Vapor-phase synthesis of uniform silica spheres through two-stage hydrolysis of SiCl4

Article abstract of DOI:10.1016/j.materresbull.2008.01.003

We report, for the first time, a vapor-phase synthesis of nearly monodispersed silica spheres 250-300 nm in size through a two-stage hydrolysis of SiCl₄. In the first stage, SiCl₄ vapor was partially hydrolyzed with water vapor in a batch reactor at 150 °C to form silicon oxychloride particles, nearly monodispersed and spherical. In the second stage, these oxychloride particles were converted into silica particles through further hydrolysis at 1000 °C in a tubular reactor, while the morphology and size after the first-stage reaction remained virtually unchanged.

Full text of DOI:10.1016/j.materresbull.2008.01.003

Materials Research Bulletin 43 (2008) 2833–2839
www.elsevier.com/locate/matresbu
Vapor-phase synthesis of uniform silica spheres through
two-stage hydrolysis of SiCl4
*
Hoey K. Park, Kyun Y. Park
Department of Chemical Engineering, Kongju National University, 182 Shinkwandong,
Kongju, Chungnam 314-701, Republic of Korea
Received 10 August 2007; received in revised form 14 December 2007; accepted 6 January 2008
Available online 15 January 2008
Abstract
We report, for the first time, a vapor-phase synthesis of nearly monodispersed silica spheres 250–300 nm in size through a two-
stage hydrolysis of SiCl4. In the first stage, SiCl4 vapor was partially hydrolyzed with water vapor in a batch reactor at 150 8C to
form silicon oxychloride particles, nearly monodispersed and spherical. In the second stage, these oxychloride particles were
converted into silica particles through further hydrolysis at 1000 8C in a tubular reactor, while the morphology and size after the
first-stage reaction remained virtually unchanged.
# 2008 Elsevier Ltd. All rights reserved.
Keywords: A. Oxides, A. Nanostructures, B. Chemical synthesis, B. Vapor deposition
1. Introduction
Submicron silicas of varying morphology and size are used in many applications, such as fillers for paints and
rubbers, for manufacturing nanocomposite materials, and as adsorbents. The preparation of monodispersed spherical
silica has received considerable attention since Stober et al. [1] reported that a hydrolysis of tetraethoxysilane in a
basic solution of water and alcohol led to the formation of monodispersed silica spheres. A number of studies followed
to investigate the particle growth mechanisms, to improve the productivity, and to control the particle size [2–5].
Another applicable method for the preparation of monodispersed spherical silica is the spray hydrolysis of SiCl4
droplets of uniform size distribution with water vapor [6]. A crucial step is to obtain uniform SiCl4 droplets. The SiCl4
droplets formed initially by nebulization are broadly distributed in size, and pass through evaporation and
condensation to make the distribution uniform in the presence of AgCl nuclei introduced with the carrier gas. An
attempt to hydrolyze tetraethoxysilane (TEOS) droplets was made in a way similar as for the SiCl4 droplets, but failed
due to the slow hydrolysis rate [7].
The vapor-phase synthesis with all reactants as vapor is among the popular routes to nanoparticles of various
oxides, including silica. The flame oxidation of SiCl4 vapor is an example. The silica from the flame oxidation, which
is called fumed silica, is not spherical; rather, it is highly aggregated and exhibits chain-like structures [8]. Lee and
Choi [9] showed that the silica aggregates could be made spherical through a laser beam irradiation; however, those
* Corresponding author. Tel.: +82 41 850 8637; fax: +82 41 858 2575.
E-mail address: kypark@kongju.ac.kr (K.Y. Park).
0025-5408/$ – see front matter # 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2008.01.003

2834
H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
Fig. 1. Illustration of the proposed method in comparison with the existing flame oxidation method.
spheres were not uniform in size distribution. TEOS, which is easier to handle than SiCl4, has been used in many
studies as a precursor for silica particles through the vapor-phase route [10–13]. None of these silicas was reported to
be monodispersed and spherical.
We report a vapor-phase synthesis of nearly monodispersed silica spheres through a two-stage hydrolysis of SiCl4.
The overall reaction of the hydrolysis is represented by SiCl4 + 2H2O = SiO2 + 4HCl. Fig. 1 shows the concept of the
proposed method in comparison with flame oxidation. In the first stage, the SiCl4 vapor is partially hydrolyzed with
water vapor at 150 8C to form intermediate oxychlorides, represented by SiOxCly(OH)z. In the second stage, the
intermediates are completely hydrolyzed to silica at a higher temperature. The reactor used in the first stage is of batch
type to allow for uniform particle growth; in the second stage, the reactor is tubular for operational convenience.
Compared to the Stober method, an advantage of the present method is that dry silica of comparable grade could be
produced directly, eliminating the filtration, washing, drying, and calcination. This advantage would be particularly
appreciable in ceramics and composites applications that require dry silica as starting material.
2. Experimental
Fig. 2 shows a schematic drawing of the experimental setup, which consists of a precursor evaporator, a tubular
furnace and a particle collector. The evaporator, made of stainless steel, is 0.5 cm in diameter and 5 cm in length, and
equipped with a septum in the center, a thermocouple, a pressure sensor, and a valve at each end. It is immersed in a
cylinder through which hot oil is circulated. A small glove box is located over the septum to protect the moisture-
sensitive precursor from being exposed to air during the injection of the precursor with a syringe into the evaporator.
More details on the apparatus are available elsewhere [14]. A predetermined volume (less than 1 ml) of liquid SiCl4
was injected into the evaporator at 150 8C. The pressure in the evaporator rose with the vaporization and leveled off at
the completion of vaporization. Avolume of water was then injected into the evaporator. The volume was so small that,
upon injection, the water hung as a drop on the syringe tip. The vaporization of the water drop finished within a few
seconds, which was confirmed with the pressure sensor.
The mixture of the two vapors was held in the evaporator for a while to induce a partial hydrolysis or the first-stage
hydrolysis leading to the formation of an intermediate product. Thus, the evaporator also worked as the reactor for the
first-stage hydrolysis. By opening the valves at both ends, the intermediate product was transported by nitrogen gas to a
quartz tube, 4 mm in diameter and 35 cm in length, for the second-stage hydrolysis at 1000 8C. After the second-stage
reactor, a 250 ml flask containing 1 N NaOH solution was installed for removal of the HCl gas formed due to the
hydrolysis.
At the reactor exit, the produced particles were collected in situ on a carbon film that coated a 200 mesh nickel grid
(Electron Microscopy Science, Model CF200-Ni) and examined by transmission electron microscopy (TEM) (JEOL,

H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
2835
Fig. 2. Schematic drawing of the experimental apparatus.
Model JEM-2010) and scanning electron microscopy (JEOL, Model JSM-6335F). The crystalline structure was
identified by analyzing the electron diffraction pattern obtained from TEM. The chlorine content of the particles was
determined with the energy dispersive X-ray of TEM.
3. Results and discussion
3.1. Particle formation in the 1st reactor
Fig. 3 shows the TEM images of the deposits on the grid after the first-stage hydrolysis for 60 s, with the 1st
reactor temperature varied from 90 to 150 8C. The volumes of SiCl4 and water injected into the evaporator were
0.5 and 0.3 ml, respectively, which were determined to have a SiCl4 vapor concentration of 10 mol% with the
molar ratio of water to SiCl4 at the stoichiometric value of 2. No distinguishable particles were obtained at 90 and
120 8C. As the temperature was increased to 150 8C, spherical particles, approximately 250 nm in diameter,
emerged. By energy dispersive X-ray analysis of the particles, the atomic ratio of Si:O:Cl was determined to be
32:32:36, suggesting that the obtained spheres were not silica, but rather oxychlorides. Ignatov et al. [15]
Fig. 3. Transmission electron microscopic images of the particles collected at the 1st reactor outlets with the reactor temperature varying from 90 to
150 8C: (a) 90 8C; (b) 120 8C; (c) 150 8C. The white scale bar represents 100 nm. The 2nd reactor temperature was held constant at 1000 8C. The
initial concentration of SiCl4 vapor in the 1st reactor was 10 mol%, and the molar ratio of H2O to SiCl4 was at the stoichiometric ratio of 2.0. The
residence times were 60 s in the 1st reactor and 0.7 s in the 2nd reactor. The reactor pressures were atmospheric.

2836
H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
Fig. 4. Scanning electron microscopic image, at a higher magnification, of a particle of rough surface shown in Fig. 3. The white scale bar represents
100 nm.
proposed the following mechanism for the gas-phase hydrolysis of SiCl4, showing potential evolution of the
intermediate silicon oxychlorides:
SiCl4 þ H2O ! SiCl3OH þ HCl
Cl3SiOH þ HOSiCl3 ! Cl3SiꢀOꢀSiCl3 þ H2O
Cl3SiꢀOꢀSiCl3 þ H2O ! Cl3SiꢀOꢀSiCl2ꢀOH þ HCl
followed by further branching, chain growth and cyclization.
Two groups of spheres can be seen in Fig. 3(c): one with a smooth surface and the other with a rough surface. A
sphere of rough surface was further investigated by scanning electron microscopy at higher magnification. As shown in
Fig. 4, that sphere was composed of numerous smaller objects of irregular shape. By EDS analysis, the chlorine
content was 13.8 at.% in the spheres of smooth surface and 54.0 at.% in those of rough surface. The spheres of rough
surface disappeared as the hydrolysis reaction time was increased from 60 to 180 s or as the molar ratio of water to
SiCl4 was increased to 6.0 by three times the stoichiometric ratio. From these experimental observations, the spheres
of rough surface were probably in the transition toward the spheres of smooth surface.
3.2. Comparison of particles between the first and the second stage reactors
Fig. 5 compares the TEM images of the particles collected at the exits of the 1st and 2nd reactors, respectively. The
operating condition was the same as for Fig. 3. The morphology and the particle size are similar between the two images.
The primary particles after the second-stage reactor are spherical and nearly uniform in size distribution. The standard
Fig. 5. Transmission electron microscopic images of the particles collected at the 1st and 2nd reactor outlets: (a) 1st reactor outlet; (b) 2nd reactor
outlet. The 1st reactor temperature was 150 8C. The other operating conditions were the same as for Fig. 3. The white scale bar represents 100 nm.

H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
2837
Fig. 6. Electron diffraction pattern of selected particles in Fig. 5(b).
Table 1
Chlorine content of the particles collected at the 1st and 2nd reactor outlets with the molar ratio of H2O to SiCl4 varied from 2 to 6
Molar ratio of H2O to SiCl4
Sampling point
Chlorine content (at.%)
2
2
1st reactor outlet
2nd reactor outlet
33.8
8.0
4
4
1st reactor outlet
2nd reactor outlet
10.6
1.07
6
6
1st reactor outlet
2nd reactor outlet
0.54
0.57
deviation for150primary particles includingthoseinFig. 5(b) was calculated to be9.8% of the numberaveragediameter.
Fine dust-like matter is scattered over the particles collected at the exit of the second-stage reactor, the reason for which is
not known. One possibility is that some of the SiClxOy(OH)z particles that passed from the 1st reactor vaporized upon
entering the 2nd reactor, which was set at the higher temperature of 1000 8C, and then precipitated through further
hydrolysis toformsmallersecondary particles orthedust-likematerial. Thecrystalline phaseoftheparticlesafterthe2nd
reactor was amorphous by selected area diffraction, as shown in Fig. 6, and their chlorine content was determined to be
0.5 at.%. The chlorine content is expected to be further reduced by replacing the second-stage reactor with a flame
reactor; a reaction temperature as high as 1800 8C is usually obtainable with the flame reactor.
Table 1 shows the variation of chlorine content of the particles after the two reactors with the molar ratio of H2O to
SiCl4 varied from the stoichiometric value of 2–6. At the molar ratio of 2, the chlorine content was 33.8 at.% after the
1st reactor and 8 at.% after the 2nd reactor. As the ratio was further increased, the chlorine content decreased
gradually; at the ratio of 6, the chlorine content after the 2nd reactor was reduced to 0.5 at.%. The gap in the chlorine
content between the two sampling points narrowed as the ratio was increased, and it became negligible at the ratio of 6.
3.3. Effect of SiCl4 concentration on particle size
An increase in the SiCl4 concentration from 10 to 20 mol%, achieved by doubling the volume of SiCl4 injected into
the evaporator, decreased the particle size of the produced silica from 300 to 250 nm, as shown in Fig. 7. In a synthesis

2838
H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
Fig. 7. Comparison of morphology and size between the silica spheres produced at the two SiCl4 concentrations in the 1st reactor: (a) 10 mol% and
(b) 20 mol%. The other operating conditions were the same as for Fig. 5. The white scale bar represents 100 nm.
of silica by oxidation of SiCl4 vapor at 1200 8C, the particle size was reported to have increased initially upon
increasing the SiCl4 concentration, showed a maximum, and then decreased with a further increase [16]. In the gas-
phase synthesis of silica using TEOS as precursor, Okuyama et al. [10] reported an increase in particle size, whereas
Smolik and Moravec [13] reported a decrease in the particle size upon increasing the precursor concentration. The
effect of the precursor concentration on the particles has thus been controversial. An increase in the precursor
concentration would increase or decrease the particle size, depending upon its influences on the nucleation rate and the
particle growth rate. In the present hydrolysis, the concentration increase probably increased the nucleation rate
predominantly, resulting in a decrease in the silicon source available per nucleus for subsequent growth, yielding
smaller particles.
4. Conclusion
The two-stage hydrolysis of SiCl4 vapor proposed in Fig. 1 was experimentally validated. Compared to the highly
agglomerated chain-like silica from the one-step flame oxidation of SiCl4 vapor, the present silica was nearly uniform
in size distribution and was spherical due to the morphology control through the low-temperature partial hydrolysis in
the first-stage batch reactor. In the second-stage tubular reactor at the high temperature, the hydrolysis continued
toward completion; the chlorine content of the particles after the first stage was reduced, while the particle morphology
was nearly retained. The vapor-phase chemical preparation of nearly monodispersed silica was demonstrated for the
first time. The characterization of the particles was not sufficient, however, due to the small volume of the particles
obtainable from an experimental run. Further studies will follow in order to extensively characterize such silica, for
which scaling up may be required to obtain particles in larger quantity.
Acknowledgments
This work was supported by the research grant, Kongju National University and by the Korea Research Foundation
Grant (KRF-2004-041-D00195).
References
[1] W. Stober, A. Fink, E. Bohn, J. Colloid Interf. Sci. 26 (1968) 62.
[2] A.K. van Helden, J.W. Jansen, A. Vrij, J. Colloid Interf. Sci. 81 (1981) 354.
[3] G.H. Bogush, C.F. Zukoski, J. Colloid Interf. Sci. 142 (1991) 1.
[4] H. Giesche, J. Eur. Ceram. Soc. 14 (1994) 205.
[5] K.S. Rao, K. El-Hami, T. Kodaki, K. Matsushige, K. Makino, J. Colloid Interf. Sci. 289 (2005) 125.
[6] A. Balboa, R.E. Partch, E. Matijevic, Colloids Surf. 27 (1987) 123.
[7] E. Matijevic, Q. Zhong, R.E. Partch, Aerosol Sci. Technol. 22 (1995) 162.
[8] G.D. Ulrich, Combust. Sci. Technol. 4 (1971) 47.
[9] D. Lee, M. Choi, J. Aerosol Sci. 33 (2002) 1.

H.K. Park, K.Y. Park / Materials Research Bulletin 43 (2008) 2833–2839
2839
[10] K. Okuyama, Y. Kousaka, N. Tohge, S. Yamamoto, J.J. Wu, R.C. Flagan, J.H. Seinfeld, AIChE J. 32 (1986) 2010.
[11] J.H. Lee, G. Beaucage, S.E. Pratsinis, Chem. Mater. 9 (1997) 2400.
[12] H.D. Jang, Aerosol Sci. Technol. 30 (1999) 477.
[13] J. Smolik, P. Moravec, J. Mater. Sci. Lett. 14 (1995) 387.
[14] J.G. Choi, K.Y. Park, J. Nanopart. Res. 8 (2006) 269.
[15] S.K. Ignatov, P.G. Sennikov, A.G. Razuvaev, L.A. Chuprov, O. Schrems, B.S. Ault, J. Phys. Chem. A 107 (2003) 8705.
[16] Y. Suyama, A. Kato, J. Am. Ceram. Soc. 59 (1976) 146.