Ceramics International 21 (1995) 265-269

Cordierite-Mullite Refractories
D. M. Ibrahim, S. M. Naga, 2. Abdel Kader & E. Abdel Salam

Ceramics Department, National Research Centre, Dokki, Cairo, Egypt

(Received 1I July 1994; accepted 30 November 1994)

Abstract: This study aims to prepare cordierite-mullite composites to be used as

refractories, mainly setters.
The cordierite-mullite percentage was kept constant (70 : 30%). Mullite and
cordierite were added in different proportions either as grog or raw materials,
fulfilling the above stoichiometric composition.
To attain the required percent of mullite in the product, a superstoichiometric
amount of alumina was added as grog or bauxite. Superstoichiometric addition
improved both physical and mechanical properties without affecting the thermal
shock resistance.
The effect of added grog percentage on the body properties is discussed.
Best mechanical, thermal and physical properties are achieved with bodies
containing 50-70% grog.

1 INTRODUCTION bauxite, as well as chemically pure MgCO,, LiF

and AlF3. The chemical analyses are given in
Cordierite-mullite composites have found their Table 1. Talc was calcined at 1000°C for 1 h and
application as refractories with recent develop- added as 1: 1 mixture of calcined and uncalcined
ments in the fast-firing techniques of ceramic to minimize preferred orientation that leads to
products during the last few decades. lamination in the product with the use of a high
Cordierite supplies resistance to thermal shock, percentage of raw talc.
while mullite provides the strength needed.‘-3 Con- Cordierite and mullite grog were prepared on
ditions of preparation of composite bodies from the basis of their stoichiometric composition, and
cordierite and mullite have been studied by differ- fired at 1350°C and 145O”C, respectively (Table 2).
ent authors.66 They found that the proper grog Once the grog was prepared, the raw constituents
gradings for cordierite and mullite were -850, + were also used to prepare a group of mixes based
224 pm and -850, + 425 ,um, respectively.(j Glass’ on the composition 70 : 30 cordierite-mullite,
or polyvinylalcohol were used as binder. Gros- respectively. The compositions were satisfied by
jeans showed that the presence of the cordierite varying the percentage of grog between 0 and 100
chamotte lowers the reaction temperature and of raw constituents (Table 3).
decreases the firing shrinkage.
The present work aims at studying the condi- Table 1. Chemical analysis of raw materials
___
tions of production, phase composition and differ- Constituent Clay Talc Bauxite
ent thermal and mechanical properties of 70: 30 (oxides)
cordierite-mullite composite used as tile setter and
SiO 44.20 62.00 6.50
prepared from local raw materials. Al,& 37.75 I.60 87.50
Fe&h 0.93 0.40 I.50
2 EXPERIMENTAL PROCEDURE TiO 1.85 - 3.70
Cao’ 0.82 0.48 0.43
MgO 0.52 30.30 -
Raw materials used in this study were: local clay from Na,O 1.15 0.32 -
Tieh District, Sinai; talc from Derhieb District, W 0.72 0.11 -
Eastern Desert; quartz from Aswan; imported L.O.I. 13.01 4.70 0.20
265
Ceramics International 0272-8842/9X$9.50 0 1995 Elsevier Science Limited, England and Techna S.r.1. Printed in Great Britain
266 D. 44. Ibrahim, S. 44. Naga, Z. Abdel Kader, E. Abdel Salam

Table 2. Batch composition of cordierite and mullite

Batch (%I Constituents

Talc Clay M&Q, Bauxite SiOz LiF AIF,

C R

Cordierite 9.38 9.38 76.60 15.24 1.90 0.57 I.00 2.0

Mullite - - 55.18 - 58.50 - - -

C: calcined; R: raw.

Table 3. Batch composition of the different mixes investigated

Mix symbol Mullite Cordierite Raw (%) Alumina as bauxite

grog (%I grog (%) (%)

-35,+70 mesh -20,+35 mesh -70,+200 mesh -200 mesh

- - - 100 -
15 35 - 50 -
21 49 - 30 -
27 45 18 10 -
30 50 20 - -

Superstoichiometric mixes

13.64 31.82 - 45.45 9.09

26.09 30.43 - 43.48 -

Two mixes representing superstoichiometry of flexural test. The thermal shock resistance test was
alumina, satisfied by the addition of either cal- carried out according to the ASTM (C-554) gen-
cined bauxite or excessive percentage of mullite eral method, and then test specimens were sub-
grog, were prepared, based on the same composi- jected to flexural test. The underload test was
tion as above (70: cordierite, 30: mullite, fulfilled carried out on selected specimens to determine the
by the mix G,). range of their application and to spot the temper-
The grain size of mullite grog fraction is chosen ature of the start of deformation.
to be between -35 and +70 mesh, while cordierite
grog has two fractions, -20: +35 and -70: +200 3 RESULTS AND DISCUSSION
mesh.6
The composite bodies were fabricated into discs, All mixes under investigation represent one and the
2 cm in diameter and 0.5 cm thick, and rectangles same stoichiometric composition. The difference
of 7 X 0.5 X 1 cm by the semi-dry press method lies in the processing of the ingredients. The addi-
under a pressure of 30 kN using molasses as a tion of various percentage of the two grogs, namely
binder. They were dried and then fired between cordierite and mullite, to the raw mix affected, to a
1200” and 1400°C with a temperature interval of great extent, the properties of the fired bodies.
50°C. The fired bodies are generally highly porous;
Bulk density and apparent porosity were deter- the apparent porosity varied between 30 and 44%.
mined according to the ASTM-C20-74. Reversible Therefore bulk density values lie between 1.53 and
linear thermal expansion was determined with a 1.94 g cme3 (Fig. 1). The increase in grog content
heating rate of S”/min up to 1000°C using an from 0 to 100% decreased the bulk density at all
automatic dilatometer.* Cordierite and mullite firing temperatures, while the addition of a super-
phases were semiquantitatively followed up by stoichiometric proportion of alumina (in the form
XRD using CaF, as an internal standard. A uni- of calcined bauxite or mullite grog) improved bulk
versal testing machine** was used to perform the density. Higher bulk densities were obtained at
1400°C for all bodies containing grog. This may
*Orton automatic dilatometer. be attributed to the dissociation of cordierite into
**Instron 1128. mullite + liquid.
Cordierite-mullite refractories 267

o Before Spalling test

x After Spalling test

I I I I I I
0 20 40 60 80 100
Grog (%)

Temperature (“C)
Fig. 2. MOR of different mixes before and after spalling test.

Fig. 1. Change in bulk density with firing temperature.

Bodies high in grog withstand 50 cycles without
showing a conspicuous change in strength, while
Addition of excess alumina, i.e. superstoichio- bodies containing a high percentage of raw mate-
metric proportion in the form of calcined bauxite rials were detrimentally affected. The crystalline
or mullite grog, improved bulk density, mechanical phases that developed, especially cristobalite,
strength and refractoriness underload of mix G,. played a role in minimizing the spalling resistance
Although the bodies are designed to achieve 70: (Fig. 2). Superstoichiometric mixes showed a drop
30 cordierite-mullite composition, this mullite in strength recorded after 50 cycles of thermal
proportion was not achieved (Table 4). The per- treatment, 30 and 39% for GA and GM, respectively
centage of mullite represents that added as grog, (Table 5).
although part of the mullite grog participated in
the crystallization of cordierite. Accordingly, the Table 5. MOR of superstoichiometric mixes
percentage of mullite recorded was in some cases
less than that started with. Mix symbol MOR (MPa)
The raw mix composition (G,) favored the crys-
Before spalling After spalling
tallization of cordierite at the expense of mullite.
The early formed mullite at 1250°C gave way to G, 15.00 IO.45
cordierite crystallization, reaching 70% at 1350°C. GM 17.60 IO.70

The reversible linear thermal expansion coeffi-

cient lies in the same range, between 1.77 and 1.79
X 10-6/“C, except for the superstoichiometric mix Underload test results (Figs 3 and 4) indicated
containing excess mullite, which showed a higher that selected samples G7 and Gg resist deformation
value of 2.103 X 10--6/“C(Table 4). The approxi- during testing up to 1225 and 13OO”C, respec-
mately constant content of cordierite phase in the tively. The T, temperature for mixes G, and GM
former mixes and mullite phase in the latter are was shifted to its higher temperature of 1325 and
the responsible factors for such behaviour. 14OO”C, respectively, compared with that of the
MOR of the different bodies containing grog original mix Gg. The increase in percentage of
was relatively low. Porosity is responsible for mullite, either added as grog or as a result of the
these low values (Fig. 2). reaction of added alumina with the siliceous
matrix, is responsible for their improvement of
refractoriness and mechanical strength.
Table 4. Coefficient of reversible linear thermal expansion
and phase composition of different bodies

Firing (Y(X lO?W Phase composition (%) 0.6 -

GS
temp. (“C)
C M Cr g
; 0.4 -.
./&dy
.o
GO 1350 1.79 70 9 10 z
2 1300
1250 1.785
1.77 71
70 21
15 -3 d
x 0.2 - .
w
(37
G; 1250 1.77 71 20 -
G0 1250 1.78 70 20 -
1350 1.769 72 20 -
o-
I I I -I
0 200 400 Temizature ,,8:; 1000 1200
1350 2.103 72 23 -

C: cordierite; M: mullite; Cr: cristobalite. Fig. 3. Refractoriness underload of mixes G, and G,.
268 D. M. Ibrahim, S. M. Naga, Z. Abdel Kader, E. Abdel Salam

I I I I I I I
0 200 400 600 800 1000 1200 1400
Temperature (“C)

Fig. 4. Refractoriness underload of superstoichiometric mixes.

Fig. 6. Prismatic well developed crystalline cordierite uniform

SEM of mixes fired at 1250°C (Fig. 5) showed in size with different orientation. (X 10000).
grains of mullite grog preserved in their original size
as identified from the fine microstructure formed Table 6. Pore size distribution, pore volume, pore surface
of 0.3 pm mullite crystals. Cordierite prisms crys- area and density
tallizing out as a reaction product at the mullite
Measurements Results
grog grain boundaries, due to the diffusion of
Mg2+ ions from the neighboring cordierite grain G, G 10 GA
or from the matrix, are shown in Fig. 6. SEM of
products of mixes containing raw proportion Total intrusion
volume I. 1836 cc/g O-0228 cc/g 0.0691 cc/g
showed a matrix formed of well crystallized pris- Total pore
matic shaped cordierite, 5-12 pm in size. area 0.7222 m2/g 0.7844 m2/g 0.1008 m2/g
The results of distribution of pores, pore vol- Median pore
diameter 6.7897 pm 2.2697 pm 3.0768 pm
ume, pore surface area measured by the Hg
Average pore
porosimeter for the end members of the studied diameter 1.0170 pm 0.1162 pm 2.2437 pm
series (G, and G,,) indicate a great variation in the Bulk density 1.6255 g/cm3 1.5566 g/cm3 1.6464 g/cm3
Apparent
size of pores developed. In spite of displaying
porosity 2.3172% 1.6136% 1.8579%
nearly the same total pore area of 0.722 and 0.78 % capillary 32.7766 2.3983 7.9943
m2/g, respectively, the pore size and pore distribu-
tion differ conspicuously. G, showed maximum
pores of capillary size, while Glo revealed a better The above results show a correlation with the
distribution of pores with a median pore size of mechanical behaviour of these mixes; poor mechani-
2.2 pm (Table 6). The superstoichiometric mix, cal strength and high thermal shock resistance.
GA, has a good distribution of pores of median According to Ravaglioli and Krajewsky9j1’ the
size (3.07 pm), but its pore area percentage was mechanical strength and thermal shock resistance
relatively low. It represents about l/7 of that of oppose each other. They are affected by pore size,
the original mix (Table 6). shape and distribution from different concepts. So,
porosity leads to low mechanical strength, but on

Fig. 5. Fine grained mullite crystals occurring in patches with Fig. 7. Well developed prismatic cordierite crystals and acicular
prismatic well crystallized cordierite. (X 10000). fine grains of mullite. (X 7500).
Cordierite-mullite refractories 269

the other hand, to better shock resistance. Rasch” 2. IKAWA, H., WATANABE, T., URABE, K. & UDA-
recorded that fine porosity is advantageous. Thus, GAWA, SH., Thermal expansion and microstructure of
cordierite and mullite composite Yogyo-Kyokai. Shi,
not less than 50% of the total pores should lie 93(12) (1985) 3&I.
between 0.5 and 5 pm, and not more than 20% 3. MUSSLER, B. H., SWAIN, M. U. & CLAUSSEN, N.,
should be greater than 20 pm. Lachman et aLI2 Dependance of fracture toughness of alumina on grain
size and test technique. J. Amer. Ceram. Sot., 65( 11)
attributed the presence of fine porosity to the utiliza- (1982) 566-72.
tion of raw talc. Meanwhile, Rasch’ and Nakahara 4. MUSSLER, B. H. & SHAFER, M. W., Preparation and
et a1.,13recorded the presence of eutectic liquid phase properties of mullitecordierite composites. Ceram. Bull.,
at the contact of talc particles with alumina giving 63(5) (1984), 705-10.
5. MONROE, D. L., WACHTMAN, Jr, J. B. & GAULT,
the fine pores which reduce the apparent porosity. C., The mechanical properties of cordieriteemullite com-
posites as a function of temperature. Proc. of the Ist
4 CONCLUSIONS Europ. Ceram. Sot. Conf (ECerS’ 89) 18-23 June, 1989,
The Netherlands. Elsevier Applied Science, London.
6. CAKIROGLU, 0. & OZGEN, S., The effect of raw
The raw mix favoured the crystallization of cordierite materials on properties of cordierite bodies - II. Proc. of
rather than mullite. The proportion of raw mix is the 2nd Europ. Cerum. Sot. Conf (ECerS’ 91) 11-14
responsible for the proportion of porosity in the September, 199 1, Augsburg. Elsevier Applied Science,
London.
capillary range that decreases the spalling resistance. 7. ANDERSON, R. M., GERHARDT, R.. WACHTMAN,
In order to improve the mechanical strength, Jr., J. B., ONN, D. & BEECHER. S., Thermal mechani-
spalling resistance and physical properties, there is an cal and dielectric properties of mullite-cordierite compos-
ites. Advances in Ceramics, 29, (1989) 265-77.
optimum range of added grog ranging between 50 8. GROSJEAN, P., Cordierite ceramics. Zntercerum., 42( 1)
and 70% cordierite. Otherwise it negatively affects the (1993) 11-l 5.
mechanical strength. The percentage of mullite 9. RAVAGLIOLI, A. & KRAJEWSKY A., Behaviour of
thermal shock resistant refractories-A chemo-physical
detected represents that added as grog.
and mineralogical study. Znterceram., 34(5) (1985) 51-6.
Superstoichiometry of alumina, added in the form 10. RAVAGLIOLI, A. & KRAJEWSKY A., Behaviour of
of mullite grog or calcined bauxite, improved both the thermal shock resistant refractories ~ A chemo-physical
physical and mechanical properties, without affecting and mineralogical study. Zntercerum.. 34(6) (1985)
23-6.
the spalling resistance. Mullite addition also raised the 11. RASCH, H., Cordieritkeramik. Silicatkeramische tech-
mechanical properties and refractoriness underload nische Keramiken, Keramische Werkstoffe, 4.1.30, l-55,
of the mixes (high temperature bearing capacity). April 1992.
12. LACHMAN, I. M., BAGLEY, R. D. & LEWIS, R. M.,
Thermal expansion of extruded cordierite ceramics. Bull.
REFERENCES Amer. Cerum. Sot., 60(2) (1981) 202-05.
13. NAKAHARA, M., HASHHIZUKA, Y., KONDO, Y. &
1. RASCH, H., Some effects of temporary liquid phases, H.4MAN0, K., Behaviour of talc in formation of
firing conditions and raw materials. CFZ/Ber. DKG., cordierite ceramics. J. Ceram. Sot. Jup., 102 (1994)
68(7/8) (1991) 33845. 18-22.