24) Rochow, E. G.; Gilliam, W. F. J. Am. Chem. Soc. 1941, 63, 798.
(25) Cypryk, M.; Sigwalt, P. Macromolecules 1994, 27, 6245-6253.
Dimerization-Cyclization Reactions of Monomer 1. Eth-
ylene-bridged monomer 1 gelled immediately under base-
catalyzed sol-gel conditions but required 1 month for gelation
under acidic conditions. It was suspected that the prolonged gel
times were due to a cyclization process that was similar to that
first observed with the propylene- and butylene-bridged mono-
mers.22 However, we were not certain whether the five-

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5415
Figure 1. Gelation times for R,ω-bis(triethoxysilyl)alkanes (0.4 M in
ethanol, 6 H2O), 10.8 mol % HCl or NaOH catalyst).
Scheme 3. Different Reaction Paths for the Hydrolysis and
Condensation of 1,2-Bis(triethoxysilyl)ethane 1
Figure 2. 29Si NMR spectra for 1 M solutions of 1,10-bis(trieth-
oxysilyl)decane in ethanol. The top spectrum is that of the unreacted
monomer. The next spectrum is that of the product of the acid-catalyzed
hydrolysis of the monomer with 1 (middle) and 2 (bottom) equiv of
water. Spectra were obtained after 4 h; chromium acetyl acetonate (5
mmol/L) was added to each solution just before analysis. T0 refers to
the monomer and its hydrolysis products. T1 refers to the peaks from
silicons with a single siloxane bond (dimer or end-groups) and their
hydrolysis products. T2 refers to peaks from silicons with two siloxane
bonds.
reactions can be seen near -60 ppm. After 1 h, the solution of
1,10-bis(triethoxysilyl)decane and 2 equiv of water formed a
gel. In comparison, the spectra of the ethylene-bridged monomer
1 and its hydrolysis and condensation products are quite different
(Figure 3). Hydrolysis affords a more complex pattern of peaks
downfield from the monomer resonance (δ ) -45.9) due to
long-range substituent effects.27 However, the most notable
difference between the spectra for the sol-gel polymerization
of 1 and those for 1,10-bis(triethoxysilyl)decane is in the
distribution of T1 and T2 peaks. Here, two peaks (δ ) -50.7
and -53.1) appear and remain near equal intensity, even at
greater extent of reaction. The chemical shift for 4, synthesized
by hydrogenating 2,2,5,5-tetraethoxy-2,5-disilaoxacyclopent-
3-ene (Scheme 4), lies downfield (δ ) -41.3) from the
resonance due to 1.28-30 Only a small peak near -41 ppm can
be observed in the spectra shown in Figure 3.
A mass spectrum of an identical solution of 1 after 10 min
of reaction shows that most of the monomer has been converted
into the bicyclic dimer or its hydrolysis products (Figure 4).
The unhydrolyzed bicyclic dimer has parent ions at m/z ) 487
([5 + H]+) and 504 ([5 + NH4]+). Ions from the hydrolysis
products of the dimer are present (m/z ) 476 and 459 for the
singly hydrolyzed dimer, 431 and 448 for the doubly hydrolyzed
membered cyclic disilsesquioxane (4), from the intramolecular
condensation reaction of 1, would be too strained to participate
or if larger rings, based on dimers of 1, would dominate the
early stages of the sol-gel chemistry (Scheme 3). Examination
of the solutions before gelation with 29Si NMR demonstrated
that the expected pattern of resonances from monomer hydroly-
sis and condensation products was not in evidence.
Typically in sol-gel chemistry, the 29Si NMR chemical shift
of peaks moves progressively downfield in increments between
1 and 3 ppm per ethoxide group replaced with a hydroxide and
upfield 5-10 ppm with each additional siloxane bond to the
silicon atom.26 To illustrate this, a series of 29Si NMR spectra
for 1,10-bis(triethoxysilyl)decane before hydrolysis (top), mono-
mer treated with 1 equiv of water (middle), and monomer treated
with 2 equiv of water (bottom), both with aqueous HCl as
catalyst, is shown in Figure 2. The group of peaks labeled T0
(middle) represents the unreacted monomer (δ ) -45.3) and
singly (δ ) -43.2) and doubly hydrolyzed (δ ) -41.2)
monomer. A second group of peaks (T1) appears upfield from
the T0 resonances, arising from silsesquioxanes with a single
siloxane bond with no hydrolyzed ethoxide groups (δ ) -52.9)
and with one ethoxide hydrolyzed to a silanol (δ ) -51.2).
The third spectrum (bottom) shows that the peaks due to the
monomer and other T0 silicons have been reduced in intensity,
while the T1 resonances have grown to be the major contributors
to the spectrum. T2 silicons from additional condensation
(27) Meyers, S. A.; Assink, R. A.; Loy, D. A.; Shea, K. J. Manuscript
in preparation.
(28) Seyferth, D.; Robison, J. Macromolecules 1993, 26, 407.
(29) Englehardt, G.; Jancke, H.; Magi, M.; Pehk, T.; Lippmaa, E. J.
Organomet. Chem. 1971, 28, 293.
(30) Burton, D. J.; Harris, R. K.; Dodgson, K.; Pellow, C. J.; Semlyen,
J. A. Polym. Commun. 1983, 24, 278.
(26) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: the physics and
chemistry of sol-gel processing, 1st ed.; Academic Press: San Diego, CA,
1990.

5416 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
Scheme 5. Reaction Pathway for the Hydrolysis and
Condensation of 1, Showing Intermediates (1a,b) Detected
by Mass Spectrometry
was not observed in the mass spectrum; the analogous 2,2,5,5-
tetramethyl-2,5-disila-1-oxacyclopentanes are significantly strained
(8-12 kcal/mol) and readily undergo ring-opening polymeri-
zation.29 It is unlikely that such a strained species would be
present under sol-gel polymerization conditions.
The dimer was identified as 5 on the basis of the 29Si NMR
data (Figure 3) and by isolating and characterizing the pure
compound from the reaction mixture; the product was shown
to have two silicon resonances at δ ) -50.3 and -52.9. T1
silicons in 5a would be expected to give rise to a peak with a
chemical shift close to that observed for 4 (δ ) -41.3). The
29Si NMR of 5 would also have T1 and T2 resonances, but
because they would be in seven-membered rings, the peaks
should lie farther upfield of the monomer and downfield of
acyclic T1 and T2 resonances (Figure 2). Furthermore, when 1
was reacted with only 1 equiV of water under acidic conditions,
intermediates in the formation of 5 were observed in the mass
spectrum (Scheme 5). Within the first hour of reaction, the major
ions in what should be the dimer region for 1 were masses at
m/z ) 589 ([1a - OEt]+), 561, and 515, which are consistent
with the acyclic dimer 1a along with ions attributable to 5 and
its hydrolysis products. The next intermediate in the formation
of 5 would result from the first of two intramolecular cycliza-
tions of the acyclic dimer to yield 1b. The presence of 1b was
supported by the presence of the parent ion with m/z ) 561
([1b + H]+) and its fragmentation product at m/z ) 515 ([1b
- OEt]+). After several hours, the ions due to intermediates 1a
and 1b had diminished, and only peaks assigned to 5 and its
hydrolysis products remained. Only traces of ions with masses
consistent with the five-membered disilsesquioxane ring were
observed.
Figure 3. 29Si NMR spectra for 1 M solutions of monomer 1 in ethanol.
The top spectrum in each series is that of the unreacted monomer. The
next spectra are those of the product of the acid-catalyzed hydrolysis
of each monomer with 1 (middle) and 2 (bottom) equiv of water. Spectra
were obtained after 4 h; chromium acetyl acetonate (5 mmol/L) was
added to each solution just before analysis. T0 refers to the monomer
and its hydrolysis products. T1cyclic and T2cyclic refer to the peaks due to
silicons with a single siloxane bond and two siloxane bonds, respec-
tively, that are part of a seven-membered disilsesquioxane ring, such
as in compound 5.
Once formed, 5 was slow to react, even in the presence of
excess water. Oligomers larger than dimers were not observed
in the mass spectrum. The reaction of 2 equiv of water with 1
was monitored for several hours, and the only significant change
in the mass spectrum was an increase in the amount of
hydrolyzed dimer, as hydrolysis products of 5 account for the
major ions in the spectrum (m/z ) 448, 420, and 374).
Apparently, the relatively slow condensation chemistry of 5
serves as a bottleneck in the acid-catalyzed polymerization of
1. The limited amount of water employed in the experiment
cannot account for the paucity of higher oligomers, since
extensive hydrolysis of 5 is detected in the mass spectrum. The
impact of the cyclization of the ethylene-bridged monomer on
polymerization and gelation is highlighted by the ease with
which 1,10-bis(triethoxysilyl)decane forms a gel under identical
conditions.
Figure 4. Mass spectrum (CI, NH3) of the hydrolysis and condensation
products of 1,2-bis(triethoxysilyl)ethane 1 with 1 equiv of water under
acidic conditions. The products are predominantly the bicyclic dimer
4 and its hydrolysis products.
Scheme 4. Preparation of 4 by Palladium-Catalyzed
Hydrogenation of
2,2,5,5-Tetraethoxy-2,5-disilaoxacyclopent-3-ene
Cyclization Reactions of Monomers 2 and 3. Polymeriza-
tion of monomers 2 and 3 (0.4 M in ethanol) with 6 equiv of
water and HCl (10 mol %) as catalyst gave transparent, colorless
solutions that took over 6 months to gel. Analyses of these
solutions indicated the presence of the six (6)- and seven (7)-
dimer, and 403 and 420 for the triply hydrolyzed dimer). Two
structures (5 and 5a) for the dimer are consistent with the
observed masses and the hydrolysis and condensation chemistry
of alkoxysilanes. The five-membered cyclic disilsesquioxane 4

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5417
Figure 5. 29Si NMR spectra for 1 M solutions of monomer 1,3-bis-
(triethoxysilyl)propane 2 in ethanol. The top spectrum is that of the
unreacted monomer. The next spectra are those of the product of the
acid-catalyzed hydrolysis of each monomer with 1 (middle) and 2
(bottom) equiv of water. Spectra were obtained after 4 h; chromium
acetoacetate (5 mmol/L) was added to each solution just before analysis.
T1cyclic refers to the peak due to the cyclic disilsesquioxane 6 that formed
from the intramolecular condenstion of 2. T2cyclic refers to the peak due
to the tricyclic dimer 8.
Figure 6. 29Si NMR spectra for 1 M solutions of monomer 1,4-bis-
(triethoxysilyl)butane 4 in ethanol. The top spectrum is that of the
unreacted monomer. The next spectra are those of the product of the
acid-catalyzed hydrolysis of each monomer with 1 (middle) and 2
(bottom) equiv of water. Spectra were obtained after 4 h; chromium
acetoacetate (5 mmol/L) was added to each solution just before analysis.
T1cyclic refers to the peak due to the cyclic disilsesquioxane 7 that formed
from the intramolecular condenstion of 3. T2cyclic refers to the peak due
to the tricyclic dimer 9.
membered cyclic disilsesquioxane monomers, respectively. In
the 29Si NMR of the reaction of 2 with 1 equiv of water under
acidic conditions, a single peak assignable to the T1 silicon
(δ ) -47.2) in 6 appeared between where T0 (monomer; δ )
-45.8) and T1 resonances are typically expected (Figure 5).
Hydrolysis and condensation of monomer 3 under identical
conditions gave a single resonance (δ ) -49.1) for 7 (Figure
6) that was shifted slightly farther upfield from that of 6, but
still farther downfield than would be expected for an acyclic or
unstrained T1 resonance. Conversions to 6 and 7 were quantita-
tive with 1 equiv of water, and the cyclic products could be
isolated in good yields from the reaction mixtures by simple
distillation. Intramolecular cyclization under acidic conditions
was also observed with 1,4-bis(triethoxysilyl)but-2-ene (10) to
give the unsaturated cyclic disilsesquioxane 11, and with 1,3-
bis(diethoxymethylsilyl)propane (13) to give the cyclic disilox-
ane 14 as a mixture of its syn and anti diastereomers.
The reversibility of the cyclization reaction under sol-gel
conditions will impact how well the cyclic structures are
incorporated into polymeric architectures. It has been shown
with tetraalkoxysilanes that the condensation reaction is es-
sentially irreversible in these types of sol-gels.31 Because
acyclic species were not detected in the mass spectra and 29Si
NMR during the formation of 6 and 7, we prepared an
isotopically labeled cyclic disilsesquioxane, 6*. If the cyclic
species was equilibrating with an acyclic one, isotopic exchange
between the 17O/18O-labeled cyclic disilsesquioxane and un-
labeled water would be detected. 17O/18O-labeled cyclic disils-
esquioxane 6* was prepared from 2 using isotopically labeled
water (H2*O; *O ) 42% 18O and 27% 17O). The reaction of 6*
with 1 N aqueous HCl, which was monitored by mass
spectrometry, showed a gradual decrease in 6* (m/z ) 296 and
297) with a concomitant increase in unlabeled 6 (m/z ) 295)
as unlabeled water was incorporated. These observations imply
that a small amount of acyclic product was generated from 6*
in the presence of water. 17O NMR spectroscopy of the same
sample also revealed the single peak from the labeled 6* (δ )
64.0) gradually being replaced by low-frequency resonances due
to labeled silanols (δ ) 30.7) and labeled water (δ ) -5.7).
Loss of 17O from 6* is consistent with exchange due to the
reversibility of the cyclization reaction.
The stability of the six- and seven-membered cyclic disilses-
quioxanes toward ring-opening is important for understanding
their impact on the polymerization of the propylene- and
butylene-bridged monomers. Treatment of neat 6, 7, and 14 with
catalytic amounts of Lewis acidic (BF3‚OEt2) or basic (Bu4-
NOH) catalysts had no effect, indicating that insufficient ring
strain was present to facilitate ring-opening polymerizations.
Synthesis and Characterization of Tricyclic Dimers. When
2 or 3 was reacted with 2 equiV of water for several hours, new
resonances at δ ) -50.4 and -53.5, respectively, slowly
(31) Prabakar, S.; Assink, R. A. J. Non-Cryst. Solids 1997, 211, 39-48.

5418 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
1
Figure 7. Expanded views of the H resonances due to the R and â
hydrogens on the cyclic disilsesquioxane ring in 7, revealing 2JHH and
3JHH couplings.
Scheme 6. Preparation of Tricyclic 9 by
Palladium-Catalyzed Hydrogenation of Unsaturated Tricyclic
12
appeared in the 29Si NMR. These resonances fall between where
acyclic or unstrained cyclics with T1 and T2 silicons would be
expected to appear (Figures 5 and 6). Even after 3 weeks of
acid-catalyzed hydrolysis of 6 with 6 equiv of water, gelation
was not observed, and the 29Si NMR spectrum was dominated
by the resonance at δ ) -50.4. In the mass spectra of identical
solutions, peaks with masses consistent with dimers of 6 and 7
were observed for each monomer. The dimers were tentatively
assigned as tricyclic compounds 8 and 9 (m/z ) 458 [M + NH4]
and 486 [M + NH4], respectively).
Verification of these tricyclic structures was achieved by
careful hydrolysis of the cyclic disilsesquioxanes 6 and 7 to
give the pure tricyclics 8 and 9, respectively. However, the most
efficient route was to prepare 8 and 9 directly from the acyclic
monomers 2 and 3. Complete conversion of the monomers to
dimers required several days in refluxing ethanol or THF with
2 equiv of water and HCl as catalyst. Tricyclic dimer 8 was
isolated as a crystalline solid directly from the reaction in
ethanol. Dimer 9 was isolated as a crude oil, and all attempts
to crystallize it from the reaction mixture failed, probably due
to the high solubility of the dimer and to the additional flexibility
in the structure imparted by the longer alkylene groups. It was
possible, however, to prepare pure 9, still as an oil, by
hydrogenating its unsaturated analogue (12; Scheme 6). Pure,
crystalline tricyclic dimer 12 was obtained directly from the
acid-catalyzed hydrolysis of Z-1,4-bis(triethoxysilyl)but-2-ene
(10) with 2 equiv of water.
Figure 8. ORTEP plots of the X-ray crystal structure of the tricyclic
dimer 8, showing the anti configuration of the six-membered cyclic
disilsesquioxane rings about the central eight-membered tetrasiloxane
ring. The methyl carbon {C(5)} in 8 was modeled at two positions
with a site occupancy factor of 0.72(4) for C(5A) and 0.28 for C(5B).
This was done because the anisotropic thermal ellipsoid was too large
to be appropriate for one site. Packing within the structure suggests
that the ends of the ethyl groups are free to move in an interpenetrating
fashion between the planes of the molecules, which would allow them
to rotate due to thermal motion, giving rise to the disorder.
3
3
and obtaining coupling constants. The large JXA and JXA′ of
13.1 Hz is indicative of an axial-to-axial coupling, while the
smaller methylene 3JHH is consistent with axial-to-equatorial and
equatorial-to-equatorial couplings. These coupling constants
indicate that the two six-membered rings in the tricyclic system
are in the expected chair conformation. The 1H NMR spectrum
of 15 exhibited chemical shifts and coupling constants for
methylene groups in its propylene bridge that were nearly
identical to those observed for 8.
While the structures of tricyclic dimers 8, 12, and 15 could
be assigned from NMR and mass spectral data, the configuration
(syn or anti) of the dimers was determined by X-ray crystal-
lographic studies.32 The structures consist of two cyclic mono-
mer units linked in an anti configuration about a distorted eight-
membered siloxane ring (Figure 8). None of the syn stereoisomer
was isolated from any of the reactions. This is in good agreement
with the calculated MM2 structures33 that predicted that the anti
diastereomer would be more stable than the syn form by 7.55
Formation of bridged tricyclic structures appears to be
common to both silsesquioxane and siloxane monomers with
three or four carbon bridging groups, as the same series of
cyclization reactions was also observed for 13. Tricyclic dimer
15, a crystalline white solid, was of interest because it possesses
no ethoxide groups, making it a relatively unreactive model
1
compound. H NMR spectra of 8, 12, and 15 are consistent
with the rigid tricyclic structure. Compounds 8 and 15 have
similar spectra which are best described as AA′BB′MX. The
(32) X-ray quality crystals of dimers 8, 12, and 15 were grown by
sublimation of the pure solids in a vacuum. The structure and atomic
numbering scheme for 8 is displayed in Figure 8, and the bond lengths (Å)
and bond angles (deg) for this structure are given in the Supporting
Information.
1
portion of the H NMR spectrum arising from the propylene
bridge of 8, shown in Figure 7, was simulated, and decoupling
experiments were performed to assist in making assignments

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5419
since the formation of cyclics under these conditions resulted
in gel times of several months (Figure 1). Under basic
conditions, monomers 1-3 and 6-8 polymerized to rigid, free-
standing gels in less than 24 h. Gels prepared from the acyclic
monomers were generally cloudy or fully opaque, while the
cyclic monomers polymerized to give colorless, transparent gels.
All of the dried gels or xerogels were obtained as glassy white
fragments.
There is no reason to expect that intramolecular condensation
of monomers 1-3 to form cyclic disilsesquioxanes does not
occur in the base-catalyzed polymerizations. Unfortunately, it
is much more difficult to study the initial hydrolysis and
condensation products under basic sol-gel conditions. With
sodium hydroxide catalysis, the first hydrolysis and condensation
reactions are the slowest, while the rates of subsequent hy-
drolysis and condensation reactions increase with the extent of
the reactions.35 Consequently, the initial condensation products
are very short-lived and difficult to observe spectroscopically.
Xerogels 1X-3X and 6X-8X were studied by CP MAS 29Si
NMR and infrared spectroscopies in attempts to determine if
the cyclic structures of 6-8 were present in these polymers.
Characterization of Polysilsesquioxanes. (a) 29Si CP MAS
NMR Spectroscopic Analyses. Because 29Si NMR chemical
shifts are sensitive to variations in the bond angles at the silicon
atom, this analytical method is a powerful tool for differentiating
between cyclic and acyclic structures. As the bond angles
become smaller in the cyclic systems, the chemical shift of the
silicon-29 resonance moves downfield. If the cyclic structures
are incorporated into the final polysilsesquioxanes, there should
be a distinct difference in the 29Si CP MAS NMR spectra of
these materials and those in which the formation of the small
cyclic structures does not occur. The solid-state CP MAS 29Si
NMR spectra of alkylene-bridged polysilsesquioxanes usually
contain three distinguishable peaks which are assigned to the
T1, T2, and T3 silicons present in the polymer.21 A typical
spectrum observed for these materials is shown in Figure 10
for a polymer prepared from 1,10-bis(triethoxysilyl)decane as
well as the spectra for the polymers 1X-3X and 6X-8X. The
decylene-bridged monomer does not form the small organosi-
loxane rings observed in 1, 2, and 3, and the T2 and T3 peaks
for this polymer are observed at -59.3 and -66.8 ppm,
respectively. However, the spectra for the polymers prepared
from the ethylene-, propylene-, and butylene-bridged monomers,
whether the monomer was cyclic or acyclic, contain a single,
broad, unresolved peak. The T3 resonance is generally the largest
peak in the 29Si NMR spectra of polysilsesquioxanes prepared
from monomers with hexylene or longer bridging groups, and
it is observed between -66 and -67 ppm. It is expected that
the unresolved peaks for the polymers prepared from 1-3 and
6-8 arise mostly from T3 silicons since they are also prepared
under basic conditions. These peaks are shifted significantly
downfield from the typical alkylene-bridged material. We
interpret this downfield shift as an indication of the presence of
the cyclic disilsesquioxane structures in the polysilsesquioxanes
1X-3X and 6X-8X.
Figure 9. Molecular models (MM2, Chemdraw 3D Plus) of the anti
and syn diastereomers of the tricyclic dimer 7 with calculated steric
energies.
kcal/mol (Figure 9). The anti configuration of the bridging
groups is the same orientation displayed in cis syndiotactic
ladder structures that have often been proposed for soluble
polysilsesquioxanes.34 While the tricyclic dimer may represent
the beginning of a ladder polymer structure, continued growth
of the polymer would be severely limited by slow hydrolysis
and condensation at the remaining ethoxysilyl functionalities;
no discrete, higher oligomers have been isolated. X-ray crystal-
lographic studies of 12 and 15 disclosed that these dimers have
virtually the same structural features as those described for 8.
Synthesis of Polysilsesquioxanes. Finding that monomers
1-3 form substantial amounts of cyclic monomers and dimers
early in the acid-catalyzed sol-gel polymerization process raises
the following question: Are cyclic structures incorporated in
the resulting xerogels? To answer this, polysilsesquioxanes were
prepared by the sol-gel polymerization of monomers 1-3 and
6-8 with aqueous NaOH as the catalyst. Formation of pol-
ysilsesquioxane gels using an acidic catalyst was not practical
The poor resolution in the 29Si NMR spectra of the polymers
prepared from 1-3 and 6-8 may be explained by a combination
of phenomena. The presence of cyclic structures in the xerogels
adds to the variety of chemical environments in which the T2
and T3 silicons may be found, compared to the simple acyclic
structures of longer alkylene-bridged materials. Also, there may
be some ring opening occurring during the polymerization of
(33) Using MM2 with ChemDraw Pro Version 4.0: Allinger, N. L. J.
Comput. Chem. 1993, 14, 755.
(34) Cis syndiotactic ladder structures have been proposed to account
for the solublity of polysilsesquioxanes which, based on the number of
siloxane bonds to each silicon, should be cross-linked and insoluble. See:
Brown, J. F. Jr. J. Polym. Sci. Part C 1963, 83. It is likely, based on careful
viscosity studies, that the polymers are highly branched structures with a
significant contribution of cyclic structures. See: Frye, C. L.; Collins, W.
T. J. Am. Chem. Soc. 1970, 92, 5586.
(35) McCormick, A. In Sol-Gel Processing and Applications; Attia, Y.
A., Ed.; Plenum Press: New York, NY, 1994; p 3.

5420 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
Table 1. Infrared ν(Si-O) Bands of Monomers and Xerogels in
the Range of 1200-950 cm-1
monomer
ν(Si-O), cm-1
xerogel
ν(Si-O), cm-
1
2
3
4
5
6
7
8
1168, 1148, 1106, 1078
1168, 1104, 1083
1168, 1104, 1083
1169, 1082, 952
1169, 1106, 1007
1169, 1105, 1084, 1006
1169, 1106, 1083, 1034
1110, 1072, 1024, 1001
1X
2X
3X
1090, 1030
1112, 1024
1127, 1054
5X
6X
7X
1115, 1018
1127, 1039
1111, 1008
cyclic and acyclic siloxanes were used to help characterize the
network architectures of xerogels 1X-3X and 6X-8X.
The prominent absorption bands in the infrared spectra in
the range 1200-1000 cm-1 for both monomers and xerogels
are listed in Table 1. The cyclic monomers, 4 and 6-8, exhibit
a very strong absorption band around 1034-950 cm-1 which
is conspicuously absent in the spectra of the acyclic monomers.
These low-frequency absorptions are consistent with ν(Si-O)
absorptions in small cyclic structures. Most of the xerogels also
show absorption bands in this range, indicative of the presence
of the cyclic structures. A polymer prepared from 1,10-bis-
(triethoxysilyl)decane, which cannot form strained disilsesqui-
oxane rings, has a very strong absorption band at 1122 cm-1
with a shoulder at 1050 cm-1 and no low-frequency absorptions
in the range of 1034-950 cm-1. This spectrum is typical of
alkylene-bridged polysilsesquioxanes which have bridging
groups of six carbons or longer.
A band at 1030 cm-1 is observed in the xerogel 1X, which
is very close to that of the seven-membered rings in 5 (1011
cm-1) and 7 (1034 cm-1). This observation suggests that the
seven-membered disilsesquioxane rings observed for 1 under
acid-catalyzed hydrolysis and condensation conditions are also
formed under basic conditions. Xerogels prepared from the
propylene-bridged monomers 2, 6, and 8 have absorption bands
at 1024, 1018, and 1008 cm-1, respectively. These bands are
also in the low-frequency range, indicative of the presence of
cyclic structures in the final polymers. The band for 2X occurs
at a slightly higher frequency than those for 6X or 8X, which
may be due to the presence of a greater fraction of acyclic
propylene-bridged moieties. Furthermore, the absorption for 8X
is shifted to a lower frequency than that for 6X, which is
expected if the more highly strained tricyclic structure 8 is
present to a greater extent in 8X than in 6X. The absorption of
1054 cm-1 for 3X is in the range (1040-1080 cm-1) of
ν(Si-O) bands expected for acyclic structures. Thus, under base-
catalyzed conditions, infrared analysis suggests that ring closure
of 3 and incorporation of the cyclic structure into the final
polymer does not seem to predominate as it did for 2. However,
Figure 10. Comparison of the solid-state 29Si CP MAS NMR spectrum
of a decylene-bridged polysilsesquioxane (top) with spectra of xerogels
(1X-3X) obtained from monomers 1-3 that are capable of forming
cyclics and with spectra of xerogels (6X-8X) prepared from cyclic
monomers 6-8.
6-8, and ring closure may not be quantitative during the
polymerization of 1-3, yielding polymer networks that are
composed of both cyclic and acyclic structures.
(b) Infrared Spectroscopic Analyses. Infrared spectroscopy
was also utilized to probe the solid-state structure of the
polysilsesquioxanes. Siloxanes typically exhibit one or more
very strong bands in the region 1130-1000 cm-1 for
ν(Si-O).36 The location of this absorption band in the infrared
spectrum is influenced by the structure of the siloxane. For
example, when the siloxane bonds are confined to a small cyclic
structure such as [R2SiO]3, the ν(Si-O) occurs at a low
frequency of 1020-1010 cm-1, while larger cyclics, [R2SiO]4-5
are observed at 1090-1075 cm-1 36-38
This same phenomenon
,
.
is observed in the series of cyclic organosiloxane compounds
2,2,5,5-tetramethyl-2,5-disilaoxacyclopentane, 2,2,6,6-tetra-
methyl-2,6-disilaoxacyclohexane, and 2,2,7,7-tetramethyl-2,7-
disilaoxacycloheptane, which exhibit ν(Si-O) bands at 920, 987,
and 1020 cm-1, respectively.36,39 All of these Si-O bands
involved in the small cyclic structures are observed at lower
frequencies than the acyclic siloxanes R3SiOSiR3, which show
xerogel 7X has a strong absorption band at 1039 cm-1
,
indicating that the cyclic structure of the starting monomer is
at least partially preserved in the final polymer. Clearly, the
choice of acyclic or cyclic silsesquioxane monomer does have
an influence on the contribution of cyclic structures in the
resulting xerogels.
(c) Surface Area Analyses. The structures of the poly-
silsesquioxanes were also studied by nitrogen sorption poro-
simetry and scanning electron microscopy to determine what
effect the structures of the starting monomers have on the surface
areas and porosity of the final polymers. All of the materials
were found to have high surface areas, ranging from 590 m2/g
for 1X to 880 m2/g for 3X (Table 2). Most of the porosity of
the xerogels resides in pores with average diameters in the lower
end of the mesoporous domain (mesopore diameter range )
Si-O absorption bands in the range of 1040-1080 cm-1 36
. We
observed a similar trend in the SiO bands for cyclic disilses-
quioxanes. These differences in the absorption bands between
(36) Launer, P. J. Infrared Analysis of Organosilicon Compounds. Silicon
Compounds Register and ReView I; Petrarch Systems: Levittown, PA, 1987;
p 70.
(37) Richards, R. E.; Thompson, H. W. J. Chem. Soc. 1949, 124.
(38) Wright, N.; Hunter, M. J. J. Am. Chem. Soc. 1947, 69, 803.
(39) Tamao, K.; Kumala, M.; Takahashi, T. J. Organomet. Chem. 1975,
94, 367.

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5421
Table 2. Nitrogen Sorption Porosimetry Studiesaof Polymers
1X-3X and 6X-8X
surface area
(m2/g)
mean pore
diameter (Å)
pore volume
(cm3/g)
polymer
1X
2X
3X
6X
7X
8X
593
712
882
807
868
785
101
81
38
32
52
32
1.50
1.45
0.84
0.64
1.12
0.63
a Surface areas were determined by BET, mean pore diameters by
BJH theory, and pore volumes determined by single-point analysis.
20-500 Å).40 The average pore diameters and pore volumes
for the polymers prepared from 1-3 decrease as the alkylene
bridge is lengthened.
The effect of the cyclic versus acyclic monomer structure on
the porosity of the final polysilsesquioxanes was examined in
greater detail through the incremental pore volume plots41 for
each polymer. These plots, derived from the desorption branch
of the nitrogen isotherm, reveal the distribution of pore sizes
and the extent to which they contribute to the pore volume. At
each step during the desorption phase of the porosimetry
experiment, the volume of liquid nitrogen removed from the
pores of the material is measured, and a corresponding pore
diameter is calculated from the relative pressure using the
Barrett-Joyner-Halenda (BJH) method.42 The distribution of
the pore sizes in the material can be observed by plotting the
measured pore volumes versus the calculated pore diameters.
Incremental pore volume plots for the propylene-bridged
polysilsesquioxanes, prepared from the acyclic monomer 2 and
the cyclic monomers 6 and 8, exhibit a stark difference in their
pore size distributions (Figure 11). In 2X, contributions to the
pore volume are observed from pores with diameters that extend
over the entire mesopore range. However, xerogels 6X and 8X
have very narrow pore size distributions, with the pore diameters
lying mostly between 30 and 45 Å. Since all three polymers
were prepared using identical conditions, these differences must
originate from the cyclic or acyclic structures of the starting
monomers. This same trend is not observed in the butylene-
bridged polysilsesquioxanes, as the porosimetry data and pore
volume plots for 3X and 7X are very similar. Both polymers
have fairly narrow pore size distributions, with a range of pore
diameters of 30-50 Å for 3X and 40-100 Å for 7X (Figure
12).
Figure 11. Incremental pore volume distribution plots of xerogels (2X,
6X, and 8X) prepared by base-catalyzed sol-gel polymerizations of
the acyclic monomer 2, cyclic disilsesquioxane 6, and tricyclic dimer
8, respectively.
The differences in porosity between 2X and xerogels 6X and
8X are reflected in the scanning electron micrographs of the
polymers. Xerogel 2X appears to have a very coarse morphol-
ogy, with the macrostructure of the polymer being comprised
of large globular particles ranging in size from 50 to several
hundred nanometers. These structural features result in an open
framework that is responsible for the larger mesopores and
macropores (“macropore” ) diameter > 500 Å) in the polymer.
Xerogels 6X and 8X have smoother and more continuous
morphologies with more compact structures, eliminating the
larger pores.
Conclusions
Novel cyclization phenomena during the synthesis of alky-
lene-bridged polysilsesquioxanes were discovered for monomers
Figure 12. Comparison of the incremental pore volume distribution
plots for xerogels prepared from the butylene-bridged monomer 3 and
its cyclic disilsesquioxane derivative 7.
(40) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area, and Porosity,
2nd ed.; Academic Press: London, 1982.
(41) Incremental pore volume plots (pore volume versus pore size) were
prepared from the BJH desorption pore size distribution data.48
(42) Barrett, E. P.; Joyner, L.; Halenda, P. P. J. Am. Chem. Soc. 1951,
73, 373.
with short alkylene bridges of 2-4 carbons (1-3). These cyclic
products were found to inhibit the polymerization of the

5422 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
resonance of solid glycine (δ ) 176.0). Since the gels in this study are
similar, these parameters were optimized on only one of the gels to
obtain spectra with a satisfactory signal-to-noise ratio. 29Si NMR spectra
were acquired using single-pulse excitation with a relaxation delay of
480 s. With 29Si T1 values ranging from 19 to 135 s, the relaxation
delay is g3T1, which is sufficient for quantitative spectra. 29Si CP MAS
NMR spectra were deconvoluted using a Lorentz-Gaussian (50:50)
fit. Mass spectra were obtained on a Fisons Autospec (CI, ammonia).
Monomer purity was determined by GC with an HP Series II 5890
and a packed column with HP-1 (cross-linked methyl siloxane), 15 m
× 0.32 mm. Microscopy studies were performed using a Hitachi S4500
field emission scanning electron microscope. The samples were coated
with 100-200 Å of chromium using a Gatan model 681 high resolution
ion beam coater. Secondary electron images were taken using 5-kV
accelerating voltage. The images were acquired digitally from the SEM
using a PGT Imix imaging system. Porosimetry measurements were
determined using a Quantachrome Autosorb-6 multiport nitrogen
porosimeter.
Monomer Preparation: 1,2-Bis(triethoxysilyl)ethane (1). A flask
was charged with vinyltriethoxysilane (45.0 g, 237 mmol), triethox-
ysilane (44.5 g, 271 mmol), benzene (200 mL), and chloroplatinic acid
(130 mg, 0.320 mmol). The flask was fitted with a drying tube loaded
with Drierite such that the reaction was exposed to dry air. The mixture
darkened after a few minutes, and it was allowed to stir for 24 h.
Benzene was removed in vacuo, leaving a dark liquid that was vacuum
distilled (0.025 Torr, 86-88 °C), yielding a clear, colorless liquid (63.7
g, 76%). 1H NMR (300 MHz, C6D6): δ ) 3.76 (q, 12H, OCH2CH3, J
) 7.0 Hz), 1.13 (t, 18H, OCH2CH3, J ) 7.0 Hz), 0.89 (s, 4H,
SiCH2CH2Si, J ) 8.2 Hz). 13C NMR (75 MHz, C6D6): δ ) 58.5 (OCH2-
CH3), 18.5 (OCH2CH3), 2.5 (SiCH2CH2Si). 29Si NMR (99 MHz,
EtOH): δ ) -45.8. IR (neat): ν ) 2975, 2927, 2887, 1483, 1443,
1409, 1391, 1366, 1295, 1168, 1147, 1103, 1083, 957, 781 cm-1. HRMS
(CI, isobutane): calcd for C14H35O6Si2 [M + H]+, 355.1972; found,
355.1972.
monomers under acidic sol-gel conditions. The ethylene-
bridged monomer formed an unusual bicyclic dimer (5) which
consisted of two fused seven-membered organosiloxane rings.
Monomers with three- or four-carbon bridges underwent a series
of intramolecular cyclizations to form a fused tricyclic dimeric
product. Many of the cyclic products, including some the
tricyclic dimers, were isolable. These cyclic products were
shown to be reactive monomers for the base-catalyzed sol-gel
synthesis of polysilsesquioxanes that incorporate the ring
structures into the network architecture. Solid-state 29Si NMR
and infrared spectroscopies indicate that cyclic structures were
preserved in the final polymers.
It was also discovered that the structures of the monomers
(cyclic or acyclic) can have a profound influence on the
porosities and morphologies of the sol-gel materials derived
from them. The synthesis of propylene-bridged polysilsesqui-
oxanes from the acyclic monomer 2 resulted in polymers with
a very broad pore size distribution, spanning several hundred
nanometers. However, when the propylene-bridged materials
are prepared from the cyclic monomers 6 and 8, the polymers
exhibit very narrow pore size distributions of just a few
nanometers. Both the acyclic and cyclic butylene-bridged
monomers yield materials with fairly narrow pore size distribu-
tions in the lower end of the mesopore regime, between 20 and
100 Å. Manipulation of the structures of the starting monomers
in polysilsesquioxane syntheses may be an important technique
for creating materials with well-defined mesoporous domains.
Experimental Section
General Methods. All reactions were run under dry argon using
standard Schlenk techniques unless otherwise noted. Ethanol was dried
and distilled off magnesium turnings under argon, and benzene was
distilled from calcium hydride. 3-Triethoxysilylpropene, 4-triethoxy-
silylbutene, and triethoxysilane were obtained from Geleste and were
used as received. Chloroplatinic acid was purchased from Aldrich and
was kept stored under argon until needed. Isotopically enriched water
(27% 17O, 42% 18O) was purchased from ISOTEC, Inc. 1,10-Bis-
(triethoxysilyl)decane was prepared by chloroplatinic acid-catalyzed
hydrosilation of 1,9-decadiene with triethoxysilane and was spectro-
scopically identical to the report in the literature.21 Z-1,4-Bis(trichlo-
rosilyl)but-2-ene was prepared by following a procedure used to prepare
the analogous E-isomer.43
Infrared spectra were obtained on a Perkin-Elmer 1750 Fourier
transform infrared spectrophotometer. 1H, 13C, and 29Si solution NMR
spectra were recorded on Bruker AM 300 and DRX 400-MHz
spectrometers. 1H NMR spectra of 8 and 11 were recorded on a DRX
400 at a resonant frequency of 400.16 MHz, using standard single-
pulse sequences on a 5-mm BB probe. The relaxation delay was 5 s,
with 8K-32K complex points. Experiments were performed at 298
K, using 16-32 scans. Iterative simulations of the 1H spectra of 8 and
11 were performed using commercial software (ACORN NMR) on a
PC to make peak assignments and to determine coupling constants.
17O NMR experiments were conducted with an AMX 400 spectrometer
at a resonant frequency of 54.25 MHz, using a 5-mm BB probe. The
relaxation delay was 200 ms, with 32-64 scan averages as a function
of time. Residual solvent peaks and tetramethylsilane were used as
internal references for 1H and 13C spectra. An external tetramethylsilane
(δ ) 0.0) sample was used as the reference for 29Si NMR spectra, and
an external H2O (δ ) 0.0) sample was used to reference 17O spectra.
Solid-state 13C and 29Si CP MAS NMR spectra were obtained with
a Bruker AMX 400-MHz spectrometer at 100.63 and 79.5 MHz,
respectively, and were acquired with magic angle spinning (MAS)
speeds of ∼5 and ∼3-5 kHz, respectively. 13C NMR spectra were
acquired using cross-polarization (CP) with a relaxation delay of 1 s
and a CP time of 2 ms. 13C referencing was performed on the carbonyl
1,3-Bis(triethoxysilyl)propane (2).22 A flask was charged with
3-(triethoxysilyl)propene (50.0 g, 245 mmol), triethoxysilane (48.3 g,
294 mmol), benzene (100 mL), and chloroplatinic acid (0.10 g, 0.24
mmol). The flask was fitted with a drying tube loaded with Drierite
such that the reaction was exposed to dry air. While being stirred for
24 h, the solution gradually turned yellow-brown. Additional chloro-
platinic acid (0.10 g, 0.24 mmol) was added to the reaction mixture
and stirred for another 24 h. Benzene was removed in vacuo, leaving
a dark liquid that was distilled (0.025 Torr, 85-90 °C) to yield a clear,
colorless liquid (52 g, 58%). 1H NMR (300 MHz, C6D6): δ ) 3.78 (q,
12H, OCH2CH3, J ) 7.0 Hz), 1.84 (m, 2H, SiCH2CH2CH2Si), 1.15 (t,
18H, OCH2CH3, J ) 7.0 Hz), 0.85 (t, 4H, SiCH2CH2CH2Si, J ) 8.2
Hz). 13C NMR (75 MHz, C6D6): δ ) 58.4 (OCH2CH3), 18.5
(OCH2CH3), 17.2 (SiCH2CH2CH2Si), 15.0 (SiCH2CH2CH2Si). IR
(neat): ν ) 2975, 2928, 2884, 1484, 1444, 1391, 1366, 1341, 1295,
1229, 1168, 1078, 960, 900, 820, 774, 710 cm-1 29Si NMR (99 MHz,
.
EtOH): δ ) -45.8. HRMS (CI, isobutane): calcd for C15H37O6Si2
[M + H]+, 369.2129; found, 369.2113.
1,4-Bis(triethoxysilyl)butane (3).21,22 To a solution of 4-(triethoxy-
silyl)butene (25.0 g, 114 mmol), triethoxysilane (24.3 g, 148 mmol),
and benzene (100 mL) was added chloroplatinic acid (0.10 g, 0.24
mmol). Within 10 min, the solution turned yellow, and an exothermic
reaction occurred. After the solution was stirred for 48 h, chloroplatinic
acid (0.05 g, 0.12 mmol) was again added, and the mixture was refluxed
for 20 h. The benzene was removed in vacuo. Distillation (95-105
°C, 0.050 Torr) afforded a clear, colorless liquid (21.1 g, 48%). The
analytical data were in agreement with previously published data for
this compound.21 1H NMR (300 MHz, C6D6): δ ) 3.78 (q, 12H, OCH2-
CH3, J ) 7.0 Hz), 1.84 (m, 4H, SiCH2CH2CH2CH2Si), 1.15 (t, 18H,
OCH2CH3, J ) 7.0 Hz), 0.85 (t, 4H, SiCH2CH2CH2CH2Si, J ) 8.2
Hz). 13C NMR (75 MHz, C6D6): δ ) 58.4 (OCH2CH3), 26.7
(SiCH2CH2CH2CH2Si), 18.6 (OCH2CH3), 10.9 (SiCH2CH2CH2CH2Si).
IR (neat): ν ) 2975, 2928, 2884, 2765, 1444, 1391, 1366, 1295, 1226,
1201, 1168, 1104, 1083, 1006, 958, 861, 820, 795, 761 cm-1
.
2,2,5,-Tetraethoxy-2,5-disilaoxacyclopentane (4). To a solution of
2,2,5,5-tetraethoxy-2,5-disilaoxacyclopent-3-ene44 (470 mg, 1.70 mmol)
and benzene-d6 in a 250-mL glass pressure vessel was added palladium
(43) Damrauer, R.; Simon, R.; Laporterie, A.; Manuel, G.; Park, Y. T.;
Weber, W. P. J. Organomet. Chem. 1990, 391, 7-12.

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5423
on carbon (10 wt %, 10 mg). The solution was thoroughly degassed,
and the vessel was pressurized with hydrogen to 50 psig. The mixture
was stirred for 4 h at room temperature. After the vessel was
depressurized and the residual hydrogen purged, the solution was filtered
through a 0.25-µm filter and the solvent removed in vacuo to give a
clear, colorless oil (4, 260 mg, 54.6%). 1H NMR (400 MHz, C6D6): δ
) 3.85 (q, 8H, J ) 7.1 Hz, OCH2CH3), 1.17 (t, 12H, J ) 7.1 Hz,
OCH2CH3), 0.91 (s, 4H, SiCH2CH2Si). 13C NMR (75 MHz, C6D6): δ
) 59.3 (OCH2CH3), 19.0 (OCH2CH3), 5.4 (SiCH2CH2Si). 29Si NMR
(99 MHz, EtOH): δ ) -41.2. IR (neat): ν ) 2976, 2929, 2894, 1484,
Tricyclic Dimer (8). A flask was charged with 2 (42.5 g, 115 mmol),
ethanol (180 mL), and 1 N HCl (4.14 mL, 230 mmol) dissolved in
ethanol (100 mL). The mixture was stirred for 24 h and then refluxed
for 5 days. The solvent was removed in vacuo and the viscous oil was
cooled to -20 °C to produce a white solid. The solid was filtered and
recrystallized from ethanol yielding white crystals. After filtration, the
crystals were washed with cold ethanol and dried in vacuo (8, 2.74 g,
10.8%, mp 83-84 °C). 1H NMR (300 MHz, C6D6): δ ) 3.83 (q, 8H,
OCH2CH3, J ) 6.9 Hz), 1.97 (dtt, 2H, SiCH2CHaxHeqCH2Si, J ) 13.6,
13.1, 2.8 Hz), 1.83 (dtt, 2H, SiCH2CHaxHeqCH2Si, J ) 13.6, 6.0, 4.4
Hz), 1.16 (t, 12H, OCH2CH3, J ) 7.0 Hz), 0.89 (ddd, 4H, SiCHeqHax-
CH2, J ) 15.2, 6.0, 2.8 Hz), 0.63 (ddd, 4H, SiCHeqHaxCH2, J ) 15.2,
13.1, 4.4 Hz). 13C NMR (75 MHz, C6D6): δ ) 58.5 (OCH2CH3), 18.4
(OCH2CH3), 16.3 (SiCH2CH2CH2Si), 12.3 (SiCH2CH2CH2Si). 29Si
NMR (79.5 MHz, C6D6): δ ) -50.8 (T2cyclic). IR (neat): ν ) 2967,
2917, 2879, 1413, 1333, 1260, 1214, 1102, 1052, 1026, 997, 932, 901,
840, 801, 744, 662, 598 cm-1. HRMS (CI, isobutane): calcd for
C14H33O8Si4 [M]+, 440.1174; found, 440.1171.
Tricyclic Dimer (9). A solution of 15 (500 mg, 1.1 mmol) in benzene
(50 mL) with 10% palladium on carbon (20 mg) was treated with
hydrogen (50 psig) for 24 h. After the solution was filtered through
Celite and the solvent removed in vacuo, 9 was isolated (360 mg, 70%).
1H NMR (400 MHz, C6D6): δ ) 3.87 (q, 8H, J ) 7.0 Hz, OCH2CH3),
1.68 (m, 8H, SiCH2CH2CH2CH2Si), 1.20 (t, 12H, J ) 7.0 Hz,
OCH2CH3), 0.92 (s, 8H, SiCH2CH2CH2CH2Si). 13C NMR (75 MHz,
C6D6): δ ) 58.9 (OCH2CH3), 24.2 (SiCH2CH2CH2CH2Si), 18.9
(OCH2CH3), 13.2 (SiCH2CH2CH2CH2Si). 29Si NMR (79.5 MHz,
C6D6): δ ) -53.5 (T2). IR (neat): ν ) 2975, 2927, 1456, 1392, 1212,
1170, 1117, 1045, 963, 848, 792, 757, 613 cm-1. HRMS (CI,
isobutane): calcd for C16H37O8Si4 [M + 1]+, 469.1556; found,
469.1556.
1445, 1392, 1296, 1226, 1169, 1082, 952, 829, 780, 718 cm-1
.
Bicyclic Dimer (5). To a solution of 1 (23.5 g, 66.3 mmol) in ethanol
(120 mL) was added 1 N HCl (2.36 mL, 131 mmol of H2O) dissolved
in ethanol (20 mL). The reaction was stirred for 24 h, and the solvent
was removed in vacuo, leaving a viscous oil. The oil was distilled (0.05
Torr, 110-120 °C), yielding a clear, colorless liquid (5, 93% by GC,
2.33 g, 7.2%). 1H NMR (300 MHz, C6D6): δ ) 3.82 (m, 12H, OCH2-
CH3), 1.13 (m, 18H, OCH2CH3), 0.98 (m, 4H, SiCH2CH2Si). 13C NMR
(75 MHz, C6D6): δ ) 58.7 (Si(T2)OCH2CH3), 58.5 (Si(T1)OCH2CH3),
18.5 (OCH2CH3), 18.4 (OCH2CH3), 4.4 (Si(T2)CH2CH2Si(T1)), 4.2
(Si(T2)CH2CH2Si(T1)). 29Si NMR (99 MHz, EtOH): δ ) -50.3
(T1cyclic), -52.9 (T2cyclic). IR (neat): ν ) 2975, 2886, 1392, 1268, 1169,
1103, 1079, 1055, 1011, 960, 757, 736, 697, 662 cm-1. HRMS (CI,
isobutane): calcd for C16H39O9Si4 [M + H]+, 487.1671; found,
487.1663.
2,2,6,6-Tetraethoxy-2,6-disilaoxacyclohexane (6).22 To a solution
of 2 (15.0 g, 40.7 mmol) in ethanol (150 mL) was added 1 N HCl
(0.733 mL, 40.7 mmol of H2O) dissolved in ethanol (50 mL). After
the solution was stirred for 24 h, the ethanol was removed in vacuo.
The remaining liquid was distilled (0.05 Torr, 47-55 °C), yielding a
clear, colorless liquid (6, 9.67 g, 81%). 1H NMR (300 MHz, C6D6): δ
) 3.85 (q, 8H, OCH2CH3, J ) 7.0 Hz), 1.78 (m, 2H, SiCH2CH2CH2-
Si), 1.23 (t, 12H, OCH2CH3, J ) 7.0 Hz), 0.71 (t, 4H, SiCH2CH2CH2-
Si, J ) 3.8 Hz). 13C NMR (75 MHz, C6D6): δ ) 58.3 (OCH2CH3),
18.2 (OCH2CH3), 17.2 (SiCH2CH2CH2Si), 11.2 (SiCH2CH2CH2Si).
29Si NMR (99 MHz, neat): δ ) -47.2 (T1cyclic). IR (neat): ν ) 2977,
2923, 2889, 1442, 1296, 1246, 1219, 1168, 1106, 1084, 1007, 968,
898, 821, 786, 705 cm-1. HRMS (CI, ammonia): calcd for C11H26O5-
Si2 [M]+, 294.1319; found, 294.1322.
17O/18O-Labeled (6*). To a solution of 2 (19.3 g, 52.4 mmol) in
ethanol (100 mL) was added H2O* (1.00 g, 52.3 mmol, 27% 17O, 42%
18O) dissolved in ethanol (10 mL). After the solution was stirred for
24 h, the ethanol was removed in vacuo. The remaining liquid was
distilled (0.065 Torr, 63-77 °C), yielding a clear, colorless liquid (6*,
10.3 g, 66%) determined to be 97.9% pure by GC. 1H NMR (300 MHz,
C6D6): δ ) 3.85 (q, 8H, OCH2CH3, J ) 7.0 Hz), 1.78 (m, 2H,
SiCH2CH2CH2Si), 1.23 (t, 12H, OCH2CH3, J ) 7.0 Hz), 0.71 (t, 4H,
SiCH2CH2CH2Si, J ) 3.8 Hz). 13C NMR (75 MHz, C6D6): δ ) 58.3
(OCH2CH3), 18.2 (OCH2CH3), 17.2 (SiCH2CH2CH2Si), 11.2 (SiCH2-
CH2CH2Si). 29Si NMR (99 MHz, neat): δ ) -47.2 (T1cyclic). 17O NMR
(54.25 MHz, 1 M 6* in EtOH): δ ) 64.0. IR (neat): ν ) 2976, 2926,
2888, 2737, 1483, 1444, 1392, 1367, 1296, 1246, 1219, 1169, 1105,
1079, 969, 898, 824, 789, 757, 738, 702 cm-1. HRMS (CI, ammonia):
calcd for C11H26O5Si2 [M]+, 294.1319; found, 294.1324.
Z-1,4-Bis(trichlorosilyl)but-2-ene. To a mixture of triethylamine
(62.0 mL, 0.440 mol) and copper(I) chloride (0.4 g) in anhydrous diethyl
ether (40 mL) were added trichlorosilane (60.3 g, 0.445 mol) and 1,
2-dichlorobut-2-ene (25 g, 0.2 mol) in diethyl ether (50 mL) at a rate
sufficient to keep the reaction at reflux (2 h). After addition, the reaction
was refluxed for an additional 2 h before the solution was filtered and
the solvent removed in vacuo to give a clear, yellow liquid. Distillation
under vacuum (85 °C, 9 Torr) afforded a colorless liquid (55.0 g, 85%).
Z-1,4-Bis(triethoxysilyl)but-2-ene (10). Z-1,4-Bis(trichlorosilyl)but-
2-ene was added dropwise to 200 mL of refluxing triethylorthoformate.
The resulting solution was heated at reflux for 24 h before removal of
the remaining orthoformate in vacuo. The desired product was isolated
1
by distillation at 132 °C/8 mTorr (10, 40.2 g, 53.0%). H NMR (400
MHz, C6D6): δ ) 5.35 (t, 2H, SiCH2CH)CHCH2Si, J ) 3.7 Hz),
3.77 (q, 12H, OCH2CH3, J ) 7.5 Hz), 1.55 (d, 4H, SiCH2CHdCHCH2-
Si, J ) 3.7 Hz), 1.17 (t, 18H, OCH2CH3, J ) 7.4 Hz). 13C NMR (100.5
MHz, C6D6): δ ) 58.1 (OCH2CH3), 18.7 (OCH2CH3), 18.5 (SiCH2CH2-
CH2Si), 17.1 (SiCH2CH2CH2Si), -4.5 (SiCH3). 29Si NMR (79.5 MHz,
C6D6): δ ) -50.4. IR (neat): ν ) 3016, 2975, 2927, 2887, 1703,
1686, 1643, 1483, 1443, 1391, 1295, 1168, 1104, 1083, 959, 883, 798,
747, 535 cm-1. HRMS (CI, isobutane): calcd for C16H37O6Si2 [M +
1]+, 381.6324; found, 381.2116.
2,2,7,7-Tetraethoxy-2,7-disilaoxacyclohept-4-ene (11). To a solu-
tion of 10 (10.2 g, 26.8 mmol) in ethanol (50 mL) was added 1 N HCl
(0.48 mL, 26.6 mmol of H2O) dissolved in ethanol (15 mL). After the
solution was stirred for 24 h, the solvent was removed in vacuo, and
the crude oil was vacuum distilled (0.05 Torr, 65-75 °C), yielding a
2,2,7,7-Tetraethoxy-2,7-disilaoxacycloheptane (7).22 To a solution
of 3 (15.0 g, 39.2 mmol) in ethanol (185 mL) was added 1 N HCl
(0.705 mL, 39.2 mmol of H2O) in ethanol (10 mL). After the solution
was stirred for 48 h, the ethanol was removed in vacuo. The remaining
liquid was distilled (0.05 Torr, 65-70 °C) to give a clear, colorless
liquid (7, 8.22 g, 68%). 1H NMR (300 MHz, C6D6): δ ) 3.83 (q, 8H,
OCH2CH3, J ) 7.0 Hz), 1.64 (m, 4H, SiCH2CH2CH2CH2Si), 1.16 (t,
1
colorless liquid (11, 6.2 g, 76%). H NMR (300 MHz, CDCl3): δ )
5.52 (m, 2H, CH2CHdCHCH2), 3.78 (q, 8H, OCH2CH3, J ) 7.3 Hz),
1.54 (m, 4H, SiCH2), 1.17 (t, 12H, OCH2CH3, J ) 7.4 Hz). 13C NMR
(75 MHz, CDCl3): δ ) 122.8 (CHdCH), 58.5 (OCH2CH3), 18.1
(OCH2CH3), 12.9 (SiCH2). 29Si NMR (79.5 MHz, C6D6): δ ) -55.8.
IR (neat): ν ) 3028, 2975, 2928, 2886, 1631, 1484, 1444, 1392, 1296,
1169, 1104, 1085, 1044, 963, 912, 804, 775, 704, 682, 595 cm-1. HRMS
(CI, isobutane): calcd for C12H26O5Si2 [M + NH4]+, 324.1662; found,
324.1662.
12H, OCH2CH3, J ) 7.0 Hz), 0.76 (m, 4H, SiCH2CH2CH2CH2Si). 13
C
NMR (75 MHz, C6D6): δ ) 58.4 (OCH2CH3), 26.7 (SiCH2CH2CH2-
CH2Si), 18.6 (OCH2CH3), 10.9 (SiCH2CH2CH2CH2Si). 29Si NMR (79.5
MHz, neat): δ ) -49.1 (T1cyclic). IR (neat): ν ) 2975, 2927, 2885,
2736, 1484, 1444, 1392, 1366, 1347, 1307, 1296, 1212, 1169, 1106,
1083, 1034, 961, 884, 849, 800, 762, 737 cm-1. HRMS (CI, isobu-
tane): calcd for C12H28O5Si2 [M]+, 308.1475; found, 308.1473.
Tricyclic Dimer (12). To a solution of 10 (33.1 g, 87.0 mmol) in
ethanol (150 mL) was added 1 N HCl (3.132 mL, 174 mmol of H2O)
dissolved in ethanol (50 mL). The mixture was refluxed for 1 week,
and then additional 1 N HCl (0.783 mL, 43.5 mmol of H2O) was added,
(44) Loy, D. A., Carpenter, J. P., Yamanaka, S. A.; McClain, M. D.;
Greaves, J.; Hobson, S.; Shea, K. J. Chem. Mater. 1998, 10, 4129-4140.

5424 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Loy et al.
and the mixture was refluxed for another week. The solvent was
removed in vacuo, and the crude oil was stored at -20 °C overnight.
A crude white solid crystallized and was filtered. The solid was
recrystallized in hot ethanol to give white needlelike crystals. The
crystals were filtered and washed with cold ethanol (3 × 1 mL) and
Si-O was initially solved in P21/c by direct methods. An absorption
correction was applied to the data using SADABS,46 and the solutions
were refined to suitable residual values. In all cases, the discrepancies
in the high-intensity reflections warranted refinement of a secondary
extinction coefficient. Finally, all non-hydrogen atoms were refined
anisotropically, but the C5 and C7 atoms of 15 showed large thermal
motions and were best modeled over two disordered sites with fractional
occupancies that were fixed in the final refinement cycles at 70:30%
after initially being allowed to refine independently. Consequently, no
hydrogen atoms could be reasonably assigned to these carbon atoms
or their neighboring carbon atoms.
Sol-Gel Processing: Polymerizations. Sol-gel polymerizations of
1-8 were performed with a monomer concentration of 0.4 M (absolute
ethanol) and with aqueous 1 N NaOH as the catalyst. The number of
equivalents of water used for each polymerization was equal to the
number of ethoxides present in each monomer (4 or 6). Gelation was
determined as the point at which the liquid would not flow. After
gelation, the samples were allowed to age for 2 weeks and then
processed to xerogels. Xerogels were prepared by crushing the wet
gels and washing with water (3 × 50 mL) and ether (2 × 25 mL) and
then drying in vacuo at 100 °C for 24 h.
1
dried in vacuo (12, 1.24 g, 6.1%). H NMR (300 MHz, C6D6): δ )
5.64 (m, 4H, CHdCH), 3.81 (q, 8H, OCH2CH3, J ) 7.4 Hz), 1.71,
1.59 (m, 8H, SiCH2), 1.17 (t, 12H, OCH2CH3, J ) 7.5 Hz). 13C NMR
(75 MHz, CDCl3): δ ) 123.5 (CHdCH), 58.6 (OCH2CH3), 18.4
(OCH2CH3), 14.1 (SiCH2). 29Si NMR (79.5 MHz, C6D6): δ ) -60.1
(T2cyclic). IR (neat): ν ) 3027, 2977, 2929, 2886, 1633, 1443, 1393,
1372, 1189, 1164, 1109, 1046, 968, 908, 837, 813, 784, 755, 701, 610
cm-1. HRMS (CI, isobutane): calcd for C16H32O8Si4 [M]+, 464.1174;
found, 464.1168.
1,3-Bis(diethoxymethylsilyl)propane (13). To a solution of 3-(di-
ethoxymethylsilyl)propene (37.2 g, 213 mmol), diethoxymethylsilane
(31.5 g, 234 mmol), and benzene (200 mL) under air was added
chloroplatinic acid (80 mg, 0.2 mmol). The mixture immediately turned
dark yellow. After the solution was stirred for 24 h, the solvent was
removed in vacuo, and the viscous oil was vacuum distilled (0.05 Torr,
1
70-78 °C), yielding a clear, colorless liquid (13, 49.9 g, 76%). H
NMR (400 MHz, C6D6): δ ) 3.70 (q, 8H, OCH2CH3, J ) 7.2 Hz),
1.70 (p, 2H, SiCH2CH2CH2Si, J ) 8.2 Hz), 1.14 (t, 12H, OCH2CH3, J
) 6.8 Hz), 0.82 (t, 4H, SiCH2CH2CH2Si, J ) 8.4 Hz), 0.15 (s, 6H,
SiCH3). 13C NMR (100.5 MHz, C6D6): δ ) 58.1 (OCH2CH3), 18.7
(OCH2CH3), 18.5 (SiCH2CH2CH2Si), 17.1 (SiCH2CH2CH2Si), -4.5
(SiCH3). IR (neat): ν ) 2974, 2926, 2878, 1483, 1444, 1391, 1339,
Polysilsesquioxane Xerogel Syntheses. 1X. A dry, 10-mL volu-
metric flask was charged with 1 (1.42 g, 4.00 mmol) and ethanol (6.0
mL). A catalyst solution of aqueous NaOH (0.432 mL, 1 N, 24.0 mmol
of H2O) and ethanol (2.0 mL) was added to the volumetric flask, and
the mixture immediately became cloudy. Enough ethanol was added
to the flask to bring the volume to exactly 10.0 mL, and the contents
were mixed thoroughly. The cloudy solution was then transferred to a
dry scintillation vial and sealed. After 24 h, a rigid, cloudy gel had
formed. The wet gel was processed to a xerogel as outlined above and
recovered as white monolithic fragments (529 mg, 100% based on
complete condensation of 1). 13C CP MAS NMR: δ ) 58.4, 17.2, 4.7.
29Si CP MAS NMR: δ ) -46.0 (T1), -56.8 (T2), -63.7 (T3). IR (KBr
1257, 1166, 1107, 1081, 953, 901, 796, 764 cm-1
.
2,6-Diethoxy-2,6-dimethyl-2,6-disilaoxacyclohexane (14). A flask
was charged with 1,3-bis(diethoxymethylsilyl)propane (49.4 g, 160
mmol), 1 N HCl (2.88 mL, 160 mmol of H2O), and ethanol (350 mL),
and the solution was stirred for 24 h. The solvent was removed in vacuo,
and the remaining oil was vacuum distilled (0.10 Torr, 30-33 °C),
yielding a mixture of syn and anti diastereomers as a clear, colorless
pellet): ν ) 2979, 2894, 1413, 1272, 1090, 1030, 909, 771, 698 cm-1
.
1
liquid (14, 29.3 g, 78%). H NMR (300 MHz, C6D6): δ ) 3.73 (q,
Anal. Calcd for C2H4O3Si2: C, 18.2; H, 3.0; Si, 42.5. Found: C, 18.9;
H, 4.2; Si, 38.3. Nitrogen sorption surface area (BET): 593 m2/g.
2X. A 10-mL ethanol solution containing 2 (1.47 g, 4.00 mmol)
and aqueous NaOH (0.432 mL, 1 N, 24.0 mmol of H2O) was prepared
in the same manner as for 1, and the slightly opaque solution was sealed
in a scintillation vial. After 1 h, a slightly opaque rigid gel formed and
was processed to a white xerogel (585 mg, 100% based on complete
condensation of 2). 13C CP MAS NMR: δ ) 57.7, 16.4, 13.3. 29Si CP
MAS NMR: δ ) -51.2, -60.8. IR (KBr pellet): ν ) 2925, 1416,
1251, 1220, 1112, 1024, 932, 903, 761, 670, 592 cm-1. Anal. Calcd
for C3H6O3Si2: C, 24.6; H, 4.1; Si, 38.4. Found: C, 24.6; H, 4.8; Si,
35.7. Nitrogen sorption surface area (BET): 712 m2/g.
3X. A 10-mL ethanol solution containing 3 (1.53 g, 4.00 mmol)
and aqueous NaOH (0.432 mL, 1 N, 24.0 mmol of H2O) was prepared
in the same manner as for 1, and the slightly cloudy solution was sealed
in a scintillation vial. After 1 h, a cloudy rigid gel formed and was
processed to white xerogel fragments (674 mg, 105% based on complete
condensation of 3). 13C CP MAS NMR: δ ) 58.3, 25.6, 16.2, 12.7.
29Si CP MAS NMR: δ ) -55.4, -64.2. IR (KBr pellet): ν ) 2931,
1410, 1225, 1127, 1054, 852, 795 cm-1. Anal. Calcd for C4H8O3Si2:
C, 30.0; H, 5.0; Si, 35.0. Found: C, 26.5; H, 5.7; Si, 31.4. Nitrogen
sorption surface area (BET): 882 m2/g.
6X. A 5-mL ethanol solution containing 6 (589 mg, 2.00 mmol)
and aqueous NaOH (0.144 mL, 1 N, 8.00 mmol of H2O) was prepared
in the same manner as for 1, and the clear, colorless solution was sealed
in a scintillation vial. After 0.5 h, a clear rigid gel formed and was
processed to white xerogel fragments (304 mg, 104% based on complete
condensation of 6). 13C CP MAS NMR: δ ) 57.8, 16.1, 13.6. 29Si CP
MAS NMR: δ ) -61.0. IR (KBr pellet): ν ) 2929, 1420, 1252,
1115, 1018, 932, 903, 762, 592 cm-1. Anal. Calcd for C3H6O3Si2: C,
24.6; H, 4.1; Si, 38.4. Found: C, 22.5; H, 5.0; Si, 35.0. Nitrogen
sorption surface area (BET): 807 m2/g.
OCH2CH3), 3.67 (q, OCH2CH3), 1.75 (m, SiCH2CH2CH2Si), 1.15 (t,
OCH2CH3), 1.12 (t, OCH2CH3), 0.572 (m, SiCH2CH2CH2Si), 0.138 (s,
SiCH3), 0.099 (s, SiCH3). 13C NMR (75 MHz, C6D6): δ ) 58.1, 58.0
(OCH2CH3), 18.6 (OCH2CH3), 17.7 (SiCH2CH2CH2Si), 15.3 (SiCH2-
CH2CH2Si), -2.1, -2.2 (SiCH3). IR (neat): ν ) 2973, 2908, 2879,
1444, 1415, 1392, 1328, 1259, 1238, 1168, 1110, 1081, 956, 902, 811,
789, 666 cm-1. HRMS (CI, isobutane): calcd for C9H22O3Si2 [M]+,
234.1107; found, 234.1107.
Tricyclic Dimer (15). To a flask containing 14 (15 g, 64 mmol)
and ethanol (150 mL) was added 1 N HCl (1.15 mL, 64.0 mmol of
H2O) dissolved in ethanol (10 mL). The mixture was stirred at room
temperature for 4 h and then refluxed for 24 h. The solvent was removed
in vacuo, and some solid crystallized from the crude oil. The mixture
was dissolved in a minimum of hot ethanol and cooled to -20 °C for
3 h, yielding needlelike white crystals. The crystals were filtered and
1
dried in vacuo (15, 3.17 g, 31%, mp 100-101.5 °C). H NMR (300
MHz, C6D6): δ ) 2.19 (dtt, 2H, SiCH2CHaxHeqCH2Si, J ) 13.4, 12.8,
2.8 Hz), 1.88 (dtt, 2H, SiCH2CHaxHeqCH2Si, J ) 13.4, 6.3, 4.2 Hz),
0.85 (ddd, 4H, SiCHeqHaxCH2, J ) 14.5, 6.3, 2.8 Hz), 0.48 (ddd, 4H,
SiCHeqHaxCH2, J ) 14.5, 12.8, 4.2 Hz), 0.20 (s, 12H, SiCH3). 13C NMR
(75 MHz, C6D6): δ ) 16.82, 16.79, -0.3. 29Si NMR (79.5 MHz,
C6D6): δ ) -13.0 (T2cyclic). HRMS (CI, isobutane): calcd for C14H33O8-
Si4 [M]+, 320.0752; found, 320.0751.
X-ray Structure Determinations. All single-crystal X-ray diffrac-
tion data were collected on a Siemens SMART CCD system at reduced
temperatures over an entire hemisphere of the reciprocal lattice. Specific
pertinent parameters for the data collection can be found in the
Supporting Information. The reflection lists for all structures were
assembled from no fewer than 900 individual images collected at a
framewidth of 0.3° in φ. The images were analyzed and the peaks
integrated by the SAINT routine, and all reflections were used to refine
the final unit cell parameters in the monoclinic cell.
7X. A 5.0-mL ethanol solution of 7 (617 mg, 2.00 mmol) and
aqueous NaOH (0.144 mL, 1 N, 8.00 mmol of H2O) was prepared in
Structure solution and refinement were accomplished using the
SHELXL-93 software package.45 For all structures, a substructure of
(45) Sheldrick, G. M. SHELXL93. Program for the Refinement of Crystal
Structures; University of Gottingen: Gottingen, Germany, 1993.
(46) Sheldrick, G. M. SADABS (Siemens Area Detector Absorption
Corrections); Siemens Energy and Automation Inc.: Madison, WI, 1996.

Cyclization in Sol-Gel Synthesis of Polysilsesquioxanes
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5425
the same manner as for 1, and the clear, colorless solution was sealed
in a scintillation vial. After 2.5 h, a clear rigid gel formed and was
processed to white xerogel fragments (344 mg, 107% based on complete
condensation of 7). 13C CP MAS NMR: δ ) 57.8, 24.7, 16.5, 12.6.
29Si CP MAS NMR: δ ) -63.7. IR (KBr pellet): ν ) 2932, 1408,
1311, 1127, 1039, 853, 797 cm-1. Anal. Calcd for C4H8O3Si2: C, 30.0;
H, 5.0; Si, 35.0. Found: C, 29.4; H, 6.0; Si, 31.2. Nitrogen sorption
surface area (BET): 868 m2/g.
8X. In a dry, 2-mL volumetric flask, 8 (352 mg, 0.800 mmol) was
dissolved in 1.5 mL of ethanol, and then the volume was adjusted to
2.0 mL with additional ethanol. In a second 2-mL volumetric flask,
aqueous NaOH (57.6 µL, 1 N, 3.20 mmol of H2O) was mixed with
enough ethanol to bring the volume to 2.0 mL. The two solutions were
mixed in a scintillation vial, and after 16 h, the clear solution had formed
a slightly opaque gel. The wet gel was processed to a white xerogel
powder (244 mg, 105% based on complete condensation of 8). 13C CP
MAS NMR: δ ) 58.0, 16.1. 29Si CP MAS NMR: δ ) -59.5. IR
(KBr pellet): ν ) 2928, 1420, 1252, 1111, 1008, 931, 902, 757, 420
cm-1. Anal. Calcd for C3H6O3Si2: C, 24.6; H, 4.1; Si, 38.4. Found:
C, 19.8; H, 5.1; Si, 34.9. Nitrogen sorption surface area (BET): 785
m2/g.
Acknowledgment. We would like to thank Bonnie McK-
enzie for the scanning electron micrographs and Dr. David R.
Wheeler for numerous discussions. Sandia is a multiprogram
laboratory operated by Sandia Corporation, a Lockheed Martin
Company, for the United States Department of Energy under
Contract DE-AC04-94AL85000.
Supporting Information Available: Tables of crystal data,
structure solution and refinement, atomic coordinates, bond
lengths and angles, and anisotropic thermal parameters for 8,
12, and 15 (PDF). This material is available free of charge via
the Internet at http://pubs.acs.org.
JA982751V