2264 Chem. Mater.

1997, 9, 2264-2268

Direct Formation of Aerogels by Sol-Gel

Polymerizations of Alkoxysilanes in
Supercritical Carbon Dioxide
Douglas A. Loy,* Edward M. Russick,
Stacey A. Yamanaka, and Brigitta M. Baugher
Sandia National Laboratories
Albuquerque, New Mexico 87185-1407
Kenneth J. Shea*
Figure 1. Schematic for preparation of aerogels. Conventional
Department of Chemistry, University of California method (top) involves polymerization of a tetraalkoxysilane
Irvine, Irvine, California 92717-2025 in alcohol to give a gel, exchange of SCCD for alcohol in gel
and, finally, supercritical drying to give an aerogel. Direct
Received May 7, 1997 polymerization in SCCD using formic acid (bottom) allows
Revised Manuscript Received July 22, 1997 direct supercritical drying to aerogels without a solvent
exchange.
Sol-gel processing has proven to be an important
method for producing amorphous, porous silica1-3 and
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

polysilsesquioxane gels.4 Commercial applications of

monolithic sol-gels have been limited, despite the
advantages of mild reaction and processing conditions,
Downloaded via BOCHUM LIBRARIES on November 10, 2022 at 11:00:06 (UTC).

due to shrinkage (and often cracking) of gels during

drying. While a variety of processing techniques have
been developed to minimize shrinkage and prevent
cracking, one of the most successful is to avoid the
capillary forces associated with air-drying altogether by
drying the gel at a temperature and pressure above the
critical point of the solvent.5,6 The resulting aerogels
retain more of the original volume of the wet gel and
are potentially useful insulating materials due to their
low density, high surface area, and low thermal con-
ductivity. However, the high temperature and pres-
sures required for supercritical drying of gels in alcohols
increase the costs and hazards associated with prepar-
ing a dry gel. Exchanging the original solvent in the Figure 2. Photograph of a monolithic silica aerogel (1 cm ×
gel for supercritical carbon dioxide (SCCD) permits 3.2 cm) prepared by polymerizing tetramethoxysilane (TMOS)
in supercritical carbon dioxide (SCCD).
drying at lower pressures and temperatures,7 but the
required additional extraction step makes this technique
ring-opening polymerizations.14,15 This research is driven
(Figure 1) less attractive. We have simplified the
by the need to reduce VOCs and hazardous organic
preparation of aerogels by eliminating the organic
solvent waste in the polymer industry. This process has
(alcohol) solvent altogether by using SCCD as the sol-
been extended to include condensation polymers such
gel polymerization solvent. In this report, we describe
as polyesters16,17 and polysiloxanes. Because SCCD is
the first sol-gel polymerizations of tetraalkoxysilanes
a relatively nonpolar medium, sol-gel polymerizations
and 1,4-bis(triethoxysilyl)benzene (BESP) in SCCD to
in SCCD using standard formulations are not possible
give silica aerogels (Figure 2) and 1,4-phenylene-bridged
due to the large quantities of water necessary for the
polysilsesquioxane aerogels, respectively.
Recently, there has been considerable progress using hydrolysis and condensation of the alkoxysilane mono-
SCCD as a solvent for free radical8-10 and Lewis acid- mers. Complete hydrolysis and condensation of tet-
catalyzed11-13 polymerizations of vinyl monomers, and ramethoxysilane (TMOS) or tetraethoxysilane (TEOS)
to silica requires a minimum of 2 equiv of H2O or 7.2%
(vol/vol) at 2 M monomer concentration. A minimum
* To whom correspondence should be addressed. E-mail: daloy@
sandia.gov and [email protected]. of 3 equiv of H2O is needed to form gels from BESP,
(1) Dislich, H. J. Non-Cryst. Solids 1986, 80, 115-21. but because hydrocarbon-bridged triethoxysilanes form
(2) Schmidt, H. J. Non-Cryst. Solids 1988, 100, 51-64.
(3) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: the physics and gels at lower monomer concentrations (0.4-0.8 M) than
chemistry of sol-gel processing; Academic Press: San Diego, 1990. do tetralkoxysilanes, this translates to only 2.2-4.4%
(4) Loy, D. A.; Shea, K. J. Chem. Rev. (Washington, D.C.) 1995, 95, H2O (vol/vol). Alcohols are needed as sol-gel solvents
1431-42.
(5) Fricke, J. Springer Proc. Phys. 1986, 6, 2-19. primarily to permit mixing of water and the nonpolar
(6) Hrubesh, L. W. Chem. Ind. (London) 1990, 824-7.
(7) Tewari, P. H.; Hunt, A. J.; Lofftus, K. D. Springer Proc. Phys.
1986, 6, 31-7. (13) Clark, M. R.; DeSimone, J. M. Polym. Prepr. (Am. Chem. Soc.,
(8) Sertage, W. G., Jr.; Davis, P.; Schenck, H. U.; Denzinger, W.; Div. Polym. Chem.) 1994, 35, 482-3.
Hartmann, H. BASF Corp., USA, 1990; p 21. (14) Mistele, C. D.; Thorp, H. H.; DeSimone, J. M. Polym. Prepr.
(9) DeSimone, J. M.; Guan, Z.; Elsbernd, C. S. Science (Washington, (Am. Chem. Soc., Div. Polym. Chem.) 1995, 36, 507-8.
D.C.) 1992, 257, 945-7. (15) Clough, R. S.; Elsbernd, C. L. S.; Gozum, J. E. PCT Int. Appl.
(10) Guan, Z.; Combes, J. R.; Menceloglu, Y. Z.; DeSimone, J. M. WO 9 601 851 A1, 1996.
Macromolecules 1993, 26, 2663-9. (16) Al Ghatta, H. A. K. ; Cobarr S. p. A. Eur. Pat. Appl. EP 269
(11) Kennedy, J. P.; Pernecker, T. U.S. Patent 5,376,744 A, 1994. 583 A2, 1988.
(12) Pernecker, T.; Kennedy, J. P. Polym. Bull. (Berlin) 1994, 32, (17) Burke, A. L. C.; Maier, G.; DeSimone, J. M. Polym. Mater. Sci.
537-43. Eng. 1996, 74, 248-9.

S0897-4756(97)00326-8 CCC: $14.00 © 1997 American Chemical Society

Communications Chem. Mater., Vol. 9, No. 11, 1997 2265

Table 1. Surface Areas (Five Point BET) and Mean Pore Diameters (BJH) for Silica and Phenylene-Bridged
Polysilsesquioxane Aerogels Prepared by (A) Sol-Gel Polymerization with Formic Acid Directly in SCCD, (B) Sol-Gel
Polymerization with Formic Acid in Ethanol, Followed by SCCD Extraction and (C) Sol-Gel Polymerization in Formic
Acid, Followed by SCCD Extractiona
equiv of (A) surface (A) mean pore (B) surface (B) mean pore (C) surface (C) mean pore
monomer formic acid area (m2 g-1) diam (Å) area (m2 g-1) diam (Å) area (m2 g-1) diam (Å)
TMOS (1.9 M) 4 308 31 481 121 739 83
TMOS (1.9 M) 3 261 92 197 121
TMOS (2.3 M) 6 268 35 533 107
TEOS (2.2 M) 6 no gel no gel no gel no gel
BESP (0.4 M) 6 471 55 no gel no gel 434 23
BESP (0.8 M) 6 586 24 18 51
a Data were obtained from 40-point nitrogen sorption isotherms obtained with a quantachrome autosorb-6 porosimeter on ground samples

that were degassed at 100 °C.

alkoxysilane monomers to give homogeneous solutions.

Thus, sol-gel polymerizations in nonpolar solvents such
as benzene or cyclohexane would be possible only
through formation of microemulsions or by generating
the water as needed from a relatively hydrophobic
precursor.
In light of recent successes in forming polyacrylates
from emulsion polymerizations in SCCD,8,18-20 our
initial attempts were directed toward formation of stable
microemulsions of water, catalyst, and alkoxysilane in
SCCD. We have successfully used this approach to
prepare silica and silsesquioxane gels in nonpolar
hydrocarbon solvents such as toluene.21 However, we
were not able to identify conditions that would permit
Figure 3. Schematic showing the autoclave used to prepare
dispersion of sufficient water into SCCD to give homo- aerogels while permitting observation of the illuminated cell
geneous, monolithic gels. For example, introduction of through a sapphire window. The autoclave was built by Walter
a phase-transfer catalyst (NR4OH‚xH2O, R ) Me, butyl) J. Krug at John Hopkins University from Nitronic steel (5.7
to a solution of the tetraalkoxysilane or BESP, or their cm o.d. × 1.59 cm i.d., 22 cm3 working volume, fitted with a
precondensed oligomers, and a surfactant (1-3%) in 1.9 cm o.d. × 1.3 cm thick sapphire window).25
SCCD gave only low yields (<10%) of silica or bridged
polysilsesquioxanes in the form of precipitates. The
surface areas of these materials were low. For example,
a phenylene-bridged polysilsesquioxane gel was ob-
tained with a surface area of 21 m2 g-1.
It was possible to avoid the problems associated with
the poor miscibility of water in SCCD and substantially
reduce gelation times for TMOS by applying the “water-
free” sol-gel polymerization technique recently reported solution of monomer and formic acid was quickly placed
by Sharp et al.22,23 Alkoxysilanes were shown to react into an autoclave (Figure 3) that was charged with
with anhydrous formic acid to give silanols that can SCCD (35-40 °C, 41.4 MPa; 10 mL final volume) with
condense to afford the siloxane bonds necessary for stirring until a homogeneous clear, slightly amber
creating a network polymer that will form a gel. The solution was observed through the saphirre observation
alcohol generated by the reaction is converted to the port (Figure 4a). Because the concentrations of silica
formate ester which could be recovered for recycling. gel precursors were necessarily high in order to form
SCCD and formic acid are completely miscible at the gels, the amount of formic acid required for the sto-
temperatures and pressures used in these experiments. ichiometric condensation was sufficiently large (23-36
TMOS or BESP mixed with formic acid (99%) readily wt % for TMOS; 13-19 wt % for BESP) for it to be
dissolved in SCCD to give clear colorless solutions that considered as a cosolvent with SCCD. Gelation, which
gelled in less than 12 h. Although TEOS also formed could be readily detected when the stir bar inside the
homogeneous solutions (2 M) with SCCD and formic autoclave was immobilized and the entire volume of the
acid (2-4 equiv), no gels were obtained in the 12 h pressure vessel was filled with a translucent gel,
experiments. occurred in under 12 h for TMOS (Figure 4b). After the
In a typical experiment (Table 1), formic acid (2-4 TMOS-derived silica gels were aged for 12-18 h, the
equiv) was mixed with TMOS or TEOS. The resulting pressure was slowly released over 8-18 h to afford
opaque white silica aerogels in nearly quantitative yield.
(18) O’Neill, M. L.; Yates, M. Z.; Johnston, K. P.; Wilkinson, S. P.; Similarly, translucent, phenylene-bridged polysilses-
DeSimone, J. M. Polym. Mater. Sci. Eng. 1996, 74, 228-9. quioxane gels were prepared by mixing BESP (0.2-0.8
(19) Yazdi, A. V.; Lepilleur, C.; Singley, E. J.; Liu, W.; Adamsky,
F. A.; Enick, R. M.; Beckman, E. J. Fluid Phase Equilib. 1996, 117,
M) with anhydrous formic acid (4-6 equiv) and dissolv-
297-303. ing the solution in SCCD. For comparison, TMOS,
(20) Canelas, D. A.; Betts, D. E.; DeSimone, J. M. Polym. Mater. TEOS, and BESP were also polymerized with identical
Sci. Eng. 1996, 74, 400-1.
(21) Martino, A.; Yamanaka, S. A.; Kawola, J. S.; Loy, D. A. Chem. quantities of formic acid while substituting methanol
Mater. 1997, 9, 423-429. or ethanol at ambient pressure and temperature for
2266 Chem. Mater., Vol. 9, No. 11, 1997 Communications

Figure 4. (a) Photograph of the interior of the autoclave showing a homogeneous solution of polymerizing TMOS, formic acid
and SCCD before gelation. The light ring is the reflection from the light source; the gray cylinder on the floor of the chamber is
a magnetic stir bar. (b) Photograph taken after gelation had occurred showing opaque white monolith that has shrunk from the
walls of the chamber.

SCCD as the polymerization solvent. Because both acid-alkoxysilane polymerizations (of TMOS and BESP)
polymerizations in SCCD and ethanol involve premixing in SCCD formed gels that, to the eye, appeared identical
the monomers with formic acid, we also prepared gels with gels prepared through conventional sol-gel tech-
from solutions of just TMOS or BESP and formic acid niques. The failure of TEOS to form a gel is consistent
with no additional polymerization solvent. Under these with the lower reactivity of the monomer; aqueous acid-
conditions, gels took at least 1 h to form. Both the gels catalyzed sol-gel polymerizations of TEOS can require
prepared in alcohols and the gels prepared with formic months for gels to form. The gels prepared in SCCD
acid alone were extracted with SCCD to give aerogels exhibited some signs of cracking while still in the
by the conventional supercritical process (Figure 2). autoclave, particularly between the gel and the walls
Unlike experiments using surfactants to form micro- of the vessel, due to syneresis. In addition, some
emulsions in SCCD which led to precipitates, the formic shrinkage, and in some cases cracking, was observed
Communications Chem. Mater., Vol. 9, No. 11, 1997 2267

Figure 5. SEM images of (a) a silica aerogel prepared by polymerization of TMOS with formic acid in SCCD, (b) a silica aerogel
prepared by the conventional polymerization of TMOS with formic acid in methanol, followed by extraction with SCCD, (c) a
silica aerogel prepared only from TMOS and formic acid followed with SCCD extraction and drying, and (d) a phenylene-bridged
polysilsesquioxane aerogel prepared by the polymerization of 1,4-bis(triethoxysilyl)benzene with formic acid in SCCD. Aerogels
were ground into a powder. Conductive silver paint was used to adhere the sample particles to a sample mount. The particles
were then coated with chrome using a high-resolution ion beam coater. Two minutes of chrome (approximately 150 Å) was typically
applied as the sample was tilted and rotated under the chrome plasma. If the samples still charged too much to image, another
coating was applied. The samples were imaged using a high-resolution Hitachi S4500 field emitter gun scanning electron microscope.
Low accelerating voltages, 1-2 kV, were used to minimize charging problems. PGT Imix imaging software was used to acquire
the images.

during drying. Similar shrinkage has been often ob- pore structure of the gel. One possible explanation of
served for aerogels prepared from BESP in organic the microstructural differences between the aerogels is
solvents even with SCCD extraction and may reflect that the SCCD is a “better” solvent than ethanol for the
formation of a very compliant polymeric gel. Bulk growing silica polymers that form the particles.
densities for silica and phenylene-bridged polysilses- Scanning electron micrographs of phenylene-bridged
quioxane aerogel monoliths were between 0.28 and 0.31 polysilsesquioxane aerogels prepared in both SCCD
g cm-3. (Figure 5d) and methanol revealed no macroporosity
The aerogels were, with the exception of a transparent analogous to that observed in the silica aerogels.
aerogel prepared from BESP, opaque due to light Transmission electron micrographs revealed the gel
scattering from the aggregate structure of the materials. structure to be that of a tightly packed aggregate of
Closer inspection using scanning electron microscopy particles between 40 and 80 Å in diameter, independent
revealed the silica aerogels to be highly porous struc- of the concentration of BESP used in the sol-gel
tures composed of interconnected particles. The silica polymerization. Both SEM and TEM micrographs are
gels prepared in SCCD consistently exhibited larger remarkably similar to those obtained with phenylene-
particle diameters (100-2000 nm; Figure 5a) and more bridged polysilsesquioxane aerogels prepared by aque-
macroporosity than those aerogels prepared by SCCD ous acid-catalyzed sol-gel polymerization of BESP
extraction of gels formed in ethanol (20-40 nm; Figure followed by SCCD extraction. Integration of the reso-
5b). Similar, mesoporous structure (20-40 nm) in the nances in the solid state 29Si CP MAS NMR spectrum
silica aerogel (Figure 5c) prepared in formic acid fol- of the phenylene-bridged polysilsesquioxane aerogel
lowed by SCCD extraction demonstrates that using polymerized in SCCD revealed that the extent of
SCCD as a polymerization solvent contributed to the condensation to siloxane bonds (70%) was comparable
2268 Chem. Mater., Vol. 9, No. 11, 1997 Communications

to gels prepared under acidic conditions in alcohol. discovery that phenylene-bridged aerogels prepared in
These similarities suggest that SCCD is as suitable a ethanol exhibited very low surface areas indicates that
solvent for forming these hybrid organic-inorganic gels there were significant differences in the interactions
as ethanol. between SCCD and ethanol with the growing polymers.
Surface analyses of nearly all of the aerogels using Our results show for the first time that sol-gel poly-
nitrogen sorption porosimetry revealed type IV iso- merizations can be conducted directly in supercritical
therms consistent with mesoporous materials (mean carbon dioxide as the polymerization solvent by using
pore diameters between 20 and 500 Å). Surface areas formic acid as the condensation reagent. SCCD appears
of the silica aerogels prepared in SCCD (260-308 m2 to be an excellent solvent for preparing highly porous
g-1) were equal to or slightly lower than those measured architectures with both meso- and macroporous struc-
for the corresponding aerogels prepared in ethanol or ture. In fact, SCCD appears to be a better solvent than
formic acid (200-500 m2 g-1). The most interesting
ethanol for preparing phenylene-bridged polysilsesqui-
result was that the pores detected by nitrogen sorption
oxane aerogels from BESP. In addition, the process can
in the silica aerogels prepared in SCCD were less than
be successfully used to generate monolithic aerogels in
100 Å in diameter and appear to lie within the aggregate
a single step from their monomeric precursors. We
particles. The macroporous structures in the SCCD
believe that this technique will greatly simplify the
silica gels evident in the electron micrograph (Figure
preparation of aerogels and open a new frontier for sol-
5a) are too large (>500 Å) to be observed by nitrogen
gel processing using SCCD or other nonpolar organic
sorption analysis. Thus, the SCCD-silica aerogels pos-
solvents coupled with nonaqueous chemistry.
sess two different populations of pores: mesopores based
on the architecture of the individual particles, and
macropores that are the interstitial voids between the Acknowledgment. We would like to thank Thomas
particles that make up the gel. Headley for the transmission electron microscopy, Bon-
In contrast, the pore size distribution found in the nie McKenzie for the scanning electron microscopy, and
aerogels (100-121 Å) prepared in methanol corresponds Roger Assink for the NMR spectroscopy. This work was
only with the smaller, interstial mesopores observed in supported by the United States Department of Energy
the SEM micrographs of the aggregate-like structures. under Contract DE-AC04-94AL85000. Sandia is a mul-
Surface areas of the BESP aerogels prepared in SCCD tiprogram laboratory operated by Sandia Corporation,
were also lower (471-586 m2 g-1) than those of aerogels a Lockheed Martin Company, for the United States
obtained by aqueous (HCl or NaOH-catalyzed) sol-gel Department of Energy.
polymerizations in organic solvents (1000-1600 m2
g-1).24 As would be expected with the fine grained
structure observed in the SEM micrographs, the mean CM970326F
pore diameters of these materials were smaller (24-50
Å) than in the silica aerogels prepared in SCCD. The
(24) Loy, D. A.; Shea, K. J.; Russick, E. M. Mater. Res. Soc. Symp.
Proc. 1992, 271, 699-704.
(22) Sharp, K. G. PCT Int. Appl. WO 9 323 333 A1, 1993. (25) Meilchen, M. A.; Hasch, B. M.; McHugh, M. A. Macromolecules
(23) Sharp, K. G. J. Sol-Gel Sci. Technol. 1994, 2, 35-41. 1991, 24, 4874-4882.