MICROSILICA-GEL BOND

CASTABLES FOR RAPID HEAT-

UP


Presented at St Louis section/RCD 48
th
Annual Symposium, St Louis, USA
in March 28-29, 2012.




Microsilica-gel bond castables for rapid heat-up.
Bjrn Myhre*, (Elkem Silicon Materials, Norway)
Abstract:
Microsilica has traditionally been used as a reactive filler in refractory castables. Recent investigations have however
shown that for some of these applications, microsilica also act as binder, giving the desired setting by a coagulation
bond (gelling). Normally that coagulation is effected by reaction between calcium ions from cement and microsilica
and the bonding is strong enough to be practically useable. This paper presents this bonding and gives examples of
results obtained under laboratory conditions. Among the very interesting properties of the gel-bond is the absence of
chemically bonded water, which yields castables that can be heated very quickly once they have been dried at low
temperature. Silica sol was also checked as a complimentary binder. The results show a positive effect on the dried
strength.

Introduction and Background



Figure 1: A view of the microsilica process from the
stoking deck of a 27MW silicon furnace producing
Elkem Microsilica 971.
Microsilica:
The word microsilica has become generic for
condensed silica fumes produced as by-products of
silicon and ferro-silicon production. Microsilica, also
known as fumed silica, volatilized silica etc. normally
originates from the loss of SiO(g) which is an
intermediate species in the production of silicon or
ferrosilicon. It is difficult to state exactly when the
term microsilica first appeared, but in the early
1980s Elkem registered Elkem Microsilica as a trade
name for their fumed silica. It is probable that this is
the origin of the generic use of microsilica.

Silicon and ferrosilicon is produced in large electrical
smelting furnaces by the reduction of quartz (SiO2)
with carbon. The raw materials are added from the
top of furnace and form a stack. The actual reduction
to metal occurs in the bottom of the stack. Quarts and
the other raw materials are added as fairly large lumps
which heats up on their voyage down the stack. The
carbon may be added as a range of sources like coal,
coke or charcoal; which one that is used is normally
the result of economic and other considerations.
Additional to the carbon, normally wood-chips are
added to improve the texture of the stack.
In the bottom of the furnace, where the temperature
exceeds 2000C, the reduction takes place through a
series of intermediate reactions producing carbon
monoxide together with other gaseous species. One of
these reactions produce the volatile species silicon
monoxide (SiO) that at temperatures above
approximately 1800C has a vapour pressure
exceeding 1 atmosphere. This gas which at that time
is under pressure, ejects upwards together with CO,
most of the SiO gets trapped as it condenses, but
some escapes and gets oxidized when it comes in
contact with the air above the stack. This is the origin
of microsilica. Of course, as a lot of energy (some 2/3
of that of Si-production) is used in the partial
reduction of quartz to silicon monoxide, this
represents a loss for the metal producer. Therefore
most metal producers aim for a minimum silica loss.
Nevertheless, some silicon monoxide will always
escape and give microsilica. Typically 10-15% of the
quarts ends up as microsilica under good furnace
operation.
It is not only silicon monoxide that leaves the furnace.
Also a number of impurities are volatilized. The
source of which are the raw materials. Typical volatile
impurities are oxides of alkalis. One of the more
significant impurities is carbon of various sources.
The carbon of the microsilica is present as several
forms: coke residues, silicon carbide, some tar related
organics and finally as carbon black.
Above the top of the furnace, which essentially is an
open construction, a hood is placed and with the aid
of huge fans, the volatiles are sucked off and forced
through a filter. Because of the suction and because
the furnace is top-charged, also larger particles like
wood-chips, coke residue and quarts dust are
collected together with the fumes.
Because of the introduction of foreign matter,
microsilica has to be beneficiated to give a high
quality product. In Figure 2, microsilica production is
schematically shown together with a view of the
Fiskaa plant in Norway: The raw materials are fed
into the top of the furnace (A) this is the stoking
deck. The volatiles are sucked off by a fan (D). To get
rid of the coarse particles, the microsilica is treated in
a pre-collector (C). The gas is then blown into a
baghouse filter and collected as raw microsilica.
Further treatment may be given to achieve desired
properties of the product. It should be mentioned that
other filtering equipment than baghouse filters have
been and are being used.
Figure 2: A view of the Fiskaa plant (Norway)
together with a simplified flowsheet of the production
of microsilica. (A) Furnace, (B) Chimney, (C) Pre-
collection, (D) Fan, (E) Baghouse filter.

Properties of Microsilica
If we grab a handful of microsilica, the appearance is
a fine powder with variable color from almost black
to slightly off-white. There is a strong correlation
between the color and the carbon content. This
correlation is not straight forward though, since the
carbon may be present as coke residue or as silicon
carbide additional to carbon black which is the
impurity influencing color. A closer examination of
the microsilica by electron microscopy reveals sub-
micron spheres with an average particle size of
approximately 0.15 micron. The surface area, as
determined by nitrogen adsorption (BET), typically
gives values around 20 m
2
/g which is equivalent to a
spherical diameter of 0.14 micron, i.e. close to the
results obtained by electron microscopy.
Chemically, microsilica consists of amorphous SiO
2

with variable purity. Without going into detail, it may
be stated that there is a close relation between the
furnace operation and the quality of the microsilica.
Hence traditionally, the purity of the microsilica tends
to be rather variable, reflecting the furnace operation
at the time it was collected. As the metal normally is
the main product and microsilica the by-product, a
variation in the performance of the microsilica should
thus be expected. Here it should be mentioned that it
is possible to stabilize microsilica quality, but the
procedures required normally adds to cost (like
lowering Si-yield) thus justifying the higher price for
such controlled products (e.g. 971 from Elkem).


Figure 3: Micrograph of a microsilica cluster showing
spherical appearance of the individual particles.

The crystalinity of microsilica is variable according to
source and beneficiation, but values below 0.3 % by
weight
1
have been reported. Unpublished work on
high purity microsilica did however not show any
traces of crystalline silica. When crystallinity is
observed in microsilica, the source is normally quartz
carry-over from the furnace charging. In some special
cases with extended residence time at high
temperatures (> approx 800C for many hours) some
of the microsilica may crystallise as well. The
crystallisation can be accelerated by impurities,
notably alkalies. Such crystallisation should for high
quality microsilica be regarded as negligible though.
Microsilica higher in impurities e.g. alkali levels
above 1% might have more crystalline matter than
purer types.
As shown in Figure 3, microsilica consists of spheres.
These have an average diameter of about 0.15 micron.
The spheres are the building units of agglomerates
that has been believed to be bonded together by
material bridges
2.
Improvements in PSD measuring
equipment have however given PSD curves (Figure 4)
that are in good agreement with the results from SEM
and BET surface measurements and the commonness
of those material bridges becomes questionable.
Figure 4:Typical particle size distribution of a well
dispersed microsilica as measured by laser diffrac
Although microsilica is dispersed
spheres as seen in Figure 3 & 4, the normal state in a
castable is probably as a mixture of dispersed spheres
and agglomerates. The more dispersed the microsilica
gets, the better will the resulting flow be (provided it
is present in sufficient amounts), particularly
flowing castables. In free-flowing castables an
increased amount of superfines (e.g. microsilica) also
acts as flow-enhancers by reducing inter
friction between the larger aggregates.
The microsilica surface
Even though the overall composition of microsilica is
SiO
2
, the surface of a microsilica particle is not plain
siloxane bonds, but is also partially hydroxylated and
hydrated. Producers of fumed silica (particularly
Evonik Degussa) have been very active in the
characterisation and modification of the silica surface
in order to tailor specific properties of precipitated
and fumed silica.
Figure 5: Visualization of the microsilica surface, dry
and with water.
Based on theoretical and experimental results
seems that the maximum density of silanol (Si
groups on the surface of silica is approximately
4.6nm
-2
.
The surface of microsilica is also covered by
silanol groups, although the presence of impurities
plays some role, measurements have indicated silanol
densities ranging from approximately 2 to 4.5nm
microsilica from silicon production. The good
correlation between theoretical values and those
found for microsilica may indicate that the impurities
of the microsilica are not smothered out on the
surface of the microsilica particles but rather as
discrete particles or dissolved in the silica. A
combination is probable.
Typical particle size distribution of a well
dispersed microsilica as measured by laser diffraction.
dispersed as individual
, the normal state in a
castable is probably as a mixture of dispersed spheres
and agglomerates. The more dispersed the microsilica
w be (provided it
is present in sufficient amounts), particularly for free-
flowing castables an
increased amount of superfines (e.g. microsilica) also
enhancers by reducing inter-particle
the overall composition of microsilica is
, the surface of a microsilica particle is not plain
siloxane bonds, but is also partially hydroxylated and
hydrated. Producers of fumed silica (particularly
Evonik Degussa) have been very active in the
characterisation and modification of the silica surface
in order to tailor specific properties of precipitated

: Visualization of the microsilica surface, dry
n theoretical and experimental results
3 ,
it
seems that the maximum density of silanol (Si-OH)
groups on the surface of silica is approximately
covered by such
groups, although the presence of impurities
plays some role, measurements have indicated silanol
densities ranging from approximately 2 to 4.5nm
-2
for
microsilica from silicon production. The good
correlation between theoretical values and those
microsilica may indicate that the impurities
of the microsilica are not smothered out on the
surface of the microsilica particles but rather as
discrete particles or dissolved in the silica. A
The presence of the silanol groups ma
microsilica easy to disperse in aqueous systems.
Depending on the pH, a fraction of the silanol groups
dissociate (to H
+
and O
-
) resulting in a negatively
charged surface. The zeta-potential can be measured
and values higher than approximately 25m
or negative) are often taken as an indication
suspensions. Figure 6 shows a measuremen
zeta potential for Elkem Microsilica 971, and we see
that the microsilica has a negative charge over the
whole range. Due to dissociation of the silanol groups,
the negative surface charge increases with pH up to
approximately pH 7. At higher pH than 7, the zeta
potential flattens out and at still higher pH the
microsilica starts to dissolve.
Figure 6: Zeta potential of Elkem M
function of pH. A 10% microsilica slurry titrated with
HCl and with zeta potential being measured using
electro-acoustic techniques (Aco
Dynamics)
It is the negative charge on the surface of the
microsilica that is the property that makes the gel
bonding possible. Such negative sites can react with
cation and if it is a polyvalent cation it may bridge
two adjacent microsilica particles. If this happens in
three dimensions, a gel of that micr
constructed. Figure 7 visualises this process in two
dimensions using Ca
2+
as the bridging cation.











Figure 7: Proposed gelling mechanism for
microsilica-gel bond. Ca
2+

microsilica particles.


-35
-30
-25
-20
-15
-10
-5
0
1 2 3 4 5
pH
Z
e
t
a
-
p
o
t
e
n
t
i
a
l

(
m
V
)

O-H
O-H H-O
H-O
H-O
O
-
O
-
O
-
O
-
OH
O
-
+
Ca
+
Microsilica
Microsilica
+
Ca
+
Ca
+
+
Ca
+
+
Ca
+
+
Ca
+
O
-
O-H
H-O
+
Ca
Microsilica
The presence of the silanol groups makes the
microsilica easy to disperse in aqueous systems.
Depending on the pH, a fraction of the silanol groups
resulting in a negatively
potential can be measured
and values higher than approximately 25mV (positive
or negative) are often taken as an indication of stable
shows a measurement of such a
icrosilica 971, and we see
that the microsilica has a negative charge over the
n of the silanol groups,
the negative surface charge increases with pH up to
approximately pH 7. At higher pH than 7, the zeta
potential flattens out and at still higher pH the
Elkem Microsilica 971 as a
function of pH. A 10% microsilica slurry titrated with
HCl and with zeta potential being measured using
acoustic techniques (Acoustosizer, Collodial
It is the negative charge on the surface of the
property that makes the gel-
bonding possible. Such negative sites can react with a
and if it is a polyvalent cation it may bridge
two adjacent microsilica particles. If this happens in
three dimensions, a gel of that microsilica is
visualises this process in two
as the bridging cation.
: Proposed gelling mechanism for
gelling adjacent
6 7 8 9
O-H
O-H
O-H
O
-
O
-
O
-
Microsilica
Ca
+
+
Ca
+
Ca
+
Castables based on this bonding concept was
presented as early as at UNITECR 1995 in Kyoto
4,5
.
At that occasion, also the accelerating effect of
combining small amounts of cement with Alphabond
was demonstrated (Table 1). It was not clear to the
authors what mechanisms was causing the setting, at
that time; it was just assumed that one was talking
about a hydraulic bond of the cement.

Table 1 Fused alumina based castables. Setting time
(days at 20C) for combinations of Alphabond and
cement. Alphabond content (mass%) in left column
and cement content (mass%) in first row.
Alphabond/Cement 0 0.25 0.50 1.00 2.00
0 10 5
0.50 1 1
1.00 18 1
2.00 13 1 1
3.00 4 1
5.00 2

At UNITECR 2011
6
, also this time in Kyoto, a further
study on basically the same system as in 1995 was
presented. At this occasion the setting had been
studied by simultaneous zeta-potential, conductivity
and pH measurements showing that there is a
simultaneous setting combined with a sequestering of
ions in the liquid phase. Figure 8.



Figure 8, Zeta-potential, conductivity and pH of a
slurry mixture of 8parts microsilica, 0.5parts cement,
0.5parts Alphabond, 0.05parts deflocculant
(Castament FS20) and 4.15parts water as a function
of time.

Based on these results, the setting mechanism
proposed in Figure 7 was formulated. The
accelerating effect that had been seen for Alphabond
additions was explained by the Alphabond serving as
nuclei for precipitation of alumina from the liquid;
thus facilitating increased dissolution of Ca
2+
from the
cement. This in turn reacts with the negatively
charged microsilica surface causing a gel to form.
To test the effect of the setting on real castables,
some compositions based on white fused alumina
with 8wt% microsilcia in combination with different
combinations of cement a Alphabond, were prepared
with 4.1% water. Top size was 5mm and dispersant
was 0.05% Castament FS20. Results from flow decay
measurement s are shown in Figure 9.


Figure 9. Flow decay of combinations of cement and
Alphabond, WFA based castables with 8 wt%
microsilica and 4.1% water.

Clearly, by adding alphabond to the castable together
with 0.5% cement, setting is accelerated. A similar
effect can also be found by using a fixed amount of
Alphabond with the cement as variable. Increased
amounts of Alphabond tend to lower flow though.
Flow and Strength.
It should not be surprising to expect that the result of
a bond consisting of a gelled microsilica should be
fairly low. But just what levels are sufficient for
demoulding and handling?
Braulio et al
7
indicates that an initial splitting tensile
strength higher than 1 MPa would be enough to avoid
crack generation during demoulding. 1MPa splitting
tensile strength is, depending on citation, equivalent
to between 1.3 to 1.4MPa M.O.R. In other words, any
M.O.R value above 1.5MPa should be sufficiently
strong.
Even though a lot of information (e.g.
8
and
9
) was
available on the hot-strength of castables based on the
microsilica-gel bond, we had surprisingly few results
from low temperatures. Therefore it was decided as
part of another investigation to measure flow and
green-strength of a chosen castable, the recipe is
given in Table 2.
As good results also were obtained with lower water
additions
10
such as 3.5% water, it was chosen to make
the mixes with both the custom 4.1, and with 3.5%
-60
-50
-40
-30
-20
-10
0
10
20
30
0.00 2.00 4.00 6.00 25.00 29.00
time[h]
[
m
S
/
c
m
]
/

[
m
V
]
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
p
H
conductivity [mS/cm]
Zeta-potential [mV]
pH value
0
40
80
120
160
200
0 2 4 6 8 10 12
Time [h]
F
l
o
w
-
v
a
l
u
e

[
%
]
0.5%Cement + 0.0%Alphabond Free-flow 0.5%Cement + 0.0%Alphabond Vibra-flow
0.5%Cement + 0.5%Alphabond Free-flow 0.5%Cement + 0.5%Alphabond Vibra-flow
0.5%Cement + 1.0%Alphabond Free-flow 0.5%Cement + 1.0%Alphabond Vibra-flow
water. For the flow measurements, the flow cone
(height 50mm) as described in ASTM C 230 was used,
but since the European standard EN 1402-4 prescribes
the use of a taller cone (height 80mm) for self-flow
measurements, both self-flow using the ASTM cone
and the EN-cone was measured.
Table 2: White fused alumina castables with
microsilica gel bond.
NCC microsilica-gel bond [weight %]
Elkem Microsilica 971U : 8
CAC cement, Secar 71 : 0,5
Alphabond 300 : 0,5
White Fused Alumina 3-5mm : 10
White Fused Alumina 0,5-3 mm : 32
White Fused Alumina 0-0,5mm : 16
White Fused Alumina -74 mic : 20
Calcined alumina,CT 9FG : 13
Dispersant, Castament FS20 : 0.05

Figure 10 shows the flow values of the castable of
Table 2 with 3.5% and 4.1% water. It is remarkable
how high the self flow is even at 3.5% water, yielding
values that normally are found for significantly higher
water additions. Here it should be mentioned that to
obtain such high flow values, the microsilica quality
plays a decisive role. It is of uttermost importance
that the microsilica is easy to disperse and that the
mixture does not contain flocculating contaminants
like easily dissolvable salts, notably polyvalent.
Lowering the water addition creates one problem
connected to extended wet-out time. At very low
water additions the wet-out becomes difficult, if the
results of Fig. 10 are taken as an example, the wetout-
time increased from 20 to 45 seconds when water was
reduced from 4.1 to 3.5%. Even lower water additions
can be used, but then wetout-time becomes a
problematic issue. In such cases different strategies
can be sought like mixing parts of the mix with the
water and afterwards adding the rest of the dry mix
etc.
Some hours after the mixing, the castable sets but will
it be possible to demould the piece? Figure 11 and 12
shows the strength green after 24 hours and after
subsequent drying at 110C.


Figure 10: Flow of the white fused alumina based
NCC base on the microsilica-gel bond. Flow values
measured with 3.5 and 4.1% water.



Figure 11: WFA based microsilica-gel bonded
castables cast with 3.5 and 4.1% water. Strength
measured green (undried) after 24hours and after
subsequent drying at 110C/24h.

The green C-M.O.R. was found to be around 2MPa
for both water additions. Although low, it should
according to Braulio et al. be possible to demould.
Drying increases the strength significantly and here
the lower water pays off yielding significantly higher
strength.

87
128
124 124
152
180
0
20
40
60
80
100
120
140
160
180
200
self-flow vibra-flow self-flow
ASTM cone ASTM cone EN cone
F
l
o
w

v
a
l
u
e
[
%
]
3.5% water
4.1% water
2.0
2.3
7.9
0
1
2
3
4
5
6
7
8
9
Green 20C Dried at 110C
C
o
l
d
-
M
.
O
.
R
.

[
M
P
a
]
4.15% water
3.5% water
5.3

Figure 12: WFA based microsilica-gel bonded
castables cast with 3.5 and 4.1% water. Strength
measured green (undried) after 24hours and after
subsequent drying at 110C/24h.

Figure 12 shows the CCS of the samples again
demonstrating the good effect of drying.

Explosion resistance
At IREFCON 2012
8
Myhre and Fan presented results
on explosion resistance using the castable
composition of Table 2. They compared the results of
the microsilica-gel bond with results using a similar
LCC composition using 6% CA-cement. The
composition is given in Table 3.

Table 3 Low-cement castable based on white fused
alumina. 4.15% water for casting.
LCC CA-cement bond [weight %]
Elkem Microsilica 971U : 8
CAC cement, Secar 71 : 6
White Fused Alumina 3-5mm : 10
White Fused Alumina 0,5-3 mm : 32
White Fused Alumina 0-0,5mm : 16
White Fused Alumina -74 mic : 16
Calcined alumina,CT 9FG : 12
Dispersant, Castament FS20 : 0.05

Based on the recipes given in Table 2 and 3, a set of
NCC (microsilica-gel bond) and LCC castables were
cast with 4.15% water. Samples were cast into 50mm
cubes as described in the Chinese standard
YB/T4117-2003.
The standard describes placing green (not dried)
samples (cubes of 50mm) into a furnace heated to a
preset temperature and inspecting the sample after 30
minutes. The temperature at which cracks forms or
explosion occurs is then reported as explosion
resistance.
After 24 hours curing at 100% relative humidity and
20C, the demoulded samples were protected from
drying before being placed into the furnace for 30
minutes. After inspecting the samples for cracks etc.,
the temperature of the furnace was increased and a
new sample inserted into the hot furnace. This was
repeated until cracks or explosions were detected.
In Table 4, the results are shown. It became quickly
clear that the undried NCC exploded into tiny
fragments already at 350C, whilst the LCC was able
to take 600C before it split in two pieces. The low
explosion resistance of the NCC surprised us since we
were convinced that the NCC should take heat-up
very well due to the lack of bonded water. Could it be
that our assumptions of only minute amounts of
bonded water were incorrect? It was quickly
suggested that low strength also could be a reason for
the steam explosion of the NCC when tested
green/undried. To check, samples were dried at 110C
before testing. Testing of the dried samples gave a
very different result. The LCC split in two pieces at
600C as it did for the undried sample, but the NCC
was virtually indestructable even at temperatures as
high as 1200C!

Table 4: Results from explosion testing of the
castables shown in Table 2 and 3.
Temp.[C]
Before drying
After drying at
110C
LCC NCC LCC NCC
200
250
300
350
400
500
550
600
800
1000
1200
:sample passed the test, :sample did not pass.

Reducing the casting water from 4.15 to 3.5% gives
benefits beyond those related to strength. It does
increase the explosion temperature. In our testing an
increase from 350 to 500C was experienced. Table
5 shows a comparison between the two.



12.4
39.4
16.8
62.2
0
10
20
30
40
50
60
70
Green 20C Dried at 110C
C
C
S

[
M
P
a
]
4.15% water
3.5% water
Table 5: Explosion resistance for NCC samples tested
green i.e. directly from the mould. Samples with
4.15 mass% and 3.5 mass% water for casting.
Temp. [C] NCC (microsilica-gel bond)

4.15 mass%
water
3.50 mass%
water
300
350
400
500
:sample passed the test, :sample did not pass.

It is seen that drying at 110C does not affect the
explosion resistance of the LCC indicating that the
bond is not considerably changed during the drying.
The fact that it tolerates temperatures up to 600C
before failing the test is probably an effect of the
good strength of the castable. The improvement of the
NCC by lowering the water may also be similar, not
only is the amount of water less, but the strength is
improved as well.
The excellent results of the dried NCC was attributed
to the low amount of residual water in the bond after
drying. To substantiate this hypothesis, a series of
gravimetric experiments were undertaken and the
following figure, Figure 13, was presented at
IREFCON 2012
8


Figure 13: Weight loss of microsilica-gel bonded
NCC as a function of thermal treatment. Castable of
Table 2 cast with 4.15% water.

According to these results, some 89% of the water
disappears at 110C, leaving only 10.84% of the
original 4.15wt% water (0.45wt%). Somewhat
surprising, it was found that as much as 19% (0.78
out of 4.15%) had disappeared before drying at 110C.
It is assumed that most of this loss came during the
handling and weighing of the samples after
demoulding.

We felt that we should have a correlation with the
LCC in order to explain and verify our hypothesis.
Therefore some additional gravimetric investigations
were performed on the LCC composition using 4.15%
water. Instead of 800C a temperature of 600C was
chosen as firing temperature since firing at 1000C
did not give any further weight loss. Figure 14 shows
the results obtained on the LCC sample.
Figure 14 shows that the residual water was as high as
27% after drying at 110C. If this difference is tied up
in hydrates that both clog the pore structure and
liberates over a narrow temperature range, it may well
result in bursting of the sample at 600C.


Figure 14: Weight loss of LCC as a function of
thermal treatment. Castable of Table 3 cast with 4.15%
water.

Silica sol
After the patents (e.g.
11
) of Magneco Metrel
concerning use of silica sol as binder in refractory
castables recently started to expire we have seen an
incrteased interest in using silica sol as binder. This
has to some extent puzzled us since basically we are
dealing with a variation of the bond system treated in
this paper, i. e. the microsilica-gel bond. The
difference is that a liquid silica sol (sometimes termed
colloidal silica) is made to gel by some gelling agent.
If used together with microsilica, it becomes very
difficult to distinguish between the effects of the
gelling of the sol or the microsilica. To check for
beneficial effects of silica sol additions, we set up a
program where parts of the microsilica was replaced
by silica sol. The silca sol was a silica of 130m
2
/g
dissolved in an aqueous liquid at 40weight%, while
compensating for the added liquid, so that silica
sol+microsilica and total water was kept constant at 8
and 4.1% respectively. It was expected that the
addition of the silica sol that was much finer than
microsilica, could complement on particle packing
and thus enhance flow. Alternatively a chance existed
that the IPS (interparticle separation), would become
too small so that the particles started to interact and


18.80%
70.36%
10.84%
Curing 20C Drying 110C Firing 800C
10 %
63 %
27 %
Curing 20C Drying 110C Firing 600C
thus influence flow negatively.
In Table 6, the experimental set up, the recipes, are
given, the compositions were made so that amount of
liquid and amount of superfine silica should be
constant at 4.1 and 8% respectively.





Table 6: Castables with silica sol. The compositions were made so that amount of liquid and amount of silica should
be constant at 4.1 and 8% respectively.
[weight %]
Elkem Microsilica 971U
8 7.8 7.2 6.4 5.6
CAC cement, Secar 71 0,5 0,5 0,5 0,5 0,5
Alphabond 300 0,5 0,5 0,5 0,5 0,5
White Fused Alumina 3-5mm 10 10 10 10 10
White Fused Alumina 0,5-3 mm 32 32 32 32 32
White Fused Alumina 0-0,5mm 16 16 16 16 16
White Fused Alumina -74 micron 20 20 20 20 20
Calcined alumina,CT 9FG 13 13 13 13 13
Dispersant, Castament FS20 0.05 0.05 0.05 0.05 0.05
Silica sol: BindZil 40/130 % 0 0,5 2 4 6
Water 4.1 3.8 2.9 1.7 0.5

Figure 15: Flow (ASTM cone) as a function of silica
sol addition. Castables with a total of 4.1% water and
8% silica according to recipes given in Table 6.
Figure 15 shows the flow values of the castables in
Table 6 as measured with the 50mm tall ASTM cone.
Up to 4% silica sol addition, the flow is unaffected,
from 4% on a decrease becomes obvious. The reason
for this is not obvious, but may be connected to the
high surface of the colloidal silica. At 4% silica sol
the colloidal silica has a surface area similar to 8-10%
microsilica. As a consequence, the IPS decreases,
possibly to a level where interactions start to become
noticeable. At what distance particle-particle
interaction starts to become problematic is difficult to
predict, but distances of 50-70nm has been suggested
in literature
12.
.
If the water in our mix (i.e. that with 4% silica sol)
is evenly distributed on the silica surface, then rough
calculations without taking interparticle porosity into
consideration, yields a layer of approximate 11nm
thickness, - or an average particle-particle distance of
22nm. This is well below the 50-70nm suggested in
10
.
With addition of more silica surface, it was
considered probable that this could result in better
strength values due to lower IPS and more silanol
groups. Figure 16 shows the cold modulus of rupture
and the cold crushing strength as measured
immediately after demoulding. The samples had been
cured for 24 hours at close to 100% relative humidity
prior to the testing.
Opposite to the expectations, no positive effect of the
silica sol additions were seen in the green state. A
possible explanation could be that the Bindzil is
stabilised in such a way that it does not react with
Ca
2+
under alkaline conditions. The rather high pH of
9.2
13
of the Bindzil indicate that some sort of pH
adjustment has been performed on it. Such adjustment
could possibly render the silica surface unreactive
under the present coagulation conditions.
124 123 123
101
72
152
155
152
148
124
0
20
40
60
80
100
120
140
160
180
0 0.5 2 4 6
F
l
o
w
-
v
a
l
u
e

[
%
]
Silica sol addition [%]
Self-flow
Vibra-flow
Figure 16:Green strength as a function of silica sol
addition. Castables with a total of 4.1% water and 8%
silica according to recipes given in Table 6.
After drying however, the picture is
Figure 17, it is seen that both MOR and CCS
increases with the addition of silica sol.
during the drying at 110C, the silica sol improves
bonding. The mechanism behind is difficult to explain,
but it has been put forward
14
that the silanol groups
on the silica surfaces may condense to bonding
siloxane and water. This condensation is illustrated in
Figure 17.
Figure 17:Dried strength as a function of silica sol
addition. Castables with a total of 4.1% water and 8%
silica according to recipes given in Table 6.







2.0 2.0 2.0
1.8
12.4
12.5
10.6
10.1
0
2
4
6
8
10
12
14
0 0.5 2 4
G
r
e
e
n
-
s
t
r
e
n
g
t
h

[
M
P
a
]
Silica sol addition [%]
5.3 5.5 5.7 5.7
39.4
43.3
43.9
0
10
20
30
40
50
60
0 0.5 2
D
r
i
e
d

(
1
1
0

C
)
s
t
r
e
n
g
t
h


[
M
P
a
]
Silica sol addition [%]
C-M.O.R.
CCS
Green strength as a function of silica sol
addition. Castables with a total of 4.1% water and 8%
silica according to recipes given in Table 6.
After drying however, the picture is changed. In
Figure 17, it is seen that both MOR and CCS
ddition of silica sol. Somehow,
during the drying at 110C, the silica sol improves
bonding. The mechanism behind is difficult to explain,
that the silanol groups
may condense to bonding
This condensation is illustrated in
Dried strength as a function of silica sol
addition. Castables with a total of 4.1% water and 8%
silica according to recipes given in Table 6.


Figure18: Suggested
12
silanol condensation during
drying of microsilica containing ULCC and NCC.
Although probable, we do not have firm evidence of
this mechanism. Nevertheless we do
strengthening following the drying. This
strengthening seems to be dependent on the amount
of silica sol so even if the silica sol gives weaker
green-bodies, it contributes to higher dried strength.
The overall conclusion may therefore be
in the following way: If it is possible to have a liquid
silica sol additional to the dry-mix, and t
in green-strength is not a problem, then it may be a
good idea to make castables where
microsilica is supplemented with silica sol.
resistance has not been tested, but is expected to yield
similar results as the non-silica sol compositions.

Hot strength
This paper is not intended to present much about the
hot-properties of this type of castable, just a few
figures taken from another source
bond systems.
It is an old experience that if microsilica
together with cement, often 5 mass% of each,
refractoriness may suffer severely. Figure
comparison between the refractoriness under load
(R.U.L.) of three castables based on the compositons
given in Table 2 and 3 with 8 mass% microsilic
two LCC differ in the type of cement used, one is a
common 70% CAC the other a
clearly seen that the reduction in CaO
leads to improved refractoriness. The big
improvement comes for our microsilica
NCC with properties that can make it
temperatures up to 1800C. All this is due to mullite
formation, a topic that has been treated extensively in
earlier
15
publications. Briefly described, the
microsilica, in absence of cement, reac
and forms mullite. The situation is quite different if
cement is present. A liquid is formed that has a
composition close to 50 mass% 70% CA
combination with 50 mass% microsilica. So if e.g. a
castable with 5 mass% microsilica and
cement is tested, then 10 mass% liquid forms at
1500C. The result is a catastrophic failure of the
1.8
10.1
9.3
4 6
C-M.O.R.
CCS
5.7 6.1
49.3
52.0
4 6

condensation during
drying of microsilica containing ULCC and NCC.
, we do not have firm evidence of
Nevertheless we do observe
the drying. This
strengthening seems to be dependent on the amount
silica sol so even if the silica sol gives weaker
bodies, it contributes to higher dried strength.
conclusion may therefore be formulated
f it is possible to have a liquid
mix, and the lowering
strength is not a problem, then it may be a
good idea to make castables where some of the
microsilica is supplemented with silica sol. Explosion
resistance has not been tested, but is expected to yield
sol compositions.
present much about the
properties of this type of castable, just a few
other source
10
with the same
that if microsilica is used
together with cement, often 5 mass% of each,
suffer severely. Figure 19 shows a
comparison between the refractoriness under load
(R.U.L.) of three castables based on the compositons
with 8 mass% microsilica. The
two LCC differ in the type of cement used, one is a
common 70% CAC the other an 84% CAC. It is
reduction in CaO (84% cement)
leads to improved refractoriness. The big
improvement comes for our microsilica-gel bond
NCC with properties that can make it applicable at
temperatures up to 1800C. All this is due to mullite
formation, a topic that has been treated extensively in
publications. Briefly described, the
microsilica, in absence of cement, reacts with alumina
and forms mullite. The situation is quite different if
cement is present. A liquid is formed that has a
composition close to 50 mass% 70% CA-cement in
combination with 50 mass% microsilica. So if e.g. a
castable with 5 mass% microsilica and 5 mass%
cement is tested, then 10 mass% liquid forms at
1500C. The result is a catastrophic failure of the
castable when temperature reaches 1500C This may
be seen in Figure 9 as the almost vertical drop at
1500C.

Fig. 19: Effect of cement on RUL for WFA based
castables with 8 mass% microsilica. LCC has
6 mass% cement.

Compared to microsilica free castables, the
microsilica-gel bond shows similar refractoriness,
Figure 20. When using the microsilica-gel bond it is
possible to obtain a refractoriness equal to or even
above the microsilica-free LCC alternative.

Fig. 20: Comparison of RUL for WFA based
microsilica-gel bond NCC (0.5 mass% cement) with 8
mass% microsilica, LCC (6 mass% cement) with 8
mass% microsilica and LCC (6 mass% cement)
without microsilica

Conclusion:
For castables are based on gelation of microsilica as
bond system, several attractive properties have been
identified. These are; very good placing properties
combined with high refractoriness and hot strength.
The green strength is low but sufficient, and the dried
strength is good. Further, the bond contains only
small amounts of bonded water, so once the free
water is removed, the castable tolerates very high
heating rates. If silica sol is used complementary to
the microsilica, improved strength was seen for the
dried samples. The green strength did not benefit
however.
In general, the gel-bond offer attractive possibilities
for advanced refractories, opportunities that may be
difficult to match with alternative bond-systems.

References:

1
K. Heggestad, J.L. Holm, K. Lnvik and B.
Sandberg, "Investigations of Elkem Microsilica by
Thermosonimetry", Thermochimica Acta, 72 (1984),
205-212
2
E.Dingsyr, M. Dstl and C. Wedberg, in "Preprint
of the Fifth European Symposium Particle
Characterization, 24-26 March 1992, Nrnberg,
Germany. publ. by Nrnberg Messe GMBH
3
R.K. Iler, The chemistry of silica, John Wiley &
Sons, New York 1979.
4
B. Myhre and K. Sunde, "Alumina based castables
with very low contents of hydraulic compound. Part I:
The effect of binder and particle-size distribution on
flow and set.", Proc. UNITECR95, Kyoto, Japan,
Nov. 19-22 1995, p. II/309-16
5
B. Myhre and K. Sunde, "Alumina based castables
with very low contents of hydraulic compound. Part
II." Strength and High-Temperature Reactions of No-
cement Castables with Hydraulic Alumina and
Microsilica, Proc. UNITECR95, Kyoto, Japan,
Nov. 19-22 1995, p. II/317-24
6
B. Myhre and H. Fan, Gel Bonded Castables
Based on Microsilica as Binder, proc.
UNITECR2011
7
M. A. L. Braulio*, V. C. Pandolfelli and C. Tontrup,
Colloidal Alumina as a Novel Refractory Castable
Binder proc. 53rd Int. Coll. on Ref. Aachen 2010 p.
111-114
8
B. Myhre and Aase M. Hundere: Substitution of
Reactive Alumina with Microsilica in Low Cement
and Ultra Low Cement Castables. Part I: Properties
Related to Installation and Demoulding Proc.
UNITECR97, New Orleans, USA, Nov. 4-7 1997, p.
43-52
9
Aase M. Hundere and B. Myhre: Substitution of
Reactive Alumina with Microsilica in Low Cement
and Ultra Low Cement Castables. Part II: The Effect
of Temperature on Hot Properties Proc.
-2
-1
0
1
2
0 500 1000 1500 2000
Temperature [C]
E
x
p
a
n
s
i
o
n

[
%
]
LCC (70% CAC)
LCC (84% CAC)
0.5% (70%CAC)
-2
-1
0
1
2
0 500 1000 1500 2000
Temperature [C]
E
x
p
a
n
s
i
o
n

[
%
]
0.5% cement 8% MS
6% cement 8% MS
6% cement , 0% MS

UNITECR97, New Orleans, USA, Nov. 4-7 1997, p.
91-100
10
Bjrn Myhre and Haibing Fan, Microsilica-gel
bond for explosion proof castables proc. IREFCON
2012, Kolkata 2012 p 71-76
11
US pat. 5147830 Composition and method for
manufacturing steel-containment equipment
Subrata Banerjee et al
12
J.E. Funk and D. R. Dinger:"Particle Size Control
for High-Solids Castable Refractories", Am. Ceram.
Soc. Bull.73[10],66-69, (1994)
13
ALBIN KLINT, Amphiphilic surface modification
of colloidal silica sols, Master of Science Thesis in
the Master Degree Programme Chemistry and
Bioscience, Department of Chemical and Biological
Engineering, Division of Applied Surface Chemistry
CHALMERS UNIVERSITY OF TECHNOLOGY,
SWEDEN
14
Li Zaigeng et al., Phase Compositions and Setting
Mechanisms of Low Cement, Ultra-low Cement and
Cement-free Castables in Proc. 2nd Int. Symp on
Refr. Beijing China, Oct. 30 - Nov. 2 1992, p.
540-547
15
B. Myhre: Lets Make a Mullite Matrix! Ref.
Appl. and News, vol 13, No 6, 2008