See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/284478995

Colloidal silica as a nanostructured binder for refractory castables

Article · November 2015

CITATIONS READS
72 4,415

4 authors, including:

Rafael Salomão Victor C Pandolfelli

University of São Paulo Universidade Federal de São Carlos
130 PUBLICATIONS   1,353 CITATIONS    638 PUBLICATIONS   6,973 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Biomaterial binders View project

Erosion Resistance of Refractory Concretes View project

All content following this page was uploaded by Rafael Salomão on 30 November 2015.

The user has requested enhancement of the downloaded file.

Feature Article . . .
COLLOIDAL SILICA AS A NANOSTRUCTURED
BINDER FOR REFRACTORY CASTABLES
M. R. Ismael, R.D. dos Anjos, R. Salomão and V.C. Pandolfelli, Materials Engineering Department -
Federal University of São Carlos, Rod. Whashington Luís, Km 235, 13565-905, São Carlos – SP, Brazil
[email protected] or [email protected]

ABSTRACT loidal silica as a refractory bonding system has been known for a while
Great efforts have been carried out in order to replace calcium alu- in the refractory industry, little systematic research has been found in
minate cement (CAC) by other types of binders in refractory castables. the open literature.
CAC contents higher than 2-3 wt% can impose difficulties for the dry- When combined with other solid particles, colloidal silica can be
ing step and for the system’s refractoriness when microsilica is pres- linked together in branched chains, in a process known as gelation [2,
ent. The present work applies a nanostructured colloidal suspension as 10], which can be induced by water removal. During the drying step,
a binder for refractory castable compositions. Colloidal silica’s (CS) the hydroxyl groups (Si-OH) on the surface of the particles generate
are stable water based suspensions that, when combined with castable siloxane bonds (Si-O-Si), which results in a three-dimensional net-
particles, provide, among other advantages, a higher drying speed and work (Figure 1) [10].
sinterability for the product. In this work, the performance of CAC, Gelation can also be induced by pH changes and a salt or water-mis-
HA (Hydratable Alumina) and CS as binder agents in refractory casta- cible organic solvent addition, named gelling agents. With the proper
bles was compared and analyzed. selection of these agents or the pH variation, the colloid is gelled
INTRODUCTION around the solid particles providing mechanical strength to the system
The reasons for the growing market of monolithic refractories, at the after drying [2, 10-11]. The resulting structure is highly permeable and
expense of traditional bricks, are their easy production, cheap installa- allows the composition to be easily dried, reducing cracks and explo-
tion and similar performance shaped products [1]. A significant sive spalling. During sintering, its high surface area (typically close to
advance in monolithics technology is the current supply of designed 200 m2/g) may induce the formation of mullite in alumina-based sys-
aggregates and dispersant additives, as well as in placing techniques.
The use of a refractory binder less susceptible to curing conditions
than the traditionally employed calcium aluminate cement (CAC) is a
subject under constant investigation [2].
Calcium aluminate cement is the most used hydraulic binder in
refractory castables compositions. Nevertheless, CAC amounts higher
than 2-3 wt% in Al2O3-CaO-SiO2 systems may induce liquid forma-
tion at lower temperatures [1, 3]. Other important concerns for the high
CAC content castables are the curing and dewatering steps, which
must be carefully conducted in order to reduce explosive spalling [1].
Among the different binding systems for cement free castables,
hydratable aluminas (HA) have attracted great attention [4-6]. HA are
low crystalline mesophase transition aluminas (e.g. ␬, θ, ρ), which,
similar to cement, harden after reacting with water [7]. During heating,
the hydrated phases lose their chemically bound water, giving rise to
the stable form of alumina (α-Al2O3), which at higher temperatures
will help to develop a strong ceramic bond [1, 7]. However, the use of
hydratable aluminas also present limitations, such as the high likeli-
hood of explosive spalling during drying, because of its less permeable
structure when compared with a cement based composition [1, 4].
Concerning all these hydraulic binders’ drawbacks, the present work
applies a nanostructured colloidal suspension as a binder for refracto-
ry castable compositions. Colloidal silica sols are stable water-based
suspensions, containing up to 50 wt% of nanometric spherical amor-
phous silica particles (8-15 nm diameter). These materials have been
broadly used in the formulations of coatings, catalysts, papers, and
anti-soil treatments as well as high temperature binders for fibrous iso-
lating and investment castings [8, 9]. Although the application of col- Figure 1. Colloidal silica gelling mechanism.

16 July/August 2006 Refractories Applications and News, Volume 11, Number 4

tem, which improves the chemical resistance and the hot-strength loidal silica employed in each case (8-14% by weight in literature, and
properties [11]. 7.5 wt % of a 40 wt% SiO2 solution, in the present study) [12].
As the use of colloidal silica in refractory castables may reduce the In order to destabilize the colloidal silica suspension and then
problems presented by the current hydraulic binders [2, 11-12], in this increase the probability for bond forming collisions, MgO sinter was
work the performance of calcium aluminate cement and hydratable added. MgO favors anionic reaction mechanisms on its surface, by the
alumina as binder agents in refractory castables was compared to this withdrawal of hydrogen ions from Si-OH groups, which, as a result,
nanostructured binder. Their rheological behavior, mechanical provides a higher siloxane bond formation, increasing the colloidal
strength, permeability, apparent porosity, drying profile, refractori- silica gelation rate [10].
ness under load and creep were compared and analyzed. Mixing was carried out by a two-stage water addition method [17].
EXPERIMENTAL PROCEDURE During the processing step, the torque (N.m) required was computer
High-alumina self-flow castable compositions were formulated recorded. The results are presented in Figure 2.
using the software PSDesigner [13]. A particle-size distribution The dry mixing (up to 60 s of mixing, Figure 2-segment a) is char-
adjusted to theoretical curves based on Andreasen’s packing model acterized by the disruption of the dry powder agglomerates and the
[14], with a distribution coefficient (q) equal to 0.21, was used. The subsequent particles’ homogenization. When part of the water amount
mixtures were composed of calcined alumina (A1000 and A17NE – (≅ 75.0 wt %) is added to the system (at 60 s), strong capillary forces
Almatis U.S.), white fused alumina (grades 4/10, 8/20, 20/40, TP 40 start to actuate, increasing the torque values required. This effect is
and TP 200 – Elfusa Brazil) and 3.0 wt% of a hydraulic binder: cal- commonly known as the castable turning point (Figure 2-segment b)
cium aluminate cement (CA14M – Almatis U.S.) or hydratable alu- [17]. After the remaining water has been added (after the turning
mina (Alphabond 300–Almatis U.S.) or colloidal silica (Bindzil 40 point), the material was mixed for another couple of minutes (Figure
wt% SiO2 – Nalco). The composition with the hydratable alumina 2-segment c) [17].
(HAB) was dispersed with a polyethyleneglycol based additive The mixing curves attained for the different systems studied showed
(SKW Polymers) and the other systems (CAC and CS) with citric that the colloidal silica containing composition (after proper disper-
acid (Labsynth Brazil). For the CS compositions, MgO sinter (d50 = sion) required lower torque values and a shorter mixing time to com-
15 µm; 98 wt% MgO) was used as a gelling agent (0.6 wt% of the plete the process than those containing CAC or HAB. It can be asso-
colloidal silica) [12]. A 4.5 wt% water content was added to the com- ciated with two main effects: 1) the greater viscosity of colloidal sili-
positions for mixing and binder hydration. Mixing was carried out ca (about 10 mPa.s) when compared to water, which keeps the parti-
under constant rotation (44 rpm) in a castable rheometer developed in cles apart, and makes the mixing process easier, and, 2) colloidal sili-
the authors’ research group [15]. ca does not promote chemical reactions during the mixing step such
Compositions of castables were cast into 40 mm x 40 mm cylindri- as calcium aluminate hydrates (CAC bonded) and alumina gel forma-
cal molds for the drying tests, mechanical strength and apparent tion (HAB bonded), that can disturb the ionic force and the viscosity
porosity measurements. For the permeability, 75 mm diameter by 22 of the suspension.
mm thick disks were prepared, whereas for refractoriness under load Mechanical Strength, Permeability and Porosity
and creep measurements, 50 mm external diameter and height and
Measurements
12.4 mm central hollowed cylinders were cast. Samples were cured
at 50ºC in an acclimatized chamber (Vötsch 2020) for 24, 72, 96 and The mechanical strength was measured according to the ASTM
168 h. C496-90 standard (Splitting Tensile Strength of Cylindrical
Concrete Specimens) in a MTS Systems, Model 810 using five
Cement containing compositions were cured in a moisture-saturat-
samples of each castable composition, immediately after curing.
ed environment (~ 100% RH), in order to favor the hydration mech-
anism. The other compositions were cured without moisture (in open Permeability was measured at room temperature using two sam-
air), inside the acclimatized chamber. For those bonded with hydrat- ples of each composition. The Darcyan (k1) and the non-Darcyan
able alumina, there was no need for moisture for strength develop- (k2) permeability constant were obtained by a polynomial fitting
ment; in fact, its presence may inhibit full strength development, par-
ticularly at the surface of the castable [16]. Those bonded with col-
loidal silica required water removal so that the gelling mechanism
could take place. The water removal decreases the inter-particle sep-
aration distance, increasing the particle collision and siloxane bond
formation [10]. After the curing time, the samples subjected to per-
meability and apparent porosity tests were first dried in silica gel for
96 hours, at the curing temperature.
RESULTS AND DISCUSSION
Mixing Process
The literature reports colloidal silica bonded castables as self-dis-
persing systems due to the nanometric size and spherical shape of sil-
ica particles [2, 12]. For the composition studied, however, this effect
was not observed, as the dispersant-free colloidal silica containing Figure 2. Mixing behavior for castables containing different hydraulic binders: a)
formulation could not be properly mixed and homogenized. This dry mixing; b) castable turning point and c) addition of the total water remaining
could be attributed to the differences between the amounts of col- and castable homogenization.

Refractories Applications and News, Volume 11, Number 4 July/August 2006 17

between the airflow and air pressure values using Forchheimer’s
equation (Equation A) [18].

(A)

where Pi and P0 are the absolute inlet and outlet air pressures,
respectively; Vs is the air velocity; L is the sample thickness; µ is
the air viscosity and ρ its density.
Apparent porosity was obtained according to the ASTM C 20-87
(immersion test), using kerosene as the immersion liquid.
Figure 3 compares CS bonded castables cured in two conditions:
open air and a moisture-saturated environment. The results indicate Figure 4. Permeability measurements (k2) as a function of curing time for differ-
that the CS system requires water removal to develop mechanical ent binders containing compositions, cured at 50ºC.
strength. In addition, Figure 3 also shows two distinct mechanical
strength behaviors. Whereas HAB and CS containing compositions
developed their maximum mechanical strength after the first 24 hours
at 50ºC, the CAC containing ones needed at least 72 more hours to
achieve its highest value. For the CAC bonded castable, this gradual
increase in mechanical strength is well reported in the literature and
associated with the time required for the dissolution/precipitation
processes involved in the hydraulic bonding mechanism [19]. The
HAB and CS bonded systems, on the other hand, have their mechan-
ical strength development related to the drying: during the water
removal, the sol is converted to dried gel, increasing the mechanical
strength [10]. Therefore, as drying is promoted, mechanical strength
is developed. An important consequence of this mechanism is the
reduction of the full processing time, since the curing step and the
mechanical strength development can occur simultaneously.
Figure 5. Apparent porosity (AP) for castables based on different binders, cured at
The binders tested display distinct permeability levels (Figure 4) 50ºC.
and these differences can be related to each consolidation mechanism
and sub-products. The CAC cured below 20ºC and the HA, during metric amount. A superior liquid content promotes the separation of
their hydration process, generate low-density products (calcium alu- the matrix particles and an inter-particle force reduction, resulting in
minate hydrates and alumina gel, respectively) that pores and reduce a more porous structure [21].
permeability [4, 20]. CS, on the other hand, does not generate Drying Behavior
hydrates [2], which, after gelation, favors the formation of an open Dewatering tests were conducted in a thermo gravimetric device
and highly permeable structure. composed by an electric furnace coupled to a digital scale [22].
As all formulations were mixed with the same water amount (4.5 wt Results correlating the samples’ temperature with their mass loss rate
%) it would be reasonable to expect similar values of apparent poros- were obtained applying continuous heating (5ºC/min) from room
ity (Figure 5). Nevertheless, due to its nanometer size, colloidal sili- temperature up to 600ºC. Mass loss was analyzed using the W param-
ca particles could behave as a liquid increasing the apparent volu- eter and dW/dt (drying rate), calculated following equations B and C,
respectively:

(B)

(C)

where, W represents the cumulative fraction of water expelled during

the heat-up per total amount of water initially present in the body; M
is the instantaneous mass during the heating stage, Mo is the initial
mass and Mf is the final mass of the tested sample. Figure 6 shows
the drying behavior for the different systems analyzed, for samples
Figure 3. Mechanical strength as a function of curing time for castables prepared cured for 24 hours at 50ºC.
with different binders, cured at 50ºC.

18 July/August 2006 Refractories Applications and News, Volume 11, Number 4

 ability to expel it, the modulus of rupture of the castable can be exceed-
ed, leading to material spalling.
Due to the difference among the castables’ permeability values
(Figure 4), the water removal was facilitated in colloidal silica con-
taining castables. Therefore, they were dried much more easily and
quickly, and likely generated lower pressure levels. The lower perme-
ability of the HAB samples required higher temperatures to expel the
water vapor from the bulk of the solid, increasing the risk of explosive
spalling during drying.
The absence of the third stage in the CS curve is a good evidence of
the non-existence of high temperature chemically bonded water in col-
loidal silica containing castable, which reduces the damage risks dur-
ing heating.
Thermo Mechanical Behavior
Thermo mechanical properties were evaluated through refractoriness
under load (RUL) and creep measurements in a NETZSCH-421
device. RUL measurements were performed in pre-fired samples (at
550ºC for 12 hours) and the analysis was carried out under continuous
heating (5ºC/min) up to 1600ºC, under a compressive load of 0.2 MPa.
Creep samples were first sintered at 1500ºC during 24 hours and then
subjected to a compressive load of 0.2 MPa at 1450ºC for 48 hours.
Figures 7 and 8 display these results.

Figure 6. Drying rate (dW/dt) as a function of heating time (5ºC/min) for casta-
bles prepared with different binders, cured at 50ºC.
Figure 7. Refractoriness under load for calcium aluminate cement (CAC), hydrat-
able alumina (HAB) and colloidal silica (CS) based compositions.
The dewatering of low permeability refractory castables, such as
CAC bonded ones, is carried out in several stages, driven by different
mass and energy transport mechanisms. In the first stage (from room
temperature up to around 100ºC) (Figure 6a-segment i), the main dry-
ing mechanism is the free water evaporation. The second stage (from
100 up to 250-300ºC) corresponds to the free water ebullition and
decomposition of low temperature hydrates and gel (Figure 6a-seg-
ment ii). The third one is usually related to the decomposition of
hydrated products, such as C3AH6 and AH3 (Figure 6a-segment iii)
[22].
The CS and HAB compositions were cured in an open atmosphere
during which some drying occurred; therefore, the amount of free
water removed and deleted by evaporation (first stage) during the
dewatering test was reduced. In the second stage, indicated by seg-
ment (ii) in Figure 6 a, b and c), the drying rate becomes highly sen-
sitive to the permeability level, as a higher amount of steam can be
Figure 8. Creep for castables containing calcium aluminate cement (CAC),
entrapped in low permeability structures, raising the pressure inside hydratable alumina (HAB) and colloidal silica (CS) for samples pre-sintered at
the castable. If the vapor rate formation is greater than the structure’s 1500ºC and tested at 1450ºC.

Refractories Applications and News, Volume 11, Number 4 July/August 2006 19

 As the samples were pre-fired at 550ºC, the deformation level on the 5. Y. Hongo, “ρ-Alumina Bonded Castable Refractories,” Taikabutsu
RUL test also includes sintering. The colloidal silica composition pre- Overseas, 9 [1] 35-38 (1988).
sented higher sinterability, as shown in Figure 7. This can be 6. R. Racher, “Improved Workability of Calcia Free Alumina Binder
explained by its high surface area, which promoted the system’s reac- Alphabond for Non-Cement Castables,” Presented at the 9th
Biennial Worldwide Congress on Refractories, November 8-11,
tivity, increasing the sintering rate. The expansion detected close to 2005, Orlando, FL.
1500ºC for the CS’s formulation is likely to be associated to the mul-
7. W. Ma and P. W. Brown, “Mechanisms of Reaction of Hydratable
lite formation [23]. Aluminas,” J. Am. Ceram. Soc., 82 [2] 453-56 (1999).
In order to detect the deformation under load, creep measurements 8. E. Lidén, S. Karlsson and B. Tokarz, “Silica Sols as Refractory
were carried out. The influence of matrix content on the creep behav- Fiber Binders,” J. Eur. Ceram. Soc., 21 [6] 795-808 (2001).
ior of castables may explain Figure 8. Grain sliding deformation 9. DUPONT, Ludox Colloidal Silica: Properties, Uses, Storage, and
mechanism is more active when a large amount of fine particles is Handling. Technical Catalog, 22p.
present. The system porosity is also related with the creep behavior, 10. R. K. Iler, “The Chemistry of Silica: Solubility, Polymerization,
which explains the smaller creep deformation for the CAC composi- Colloid and Surface Properties, and Biochemistry,” New York,
Wiley, 866p (1979).
tion.
11. S. Banerjee, “Versality of Gel-Bond Castable/Pumpable
CONCLUDING REMARKS Refractories,” RAN, 6 [1] 1-3 (2001).
The results showed that the use of colloidal silica in refractory 12. S. Banerjee, et al., “Composition and Method for Manufacturing
castable compositions can be advantageous in comparison with the Steel-Containment Equipment,” Patent number 5,147,830. Sept.
15, 1992.
refractory cement and the hydratable alumina. Lower torques values
were required during the mixing step and an earlier turning point was 13. R. G. Pileggi, F. Ortega, R. Morabito, S. L. Vendrasco and V. C.
Pandolfelli, “Desenvolvimento e aplicação de um software que
attained, compared to the other compositions analyzed. The mechani- automatiza o processo de combinação de matérias-primas na
cal strength values as a function of curing time for the CS bonded sys- obtenção de produtos cerâmicos,” (“Development and Application
tem showed a lower sensitivity to curing conditions, as a shorter dry- of a Software Designed to Combine Different Raw Materials in
ing time was necessary to attain suitable properties. This effect can be Order to Obtain Ceramic Products.”), Ceramica (Brazil), 44 189-95
technologically useful, as it provides a reduction on the full processing (1998).
time. 14. I. R. Oliveira, A. R. Studart, R. G. Pileggi and V. C. Pandolfelli,
“Dispersão e Empacotamento de Partículas–Princípios e
The gelling mechanism of colloidal silica containing composition Aplicações em Processamento Cerâmico,” (“Dispersion and
generated a high-permeability structure, facilitated water removal, pre- Packing of Particles–Basic Principles and Applications to Ceramic
venting spalling. The drying process was favored due to the water loss Processing”), Fazendo Arte Editorial, 224p (2000).
which occurred during curing in open air and the absence of higher 15. R. G. Pileggi, V. C. Pandolfelli, A. E. Paiva and J. Gallo, “Novel
temperature chemically bonded water. Rheometer for Refractory Castables,” Am. Ceram. Soc. Bull., 79 [1]
54-58 (2000).
The nanometer-sized colloidal silica particles also improved the sys-
16. Global Product Data Alphabond 300–Almatis Technical Catalog.
tem’s reactivity, increasing the castable sinterability, promoting mullite
17. R. G. Pileggi, A. R. Studart, V. C. Pandolfelli, et al., “How Mixing
formation. Nevertheless, the porosity level developed was too high. Affects the Rheology of Refractory Castables Part I,” Am. Ceram.
Although the colloidal silica is a nanosized system, the results Soc. Bull., 80 [6] 27–31 (2001).
obtained showed that their contribution to the packing was low. 18. M. D. M. Innocentini, A. R. F. Pardo, V. R. Salvini and V. C.
Therefore, superior mechanical strength values could be achieved with Pandolfelli, “How Accurate is Darcy’s Law for Refractories,” Am.
an adequate particle-size distribution, which would result in lower Ceram. Soc. Bull., 78 [11] 64–68 (1999).
porosity and consequently in high-quality performance refractory 19. J. R. Garcia, I. R. Oliveira and V. C. Pandolfelli, “Processo de
castables. This will be the subject of our next publication on colloidal hidratação e os mecanismos de atuação dos aditivos aceleradores e
silica based castables. retardadores de pega do cimento de aluminato de cálcio.”
(“Hydration Process and the Mechanisms of Retarding and
ACKNOWLEDGEMENTS Accelerating the Setting Time of Calcium Aluminate Cements.”),
submitted to Ceramica (Brazil) 2005.
The authors would like to thank the Brazilian research funding insti-
20. F. A. Cardoso, M. D. M. Innocentini, M. M. Akiyoshi and V. C.
tutions CAPES and FAPESP for supporting this work; and ALMATIS Pandolfelli, “Effect of Curing Time on the Properties of CAC
(Brazil), MAGNESITA S.A. (Brazil), Forza (Brazil) and Elfusa Geral Bonded Refractory Castables,” J. Eur. Ceram. Soc., 24 [7] 2073-
de Eletrofusão Ltda (Brazil) for the raw materials supplied. 2078 (2004).
REFERENCES 21. F. T. Ramal, Jr., R. Salomão and V .C. Pandolfelli, “Water Content
and Its Effect on the Drying Behavior of Refractory Castables,”
1. W. E. Lee, W. Vieira, S. Zhang, K. G. Ahari, H. Sarpoolaky and C. RAN, 10 [3] 10-13 (2005).
Parr, “Castable Refractory Concrete,” Intern. Mat. Rev., 46 [3] 145-
167 (2001). 22. M. D. M. Innocentini, F. A. Cardoso, M. M. Akiyoshi and V. C.
Pandolfelli, “Drying Stages During the Heat-up of High-Alumina,
2. S. Banerjee, “Monolithic Refractories–A Comprehensive Utra Low Cement Refractory Castables,” J. Am. Ceram. Soc., 86
Handbook,” Singapore: World Scientific/The American Ceramic [7] 1146-1148 (2003).
Society, 311p (1998).
23. A. M. Hundere, B. Myhre and B. Sandberg, “Mullite Bonded
3. S. Banerjee, “Recent Developments in Monolithic Refractories,” Castables,” Proceedings of the International Symposium on
Am. Ceram. Soc. Bull., 77 [10] 59-63 (1998). Refractories, Haikon, China, 12-15, November 1996. R AN
4. F .A. Cardoso, M. D. M. Innocentini, M. F. S. Miranda, F .A. O.
Valenzuela and V. C. Pandolfelli, “Drying Behavior of Hydratable
Alumina-Bonded Refractory Castables,” J. Eur. Ceram. Soc., 24 [4]
797-802 (2004).

20 July/August 2006 Refractories Applications and News, Volume 11, Number 4

View publication stats