Production of Monodisperse Colloidal Silica Spheres: Effect of Temperature
INTRODUCTION
In experimental studies of colloidal phenomena, it is
often desirable to use particles which are spherical and of
uniform size. Such monodisperse colloidal suspensions
yield results which are not biased by shape and size dis-
tribution effects and hence are more amenable to physical
interpretation.
In 1968, Strber et al. (1) reported a reliable yet simple
method for producing spherical silica particles in the mi-
cron size range. The method involves reacting alkyl silicates
with water in a mutual short-chain alcohol solvent using
dissolved ammonia as a morphological catalyst. In general,
they found that those factors which accelerated the rate
of reaction tended to produce smaller particles. Thus, by
varying the alkyl group of both the silicate and the alcohol,
they were able to manufacture quasi-monodisperse silica
particles having average diameters in the range of 0.05-
2.0 #m. All reactions were carried out at 22°C.
Since the publication of this pioneering work, silica par-
ticles made by the method of StSber et al. have been used
as model colloids in a large number of experimental in-
vestigations (e.g., (2-6)). However, the method still has
two notable drawbacks. First, it produces larger particles
(> 1 ~tm) which are occasionally nonspherical and often
unacceptably nonuniform. Second, to achieve a range of
sizes, it requires the use of alkyl silicates other than tetra-
ethyl orthosilieate (TEOS), which is the only tetraester of
silicic acid commercially available in a pure form. There-
fore, in the present investigation, an attempt was made to
produce monodisperse silica spheres with a range of sizes
by reacting only TEOS and water. In this case, the reaction
rate was altered by varying the temperature (-20 to 60°C)
and the alcoholic solvent (methanol, ethanol, n-propanol,
and n-butanol).
EXPERIMENTAL
Materials
Methanol, n-propanol, and n-butanol (AnalaR grade,
BDH), as well as absolute ethanol (Stanchem), were used
as solvents.
Distilled water was deionized just prior to use.
Tetraethyl orthosilicate (reagent grade, Fisher) was twice
distilled before use in the experiments to ensure maximum
sphericity and monodispersity as reported by Bowen (7).
Anhydrous ammonia (99.9%, Union Carbide) was em-
ployed as the catalyst/stabilizer.
290
0021-9797/87 $3.00
Copyright © 1987 by Academic Press, Inc.
All fights of reproduction in any form reserved.
Procedures
In the present study, all silica particles were produced
by reacting tetraethyl orthosilicate with water in alcohol
solutions which were normally saturated with ammonia.
The general sequence of steps was similar to that employed
by Bowen (7) and is described below.
Measured volumes of alcohol and freshly distilled and
deionized water were pipetted into a 120-ml glass bottle
with a teflon-lined screw-on cap. The bottle was then po-
sitioned in a specially constructed Plexiglas jacket. An eth-
ylene glycol-water mixture was circulated through the in-
sulated jacket from a refrigerated constant temperature
bath (NESLAB Endocal RTE-9) to maintain the reaction
temperature to within _0.1 °C of the desired value. The
contents of the bottle were agitated continuously by a
magnetic stirrer. After thermal equilibrium was attained,
anhydrous ammonia gas from a cylinder was bubbled into
the water-alcohol mixture through a glass capillary until
saturation was reached (indicated by no further increase
in volume). At this juncture, the TEOS, which had been
maintained at the same temperature, was injected into the
bottle using a glass syringe. The bottle was then capped
tightly to prevent ammonia from escaping. For all exper-
iments, the volume ratio of TEOS:water:alcohol was
1:5:25, a ratio found by Bowen (7) to yield optimal results
for the TEOS, water, and n-propanol system.
After an invisible hydrolysis reaction in which silicic
acid was formed, the mixture suddenly became opalescent
signifying the onset of silica precipitation. The precipitate
appeared almost instantaneously at 60°C but only after a
lapse of about 15 min at -20°C, indicating that the hy-
drolysis reaction rate was considerably reduced as the
temperature was lowered. As a consequence, the contents
of the reaction bottle were sampled at 1, 2, 3, 5, and 9 h
for reaction temperatures of 60, 40, 20, 0, and -20°C,
respectively, preliminary experiments in which average
sizes were measured at times shorter than these having
indicated that the reactions had already reached comple-
tion in all cases. Clean glass capillaries were usedto transfer
a droplet of the suspension to carbon-coated electron mi-
croscope carrier grids. The samples were allowed to dry
and then micrographs were taken at a number of random
locations on the grid using a Hitachi HU-11A transmission
electron microscope. The photographs, along with those
of a calibration standard (54864 fines/in., Ladd Research
Industries), were then analyzed automatically by means
of a Leitz TA5-PLUS image analyzer. Between 100 to 400
Journal of Colloid and Interface Science, Vol. 118, No. 1, July 1987
 NOTES 291

FIG. 1. Electron micrograph of silica particles obtained using ethanol as the solvent and a temperature
of O°C.

particles were measured in order to determine the average 99.9

size and standard deviation of each sample.

99.0
RESULTS AND DISCUSSION
.J

Figure 1 shows an electron micrograph of one of the ~-

batches of larger silica particles, obtained in this case using a.~ 90.0
the solvent ethanol and a temperature of 0°C. Figure 2
gives the cumulative size distribution curve of the same ___
sample plotted in terms of normal-probability coordinates. -'tn
L=J 50.0

/
The two figures demonstrate that the particles are normally zD
distributed, reasonably spherical and have a relatively
narrow range of sizes. This sample has an average diameter
of 1.37 #m and a standard deviation of 0.02 #m, indicating ,x 10.0
that only 5% of the particles differ from the median size
o
by more than 3%. Similar results were realized with the
other large particle batches; generally speaking, the smaller o_
particles were even more monodisperse and spherical. 1.o
The number average particle sizes obtained as a function
of temperature and type o f solvent are listed in Table I o.1
1.32 1.34 1.3s t~ 1.40 1.42
and plotted as Fig. 3. Several runs were repeated on dif- DIAMETER (urn)
ferent days under otherwise identical experimental con-
ditions. These showed excellent reproducibility. Because FIG. 2. Normal-probability plot of the cumulative size
of water miscibility considerations, the alcohols used as distribution curve of the particle batch shown in Fig. 1.
Journal of Colloid and Interface Science, Vol. 118, No. 1, July 1987
292 NOTES

TABLE I

Average Diameter and Standard Deviation of Silica Spheres Obtained Using Different Temperatures and Solvents

Temperature (*C)

Solvent -20 0 20 40 60

Methanol 1.01 ___0.04 0.95 ___0.07 0.44 + 0.02 0.23 _+0.02

0.44 _ 0.03 a 0.21 _ 0.02 a

Ethanol 1.87 ± 0.11 1.37 _ 0.05 0.54 _ 0.03 0.26 ± 0.02

0.55 _ 0.03 a
0.53 + 0.03 b

n-Propanol 1.67 _ 0.05 1.05 _ 0.05 0.46 _ 0.02 0.27 + 0.02 0.25 ± 0.01
0.46 ± 0.03 b 0.29 + 0.02 a

n-Butanol 0.88 ± 0.03 0.62 + 0.02 0.35 + 0.01 0.26 + 0.02

"Duplicate experiment.
b Water-alcohol mixture saturated with ammonia at 40°C.

solvents in the present study were restricted to those having increasing temperature. This hypothesis was tested by
low molecular weights and straight hydrocarbon chains. conducting two experiments (one with ethanol and one
Even so, it was not possible to employ n-butanol below with n-propanol) in which the alcohol/water mixtures were
0°C. Also, excessive evaporation prevented particle pro- saturated at 40°C and then cooled to 20°C before injecting
duction with methanol and ethanol at temperatures ex- the TEOS. As can be seen from Table I and Fig. 3, the
ceeding 40°C. results of both tests were virtually the same as those ob-
The most dominant feature of Fig. 3 is that, regardless tained in corresponding solutions saturated at 20°C, in-
of which alcoholic solvent is used, the average size o f the dicating that the ammonia concentration near saturation
produced particles varied inversely with temperature. One levels does not significantly influence particle size.
possible explanation is that this trend is related to the sat- A more plausible explanation is that the trend is asso-
uration concentration of ammonia which decreases with dated with the hydrolysis reaction rate, which was observed
to decrease significantly at lower temperatures. This re-
2.0 i i lationship between particle size and reaction rate is in ac-
Zx cord with the earlier observations of Stfber et al. (1) and
Solvent:
can be interpreted in terms of the limited self-nucleation
m\ 0 Methanol
model ofmonodisperse particle production first proposed
~ z= Ethanol
by LaMer and Dinegar (8). The model hypothesizes the
~ ~ ~zx 13 Propanol
existence of a critical concentration, C ' t , above which nu-
13c ~ 0 Butanol
cleation proceeds at an appreciable rate and below which
nucleation is negligible. C*t is significantly greater than
- - 1.0 the saturation concentration, C~t, above which sponta-
E3
bJ neous growth will occur providing that nuclei are already
(3 available. Thus, as the hydrolysis of TEOS proceeds, silica
13E nuclei begin to form only after the silicic acid concentration
>
exceeds C*t. As soon as a sufficient number of nuclei are
present, particle growth permanently reduces the super-
saturation of Si(OH)4 below C*t. From this point onward,
only particle growth can occur and, providing that the
0.0 , , i I i , , I , l , self-nucleation period is limited to a very short initial out-
-40 0 40 80
TEMPERATURE (°c) burst, the final sol will be monodisperse. If the reaction
rate is markedly reduced, it thus seems reasonable that a
FIG. 3. Average diameters of silica particles obtained as much smaller number of nuclei will be required to relieve
a function o f temperature and type of solvent used. The the high supersaturation of silicic acid. As a consequence,
two filled-in symbols represent cases where the water-al- since the same amount of TEOS is reacted in each exper-
cohol mixture was saturated with ammonia at 40°C before iment, the average size of the particles must increase as
the reaction was carried out at 20°C. the temperature is diminished.
Journal of Colloid and InterfaceScience, Vol. 118, No. 1, July 1987
 NOTES
The relationship between particle size and the different
alcohols employed is far more complicated. In general,
particles produced in ethanol and n-propanol exhibited
the strongest temperature dependence, especially at low
temperatures. At elevated temperatures, the particle size
varied directly with the chain length of the alcohol but,
under cooler conditions, this tendency appeared to reverse
itsclfexcept in the case of methanol. This complex behavior
is not easily explained but it is probably related to the
different effects temperature has on such solution properties
as solubility and viscosity.
The most important result demonstrated by this study
is that, by using a single alkyl silicate (commercially avail-
able TEOS) along with a short-chain alcoholic solvent
(ethanol or n-propanol), and by simply varying the reaction
temperature, it is possible to produce uniform silica spheres
over a wide range of sizes (0.2-2.0/~m). The particles made
in this fashion are relatively monodisperse, with the size
distribution becoming wider as the temperature is in-
creased. The best results are obtained when the apparatus
is maintained as clean as possible and reagents of maxi-
mum purity are employed.
ACKNOWLEDGMENT
The authors are indebted to the Natural Sciences and
Engineering Research Council of Canada for continuing
financial support.
293
REFERENCES
1. Stfber, W., Fink, A., and Bohn, E., J. Colloid Interface
Sci. 26, 62 (1968).
2. Bauer, E. J., Reagor, B. T., and Russell, C. A., ASH-
RAEJ. 15, 53 (1973).
3. Kirkland, J. J., J. Chromatography 185, 273 (1979).
4. Bowen, B. D., and Epstein, N., J. Colloid Interface
Sci. 72, 81 (1979).
5. Kops-Werkhoven, M. M., and Fijnaut, H. M., J.
Chem. Phys. 74, 1618 (1981).
6. Van Helden, A. K., Jansen, J. W., and Vrij, A., J.
Colloid Interface Sci. 81, 354 (1981).
7. Bowen, B. D., Ph.D. Thesis, The University of British
Columbia, Canada, 1978.
8. LaMer, V. K., and Dinegar, R., J. Amer. Chem. Soc.
72, 4847 (1950).
C. G. TAN
B. D. BOWEN
N. EPSTEIN
Department of Chemical Engineering,
The University of British Columbia,
Vancouver, British Columbia V6T 1 W5, Canada
Received September 19, 1986
Journal of Colloid and Interface Science, Vol. 118, No. 1, July 1987