-
Herbert Giesche
New York State College of Ceramics at Alfred University, School
of Ceramic Engineering and Science,2 Pine Street, Alfred, New York
14802-1296, USA
(Received 6 December 1993; revised version received 23 February
1994; accepted 8 March 1994)
Abstract Struktur. Bei hiiheren Wassergehalten war
dieTeilchengriij3enverteilung enger und die Formsphiirischer.
Die Ergebnisse legen elnen Wechsel im Reaktion-smechanismus
nahe. Zu Beginn lagern siGh nanome-ter groj3e Primiirpartikel zu
Agglomeraten zusam-men und ab einem bestimmten Zeitpunkt erfolgt
dasPartikelwachstum durch Kondensation
monomererKieselsiiureeinheiten an der Partikeloberfliiche.
II existe actuellement plusieurs modeles permet-tant de decrire
la formation de particules monodis-persees de silice par hydrolyse
et condensation con-trolies du tetraethylorthosilicate (TEOS) dans
desmelanges eau/NH) et ethanol. L 'etude presenteemontre des
resultats obtenus pour une gamme ireslarge de conditions
experimentales. La croissancedes particules est liee a Ja cinetique
d'une reactiondu premier ordre, elle-meme dependante de la
con-centration en TEOS. Les ordres reactionnels pourdes solutions
contenant de l'eau et de l'ammoniacsont respectivement de 0.97 et
1.18. Seuls desmonomeres (70-90%) et des dimeres (10-25%)
desilicates se forment, et l'energie d'activation est de27kJ mol-I.
(6'5 kcal mol':-I). Les particules presen-tent une structure
interne ultra-microporeuse. Onpeut egalement observer une
distribution plus serreeet une sphericite plus marquee des
particules,lorsque la concentration en eau augmente.
L'inter-pretation des resultats est la suivante: presence
d'unmecanisme d'agregation de noyaux primaires detaille
nanometrique (durant la premiere partie de lareaction) suivi d'un
mecanisme de croissance addi-tionnelle de monomeres.
Currently there are several models discussed to de-scribe the
formation of monodispersed silica parti-clesby the controlled
hydrolysis and condensationof tetraethylorthosilicate (TEaS) in
mixtures ofwater/ammonia and ethanol. The present study willshow
results on an extended range of experimentalconditions.
Particle growth followed a first-order reactionkinetic with
respect to TEaS concentration. Reac-tion orders for the ammonia and
water content were0.97 and 1.18, respectively. Only monomeric(z
70-90%) and dimeric (z 10-25%) silicate unitswere detected during
the growth reaction and an ac-tivation energy of 27 kJ mol-l (6.5
kcal mol-l) wasindicated. Particles revealed an
ultramicroporousinternal structure. A narrower size distribution
anda more spherical shape were observed at higherwater
concentrations.
The results suggested an aggregation mechanismof nanometer-sized
primary nuclei during the firststage of the reaction followed by a
monomer addi-tion growth mechanism thereafter.
Zur Zeit werden verschiedene M odelle unter-sucht, welche die
Bildung monodisperser Si02-Teilchen durch Hydrolyse und
Kondensation vonTEOS in WasserlAmmoniak und Ethanol Gemi-schen
beschreiben. Die vorliegende Arbeit zeigtErgebnisse in einem
erweiterten experimentellenBereich.
Das Partikelwachstum verlief entsprechend einerReaktion erster
Ordnung beziiglich der TEOSKonzen tra tion. Die Reak(ionsordnung
beziiglichder Ammoniak- und Wasser-Konzentration war0.97
bzw. 1.18. Wtihrend der Reaktion wurdenvornehmlich monomere
(~70-90%) und dimere(~10-25%) Kieselstiure-Einheiten beobachtet.
DieAktivierungsenergie betrug 27 kJ mol-l (6.5 kcalmol-I). Die
Partikel zeigten ein ultramikroporose
1
Introduction
Already in 1956 Gerhard Kolbe! observed theformation of
spherical silica particles, whentetraethylorthosilicate was'
hydrolysed in the
189
Journal of the European Ceramic Sociery 0955-2219/94/$7.00 @
1994 Elsevier Science Limited, England. Printed inGreat Britain
-
lytical grade. Some experiments required a higherNH3
concentration in relation to the permittedwater content. In order
to achieve this, part of theregular ethanol was replaced by ethanol
saturatedwith ammonia. The latter was prepared by bub-bling NH3
through the ethanol at 195 K (-78°C)until the ammonia concentration
reached about 6mol dm-3. All ammonia concentrations werechecked in
the final solutions (prior to the precipi-tation) by titration with
HCl. The water andTEOS concentrations were calculated from
theweight fractions of the different portions.
The precipitation proceeded as follows. MixtureA-alcohol, water
and ammonia-and mixtureB- TEOS in alcohol (1: 1 volume ratio
)-cwereheated in closed containers to the desired tempera-tures.
Then mixture A was put in a round-bottomed flask, equipped with a
paddle stirrer anda condenser and placed in a constant
temperaturebath. Thereafter mixture B was added undervigorous
stirring (1000 rpm). The mixing wasstopped after 15 s. In spite of
the constant temper-ature bath, the temperature in the centre of
thereaction vessel increased slightly by up to 2 Kduring the first
few minutes after the addition ofTEOS, which was due to the
exothermic characterof the reaction.
presence of ammonia. In 1968 Stober et al.2 tookup and improved
this process to produce excep-tional monodispersed silica
particles. The finalparticle size could be controlled reasonably
wellwithin a wide range, and particles of up to 1.5 ,urnwere
prepared. Irrespective of this result it wasjust a few years ago
that a renaissance of this pro-cess took place, when a number of
scientistsstarted to use these powders as model systems invarious
areas; e.g. light-scattering studies,3-6 therheology of
dispersions,7-11 formation of orderedsediments, 1 1-14 or the
sintering of model powder
compacts.I3-16 Despite the increased number of ap-plications of
these 'Stober' silica powders, the re-action pathway still remained
vague, due to therelative fast reaction rate and various reaction
in-termediates, until very recently the nucleation andgrowth
process was studied in more detail.13.17-28At present the process
is not resolved completely,but all interpretations of the
experimental resultsfocus on two models: a monomer-addition and
anucleation-aggregation growth model. A combi-nation of both models
will probably best describethe entire process. Switching from a
nucleation-aggregation mode to a monomer-addition processas the
reaction proceeds and depending on theoverall starting
conditions.
The first part of the present paper will reveal aseries of
well-defined experiments describing thesilica precipitation
process. The influences of thestarting reaction conditions on
particle properties(size, size distribution and shape) and
reactionkinetics (reaction rate and induction period) werestudied
over a wide range. The results were com-pleted by gas
chromatography and gas adsorptionexperiments as well as a
statistical analysis ofseveral reaction parameters. In a second
part, acontrolled growth procedure will be presented,which allows a
given particle size of up to several,urn in diameter having an even
narrower size dis-tribution to be prepared exactly. Moreover, a
con-tinuous production process of such mono-dispersed silica
particles will be described.
2.2 Light scatteringThe growth of particles in the solution
wasfollowed with a very simple light scattering device.An
IR-sensitive photoelectric barrier was used forthis purpose. The
emitter and detector were cutand adjusted to fit into two NMR
tubes. Figure 1shows the schematic drawing of the device. Emit-ter
and detector were positioned at an angle of 900and an additional
shielding prevented any directoptical path between them. The device
fitted intothe reaction flask and the particle growth reactioncould
be recorded in situ. After a careful calibra-tion with various
suspensions having a knownparticle size and concentration, the
actual particlesize and concentration during the reaction or
thereaction conversion, could be calculated from thedetector
signal. The light-scattering intensity wasapproximately
proportional to the particle massover most of the evaluated range.
Further detailsare listed elsewhere.13
2 Experimental
2.1 Particle preparation (one step)Tetraethylorthosilicate
(TEOS; Wacker, Burghausen,Germany) was distilled immediately before
use.Prior to the distillation, CaO was added in orderto capture HCI
impurities, which often remainedfrom the synthesis of TEOS. All
other chemicalswere used without further purification. Ethanol(E.
Merck, Darmstadt, Germany) was denaturedwith methyl ethyl ketone.
Ammonia solution(25%; E. Merck, Darmstadt, Gennany) was ana-
2.3 TrimethylsilylationTrimethylsilylation, also called
'end-capping', is awell-established characterization method for
lowmolecular weight silicate units?9-32 The reactiontransforms
Si-OH groups into non-polar (hydro-phobic) Si-O-Si(CH3)3 groups.
The latter groupsare very stable with temperature or towards
anyfurther reaction. Due to the hydrophobic charac-
-
equipped with a capillary column, OV-l, and aflame ionization
detector was used for the GCanalysis. The carrier gas was nitrogen
at a pres-sure of 0.077 MPa and the Fill (flame ionizationdetector)
gases, hydrogen and oxygen, wereadjusted to pressures of 0.05 MPa
and 0.10 MPa,respectively. The temperature of the injectionblock
was 573 K (300°C) and that of the detectorwas 623 K (350°C). The
best separation of thedifferent silicate species was achieved by a
temper-ature program starting at 388 K (115°C) and in-creasing at a
heating rate of 10 K min-1 up to afinal temperature of 573
K(300°C). The retentiontimes and Fill response factors were in
accor-dance to results by Garzo et a/.3D and Schubert,32who worked
at the same GC conditions. Thelatter results could be used to
verify and charac-terize all chromatographic peaks.
\ \
Emitter Detector(a)
2.4 Transmission electron microscopy (TEM)TEM micrographs were
obtained at a Philips(Kassel, Germany) Model EM 300 electron
micro-scope. The acceleration voltage of the electronbeam was. 80
kV. TEM was employed to deter-mine the particle size and structure
of the silicabeads. Copper carrier grids were covered with acarbon
film. One drop of the properly diluted par-ticle dispersion was put
onto these grids anddried. In order to improve the accuracy of
themagnification, the TEM was calibrated with agrating replica
prior to the measurements.
The size and shape of the observed particleswere evaluated from
the micrographs with thehelp of an image analysing system,
KontronVideoplan System (Kontron, Munich, Germany).On a magnetic
board the particle profiles weretraced with a pen. Particle size
and shape was cal-culated from these computerized data (the
pro-jected particle area). The mean particle size wasthe equivalent
of a circle having the same area andthe shape factor was the axial
ratio of an ellipse,having the same inertial moment as the
copiedstructure. Over 200 particles were measured foreach sa~ple
and the standard deviation was calcu-lated from the fit of these
data to a Gaussianprobability distribution. Due to the narrow
sizedistributions of almost all samples, the number ofanalysed
particles per sample was sufficient toguarantee accurate values.
However, the measur-ing method had a certain error in itself, which
waschecked with the help of a tesi sample. Circles ofexactly 3 cm,
2 cm, and 1.5 cm in diameter weremeasured 100 times. The absolute
value for thestandard deviation was about :tQ.016 cm for allcircle
sizes, corresponding to relative errorsbetween :tQ.5% and ::!:1.1%.
Therefore all TEMmicrographs were enlarged to result in
particle
Fig.
(b)
.Sectional drawing, (a) horizontal and (b) vertical, ofthe light
scattering device.
ter of the end-capped reaction products, the lattercan be
separated by extraction with non-polarsolvents. A qualitative and
quantitative gaschromatographic analysis followed.
The experiments shown here were performed inan analogous way as
described by Bechtold eta/.3! After various periods of time from
the startof the hydrolysis, samples were taken from thereaction
mixture and added under vigorousstirring to the same volume of
trimethylchlorosi-lane. The mixture was cooled in an ice bath,
sincethe reaction had a very exothermic character. Twolayers formed
and the water-insoluble productswere extracted with methylene.
chloride, washedwith water and dried with Na2S04' After
filteringoff the Na2S04, methylene chloride (solvent)and
hexamethyldisiloxane (the self-condensationproduct of
trimethylchlorosilane) were remvedby evaporation. The 'frozen'
particle sizes couldbe inspected by TEM in the so-prepared
samplesand the low molecular weight building units
werecharacterized by gas chromatography (GC).
A gas chromatograph 4300 (Carlo Erba)
-
diameters of 2 to 3 cm on the photographs, whichreduced the
error of the method to less than 1%.Nevertheless, the latter effect
has to be taken inmind for some of the samples having a
standarddeviation of 2% or less.
volume could be determined with an accuracy of::to. 03 cm3,
corresponding to a relative error of0.5-1 %.
The apparent water density, dapp(H2O), wasmeasured at 298 K
(25°C), using a small pyc-nometer of ,.. 0.8 cm3 volume. The
reproducibilityof these density measurements was about ::tO~
15%.
3 Results and Discussion
The silica precipitation reaction was studied overthe following
range of starting conditions: TEOS0.1-0.4 mol dm-3; NH3 0.8-4.2 mol
dm-3; H2O3.0-13.0 mol dm-3, and reaction temperature293-333 K
(20-60°C). Table 1 presents the experi-mental results in terms of
particle properties andreaction kinetics. The systematic set-up of
thereaction conditions allows for a statistical analysisof all
parameters and various combined effects.The factorial analysis was
performed accordingto the general procedure described by
Davies.34Further details are described elsewhere. 13 The
parameters were analysed in terms of a linear orquadratic effect
and the various combined effectsof these parameters on the observed
characteristicproperty. Tables 2 and 3 summarize the results ofthe
calculations. As shown earlier, the precipita-tion reaction was
relatively independent of theexperimental procedure (e.g. mixing
time orstirring speed), provided that the reactants werethoroughly
mixed within a reasonable time (lessthan the induction period,
described in furtherdetail later). Nonetheless, the highest degree
ofreproducibility was assured, when the procedurewas performed in a
very controlled way, as de-scribed in the experimental part. As an
example,
2.5 Gas adsorptionNitrogen sorption isotherms were determined
at77 K (-196°C), using a microbalance (model 4102or 4434;
Sartorius, Gottingen, Germany). Prior tothe nitrogen adsorption
samples were degassed at473 K (200°C) until the sample mass
remained con-stant (10 h, final vacuum lQ--4 Pa). Nitrogen was
of99.999% purity (Linde, Dusseldorf, Germany). Thesaturation vapour
pressure of nitrogen was mea-sured close to the sample, employing a
speciallyconstructed device13.33. The pressure was. recordedwith a
piezo-resistive pressure transducer (Model4043Al; Kistler,
Ostfildern, Germany). The specificsurface area was calculated from
the nitrogen sorp-tion data according to the BET equation, using
amolecular cross-sectional area of 0.162 nm2 for thenitrogen
molecule.
Sorption measurements of helium, argon, kryp-ton and xenon
(purity> 99.99%, Messer,Griesheim, Germany) were performed in a
similarway on the dried and calcined silica samples at298 K (25°C),
using a magnetic suspension bal-ance (Model 4201; Sartorius,
Gottingen, Ger-many) and a capacitive-resistive membrane pres-sure
transducer (Model Baraton 220B; MKS).
2.6 Apparent helium and water densitiesThe apparent He density,
dapp(He), was deter-mined on a helium pycnometer (Model
930;Beckman, Munich, Germany). The total volumeof the sample cell
was ::= 10 cm3 and the sample
800 I:
/
Ec
---600~QI...QI 400E=is 200
_ ::"""""" 2 0. -'" 00 ~ 2.5 .~. ~
O~ ~~0 / 0.8 ~~«"\ ~O -
3 8 13 IW . I I d .3 It'
ater concentration mo m 293 313 333
Fig. 2. Three-dimensional plot of the final particle diameter
versus the water and ammonia concentration of the reaction
mixtureat a constant TEOS concentration of 0.2 mol dm-3. The left,
middle and right bars of each data triple indicate results at
reaction
temperatures of 293, 313, 333 K (20, 40, 60°C),
respectively.
-
Table 1. Particle properties (particle diameter, standard
deviation of the size distribution and shape factor) and
experimental dataof the reaction kinetics (first-order reaction
rate, kTEOS, and induction time) listed for the various reaction
mixtures
Reaction conditions Particle properties
H2O(mol dm-3)
NH3(mo/dm-3 )
TEOS(mol dm-3)
Temperature(K)
Diameter(nm)
Shapefactor
kTEOS( 10-3 S-J)
Reaction kinetics
Induction time(JoJ s)
Standard deviation(nm)
3333333888
1313131313131333388888
1313133333333888
13131313131313
0.80.80.80.80.80.80.80.80.80.80.80.80.80.80.80-80.82.52.52.52.52.52.52.52.52.52.52.54.24.24.24.24.24.24.24.24.24.24.24.24.24.24.24.24.2
0-10-20-40-20-10-20-40-20-20-20-10-20-40-20-10-20-40-20-20-20-20-20-20-20-20-20-20-20-10-20-40-20-10-20-40-20-20-20-10-20-40-20-10-20-4
293293293313333333333293313333293793293313333333333293313333293313313313333293313333293293293313333333333293313333293293293313333333333
27426716114688.78786.3
665309186283555692310166240224510311185633550572537260
550320261588660660421242300395730534310417602600466271283308
14.226.725.612.610.08.58.3
25.015.210.531.715.516.012.415.312.89.0
28.025.014.811.311.712.110.920.011.713.8
16.722.740
0.9590.9530.9240.9360.9020.8830.8900.9500.9600.9630.9220.9630.9490.9610.9460.9600.9650.9250.8960.9350.9640.9600.9550.9580.9670.9620.9680.9680.9440.880
Agglomerate0.9290.9600.9430.9270.8700.9550.9610.9490.963
Agglomerate0.9540.9620.9660.967
0.166 .63
0.124 15
0.85 0.79
0.2540.345}.07
0-2900-3000-250
0.61 0.176
89 0.121
2.14 0.101
0.650.4790.970.711.662.381.754.202.252.997.40
0.4320.3910.2870.1530.0740.0860.0790.0700.1080.0520.034
0.373 0.300
25.710.830.549.85032.015.08.29.8
09 0.198
0.92 0.188
1.112.296.3
0.121
0.0580.057
2.12 0.078
8.09.111.48.6
5.9 0.0408
13.4 0.0273
one experiment was repeated five times and theresulting mean
particle diameter was consistentwithin :t2.5%.35
combined effect of ammonia (lin)/water (lin) con-centrations was
suggested with a probability of>95%, whereas effects of water
(sq)/temperature(lin) or water/TEOS (lin) were observed at a
levelof only 90%. All other parameters or combina-tions seemed to
be of minor importance. Irrespec-tive of these conclusions, the
statistical analysiscan only show that certain parameters
influencethe particle size, but it can not tell how much orin which
direction the size will be influenced. Thethree-dimensional graph
in Fig. 2 can give a muchbetter answer to the last question. With
decreasingtemperature or increasing ammonia concentrationparticles
became larger, whereas the effect of the
3.1 Particle sizeThe statistical analysis (see Tables 2 and 3)
indi-cated the significance of several experimental pa-rameters on
the final particle size. The reactiontemperature (lin; linear)
ammonia concentration(lin), and water concentration (sq;
quadratic)clearly influenced the particle size, as shown bythe high
variance ratios (above a level of signifi-cance of 99%). The
importance of the water con-centration (lin), TEaS concentration (I
in) and the
-
Table 2. Factorial analysis of the experiments shown in
Table
Reciprocal effect Variance ratio
Induction timeShapeSize
42-180.058.70
20.92166.31
1.7910.740.940.790.400.410.081.62.0.570.095.080.020.921.20
Standard deviation kTEOS
NH3(lin) 2.26NH3(sq) 1.20H2O(lin) 28.76H2O(sq)
1.47Temperature(lin) 2.61Temperature(sq) 0.22NH3(lin)/H2O(lin)
0.11NH3(sq)/H2O(lin) 0.16NH3(lin)/H2O(sq) 2.27NH3(sq)/H2O(sq)
1.58NH3(lin)/Temperature(lin) 13.70NH3(sq)/Temperature(lin)
0.34NH3(lin)/Temperature(sq) 0.25NH3(sq)/Temperature(sq)
2.70H2O(lin)/Temperature(lin) 0.00H2O(sq)/Temperature(lin)
4.08H2O(lin)/Temperature(sq) 0.04H2O(sq)/Temperature(sq)
0.78NH3(lin)/H2O(lin)/Temperature(lin) 7.04
The value pf the variance ratip indicated the prpbability of a
reciprocal effect on the evaluated parameter. Experimental
parame-ters: TEOS concentration, 0.2 mol dm-3; ammonia
concentration, 0.8, 2.5, 4.2 mol dm-3; water concentration, 3.0,
8.0, 13.0 moldm-3; temperature, 293, 313, 333 K. The abbreviations
(lin) and (sq) represent linear and quadratic reciprocal
effects.The different levels of significance are: low (a = 0.90)
for a variance ratio >3.59; high (a = 0.95) for a variance ratio
>5.59; veryhigh (a = 0.99) for a variance ratio> 12.20.
96.710.24
[55.492.97
[19.576.09
51.610.140.280.53
41-790.321.950.53
57.360.022.350.12
31-23
92.701.5'66
1.41.,95
24'74
1.4'590.21
68.8711.711.2.32
1.735.271..1.1
0.020.048.000.012.820.017.38
7.101.06
10.201.126.210.715.520.271.592.883.091.020.080.000.093.230.020.280.04
water concentration clearly indicated a maximumof the particle
size around a concentration of 8mol dm-3. The results shown here
are in agree-ment with the original work of Stober et al.2 andother
groups.21,36,37 However, the results presentedhere cover a much
broader range of conditions(e.g. the influence of the reaction
temperature wasnot studied in such a systematic way by others).
3.2 Particle size distributionIn a similar way the influence of
the reaction con-ditions on the width of the particle size
distribu-tion (standard deviation) was analysed. The latterwas
controlled by the same parameters, but theireffect was less
pronounced, as shown by the lowervalues of the variance ratios (see
Table 2). Figure3 illustrates the tendency towards a narrower
sizedistribution at higher water concentrations. How-ever, Fig. 3
does not allow the influence of otherparameters like the
temperature, the ammonia orTEOS concentrations to be interpreted;
even so,statistics indicated their effect on a confidencelevel of
> 99 or 95% (see Tables 2 and 3). A lowerlimit of about :tlO nm
was observed for the abso-lute value of the standard deviation,
irrespectiveof the mean particle size of the samples (valid onlyfor
the samples shown here!). Yet it has to bementioned that the size
distributions could besmaller than :tlO nm, when the mean particle
sizefell below lOO nm at even lower ammonia orwater concentrations,
or when a controlled growthprocess was applied,13.38 as described
in part twoof this study.
Table 3. Factorial analysis of the experiments shown in Table
I
Reciprocal effect Variance ratio
Size Standard deviation Shape
23.'3.
46.4.1
0.1
ll.j0.'O.0'10-1
~.,O.'001
O.
6.215.368,617'350.236'980.020.01
11'640.142.601.28
'0.950.43
1.145.693.327-351.550.08
12-431.457-260.421,131.27
'5.561.93
3.3 Particle shapeThe 'shape factor' of the silica particles was
evalu-ated from the TEM micrographs as described inthe experimental
part. TEM pictures could giveonly a two-dimensional view of the
particles and itis probable that the actual shape was
somewhatdifferent from the two-dimensional value. How-
NH3H2OTemperatureTEOS (lin)TEOS
(sq)NH3/H2ONHfiemperatureH2O/Temperature'NH~EOS (lin) ,','
NH3/TEOS{sq)"H2O/TEOS l~iJ\)"H2O/TEOS (~q) .,Temperature/TEOS
(lin)Temperature/TEOS- (:Sq):
, ' ,
Th~ y:a~ue or the variance ratio indicated the probabi.lity of
areciprocal eftec(on the evaluated parameter.
Experimentalparameters: TEOS concentration, 0-1, 0.2, 0.4 mol dm-3;
am-monia concentration, 0.8, 4-2 mol dm-3; water concentration,3.0,
13-0 mol dm~3; temperature, 293, 333 K. (lin), Linear;(sq),
quadratic.The different levels of significance are: low (a = 0.90)
for avariance ratio >2.73; high (a = 0.95) for a variance
ratio>3.74; very high (a = 0.99) for a variance ratio
>6.71.
763912
5659~37430[)6[)345176951
-
/
//
Ec 50---
/f
40
30
~I~
c0
"..~">Q)
"0
"E~"0C~...(/)
~
c,' o~ 2.5 .~c; ~
~ ~~0.8 ~~O ~o'~ \ "
20
10
Ta3 8 13
.-3 I .\Water CQncentratlon / mol dm 293 313 333 K
Fig. 3. Three dimensional plot of the standard deviation (number
distribution of the particle diameter) versus the water and
am-monia concentration of the reaction mixture at a constant TEOS
concentration of 0.2 mol dm-3. The left, middle, and right bars
of each data triple indicate results at reaction temperatures of
293, 313, 333 K (20, 40, 60°C), respectively.
tion (I in) were also significant at a level of >95%.Figure 4
presents the results in a similar three,.di-mensional graph as was
shown before for themean particle size values of the samples. It'
wasobvious from this diagram that the produced par-ticles tended to
be more spherical at higher waterconcentrations. Nonetheless, the
results have to beinterpreted with great caution, since the shape
wasclosely related to and 'influenced by' the mean.particle size.
In general, small particles « 1 00 nm0 (diameter» tended to be not
as spherical asparticles of larger sizes. ,The explanation was
verysimple, inasmuch as inhomogeneities appearingduring the early
stages of the particle precipitationcan be equalized during the
latter growth process.On the other hand, conditions which
producedlarger sizes (>600 nm 0) were often accompanied
ever, in contrast to needles or platelets, there wasno preferred
particle orientation to worry about,when these particles were lying
on a flat surface.Due to the (relatively) spherical shape of nearly
allparticles, only minor differences between the mea-sured and
actual shape were expected.
The particle shape could deteriorate from thatof a perfect
sphere due to agglomeration processesor inhomogeneous conditions
during the growthperiod of the particles. As indicated in the
facto-rial analysis, the water concentration (lin),
TEOSconcentration (lin) and the combined effect of am-monia
(lin)/temperature (lin) or ammonia/TEaSconcentration (lin) had a
highly significant infl-uence (>99%) on the shape. Furthermore
the threeparameter interaction of water
(lin)/ammonia(lin)/temperature or temperature/TEOS concentra-
///
(
(
(
(
Q) (
i"(.c(/) (
I-
/
/;
III
2 (,.£~
(,0 ~2.5 .~. b-~
A(" ~
~- 0'...~ \ ~ (-
~IIJ
3 8 13 1111.-3 I .,
Water concentration I mol dm 293 313 333 K
Fig. 4. Three-dimensional plot of I-shape parameter versus the
water and ammonia concentration of the reaction mixture at
aconstant TEOS concentration of 0.2 mol dm-3. The left, middle and
right bars of each data triplet indicate results at reaction
temperatures of 293, 313, 333 K (20, 40, 60°C),
respectively.
).14
).12
).10
).08
).06).04
).02
-
by agglomeration and the latter process, likewise,deteriorated
the particle shape.
3.4 Particle structureThe silica powders were X-ray amorphous,
evenafter a calcination at 1373 K (1 100°C) for severaldays. Only
at calcination temperatures above 1573K (1 300°C) was
a-cristobalite formed.
DT A and TG analysis indicated the loss ofadsorbed water (::= 5
wt %) below 473 K (200°C)and an increasing dehydroxylation process
(::= 5 wt%) at higher temperatures (up to 1073 K (800°C)).
Moreover, considerable amounts of ammoniawere strongly bound to
or incorporated in thesilica particles. For example, a significant
shift ofthe pH of an aqueous dispersion was noticed withtime. This
process was caused by the release ofammonia from the silica
particles. Several washingcycles or a calcination of the powder at
623 K(550°C) reduced the pH shift, but it could not beeliminated
completely. An explanation for thisbehaviour would be the
incorporation of ammo-nia in the silica particles during their
synthesis.The very slow release process may be attributed toa
strong adsorption potential andlor to exceed-ingly small pore
openings (of the same size asNH3) in the particles, hindering the
escape of theammonia molecule from the particle core.
In a similar way nitrogen sorption measure-ments-also indicated
an ultramicroporous structureof the particles.13,39 Nitrogen
sorption isothermswere determined point by point by measuring
the'equilibrium' adsorbed mass of nitrogen at therespectivead]usted
relative pressures. However, itproved to be difficult to determine
the trueequilibrium values. At a chosen gas pressure theadsorbed
amount reached an approximate plateauvalue within a few minutes;
nevertheless, a veryslow but continuous further uptake was
noticedover more than two days. In addition, the de-sorption branch
of the isotherm did not meet theadsorption branch at low relative
pressures, indi-cating the inaccuracy of the determined datapoints.
In consequence, the calculated BETsurface areas were somewhat
larger than the cal-culated geometrical surface area of the silica
parti-cles. However, a better correlation was noticedwhen the
specific surface area was calculated frommercury porosimetry'
measurements39 or in caseswhen the adsorption data were determined
withina short time (less than 1 h for all data points inthe BET
region; relative pressure: PI Po < 0-6).Likewise, sorption
studies, which used differentinert gases at room temperature, also
indicatedultramicropores.40 The adsorbed amounts and thekinetics of
the adsorption process indicated thatsome gas molecules were
allowed to penetrate
the particles whereas others were excluded. Forexample, the
adsorbed amount of helium or argonwas roughly 10 to 100 times more
than theamounts of krypton or xenon under the sameconditions,
suggesting a cut-off pore size of about0.3 nm.
Another fact which can be an indication for theporous structure
of the particles was the density ofthe powders. Densities
determined with the heliumor water pycnometer were ~ 2.0 g cm-3 for
alloriginal silica samples. On calcination at ~1073 K(800°C) it
increased to the literature value ofamorphous silica (2.2-2.25 g
cm-3). Since no crys-talline phases developed during this
calcination,the density change was obviously due to some
re-arrangement within the particles, which, on theother hand, was
only possible when the originalparticles had a porous structure.
The low densityof the original Stober silica particles was
alsoobserved by several other groups. 3. 10,21,36 For
example, van HeIden et al.36 reported densities ofas low as 1.61
g cm-3 for some of their silica sam-ples. The latter density would
suggest a porosityof about 25-30 vol. %.
Last but not least, some of the silica samplesdisplayed a
particulate structure on TEM micro-graphs. An example is shown in
Fig. 5. The size ofthe subunits was well below 10 nm, but it
variedfrom sample to sample and quite often no struc-ture was
visible at all.
3.5 Precipitation reactionIn principle, the reaction of TEOS to
form silicacan be divided into two major steps: the hydro-lysis and
the condensation reaction. It is nowthe purpose of the following
paragraphs to givesome further details of this reaction; even
so,the complete reaction is still far from a fullunderstanding,
since there are numerous reactionintermediates and possible
reaction pathways toaccount for.
3.5.1
Reaction intermediatesIn this study the trimethylsilation
technique wasused to determine the reaction intermediates'.
Lowmolecular weight silicate units of up to 20 Si unitscan be
identified by gas chromatography in thisway. Furthermore the
overall particle growthreaction was stopped, since all reactive
sites(hydroxyl groups) on the surface of existing nucleiwere
blocked by the trimethylsilyl groups.
Several reaction mixtures, according to Table I,were studied in
further detail by the trimethylsila-tion technique. Irrespective of
differences in theoverall reaction rates, the percentages among
thelow molecular weight units did not change verymuch with reaction
time and no characteristic
-
3.5.2 Particle growthThe particle size was 'frozen' immediately
afterthe addition of (CH3)3SiCI. The particle growthreaction could
be followed by stopping the reac-tion after various times. The TEM
micrographs ofthe latter samples indicated a monomodal particlesize
distribution for all reaction times and nosmaller subunits or
nuclei were observed. Figure 6presents a plot of the particle size
versus the reac-tion time for one sample. A first-order reactionwas
expected .and the corresponding plot of theconverted data (second
y-axis in Fig. 6) resulted ina linear relation of the data points,
thus verifyingthe previous assumption.
Fig. 5. Transmission electron micrographs of silica
particles,demonstrating the nanometer-sized substructures.
(Original
magnification x 666000.)
trend was noticed. Probably differences existed forthe very
first reaction period « 1 min), but it wasdifficult to take samples
during this time and thescattering of the obtained data prohibited
any fur-ther conclusions. In particular, 70-90% monomer,10-25%
dimer,
-
400Legend (see tab.1 ):
13.0/4.2/0.2/333
13.0/4.2/0.1/293
3.0/0.8/0.1/333
=!~>200
.~..C~.5150=.5~ 100::.u..~ 501
8.0/0.8/0.2/333
r;
0
~~
I/
!
00 100 200 300 400
Time I minFig. 7. Light-scattering intensity (reaction
conversion) as a function of time. The legend refers to the
reaction conditions as.setout in Table 1. The numbers represent
H2O, NH3, TEOS conc. and temperature. The inserted diagram shows
the curves on an
expanded time scale, demonstrating the large differences in the
induction period.
always increased. The corresponding three-dimen-sional graph
(Fig. 8) of the reaction rate versuswater and ammonia concentration
at the threetemperatures clearly demonstrates this
relation-ship.
Likewise the data were taken to determine thereaction orders
relative to the water and ammoniaconcentration. The logarithm of
the reaction ratewas plotted versus the logarithm of the water
orammonia concentration respectively. Each set ofthree data points,
which were obtained under
otherwise equal conditions, were connected by aline and the
average value of the slope of all ofthese curves corresponded to
the reaction order ofthe viewed parameter. A reaction order of
0.97and 1.18 was observed for the ammonia and thewater
concentration, respectively. The latter valuesagreed well with the
data given by Harris et al.19ammonia: 0.9 and water: 1.5.
In a similar way the activation energy of theparticle growth
reaction was determined by plot-ting the logarithm of the reaction
rate versus the
/
:0 @~0 16 / /...14 '/ /)12 //~ 10f ,-
i 81= 6
~
; / ..8 ..8 ,II0 1...1...1...,...,...,...1...,...,...,.,
,...,.., ,..-'
3 8 13
4.2O'
o~2.5 CI ~.~~ ~
o~ b"0.8 ~~ ~c'
~ \
/
~ I~~
,.8
c0
.~u~Q)~
.-oS I .,Water concentration / mol dm 293 313 333 K
Fig. 8. Three-dimensional plot of the particle growth reaction
rate, kTEOS' versus the water and ammonia concentration of the
re-action mixture at a constant TEOS concentration of 0.2 mol dm-3.
The left, middle and right bars of each data triplet indicate
results at reaction temperatures of 293, 313, 333 K (20, 40,
60°C), respectively.
-
small systematic difference towards higher valueswas seen in the
data of van Blaaderen et al.,28whereas the reaction rates published
by Bogush &Zukoski25 always exhibited slightly lower valuesthan
calculated. On the other hand, within eachset of literature data
the linear relation betweenexperimental and calculated values was
excellent,as indicated by the correlation coefficients of r =0.997
and 0.994, respectively. Nonetheless, thegeneral agreement was very
remarkable, since theliterature data were obtained by completely
differ-ent techniques and in addition, the reaction ratesused for
the evaluation of eqn (1) and the valuesof the constants, k' and Ea
were determined by thevery simple light-scattering device described
in theexperimental part.
inverse of the reaction temperature (in K). Theobserved value of
EA = 27 kJ mol-I (6.5 kcalmol-I) fell within the range of various
literaturedata41.43-47 for the polymerization or
precipitationreaction in silica systems.
The growth reaction rate can now be summa-rized as follows.
The constant, k', was calculated from the reac-tion rates giv~n
in Table 1 according to eqn (1).Its average value was: k' = 2.36
s-l(mol dm-3)-2'15.Moreover, reaction rates published in the
litera-tureI7.19.25.28 were compared with the
correspondingvalues
calculated from eqn (1), shown in Fig. 9.Although the overall
agreement was adequate, a
3.5.4 Induction periodThe induction period was evaluated from
thelight-scattering curves as well. It was defined asthe time
interval until the scattering intensity devi-ated from its starting
value (derived by extrapola-
5
"';'"U)M
I
0
3
2 .I
-co~~G)
E--..G)Co
1
0.5
8,..>< 0.3Q)
I(/) 0.20UJI-
~
..
.~
0.110.1
0.2
2 3
10-351
50.3 0.5 1
kTEOS -calculated I
Fig. 9. Plot of SiO2 particle growth or TEOS hydrolysis rates,
as determined experimentally by Bogush & Zukoski,26
vanBlaaderen et al.,28 Harris et a[.l9 or Matsoukas &
Gularil7.18 versus the corresRonding calculated values according to
eqn (I).Dotted lines represent the linear regression of the data of
Bogush & Zukoski 6 and van Blaaderen et al?8 having
correlation
coefficients r = 0.994 and 0.997, respectively.
X [H20]1'18 [NH3]O.97 [TEOS]
-
/
/~ 2.0
r-1.5G)
E~ 1.0'88 ... 4.2c,.
O~2.5 ..e.Ci ~~ b-~
0.8 ~~O ",0'~ \'~
/
8..III
I.., leeO.FI
tion). This time varied over a wide range from 27 sup to about
30 min. Influences on the inductionperiod were evaluated by a
statistical analysis, asshown in Table 2. The linear water and
ammoniaeffect and their combination were of highest signi-ficance
as shown by their very high varianceratios, but clearly many other
parameters or com-binations thereof were also of importance
(atprobabilities of >99% or >95%, see Table 2). Dueto the
various quadratic and combined effects,being of:a certain
statistical importance, the over-all conclusions were not as clear
cut as the analo-gous results for the reaction rate. The
three-dimensional plot (Fig. 10) of the induction periodversus the
water and ammonia-concentration ortemperature graphically
illustrated the main re-sults. The induction period decreased with
increas-ing water and ammonia content or temperature. Acomparison
of Figs 8 and 10 now clearly demon-strated the inverse behaviour of
induction periodand reaction rate. This relation was
understand-able, since there was no indication that theseparameters
would be controlled by completelydifferent reaction mechanisms. It
was obvious thata fast overall reaction would most probablyshorten
the 'nucleation' or 'induction' period too.
exaIt)ple of the base-catalysed silica reaction, is anideal
example to study the formation of mono-dispersed particles. At
present there are two majormodels to explain the particle formation
processin the Stober silica system. First, there is themonomer
addition growth model, IS in analogy to
a 'LaMer' precipitation diagram,4S which describesthe nucleation
by exceeding the supersaturationlimit and the growth by
condensation ofmonomeric silicic acid on the surface of the
exist-ing particles (nuclei). It is mainly a chemical typeof
explanation, focusing on the hydrolysis andcondensation rate and
the solubility of silicic acid.Second, nucleation and growth of
particles ismodelled by a controlled aggregation mechanismof
sub-particles, a few nanometer in size}6 Col-loidal stability,
nuclei size, surface charge anddiffusion and aggregation
characteristics are theimportant parameters in this model. The
twomodels seem to be nearly incompatible in severalaspects and it
should be easy to distinguish be-tween them. However, there were a
number of ob-servations, shown here or in other publications,which
clearly support either the first or the secondmodel.
3.5.5.1
Monomer addition growth model. In thefollowing section,
experimental results of thepresent study will be discussed in
relation tothe previously mentioned models and comparedwith
literature data. For example, the particlegrowth could be best
described by a first-order re-action with respect to the TEOS
concentration.Even so, there are several publications favouring
asecond-, third-, or fourth-order reaction for thepolymerization of
silicic acid (for further details,see ller49 or Rothbaum &
Rohde41). Most of
3.5.5 Comparison with modelsMany papers have been published
concerning thehydrolysis and condensation reaction in thesilicate
system. The majority of them have beendealing with the reaction
under acidic conditions,when silica polymers or gel structures are
formed.On the other hand, the base-catalysed reaction,especially in
the presence of ammonia, has beenstudied only recently in more
detail.17-28 In addi-tion, the Stober process, as the most
prominent
o. o~' , .I .., I .., I , ..I ., , I .., I .., I , , , I .., I
., , I .., I ., .1 ...I , ..I ...I ., , I ., .I ...I , ..I
.,..7'
3 8 13. I -3 I .,
Water concentration mol dm 293 313 333 K
Fig. 10. Three-dimensional plot of the induction time,
determined by extrapolation from the light-scattering curves,
versus thewater and ammonia concentration of the reaction mixture
at a constant TEaS concentration of 0.2 mol dm-3. The left,
middle
and right bars of each data triplet indicate results at reaction
temperatures of 293, 313, 333 K (20, 40, 60°C), respectively.
-
201
the recent publications, 17-19,25,28 concerning the'Stober'
silica precipitation, agreed with the resultsshown here and
describe a first-order hydrolysis ofTEOS as the rate-limiting
process in the silicaparticle precipitation. The second reaction
step,the condensation reaction, was found to be fasterby at least a
factor of three.25,26,28
Moreover, the appearance of an induction periodwould also favour
a LaMer-type mechanism, sinceit will take a certain time for the
hydrolysis ofTEOS to exceed the equilibrium or the
criticalnucleation concentration of silicic acid in thesolution
followed by precipitation of silica. Yetthe latter hypothesis is
somehow incompatiblewith the observations of an induction period
inseeded growth experiments by van Blaaderen etal.28 or Philipse.5
In these experiments silica seedswere prepared and grown in the
same solution,and therefore the silicic acid concentration shouldbe
already at or above its equilibrium value, andconsequently, as soon
as some further TEOSwas hydrolysed, the condensation reaction or
theparticle growth reaction should take place.Nonetheless, an
induction period was again ob-served. This discrepancy between
expected andobserved behaviour can be explained. Fleming46studied
the diSsolution and precipitation rate ofsilica in dependence on
the free silicic acid concen-tration near its solubility
equilibrium. His resultsindicated 0. that the precipitation rate
decreased
linearly with decreasing silicic acid concentration.However, the
point where the deposition ratedropped to nearly zero was observed
at a silicaconcentration which was about twice its equilib-rium
value. Thereafter a very slow drift towardsthe final equilibrium
was noticed. Fleming ex-plained this drift by a different
solubility of freshand aged silica. It follows for the case of
seededgrowth experiments that such ~ drift will occur ina similar
way after the first step, the preparationof the seed particles.
Obviously it will take acertain amount of TEOS to be hydrolysed,
beforethe concentration of free silicic acid reaches the'first
equilibrium' again and the precipitation orthe particle growth
starts after exceeding thisvalue.
3.5.5.2 Aggregation growth model. On the otherhand, there were
several results which referred toan agglomerated structure of the
particles. Themost prominent evidence was the structures seenin
some of the TEM micrographs (see Fig. 5), butalso the measured
densities (up to 30% lower thanthe theoretical value), the
incorporation of ammo-nia in the particles, the high specific
surface area,the slow kinetics in nitrogen adsorption
measure-ments, or the observed cut-off size in adsorption
experiments using different inert gases.40 All ofthese were
clear indications of an ultramicro-porous structure of the silica
powders. Thequestion remained, how such a porous structuredeveloped
during the synthesis.
The simplest explanation was given by an ag-gregation growth
model, described for example byBogush & Zukoski26 or Philipse.5
Small silicanuclei of several nanometer in size form continu-ously
during the hydrolysis and condensationreaction of TEOS. These
nuclei are colloidallyunstable and they will coagulate and form
largerunits. These are stable above a critical size andtheir
further growth is accomplished by additionof subsequent primary
nuclei to the existing'particle' surface of the agglomerates. The
numberand size of these units will be determined not onlyby the
reaction kinetics, but also by differentparameters effecting the
dispersion stability, likethe ionic strength of the solution,
temperature,charges on the particle surface, pH, solvent
prop-erties (viscosity, dielectric constant, etc) and so on.
Bogush & Zukoski25,26 could explain several oftheir
observations with this model. For example,the final particle size
increased from 343 nm to710 nm 0, when the precipitation
mixturecontained 10-2 mol dm-3 NaCl; even so, thereaction rate
remained nearly constant at kTEOS =1 X 10'-3 S-I. This observation
was explained bythe lower colloidal stability in the system
contain-ing the higher NaCl concentration (higher ionicstrength).
Due to this, a smaller number of aggre-gate units were formed at
the beginning and con-sequently particles grew larger in size.
Further evidence for such a picture of particlesmade by
aggregation of primary nuclei was givenby results of van HeIden et
al.36 They repeatedlyobserved a second peak in SAXS measurementson
Stober silica particles. The authors questionedwhether this result
was real or an artefact. Theposition of the peak was always noticed
around astructural size of 1 nm. The latter value for thecritical
size of silica nuclei was in very goodconformity with calculations
by Makrides et al.45Depending on the chosen values for the
interfacialenergy (silica/solution), the silica concentrations,and
the temperature, the authors calculated acritical size of about 1
to 2 nm.
The formation of these primary nanometer-sizedsubunits as an
intermediate step in the growthreaction could also account for the
observedinduction period. The condensation reaction canproceed
parallel to the hydrolysis reaction fromthe very first beginning
and still it will take acertain time (induction time) to produce
theprimary subunits, which then may agglomerate oradhere to the
surface of seed particles.
-
On the contrary, results of the trimethylsilationexperiments
presented here, stayed in clearcontrast to an aggregation growth
model. Mostlymonomer, some dimer and only insignificantamounts of
higher oligomeric silicate units weredetected throughout the whole
precipitation reac-tion. This opposed the production of
nanometer-sized nuclei as requested by the aggregationgrowth model,
since at least a certain amount ofhigher oligo-meric units should
be detected, repre-senting the intermediate steps towards these
pri-mary nuclei. The trimethylsilation data obviouslyfavoured the
monomer addition growth model.However, it has to be noticed that
these resultswere obtained at reaction times when the first
for-mation of particles was already passed (after theinduction
period).
The induction period was inversely related to thereaction rate
and clearly influenced by the same
parameters.The results were compared with particle forma-
tion models and the following mechanism ap-peared to be most
realistic. Nanometer-sized sub-units were formed during the very
first reactionstage, followed by a controlled aggregation tolarger
units. The colloidal stability and aggrega-tion rate of the primary
nuclei determined thenumber and size of the particles. However,
thegrowth mechanism changed at a later stage in thereaction and
thereafter particles grew solely bycondensation of monomeric and
dimeric silicateunits at the particle surface. The point at
whichthe reaction mechanism changed, depended verymuch on the
overall reaction conditions. Similarconclusions were drawn by van
Blaaderen et al.28and Harris et al.19 They also concluded that
theparticle formation started with the aggregation ofsiloxane
substructures and proceeded furtherthrough a surface
reaction-limited condensation ofhydrolysed monomers or small
oligomers. Harriset al.19 assumed the change in mechanism at a
par-ticle size of about I 00 ~m in their experiments.
3.5.5.3 Eden growth model. A different explana-tion for the
development of a porous structure isgiven by Keefer and coworker
.50,51 The nucleationand random growth of clusters from partially
hy-drolysed monomers was simulated. The Edengrowth model, used in
that study, resulted in moreor less porous structures. The fractal
structuresdepended strongly on the assumed degree of hy-drolysis of
monomeric silica units. Fully hydro-lysed monomers resulted in
relative dense particles,whereas the structure became more open
withonly partly hydrolysed silicate building units.
4 Conclusion
Acknowledgements
The author would like to thank Professor K. K.Unger for an
introduction to this fascinating fieldof science and is grateful
for his guidance and sup-port throughout this work. Finally the
DeutscheForschungsgemeinschaft is thanked for theirfinancial
support through Grant Number Un26/25-1.
References
The Stober silica preparation process was studiedover a wide
range of experimental conditions(TEOS, NH3, H2O concentration and
tempera-ture). Particle properties-size, distribution, shapeand
structure-as well as reaction kinetics-growth rate and induction
period-were evaluatedby a statistical analysis. The mean particle
sizewas influenced by all experimental parameters,whereas the
effect on size distribution and shapeof the particles was much less
pronounced. Allsamples were relative monodisperse, but a
trendtowards a narrower distribution or a more spheri-cal shape was
observed at higher water concentra-tions. The particles showed an
ultramicroporousinternal structure, as revealed by TEM micro-graphs
or adsorption measurements.
Analysis of the reaction kinetics revealed thatonly monomeric
and some dimeric silicic acid wasdetected as intermediate silicate
units in the solu-tion. The particle growth could best be
describedby a first-order process relative to the TEOS
con-centration. The reaction orders relative to thewater and
ammonia concentration were 1.18 and
1. Kolbe, G., Das komplexchemi~che Verhalten der Kiesel-saure.
Dissertation, Friedrich-Schiller Universitat Jena,1956.2. Stober,
W., Fink, A. & Hohn, E., Controlled growth ofmonodisperse
silica spheres in the micron size range. J.Colloid Interface Sci.,
26 (1968) 62-9.
3. Van HeIden, A. K. & Vrij, A., Contrast variation in
lightscattering: silica spheres dispersed in apolar solvent
mix-tures. J. Colloid Interface Sci., 76 (1980) 418-33.
4. Philipse, A. P. & Vrij, A., Polydispersity probed by
lightscattering of secondary particles in controlled growth
ex-periments of silica spheres. J. Chern. Phys., 87
(1987)5634--43.
5. Philipse, A. P., Quantitative aspects of the growth of
-
crometre particles. Faraday Discuss. Chem. Soc., 90(1990)
345-57.
24. Kim, S. & Zukoski, C. F., A model of growth by
hetero-coagulation in seeded colloidal dispersions. J. Colloid
In-terface Sci., 139 (1990) 198-212.
25. Bogush, G. H. & Zukoski, IV, C. F., Studies of the
kinet-ics of the precipitation of uniform silica particles
throughthe hydrolysis and condensation of silicon alkoxides.
J.Colloid Interface Sci., 142 (1991) 1-18.
26. Bogush, G. H. & Zukoski IV, C. F., Uniform silica
parti-cle precipitation: an aggregative growth model. J.
ColloidInterface Sci., 142 (1991) 19-34.
27. den Ouden, C. J. J. & Thompson, R. W., Analysis of
theformation of monodisperse populations by homogeneousnucleation.
J. Colloid Interface Sci., 143 (1991) 77-84.
28. van Blaaderen, A., van Geest, J. & Vrij, A.,
Monodis-perse colloidal silica spheres from tetraalkoxysi1anes:
par-ticle formation and growth mechanism. J. Colloid Inter-face
Sci., 154 (1992) 481-501.
29. Calhoun, H. P. & Masson, C. R., In Reviews on
Silicon,Germanium, Tin and Lead Compounds, ed. M. Gielen.Freund
Publishing House, Tel-Aviv, 1981, p. 153.
30. Garzo, G., Hoebbel, D., Ecsery, Z. J. & Ujszaszi, K.,
Gaschromatography of trimethylsilylated silicate anions-separation
with glass capillary columns and new aspectsin derivatization. J.
Chromatography, 167 (1978) 321-36.
31. Bechtold, M. F., Vest, R. D. & Plambeck Jr, L.,
Silicicacid from tetraethyl silicate hydrolysis. Polymerizationand
properties. J. Am. Chem. Soc., 90 (1968) 4590-8.
32. Schubert, P., Der Einfluss der Silikatquelle auf die
Syn-these von ZSM-5-neue Wege zur Synthese und Charak-terisierung.
Dissertation, Universitat Mainz, 1985.
33. Straube, B., Kritische Untersuchung und Standard-isierung
physikalisch-chemischer Messmethoden zurErmittlung der spezifischen
Oberflache und Poren-verteilung von Adsorbentien. Dissertation,
UniversitatMainz, 1985.
34. Davies, O. L., The Design and Analysis of
IndustrialExperiments. Oliver and Boyd, London, 1967.
35. Giesche, H., Untersuchung fiber die Herstellung
undAgglomeration von monodispersen Kieselsauresolen.Diplomarbeit,
Mainz, 198.3.
36. van HeIden, A. K., Jansen, J. W. & Vrij, A.,
Preparationand characterization of spherical monodisperse silica
dis-persions in nonaqueous solvents. J. Colloid Interface Sci.,81
(1981) 354-68.
37. Tan, C. G., Bowen, B. D. & Epstein, N., Production
ofmonodisperse colloidal silica spheres: effect of tempera-ture. J.
Colloid Interface Sci., 119 (1987) 290-3.
38. Unger, K. K., Giesche, H. & Kinkel, J. N.,
KugelformigeSi02-Partikel. DE Patents 3534143 AI, 25 September1985;
3616133 AI, 14 May 1986; US Patents 4911903,27March 1990,4775520,4
October 1988.
39. Giesche, H., Unger, K. K., Muller, U. & Esser, U.,
Hys-teresis in nitrogen sorption and mercury porosimetry
onmesoporous model adsorbents made of aggregatedmonodisperse silica
spheres. Colloids Surfaces, 37 (1989)93-113.
40. Reichert, H., Adsorptionsuntersuchungen an
Si02-Ag-glomeraten. Diplomarbeit, Universitat Mainz, 1988.41.
Rothbaum, H. P. & Rohde, A. G., Kinetics of
silicapolymerization and deposition from dilute solutionsbetween 5
and 180°C. J. Colloid Interface Sci., 71 (1979)533-59.
42. Nielsen, A. E., Kinetics of Precipitation. Pergamon
Press,New York, 1964.
43. Bohlmann, E. G., Mesmer, R. E. & Berlinski, P.,
Kinet-ics of silica deposition from simulated geothermal
brines.Soc. Petroleum Eng. J., (1980) 239-48.
44. Weres, 0., Yee, A. & Tsao, L., Kinetics of silica
poly-merization. J. Colloid Interface Sci., 84 (1981) 379-402.
45. Makrides, A. C., Turner, M. & Slaughter, J.,
Condensa-tion of silica from supersaturated silicic acid solutions.
JColloid Interface Sci., 73 (1980) 345--67.
(charged) silica spheres. Colloid Polym. Sci., 266
(1988)1174-80.
6. Philipse, A. P. & Vrij, A., Preparation and properties
ofnonaqueous model dispersions of chemically modified,charged
silica spheres. J. Colloid Interface Sci., 128 (1989)121-36.
7. van der Werff, J. C. & de Kruif, C. G., Blom, C.
&Mellema, J., Linear viscoelastic behavior of dense hard-sphere
dispersions. Phys. Rev. A, 39 (1989) 795-807.
8. van der Werff, J. C. & de Kruif, C. G., Hard-sphere
col-loidal dispersions: the scaling of rheological propertieswith
particle size, volume fraction, and shear rate. J.Rheol., 33 (1989)
421-54.
9. Jones, D. A. R., Leary, B. & Boger, D. V., The rheologyof
a concentrated colloidal suspension of hard spheres. J.Colloid
Interface Sci., 147 (1991) 479-95.
10. Chen, M. & Russel, W. B., Characteristics of
flocculatedsilica dispersions. J. Colloid Interface Sci., 141
(1991)564-77.
11. Ackerson, B. J., Shear induced order and shear process-ing
of model hard sphere suspensions. j. Rheol., 34(1990) 553-90.
12. Emmett, S., Lubetkin, S. D. & Vincent, B., The growthof
ordered sediments of monodispersed hydrophobic sil-ica particles.
Colloids Surfaces, 42 (1989) 139-53.
13. Giesche, H., Modellporenkorper aus monodispersenspharischen
SiO2-Partikeln: Herstellung, Charakter-isierung und
Sinterverhalten. Dissertation, UniversitatMainz, 1987.
14. Sacks, M. D. & Tseng, T.-Y., Preparation of SiO2
glassfrom model powder compacts: I, formation and charac-terization
of powders, suspensions, and green compacts.J. Am. Ceram. Soc., 67
(1984) 526-32.
15. Sacks, M. D. & Tseng, T.-Y., Preparation of SiO2
glassfrom model powder compacts: II, sintering. J. Am.Ceram. Soc.,
67 (1984) 532-7.
16. Giesche, H. & Unger, K. K., Sintering of
monodispersedsilica. In Proceedings of Ceramic Powder Processing
Sci-ence, 2nd International Conference, Berchtesgaden, 1988,ed. H.
Hausner, G. L. Messing & S. Hirano. DeutscheKeramische
Gesellschaft, Koln, 1989, pp. 755-64.
17. Matsoukas, T. & Gulari, E., Dynamics of growth of
sil-ica particles from ammonia-catalyzed hydrolysis of
tetra-ethyl-orthosilicate. J. Colloid Interface Sci., 124
(1988)252-61.
18. Matsoukas, T. & Gulari, E., Monomer-addition growthwith
a slow initiation step: a growth model for silica par-ticles from
alkoxides. J. Colloid Interface Sci., 132 (1989)13-21.
19. Harris, M. T., Basaran, O. A. & Byers, C. H.,
Theprecipitation dynamics of silica particles. In
CeramicTransactions, Vol. 12, Ceramic Powder Science III, ed. G.L.
Messing, S.-I. Hirano & H. Hausner. The AmericanCeramic Society
Inc., Westerville, OH, 1990, pp. 119-27.
20. Bogush, G. H., Dickstein, G. L., Lee, P., Zukoski, K.
C.& Zukoski IV, C. F., Studies of the hydrolysis and
poly-merization of silicon alkoxides in basic alcohol solutions.In
Mater. Res. Soc. Symp. Proc., Vol. 121, Better Ceram-ics Through
Chemistry III, ed. Jeffrey C. Brinker, D. E.Clark & D. R.
Ulrich. Materials Research Society, Pitts-burgh, PA, 1988, pp.
57-65.
21. Bogush, G. H., Tracy, M. A. & Zukoski IV, C.
F.,Preparation of monodisperse silica particles: control ofsize and
mass fraction. J. Non-Cryst. Solids, 104 (1988)95-106.
22. Bogush, G. H., Brinker, C. J., Majors, P. D. & S~ith,
D.M., Evolution of surface area during the controlledgrowth of
silica spheres. In Materials Research SocietySymposium Proceedings,
Vol. 180, Better CeramicsThrough Chemistry IV, ed. B. J. J.
Zelinski, J. C. Brinker,D. E. Clark & D. R. Ulrich. Materials
Research Society,Pittsburgh, PA, 1990, pp. 491-4.
23. Look, J.-L., Bogush, G. H. & Zukoski, C. F.,
Colloidalinteractions during the precipitation of uniform
submi-
-
46.
Fleming, B. A., Kinetics of reaction between silicic acidand
amorphous silica surfaces in NaCI solutions. J.Colloid Interface
Sci., 110 (1986) 40-64.47. Brady, A. P., Brown, A. G. & Huff,
H., The polymeriza-tion of aqueous potassium silicate solutions. J.
ColloidInterface Sci., 8 (1953) 252-76.
48. LaMer, V. K. & Dinegar, R. H., Theory, production
andmechanism of formation of monodispersed hydrosols. J.Am. Chern.
Soc., 72 (1950) 4847-54.
49.
lIer, R. K., The Chemistry of Silica. Wiley, New York,1979.
50. Keefer, K. D., Growth and structure of fractally roughsilica
colloids. In Better Ceramics through Chemistry II,Mat. Res. Soc.
Symp. Proc., Vol. 73, ed. J. C. Brinker, D.E. Clark & D. R.
Ulrich. Materials Research Society,Pittsburgh, PA, 1986, pp.
295-304.
51. Keefer, K. D. & Schaefer, D. W., Growth offractallyrough
colloids. Phys. Rev. Lett., 56 (1986) 2376-9.
