Journal of Minerals & Materials Characterization & Engineering, Vol. 3, No.2, pp 81-89, 2004
jmmce.org Printed in the USA. All rights reserved
81
Study On Application Of Alumina In High-Purity
Andalusite Based Refractory
Guihua Liao
1
, Keying He
1
, Liusheng Li
1
, Mingxue Jiang
2
1, Department of Materials Engineering, Luoyang Technology College,
Luoyang 471003, China
2, College of Material Science & Engineering , Xi’an University of
Architecture & Technology, Xi’an 710055, China
Abstract:
Effects of some kinds of alumina on physical properties of high-purity andalusite
based refractory were studied, and, effects of type and amount of added alumina on phase
composition and microstructure of the refractory were investigated by XRD and SEM.
The results revealed that the sintering property of the refractory sample without added
alumina was better than that of one with added alumina, whereas the effect of a selected
sample of added alumina on the sintering property of the high-purity andalusite based
refractory was related to the purity and activity of the added alumina sample. The SiO
2
glass phase originated from the decomposed andalusite aggregate cannot wholly react
with added alumina to form secondary mullite. Moreover, when the quantity of added
alumina amounted to about 15% (mass percent), more extensive secondary mullitization
and greater expansion will be observed in the experimental sample.
Keywords: andalusite based refractory; alumina; secondary mullitization; physical
properties
Andalusite (Al
2
O
3
SiO
2
) is a member of the alumino-silicate minerals. Because
of having excellent high temperature volume stability, mechanical strength, thermal
shock resistance and creep resistance, the refractories made from andalusite have been
used in many industrial fields, such as metallurgical industry, building materials industry,
and others
[1-7]
. Although mullitization and sintering properties of andalusite concentrate
powder have been studied by some scholars
[8-11]
, the reports about applied research of
alumina in high-purity andalusite based refractory have not been seen. Aiming at the
development of high performance andalusite based refractory products, this paper probes
the effects of the addition of selected samples some of alumina on microstructure, phase
composition and specified physical properties of high-purity andalusite based refractory
(hereafter abbreviated to HPABR).
1. Experimentation
The chemical composition of main raw materials adopted in experiments are
listed in Table 1. The high-purity andalusite listed was from South Africa, and its coarse
concentrate had been calcined before being used, the calcined condition was
1500℃×3h. Concerning the added alumina raw materials, all of those listed except

82
Liao Guihua, He Keying, Li Liusheng and Jiang Mingxue Vol. 3, No.2
active alumina are less than 0.044mm in particle size, whereas active alumina is a
micropowders that has a particle size of less than 5μm. The experiment included two
parts. At first, equivalent amount of sintered alumina, fused brown corundum, bauxite
based corundum and active alumina were added into the matrix of samples respectively,
and the effects of different samples of alumina on physical properties of HPABR were
investigated. Second, a selected sample of alumina was fixed, and the effects of this
addition of the fixed alumina on physical properties of HPABR were further investigated.
By way of adopting an oil press, green bodies of all samples were formed under pressure
of 200 MPa. After being dried for about 24 hours, the green bodies were fired in an
electric furnace under 1500°C and for 3 hours of soaking time.
Table.1 chemical composition of main raw materials (%)
Raw materialSiO
2
Al
2
O
3
Fe
2
O
3
TiO
2
CaO MgO K
2
O Na
2
0
Andalusite38.60 59.39 1.01 0.03 0.22 0.23 0.15 0.06
Sinerted alumina0.18 99.41 0.12 — 0.15 0.13 0.05 0.09
Fused brown corundum0.63 95.42 0.37 2.23 0.13 0.08 0.31
Bauxite based corundum0.29 98.59 0.16 0.42 0.10 — 0.07 —
Active alumina0.12 99.52 0.08 — 0.06 0.05 — 0.03
After the fired samples cooled down, the bulk density, porosity, crushing strength,
firing expansion and refiring expansion of them were tested. The creepage of a part of the
samples were also tested, and microstructure of the selected samples were observed by
SEM. In addition, the phase composition of a part of the samples were tested by XRD,
and the instrument used was Japanese D/max-3c model diffractometer. When the
diffractometer tested samples, it had 35 KV working voltage and 40 mA working current,
and scaned the samples by CuK
α
lines.
2. Results and Discussion
2.1 Effects of different kinds of alumina added in matrix on properties of HPABR
Effects of different samples of
alumina added to the matrix on bulk
density, porosity, crushing strength of
samples are shown in Fig.1 and Fig.2
respectively, effects of the samples with
alumnia on firing expansion and refiring
expansion of the samples are shown in
Fig.3. and the phase composition of fired
samples are listed in Table 2. The samples
were: Symbol J
0
without added alumina, J
1
with sintered alumina added, J
2
with fused
brown corundum added, J
3
with bauxite
based corundum added and J
4
was the
sample with active alumina added.
Fig.1 and Fig.2 show that adding different samples of alumina in matrx had
Fig.1. Bulk density and apparent porosity of fired
samples added with different alumina

Vol. 3, No.2 Study on Application of Alumina in High-purity Andalusite Based Refractory
83
different effects on sintering property of
HPABR sample. As shown in the figures,
sample J
0
without added alumina had
better sintering property, while the
sintering property of other samples added
with alumina was inferior to sample J
0
. In
addition, the sintering property of sample
J
1
, which was added with sintered
alumina, almost was equivalent to that of
sample J
3
added with bauxite based
corundum, whereas the sintering property
of sample J
2
added with fused brown
corundum was better than that of sample J
1
and J
3
, and, the sintering property of
sample J
4
added with active alumina was
inferior to that of sample J
1
and J
3
.
The causes of the results
mentioned above should lie in differences
in purity, activeness of added alumina, and
in difference in intensity of secondary
mullitization effect caused by these
factors. The reason why sample J
0
without
added alumina had a better sintering
property is that, when andalusite in sample
J
0
was heated under 1300°C ~1500°C and
transformed into mullite with a capillary
network simultaneously, about 15%~19% (of andalusite mass) silica-rich glass would be
formed in it
[11]
, the silica-rich glass originating from andalusite mullitization would react
with other impurities in sample J
0
and form low melting liquid phase further,
consequently, the sintering state of sample J
0
was improved. Instead, in those samples
added with various alumina, because added alumina would react with the silica glass
originating from andalusite mullitization to form secondary mullite. While this reaction
was followed with volume expansion of 7%~8%
[12]
, which hindered sintering of the
samples. Therefore, it was conclude that the sintering property of the samples with
various alumina added was inferior to that of sample J
0
. The firing expansions of samples
shown in Fig.3 also prove this conclusion that except for sample J
0
, the firing linear
changes of all the other samples showed volume expansion, and only sample J
0
showed
volume shrinkage. It was concluded that sintered alumina and bauxite based corundum
had equivalent content of impurities (Table 1) and the same fineness of powders, so they
had equivalent reaction activity, consequently sample J
1
and J
3
had equivalent sintering
property. That sample J
4
had inferior sintering property was because active alumina
added to it resulted in a more intense secondary mullitization effect under high
temperature, thus it was more difficult for sample J
4
to be sintered. This can be verified
by the firing expansion of fired samples shown in Fig.3 and the phase composition of
fired samples shown in Table 2. Sintering property of sample J
2
was better than that of
Fig.2. Cold crushing strenghth of fired
samples added with different alumina
Fig.3. Firing and refirinng linear change of
samples added with different alumina

84
Liao Guihua, He Keying, Li Liusheng and Jiang Mingxue Vol. 3, No.2
sample J
1
and J
3
, but mullite content in sample J
2
was less, this might probably relate to
higher content of alkali oxide impurities in fused brown corundum. Wang Jinxiang et al
has researched effect of K
2
O on phase composition of sintered bauxite based material
[13]
.
His research results showed that increase of K
2
O content in the material would not only
result in increase of glass phase in it, but at the same time, also cause decomposition of
mullite in it.
Table.2 Phase composition of fired samples added with different alumina
Minerals
J
0
J
1
J
2
J
3
J
4
Andalusite
1618291519
Mullite
6467566670
Corundum
—1—4—
Glass
2013151211
In addition, the diffraction intensity
of the strongest peak (110) of andalusite in
fired sample J
0
~J
4
was shown in Fig.4. It
can be seen from Fig.4 and Table 2 that,
compared with other fired samples, fired
sample J
2
contained much andalusite
mineral. This is in contradiction with
research conclusions of other researchers
[9-
10]
, who found that existence of impurities,
especially alkali impurities, would cause
further decomposition of andalusite. The
causes of these results need to be probed
further.
SEM photographs of matrix area of
fired sample J
0
, J
1
and J
4
are shown in
Fig.5. By way of comparision and analysis, it can be seen that all gaps between grains in
matrix area of sample J
0
are filled with glass phase, and it appears similar grains
dispersively existing in glass matrix. Whereas SEM photographs of sample J
1
and J
4
manifest the borders between grains in the matrix area as more vague because of the
formation of secondary mullite, which was caused by the reaction between added
alumina and silica glass originating from andalusite decomposition. It could be said that
secondary mullite formed a bridge-link between original grains in the matrix area of
sample J
1
and J
4
, especially in matrix area of sample J
4,
the bridge-link formed by
secondary mullite appears to be tighter.
2..2 Effects of amount of added alumina on properties of HPABR
2.2.1 Definition of amount of alumina added in the matrix
Fig.4. XRD patterns of sample J
0
~J
4

Vol. 3, No.2 Study on Application of Alumina in High-purity Andalusite Based Refractory
85
In the light of experimental results
presented above, it was decided that active
alumina would be used for studying
further effects of varying the amount of
added alumnia on properties of HPABR.
In order to promote the generation of silica
glass from mullitized andalusite to be
transformed into secondary mullite, it is
necessary to control the appropriate
amount of alumina existing in the sample.
As stated above, when andalusite crystal
wholly was transformed into mullite with
a capillary network under high
temperature, about 15%~19% (mass
percentage) of silica glass would be
formed, however most of the silica glass it
would be trapped in the capillary network
of the neoformed mullite, whereas only
about 3.5%(mass percentage) of the silica
glass would not be accommodated in the
capillary network and would be squeezed
out. This “free” silica glass expelled from
the neoformed mullite grains easily
reacted with added alumina to form
secondary mullite. Therefore, if 18% silica
glass originated from mullitization of
andalusite powder and 3.5% silica glass
came from being expelled from mullitized
andalusite aggregate and these combined
amounts of silica glass could participate in
the reaction with added alumina to form
secondary mullite, then the amount of
alumina added in a sample can be determined or defined by the total amount of andalusite
aggregate and powder in the sample. On this account, the formula of each sample with a
different amount of added active alumina was designed as shown in Table 3, the amount
of active alumina added in sample J
6
was approximately equal to amount (≈16%)of
alumina needed for making silica glass expelled from mullitized andalusite aggregate and
powder to react with it and to form secondary mullite.
Table.3 Formula of samples added with alumina
Raw materialJ
5
J
6
J
7
J
8
Andalusite aggregate60606060
Andalusite powder28231813
Active alumina10152025
Bond2222
J
0
(1000×)
J
1
(1000×)
J
4
(1000×)
Fig.5 SEM photographs of matrix of
sample J
0
~J
1
and J
4

86
Liao Guihua, He Keying, Li Liusheng and Jiang Mingxue Vol. 3, No.2
2.2.2 Effects of addition amount of active
alumina on properties of HPABR
Effects of addition amount of
active alumina in the matrix on bulk
density, porosity and crushing strenghth of
HPABR sample are shown in Fig.6 and in
Fig.7 respectively, and effects of it on
firing expansion and refiring expansion of
the samples are shown in Fig.8.
It can be seen from Fig.6 that the
bulk density of the samples increased with
increment of additional amount of active
alumina micropowders; and, that there was
not much difference in porosity amongst
other samples except sample J
6
, and only
the porosity of sample J
6
was a little
higher. The change of crushing strength
was from the former decrement to latter
increment, which took sample J
6
as a turn
point (Fig.7). This also was in accordance
with the firing expansion case of samples.
Compared with sample J
7
and J
8
, sample J
6
had larger porosity and lower crushing
strength strenghth. It was probably
because the amount of active alumina
added in sample J
6
correctly resulted in
both the largest secondary mullitization
effect and having no remainder of active
alumina micropowders to impel further
sintering of sample J
6.
Because of the
lower amount of active alumina
micropowders in sample J
5
, the secondary
mullitization effect in it was less than that
in sample J
6
. Thus, sample J
5
showed
smaller firing expansion, and at the same
time, as a result of much more glass phase
present amongst mullite grains and firmly
sticking them together, it also showed
higher crushing strength. Table 4 shows phase composition of each fired sample, which
was obtained by way of XRD analysis (XRD patterns are omitted). It can be seen from
Table 4, that there was a certain amount of corundum phase in sample J
7
and J
8,
and the
amount of corundum existing in them was approximately equal to the amount gained by
subtracting alumina amount (≈16%) consumed in secondary mullitization process from
the amount of active alumina added in them initially.
Fig.6 Effects of addition amount of active alumi
n
on bulk density and apparent porosity of fired
samples
Fig.7 Effect of addition amount of active alumina
on cold crush strenghth of fired samples
Fig.8 Effects of addition amount of active alum-
ina on firing and refiring linear change of
samples

Vol. 3, No.2 Study on Application of Alumina in High-purity Andalusite Based Refractory
87
Table.4 Phase composition of fired samples with different addition amount of active alumina powder
Mineral
J
5
J
6
J
7
J
8
Andalusite
16171910
Mullite
65686769
Corundum
——29
Glass18131111
The experimential result mentioned
above proves that the inference made in
the preceding paragraph is correct, that is,
in the sintering process of andalusite based
products, most of the whole silica glass
(≈18%) originated from the mullitization
of andalusite powder and a little silica
glass (≈3.5%) was expelled from the
mullitized andalusite aggregate which
participated in the secondary mullitization
reaction, whereas the silica glass trapped
in capillary crevices in initial mullite
grains could hardly react with added
alumina to form secondary mullite.
Therefore, if the intent is to avoid
corundum phase arising in sample, the
additional amount of alumina should be
defined in the light of amounts of
andalusite aggregate and powder in
sample. Fig.9 has shown the effect of
addition amount of active alumina
micropowders on creepage of fired
samples, the condition of creep experiment
was 1400°C ×50h, 0.2MPa. It can be seen
that, increasing the amount of added
alumina, the creepage of fired samples
decreased at first, then increased again,
and, the samples with 15%~20% added
alumina had lower creepage. Although
presently there is a research conclusion
[14]
that indicates the multiphase material
composed simultaneously of mullite and
corundum had better creep resistance than
the monophase material composed only of
mullite or corundum, however, this was
only as far as the material nearly without
glass phase is concerned. So far as this
experimental work is concerned, the fired samples contained a significant quantity of
glass phase (Table 4), so the conclusion mentioned above is not suitable to this
Fig.9 Effect of amount of alumina added on
cree
p
a
g
e o
f
sam
p
les
J
5
(1000×)
J
8
(1000×)
Fig.10 SEM photographs of matrix of sample J
5
and J
8

88
Liao Guihua, He Keying, Li Liusheng and Jiang Mingxue Vol. 3, No.2
experimental result. Contrarily, because of bad creep resistance of granular corundum
grains, in the glass phase existing in grain borders, granular corundum grains would slide
easily when they bear a compression stress for long time under high temperature, and
consequently would result in a larger creep effect. Fig.10 shows SEM photographs of
matrix area of fired sample J
5
and J
8
. Apparently, many granular corundum grains
dispersively existed in the matrx of sample J
8
, while there were less granular corundum
grains in the matrx of sample J
5
, and the products generated in secondary mullitization in
it made the joining degree between grains higher.
3. Conclusion
1. The effect of various samples of alumina powder on sintering property of
HPABR is relevant to purity and activeness of it. The sintering property of samples with
added active alumina micropowders is inferior, that of samples with added fused brown
corundum is better, while that of samples with added sintered alumina or bauxite based
corundum would be graded between the two above.
2. In the sintering process of andalusite based material, only about 18% silica
glass originating from mullitization of andalusite powder and about 3.5% silica glass
expelled from mullitized andalusite aggregate can be available to participate in secondary
mullitization reaction.
3. As far as this experimental work is concerned, when the amount of added
active alumina in sample is about 15%, the sample had the largest secondary mullitization
effect and porosity, firing expansion, and lower crushing strenghth.
4. Comprehensively taking demands of various properties of HPABR into
account, we think that a suitable additional amount of alumina in it should be 10%~20%.
References:
[1] Dubreuil P. et al，Use of Andalusite Refractories in Ferrous Metallurgy，Refra.
Ind. Ceram. 1999, Vol.40 (6)：252~259
[2] R.B. Bartelink et al，Andalusite—25 years on. (A Report of the Rhino
Andalusite Mines Company)，1996.9
[3] F. T. Niemeijer et al，New Refractories for Torpedo Ladles，Steel Times
International，1999，Vol.14 (6)：38
[4] Lionel Rebouillat et al ，Andalusite-Based High-Alumina Castables，J. Am.
Ceram. Soc. 2002，Vol.85 (2)：373~378
[5] Manfred Koltermann，Andalusite Bricks for Torpedo Ladles，Steelmaking
Conference Proceedings，1979，Vol.62： 241~245
[6] YANG Zhifu.Overseas Refractory (in Chinese), 2001, (3): 32-37
[7] William H.McCracken et al，Annual Minerals Review—
Andalusite， Am.Ceram.Soc.Bull. 1997，Vol.77 (6)：85~86

Vol. 3, No.2 Study on Application of Alumina in High-purity Andalusite Based Refractory
89
[8] WEN Hongjie. Naihuo Cailiao (in Chinese), 1995, 29(5): 247-250
[9] LI Bowen. J. Mineral Petrol (in Chinese), 1993, 13(3): 36-41
[10] LIN Binyin. Naihuo Cailiao (in Chinese), 1990, (1): 28-32
[11] Jean-Pierre Ildefonse et al ，Mullitization of Andalusite in Bricks and
Castables[A]. IN: UNITECR'97. Proc. Unified Int. Tech.Conf. on
Refractories[C]，New Orleans, USA:1997. 899~908
[12] He Jin-zhou et al，The Study on Al
2
O
3
—SiO
2
Based Castables Containing
Andalusite[J]，China’s Refractories (in English)，1994, Vol.3 (2)：14~18
[13]WANG Zetian, YAN Xingjian. Zhong Xiangchong’s Select Treatise on
Refractories (in Chinese). Beijing: The Metallurgy Industry Press, 1991. 236-
237
[14] LI Tingshou. Naihuo Cailiao (in Chinese), 1986, (3): 1-8