Synthesis of pure cementitious phases by sol-gel process as precursor

Article abstract of DOI:10.1002/zaac.200400090

Calcium silicates and aluminates are the main constituents of ordinary Portland cement (OPC) and calcium aluminate cements (CAC) and therefore the pure phases are of great importance for the investigation of interactions between binder and additives or admixtures. Additionally, investigations on clinker phases doped with foreign ions enable the improvement of the performance of cements. For this purpose great amounts of pure phases are needed. These phases are hard to synthesize via a solid state reaction of solid educts. Thus there is a need for a new, more efficient route to synthesize these phases. The sol-gel process as precursor provides an alternative to the conventional method. In this paper experimental evidence is presented for an improved synthesis of calcium silicates and aluminates via sol-gel processes, the characterisation of these clinker phases and their hydration behaviour.

Full text of DOI:10.1002/zaac.200400090

Synthesis of Pure Cementitious Phases by Sol-Gel Process as Precursor
Dietmar Stephan* and Patrick Wilhelm
Garching bei München, Anorganisch-chemisches Institut der Technischen Universität München
st
Received March 31 , 2004.
Abstract. Calcium silicates and aluminates are the main constitu-
ents of ordinary Portland cement (OPC) and calcium aluminate
cements (CAC) and therefore the pure phases are of great import-
ance for the investigation of interactions between binder and addi-
tives or admixtures. Additionally, investigations on clinker phases
doped with foreign ions enable the improvement of the perform-
ance of cements. For this purpose great amounts of pure phases
are needed. These phases are hard to synthesize via a solid state
reaction of solid educts. Thus there is a need for a new, more ef-
ficient route to synthesize these phases. The sol-gel process as pre-
cursor provides an alternative to the conventional method. In this
paper experimental evidence is presented for an improved synthesis
of calcium silicates and aluminates via sol-gel processes, the charac-
terisation of these clinker phases and their hydration behaviour.
Keywords: Sol-gel processes; Calcium silicates, Calcium aluminates;
Gels; Solid-state chemistry
Synthese reiner Klinkerphasen über Sol-Gel Prozesse als Vorstufen
Inhaltsübersicht. Calciumsilicate und -aluminate sind die Hauptbe-
standteile von Portlandzement (OPC) und Tonerdezement (CAC).
Die reinen Phasen sind von großer Bedeutung für die Untersu-
chung von Wechselwirkungen des Bindemittels mit Zusatzstoffen
und -mitteln. Darüber hinaus ermöglicht die Untersuchung von
mit Fremdionen dotierten Klinkerphasen die Leistungsfähigkeit
von Zementen zu steigern. Für diese Untersuchungen werden große
Mengen der reinen Phasen benötigt, die sich jedoch nur unter ho-
hem Aufwand über Festköperreaktionen aus den festen Edukten
synthetisieren lassen. Deshalb war es notwendig einen neuen, effi-
zienteren Syntheseweg zu finden, mit dem diese Phasen leicht zu-
gänglich sind. Der Sol-Gel Prozess als Vorstufe für den Brennpro-
zess stellt eine Alternative zur konventionellen Methode dar. In die-
sem Artikel werden experimentelle Daten zur optimierten Herstel-
lung reiner Calciumsilicate und -aluminate, die Charakterisierung
dieser Klinkerphasen sowie deren Hydratationsverhalten beschrie-
ben.
1
Introduction
ite, CaAl O ), mayenite ((CaO) (Al O ) ), belite and fer-
rite.
4
7
12
2
3 7
Portland cement has been known now for more than 100
years and with a world production of 1647 · 10 tons in 2001
Very high temperatures are needed to produce OPC and
CAC clinker and even at 1450 °C only some of the compo-
nents melt. The partial melting is necessary for the rapid
reaction of the constituents to OPC and CAC clinker in
industrial processes. This production process is the reason,
why CAC and OPC contain different clinker phases, even
though not all of them contribute to the strength develop-
ment of the cement. Because of the appearance of a mixture
of multiple binary and ternary phases, the OPC and CAC
clinker systems are quite complex and therefore the use of
pure doped or undoped clinker phases for fundamental re-
search in cement chemistry is of great importance. They
can for example be used for investigations on the hydration
processes of one well defined clinker phase with various ad-
ditives or admixtures or for the prediction of the influence
of minor elements on the reactivity of cement. Normally
these pure phases are synthesized by a solid-state reaction
of the carbonates or oxides at high temperatures. For the
most essential phases this process takes several hours up to
days with several intermediate grindings [2Ϫ4]. The forma-
tion of the clinker phases can be accelerated by packing the
different constituents very close. This is done by forming
pellets from the raw material at very high pressure [5, 6].
6
[
1] it is the chemical product, that is produced in the highest
amount by far. This is due to the outstanding chemical and
mechanical properties for building purposes. In ordinary
Portland cement clinker (OPC) tricalcium silicate
(Ca (SiO )O), dicalcium silicate (Ca (SiO )), tricalcium alu-
3 4 2 4
minate (Ca (Al O )) and tetracalcium aluminate ferrite
9
ϩ
6
18
3
(Ca (Al,Fe ) O ) are the main constituents. In the mul-
2 2 5
tiple phase system of OPC clinker, these four phases always
contain minor and trace elements in different concen-
trations and the terms alite, belite, aluminate and ferrite are
used for the doped versions of the pure phases. In calcium
aluminate cements (CAC) monocalcium aluminate
(CaAl O ) is the most important phase. CAC can also con-
2 4
tain other phases such as monocalcium dialuminate (gross-
*
Dr. Dietmar Stephan
Technische Universität München
Lehrbereich anorganische Chemie
Lichtenbergstrasse 4
D-85747 Garching
E-Mail: dietmar.stephan@bauchemie-tum.de
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483
DOI: 10.1002/zaac.200400090
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
1477

D. Stephan, P. Wilhelm
Via this well established method it is possible to produce
pure phases, but the amount of clinker phases produced is
too small for extensive studies. Hence fundamental research
is often done without pure clinker phases, but with the com-
plex multi phase system of the cement, which makes in-
terpretation of the data difficult and sometimes impossible.
In order to provide a more efficient way for the synthesis
of pure phases, the sol-gel process was introduced and opti-
mized. This process has already been established for the
synthesis of very homogenous types of glass and ceramics
at lower temperatures than normally used and there have
also been made some efforts to produce pure clinker phases.
In all those cases the synthesis was done with sols of SiO2
or Al O and Ca(NO ) was used as a source for CaO in
2
3
3 2
the later product. Goktas et al. [7] and Kerns et al. [8] syn-
thesized amorphous calcium aluminates of different nom-
inal composition via a sol-gel process, where they used alu-
minium sec-butylate (Al(OC H ) ) to produce the colloidal
4
9 3
Al O as a starting material. Calcium silicate glasses were
2
3
Figure 1 Schematic steps for the synthesis of cementitious phases
via a sol-gel process. Colloidal silica or alumina is mixed with
Ca(NO ) and the pH is adjusted to a proper value. During gel-
3 2
prepared by Saravanapavan et al. [9] and Laudisio et al. [10,
1]. They prepared in a first step colloidal SiO as a starting
1
2
material from tetraethoxyϪsilicate (TEOS). Yoshizawa et al.
12] used TEOS and Ca(NO ) as starting material in a
ation a network is formed. When the sample is heated, in a first
step water evaporates to form a xerogel and Ca(NO ) precipitates
[
3
2
3
2
sol-gel process to synthesize Ca (SiO ) at a temperature of
onto the surface of the network. In a second step Ca(NO ) decom-
2
4
3 2
7
00 °C. Nettleship et al. [13] and Hong et al. [14] used the
poses to CaO and reacts with the network to form the cemen-
titious phase.
Peshini-resin process to synthesize Ca (SiO ). Roy et al. [15]
2
4
were the first to synthesize clinker phases via a sol-gel pro-
cess using a commercially available silica sol. A detailed ki-
netic study of Ca (SiO )O formation from sol-gel precur-
3
4
sors was published by Wang et al. [16].
In this paper the synthesis of some of the most important
clinker phases from commercially available colloidal SiO2
and Al O using a sol-gel process as precursor is described.
2
3
The aim was to synthesize Ca (SiO ), Ca (SiO )O,
2
4
3
4
Ca (Al O ), CaAl O and CaAl O in a more efficient way
9
6
18
2
4
4
7
and without stabilising foreign ions. The main phases of
OPC Ca (SiO )O and Ca (SiO ) as well as the main phase
3
4
2
4
of CAC CaAl O were investigated in detail.
2
4
2
Results and Discussion
The synthesis of pure clinker phases via a sol-gel process as
precursor is a series of different steps. Figure 1 shows the
schematic steps from a sol to the final product. The Under-
standing of every single step is of importance to optimize
the synthesis. In the following chapters of this paper exper-
imental evidence is presented that gives a better under-
standing of the synthesis and finally leads to an optimized
procedure for synthesis.
Figure 2 Change in zeta potential for silica and alumina sol with
the pH. At a low pH some of the hydroxyl groups on the oxidic
surface are protonated and therefore the zeta potential is positive.
At higher pH values some of the hydroxyl groups are deprotonated
and the zeta potential becomes negative.
pH of the original solution. The iso electric point (i.e.p.) of
SiO was found at pH 1.4 and for Al O it was 9.8. The
2
2
3
measured values are very close to the ones of Agrawal et al.
17] (1.6 for SiO ) and Healy et al. [18] (8.9 for Al O ). In
[
2
2
3
2
.1 Change in zeta potential and rheological
the case of Al O this is essential for the conditions of gel-
2
3
behaviour of sols with pH
ation, because the gelation takes place fastest near the i.e.p..
The coagulation of the particles can also be monitored by
testing the viscosity of the sample. The change in viscosity
of Al O in water at different pH. confirms the result of
For the stability of most electrostatic stabilized inorganic
sols, the zeta potential plays a major role. In Figure 2 the
change in zeta potential of Al O and SiO with the pH is
2
3
2
2
3
shown. The highest zeta potential of the sols is found at the
measured zeta potential (Fig. 3).
1478
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
zaac.wiley-vch.de Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483

Synthesis of Pure Cementitious Phases by Sol-Gel Process as Precursor
Figure 4 Change in viscosity for colloidal alumina in pure water
Figure 3 Change in viscosity and zeta potential for colloidal alum-
ina in water at different pH. With lower values of the zeta potential
the repulsive coulomb forces between particles decrease and the
attractive van der Waals interaction of the particles get more im-
and in a 0,95 mol/L solution of Ca(NO ) . The high ionic strength
3
2
in the solution result in a steeper drop of the potential and there-
fore in lower repulsive forces that gives higher viscosity. The ad-
sorption of ions on the surface also changes the pH for the i.e.p.
and therefore shifts the position of maximum viscosity.
2 3
portant. At the i.e.p. the zeta potential of Al O becomes zero re-
sulting in the strongest attraction of the particles and therefore the
highest viscosity. For alumina the gelation is fasted near the i.e.p.
2.2 Optimizing conditions for gelation
There are several ways to form a gel from a stable sol. One
is to coagulate the sol at the i.e.p. (change in pH), the other
is to coagulate the sol by adding an electrolyte. Additional
to the pH and the electrolyte concentration, the gelation is
influenced by the concentration of the sol and the tempera-
ture. Iler has shown, that the optimum pH for the shortest
gelation time of a sol can vary according to the concen-
tration of soluble ions [19]. Therefore the time of gelation
of an alumina sol in a solution of Ca(NO ) was measured
3
2
at different pH values. As can bee seen from the change in
viscosity with pH for Al O in pure water and in Ca(NO )
3 2
Figure 5 Gelation for colloidal silica in a 4 mol/L solution of
Ca(NO3)2 at different pH. Colloidal silica is very stable and the
i.e.p. is not the point where the gelation is fasted. Therefore the test
for the dependence of the gelation time on pH had to be performed
over a longer period of time.
2
3
solution, that is presented in Figure 4, addition of
Ca(NO ) enhances the viscosity of the alumina sol, be-
3
2
cause the potential drop of the particles is enhanced by the
addition of electrolyte and thus the zeta potential is low-
ered. The maximum of the viscosity is shifted from pH 9.5
in pure water to 6.5 in a 0.95 mol/L solution of Ca(NO ) .
2.3 Degree of gelation
3
2
In the case of the used silica sol, an instant change in
viscosity for different values of pH could not be observed.
The reason for this is the high stability of the silica sol. This
made it necessary to monitor the change in viscosity for
the process of gelation over several hours. The results are
reported in Figure 5 and an examination of the data reveals,
that the optimum pH for the gelation of silica in the pres-
ence of a 4 mol/L solution of Ca(NO ) is at pH 5.5. The
process of gelation can also be accelerated by increasing the
temperature. 70 °C proved to be a good compromise be-
tween short gelation time and easy handling of the sample.
The conditions for the synthesis of gels by sol-gel processes
for different cementitious phases are outlined in Table 1.
While a sol is converted to a gel, the growing agglomerates
have the same concentration of silica and water as the sur-
rounding solution. These aggregates, the so called “gel
phase” do not sediment and are not visible, because the
density and the refractive index are the same as that of the
remaining sol. For the same reason only a slow increase in
viscosity can be measured before the sol converts into a gel.
There are also only little changes in other properties up to
a point, where the viscosity begins to increase rapidly (see
Figure 6) and the gelation occurs at the “gel point” [19].
The degree of gelation was measured by using the
method described by Iler [19]: A sample of a sol in the pro-
cess of gelation was diluted threefold using a solution of
3
2
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483
zaac.wiley-vch.de
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
1479

D. Stephan, P. Wilhelm
Table 1 Conditions for the sol-gel process
Ca
3
(SiO
4
)O
Ca
2
(SiO
4
)
C
9
(Al
6
O
18
)
CaAl
2
O
4
4 7
CaAl O
Ca(NO
SiO / (mol/L)
Al / (mol/L)
pH
approx. time for gelation / h
3
)
2
/ (mol/L)
4
4
Ϫ
5.5
11
4
4
Ϫ
5.5
8
1.89
Ϫ
1.89
8.0
8
1.89
Ϫ
1.89
6.5
4
1.89
Ϫ
1.89
8.0
2
2
O
2
3
2 4
Table 2 Semi-quantitative composition of the Ca (SiO ) samples
fired at different temperatures for 18 h (without amorphous SiO
2
)
T / °C
1140
2.2
0
95
3.30
1230
1.3
0
90
3.36
1500/1230
Ϫ
40
60
CaOfree / wt.-%
γ-Ca
β-Ca
Density / (g/cm )
2
(SiO
4
) / wt.-%
2
(SiO
4
) / wt.-%
3
3.12
Figure 6 Correlation between degree of gelation and viscosity. At
the gel point a marginal improve in the degree of gelation gives an
intense rise in viscosity.
Ca(NO ) of the same concentration and pH as used in
Figure 7 Polymorphic transformation of Ca (SiO ). Only fine par-
3
2
2
4
the sol-gel process. Thereby the solution stays stable during
centrifugation. The mixture was centrifugated under a grav-
ity insufficient to settle single colloidal particles (approx.
ticles can form the hydraulic active metastable β
L
-modification of
Ca
2
(SiO
4
), that is of interest for hydraulic binders. If the coarser
β
H
-Ca
2
4
(SiO ) is formed during the synthesis, this is irreversibly
transformed to the non-hydraulic γ-modification during cooling.
10000 g), but during the centrifugation the already gelled
part of the sol settles because of its higher density in com-
parison to the non-gelled part of the sol, which is dissolved
in the added electrolyte solution. The degree of gelation is
calculated from the weighted gelled part of the silica sol
after discarding the supernatant. This test could not be
done with the Al O sol, because it stays not stable during
2
.4.1 Dicalcium silicate
Dicalcium silicate was synthesized using different tempera-
ture programs with a maximum temperature of 1140, 1230
or 1500 °C respectively. All samples were burned for 18 h
with one intermediate grinding. The sample reacted at
2
3
the centrifugation.
1
500 °C was additionally tempered at 1230 °C for several
hours to decompose the Ca (SiO )O. After the burning the
3
4
2
.4 Phase composition and reactivity of the
semi-quantitative composition of the three samples was
analysed and showed clear differences that can be seen from
the data presented in Table 2.
synthesized cementitious phases
During the burning of the cementitious clinker phases dif-
ferent intermediates can occur, depending on the chemical
and stoichiometric composition. The crystalline products
can be analysed by XRD and for some phases it is also
possible to find different modifications. For some phase
mixtures a semi-quantitative analysis was done, but es-
pecially in the case of calcium silicates the results were
not very reliable. Here the amount of Ca (SiO ) and
Figure 7 shows the polymorphic transformation of Ca2-
(SiO ). All transformations are reversible, except for the one
4
from γ- to β-Ca (SiO ). γ-Ca (SiO ) is the thermo-
2
4
2
4
dynamically stable modification at ambient temperature,
but it is barely hydraulic and thus is of no interest for hy-
draulic binder.
The hydraulic active β-Ca (SiO ) normally transforms to
2
4
2
4
Ca (SiO )O was quantified according to the content of free
γ-Ca (SiO ) during cooling and β-Ca (SiO ) is only stable
3
4
2 4 2 4
lime detected by the Franke method [20] (sensitivity ±
.05 %, detection limit 0.1 %).
at ambient temperature if foreign ions are incorporated into
the crystal lattice. The transformation of Ca (SiO ) without
0
2
4
1480
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
zaac.wiley-vch.de
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483

Synthesis of Pure Cementitious Phases by Sol-Gel Process as Precursor
foreign ions can be inhibited if the Ca (SiO ) is quenched
conventional method, the mixture had to be fired at
1500 °C for at least 48 hours with four to five inter-
mediate grindings.
2
4
[
21] or if fine αЈ -Ca (SiO ) is produced, which transfers to
L
2
4
stable β -Ca (SiO ) [22]. Therefore the particles have to be
L
2
4
very small. Locher [23] reports, that the critical crystal size
must be below 5-10 µm. The results of the own study show,
that it is possible to produce pure β-Ca (SiO ) by the sol-
gel process, if the right temperature is used for the synthesis.
In the samples synthesized at a temperature up to
All samples of Ca (SiO )O synthesized via a sol-gel pro-
3
4
cess were fine powders (d ഠ 16 µm) and could be hydrated
50
without previous grinding. Short additional grinding gave
even finer particles, which were much more reactive. This
shows a major difference between the sol-gel samples and
the conventional ones, which were hard and compact after
the synthesis and needed time and energy consuming grind-
ing. The medium particle size d50 of a sol-gel Ca (SiO )O
2
4
1
230 °C, no γ-modification of Ca (SiO ) was found. Only if
2
4
the temperature during the burning process was as high as
1
500 °C, γ-Ca (SiO ) was detected in the product. This is
2
4
3
4
most likely, because at high temperatures small crystals can
ground for 2 min in a planetary ball mill was about 6.5 µm,
grow together to form coarse αЈ -Ca (SiO ) that transforms
whereas an extensively ground conventional Ca (SiO )O
sample (12 min) had an particle size of 8.4 µm.
L
2
4
3
4
into β -Ca (SiO ) and finally to γ-Ca (SiO ) during cool-
H
2
4
2
4
ing. In contrast to that Nettleship et al. [13] and Hong et al.
14] and Lee et al. [24] who used other methods to synthe-
size Ca (SiO ), also found β-Ca (SiO ) as the main phase
The differences in reactivity between the sol-gel and the
conventional Ca (SiO )O can best be shown through the
[
3
4
differences in the heat of hydration and the degree of hy-
dration (see Figures 8 and 9). The reactivity of the conven-
tional sample is higher during the first period of hydration
(with an extra peak after about 2 h caused by a small
amount of free lime), whereas after an increased dormant
period with a moderate rate of heat evolution for seven
hours, the sol-gel sample then shows a very high reactivity
and after 15 hours the heat of hydration of the sol-gel
sample is higher than the one of the conventional sample.
The differences in reactivity could also be monitored by the
degree of hydration α, which was measured by the loss of
ignition at 950 °C. The sol-gel sample hydrated for 28 days
showed no unhydrated Ca (SiO )O in the XRD analysis
2
4
2
4
up to a reaction temperature of 1200 °C, but above that
temperature γ-Ca (SiO ) was the main phase and at
2
4
1400 °C it was the only existing phase. This shows, that
crystals formed during the sol-gel process are much finer
than the ones formed by other routes of synthesis. If that
was not the case, all Ca (SiO ) should have been trans-
2
4
formed into the γ-modification.
The samples containing different concentrations of the
two modifications of Ca (SiO ) could also be differentiated
2
4
3
by density (β-Ca (SiO ) ϭ 3.33 g/cm . γ-Ca (SiO ) ϭ
2
4
2
4
3
2
.97 g/cm ) and isothermic thermocalorimetry. The heat
evolution of samples only containing the β-modification is
distinctly higher than that of the samples containing both
modifications. In overall the hydraulic activity of Ca (SiO )
3
4
and therefore this sample was taken as reference for the
degree of hydration (α ϭ 1). In the case of the conventional
Ca (SiO )O α was 0.7 after 28 days of hydration and this
2
4
is much lower than the one of Ca (SiO )O.
3
4
3
4
again indicates that the sol-gel sample is more reactive com-
pared to the conventionally produced Ca (SiO )O.
3
4
2.4.2 Tricalcium silicate
In the binary system CaO-SiO the phases Ca (SiO )O, Ca-
2
3
4
2
.4.3 Tricalcium aluminate, monocalcium aluminate
2(
SiO ) and CaO can occur under the used conditions. In
4
and monocalcium dialuminate
the absence of foreign ions the triclinic I modification of
Ca (SiO )O is the only one that exists as a metastable form
Cubic tricalcium aluminate was produced by firing the
samples from the sol-gel process two to three times for 14
hours at 1260 °C with intermediate grindings. Table 4 shows
the sample composition. Only mayenite and free lime were
detected as by-products. After three firings the content of
free lime was between < 0.05 and 0.5 wt.-%. After grinding
the final product, no differences could be found in the par-
ticle size distribution compared to conventionally produced
Ca (Al O ). Furthermore the hydration activity of both
3
4
at ambient temperature [21]. A series of samples was syn-
thesized at different temperatures for different times. The
resulting semi-quantitative phase composition is shown in
Table 3. In the temperature range between 800 and 1250 °C
β-Ca (SiO ) and free lime were the only crystalline phases
2
4
that could be detected by XRD. The amount of free lime
measured by the Franke method [20] was always below the
theoretical value of 24.5 wt.-%. The reason for this is, that
Ca (SiO ) can intercalate excess CaO into its crystal struc-
9
6
18
the sol-gel and the conventional Ca (Al O ) was the same.
2
4
9
6
18
ture [25]. At temperatures above 1250 °C free lime was
This is due to the fact that the by-product mayenite melts
partially and thereby the product loses the fine structure of
the gel-network.
bound by Ca (SiO ) to form Ca (SiO )O.
2
4
3
4
After firing the samples for 14 hours at 1500 °C a free
lime content between 3.5 to 6.0 wt.-% was found. After
grinding the samples for 10 min in a planetary ball mill and
burning them for another 14 h at 1500 °C, the content of
free lime was between < 0.05 to 0.5 wt.-%. All the samples
that had been fired three times had less than 0.05 wt.-%
of free lime. In order to synthesize pure Ca (SiO )O via a
The composition of a CaAl O sample which was fired
2
4
at different temperatures for different times is shown in
Table 5. To produce pure CaAl O , a firing time of no more
2
4
than 28 hours with just one intermediate grinding was
necessary. Free lime could only be detected up to a firing
temperature of 1000 °C. After 28 hours of firing at 1500 °C
3
4
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483
zaac.wiley-vch.de
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
1481

D. Stephan, P. Wilhelm
Table 3 Sample composition of the 3CaO ϩ SiO
2
2
system; SiO from colloidal silica can not be detected by XRD
T / °C
600
30 min
11.7
Ϫ
700
800
900
1200
30 min
22.7
Ϫ
1300
30 min
20.9
13.8
65.3
1400
30 min
16.5
31.9
51.6
1500
30 min
4.1
83.1
12.8
1500
14 h
3.7
84.7
11.6
1500
b)
t
30 min
22.4
Ϫ
30 min
22.6
Ϫ
30 min
22.8
Ϫ
28 h
<0.01
> 99
Ϫ
CaOfree / wt-%
a)
Ca
Ca
3
(SiO
(SiO
4
)O (TI) / wt.-%
2
4
) (αϩβ) / wt.-%
Ϫ
75.4
75.4
75.4
75.4
a)
Ca
2
4
(SiO ) can intercalate CaO into its crystal lattice
b)
The sample was ground after 14 h
Table 5 Sample Composition of the CaO ϩ Al
2
O
3
system at differ-
2 3
ent temperatures and reaction times; Al O from colloidal alumina
can not be detected by XRD
T / °C
t
800
1000
1200
1400
1500 1500
a)
30 min 30 min 30 min 30 min 14 h 28 h
CaOfree / wt.-%
5.4
1.4
Ϫ
Ϫ
Ϫ
11.5
73.3 > 99
11.5
2.92 2.98
Ϫ
Ϫ
(
CaO)12(Al
2
O
3
)
7
/ wt.-% 94.6
19.0
40.5
39.1
Ϫ
25.2
57.3
17.5
Ϫ
20.5
64.2
15.3
Ϫ
CaAl
CaAl
2
O
O
4
/ wt.-%
/ wt.-%
Ϫ
Ϫ
Ϫ
4
7
Ϫ
3
Density / (g/cm )
a)
The sample was ground after 14 h
Table 6 Sample composition of the CaO ϩ 2Al
ferent reaction times at 1500 °C; amount of Al
from the overall composition
2
O
O
3
system for dif-
was calculated
2
3
t / h
CaAl
CaAl
14
26
4.4
92.7
2.9
3.00
35
1.7
97.2
1.1
2.97
48
0.7
98.8
0.5
2.95
61
Ϫ
100
Ϫ
2.92
Figure 8 Overall heat of hydration for conventional and sol-gel
Ca (SiO )O for one week. Up to 14 h the heat of hydration is
higher for the conventional Ca (SiO )O.
2
O
O
4
/ wt.-%
/ wt.-%
17.9
70.4
11.7
3.04
3
4
4
7
3
4
Al O / wt.-%
2 3
3
Density / (g/cm )
the by-products CaAl O and (CaO) (Al O ) combined to
4
7
12
2
3 7
form pure CaAl O . As in the case of Ca (Al O ), the
2
4
9
6
18
ground products of the conventional and the sol-gel method
did not show any significant differences in particle size dis-
tribution or hydraulic activity. This, again, was a result of
the partial melting of the mayenite during the synthesis.
For the synthesis of CaAl O the longest firing time and
4
7
the highest number of intermediate grindings were neces-
sary. The results are given in Table 6. Because Al O from
2
3
colloidal alumina can not be detected by XRD, the stated
amount of Al O in Table 6 was calculated from the overall
2
3
composition. After 61 hours at 1500 °C and four intermedi-
ate grindings pure CaAl O7 could be synthesized which
4
showed hardly any hydraulic activity.
Figure 9 Heat of hydration for conventional and sol-gel Ca
during three days. For the sol-gel Ca (SiO )O the induction period
is longer, but the maximum heat evolution is very high.
3 4
(SiO )O
3
4
Experimental Part
Ludox TM50 (Grace Davison) was used as colloidal silica source
and Aerodisp W630 (Degussa Aerosil) as colloidal alumina source.
The calcium nitrate was prepared by dissolving CaCO
3
(p.a.) with
Table 4 Sample composition of the 3CaO ϩ Al
ent reaction times at 1260 °C
2
O
3
system at differ-
a stoichiometric amount of HNO (65 %, p.a.).
3
The particle size of the silica sol was measured using acoustic at-
tenuation spectroscopy. The zeta potential of the silica and alumina
sol was calculated from the colloidal vibration current. An acoustic
sound wave is applied to the sample and a resulting electric field
or current, the so called “colloidal vibration current” (CVI) is
t / h
CaOfree / wt.-%
14
2.5
81.5
16.0
28
1.2
86.8
12.0
42
0.4
97.3
2.3
Ca
9
(Al
6
O
18) / wt.-%
(
CaO)12(Al O )
2 3 7
/ wt.-%
1482
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
zaac.wiley-vch.de
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483

Synthesis of Pure Cementitious Phases by Sol-Gel Process as Precursor
measured. Both methods were performed by using a DT 1200 from
Dispersion Technology and are described in detail by Dukhin et al.
[10] G. Laudisio, M. Catauro, A. Costantini, F. Branda, Thermo-
chim. Acta 1998, 322, 17.
[
26, 27]. The particle size distribution of the synthesized phases was
[11] G. Laudisio, F. Branda, Thermochim. Acta 2001, 370, 119.
[12] J. Yoshizawa, H. Matushita, K. Iwamoto, A. Katsui, J. Adv.
Science 2001, 13, 52.
measured by laser diffraction using the laser particle analyser
CILAS 1064 from Cilas. XRD was done on a Philips PW 1710
series automated powder diffractometer with Cu-K
α
radiation. The
[13] I. Nettleship, J. L. Shull, W. M. Kriven, J. Eur. Ceram. Soc.
1993, 11, 291.
[14] S.-H. Hong, J. F. Young, J. Am. Ceram. Soc. 1999, 82, 1681.
[15] Della M. Roy, S. O. Oyefesobi, J. Am. Ceram. Soc. 1977, 60,
178.
semi-quantitative composition of the clinker phases was calculated
by using the Full Pattern quantification software ADM V6 from
Wassermann. The heat of hydration was measured with an iso-
thermic DSC (TAM Air, Thermometrics) and the amount of un-
bound CaO (free lime) was quantified by using the Franke
method [20].
[16] Y. Wang, W. J. Thomson, J. Mater. Science 1996, 31, 1319.
[
[
[
17] D.C. Agrawal, R. Raj, C. Cohen, J. Am. Ceram. Soc. 1990,
3, 2163.
18] T. W. Healy, G. R. Wiese, D. E. Yates, B. V. Kavanagh, J.
Colloid Interface Sci. 1973, 42, 647.
19] R. K. Iler, The Chemistry of Silica, Wiley-Interscience, New
York 1979.
7
Acknowledgement. The authors are very grateful to Prof. Dr. J.
Plank for the support of the work, helpful discussions and the ad-
mission to the group.
References
[
[
20] DIN EN 159-2: Building lime Ϫ Part 2: Test methods, p 10.
21] H. Pöllmann, in: Structure and performance of cements, J.
Bensted, P. Barnes (Eds), Spon Press, London 2002, p 25.
22] E. M. Gartner, J. F. Young, D. A. Damidot, I. Jawed, in: Struc-
ture and performance of cements, J. Bensted, P. Barnes (Eds),
Spon Press, London 2002, p 57.
[
[
[
[
1] Zahlen und Daten, Ausgaben 1980-2002, Bundesverband der
Deutschen Zementindustrie e.V. 2003.
2] D. Stephan, H. Maleki, D. Knöfel, B. Eber, R. Härdtl, Cem.
Concr. Res. 1999, 29, 545.
3] D. Stephan, H. Maleki, D. Knöfel, B. Eber, R. Härdtl, Cem.
Concr. Res. 1999, 29, 651.
4] J. Neubauer, Habilitationsschrift, Friedrich-Alexander-Univer-
sität Erlangen 1998.
[
[23] F. W. Locher, Cement. Principles of Production and Use, Verlag
Bau und Technik, Düsseldorf 2003, p 35.
[24] S.-J. Lee, E. A. Benson, W. M. Kriven, J. Am. Ceram. Soc.
1999 82, 2049.
[25] H. Kühl, Zement-Chemie Band II Ϫ Das Wesen und die Her-
stellung der hydraulischen Bindemittel, Verlag Technik, Berlin
1951, p 209.
[26] A. S. Dukhin, P. J. Goetz, Ultrasound for characterizing col-
loids, particle sizing, zeta potential, rheology, Elsevier, Am-
sterdam 2002.
[27] A. S. Dukhin, P. J. Goetz, Adv. Colloid Interface Sci. 2001,
92, 73.
[
[
5] B. M. Mohamed, J. H. Sharp, J. Mater. Chem. 1997, 7, 1595.
6] B. M. Mohamed, J. H. Sharp, Thermochim. Acta 2002, 388,
1
05.
7] A. A. Goktas, M. C. Weinberg, J. Am. Ceram. Soc. 1991, 74,
066.
8] L. Kerns, M. C. Weinberg, S. Myers, R. Assink, J. Non-Cryst.
Solids 1998, 232, 105.
[
[
[
1
9] P. Saravanapavan, L. L. Hench, J. Non-Cryst. Solids 2003,
318, 1.
Z. Anorg. Allg. Chem. 2004, 630, 1477Ϫ1483
zaac.wiley-vch.de
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim
1483