Full text of DOI:10.1557/JMR.2000.0198

Phase evolution during sintering of mullite/zirconia
composites using silica-coated alumina powders
V. Yaroshenkoa) and D.S. Wilkinson
Department of Materials Science and Engineering, McMaster University, 1280 Main Street West,
Hamilton, Ontario L8S 4L7, Canada
(Received 28 April 1999; accepted 13 March 2000)
Mullite-based composites can be made by an in situ reaction process using
silica-coated alumina (SCA) powder as a mullite precursor. In this paper we present
the combined effects of zirconia and premullite seeds on the crystallization process and
microstructure development. When zirconia is added without seeding, mullite
formation proceeds through the formation of transient zircon. This phase provides a
lower energy barrier for mullite nucleation and thus lowers the mullitization
temperature. The presence of yttria as a stabilizer in zirconia reduces the activation
energy for zircon formation and thus promotes the transient reaction. The addition of
premullite seeds results in the nucleation of mullite from alumina and silica, and zircon
does not form. At low seeding levels mullitization remains nucleation-controlled;
however, once the seeding level exceeds 1–2%, this is no longer the case.
I. INTRODUCTION
drawbacks associated with the high specific surface and
low bulk density of these ultrafine powders. In particular,
low green density of the compacts results in high shrink-
age during sintering, which prevents the use of near-net-
shape processing technologies.
Mullite is used for several important high-temperature
structural applications. It combines creep resistance and
strength retained to high temperatures with chemical and
thermal stability, low thermal expansion, and heat con-
ductivity.1 The low diffusion coefficient in mullite is
responsible for both its excellent creep resistance and
poor sinterability.2 A number of approaches have been
developed recently to improve the latter, including the use
of mullite precursors rather than pure mullite as the starting
materials. This enables densification before the forma-
tion of mullite and at a much lower temperature than that
needed to sinter mullite powder. In this processing route
the formation and crystallization of mullite is combined
with sintering in a single “reaction sintering” process.
Thus the proper control of composition and process vari-
ables is critical for both the microstructure and densifi-
cation behavior of the material.
There are a number of factors that influence the reac-
tion kinetics in this system3–5 and thus the resultant mi-
crostructure. For example, because of the poor diffusivity
of both Si and Al ions in mullite, fine scale mixing of
components is critical to reduce diffusion distances and
increase the sintering driving force. Thus, much work has
been devoted to the development of various sol-gel proc-
essing routes which enable component mixing over
atomic (monophasic precursors) to nanoscale (diphasic
precursors) dimensions.1,6 There are, however, some
An alternative approach has been developed by Sacks
et al., which addresses these concerns.7,8 This is based on
the sintering of crystalline ␣-alumina particles coated
with amorphous silica layers. During the sintering of
silica-coated alumina (SCA) particles mullite formation
occurs at higher temperatures than for sol-gel powders.
(This is due to the particle size, which is usually more
than 1 order of magnitude larger.) Literature data suggest
that mullite nucleation and growth in microcomposite
silica-coated alumina (SCA) powders does not take place
at the alumina/silica interface (as would normally be ex-
pected for a mixture of silica and alumina powders).
Instead, it occurs inside the siliceous phase by a process
similar to that which occurs in a sol-gel-derived diphasic
precursor.9,10 This volume reaction makes it feasible to
control the crystallization process and modify the micro-
structure through the incorporation of seeds for addi-
tional crystal nucleation. This possibility has been
demonstrated for sol-gel-derived particles and has
proved to be effective for the reduction of grain size in
the resultant microstructures.11,12
Seeding offers even greater potential beliefs for mi-
crocomposite (e.g., SCA) powders, which have a ten-
dency to develop a coarse-grained microstructure.
Usually the final mullite grains are significantly larger
than the starting powder particles.8 This suggests that
without seeding mullite initiates from a relatively small
a)Present address: Ferro Corporation Technical Center, 7500 East
Pleasant Valley Road, Independence, Ohio 44131.
1358
J. Mater. Res., Vol. 15, No. 6, Jun 2000
© 2000 Materials Research Society
http://journals.cambridge.org
Downloaded: 15 Mar 2015
IP address: 128.218.248.200

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
number of nuclei. The seeding of crystal nuclei with fine
premullite particles reduces the starting temperature and
interval of mullitization. Also, a significant decrease of
grain size in the final microstructure was observed.10
However, as Sacks et al.10 noted, further investigation is
needed to determine which steps control the reaction ki-
netics and how this affects the development of the final
microstructure.
Another way to modify the properties of mullite ce-
ramics is through the addition of zirconia. Apart from
improving the strength and toughness of the material,
zirconia is an active sintering additive, which affects both
the sintering behavior and mullite formation processes in
mullite precursor powders.13,14 The mechanisms behind
these effects on the mullite reaction and on sintering are,
however, still obscure.
The current paper complements work presented in a
related publication.15 That paper focuses on the influence
of sintering conditions and seeding on densification
kinetics, before and after the mullite crystallization. Den-
sification curves also provided information on crystalli-
zation behavior itself. Thus the mullitization process was
found to be promoted by seed additions up to a critical
concentration, above which the kinetics become inde-
pendent of seed concentration. Still further increases
of seed concentration affect the grain morphology,
which tends to change from equiaxed to acicular.
Some data on microstructure and properties of mullite–
zirconia composites derived from SCA powders are also
presented.15
Taking into account the complex effects caused by
phase composition and processing parameters, more un-
derstanding is needed to optimize the structure and
properties of mullite-based materials. This paper pre-
sents the results of a more detailed study of the phase
evolution and reaction sequences during the sintering of
mullite/zirconia composites. The starting material con-
sisted of microcomposite powders of silica-coated alu-
mina, to which were added zirconia inclusions and
nanoscale seeds.
quently stirring for 24 h in a rotating plastic container.
The coated particles were collected by filtration, washed
with de-ionized water, and dried at 90–120 °C for 72 h.
Mullite or mullite/zirconia composites containing
15 vol% zirconia have been studied. Both unstabilized
and partially stabilized zirconia, grade TZ-0Y and grade
TZ-3Y (3 mol% Y2O3) (Tosoh Ceramic Division, Bound
Brook, NJ) were used. The SCA and zirconia powders
were dispersed separately by ball milling in water with
alumina media. The steric dispersant, ammonium poly-
methacrylate (Darvan C, R.T. Vanderbilt Co., Inc., Nor-
walk, CT) was added for stabilization of the zirconia
suspension. The zeta potential of the silica coating (30
mV in neutral media) was found to be sufficient for
electrostatic stabilization of an SCA particle suspension
in neutral water. After preliminary milling and mixing
for 24 h the individual powder suspensions were mixed
in a ball mill for an additional 24 h. The suspension of
seed particles (when used) was added at this stage.
An amorphous sol-gel-derived mullite precursor pow-
der (DeGussa Inc., Ridgefield Park, NJ) with a specific
surface area of almost 100 m2/g and a D50 particle size of
80 nm was chosen to study the seeding process. The
characteristics and preparation of these seeds is discussed
elsewhere.15
The densification kinetics have been studied using
high temperature dilatometry (Theta Industries, Inc., Port
Washington, NY) up to 1550 °C. The heating rate above
1000 °C was 2 °C/min with a hold at 1550 °C for 1 h.
The samples for microstructure observation were heated
to the sintering temperature at 5 °C/min and sintered in
air, while embedded in an alumina powder bed, at
1550 °C for 4 h.
Sintered samples were polished using standard cer-
amographic techniques and thermally etched at a tem-
perature 100 °C below the sintering temperature. The
etched samples and fracture surfaces were investigated
by scanning electron microscopy (SEM). The reactions
occurring during the sintering process were studied using
a combination of differential thermal analysis (DTA) and
x-ray diffraction (XRD) using Cu K␣ radiation. DTA was
performed (Netzsch STA-409) at heating rates of 5 and
10 °C/min. Calculation of crystal lattice parameters was
conducted by powder XRD. Twenty reflections were
measured using an internal standard of Si powder.
II. EXPERIMENTAL PROCEDURE
The preparation of microcomposite SCA particles was
similar to that described elsewhere.15 The process in-
volved precipitation of a (hydrous) silica coating [Tetra-
ethyl Orthosilicate (TEOS), Fisher Scientific, Unionville,
ON, Canada] on ␣-alumina particles (A16-SG, Alcoa
Industrial Chemicals, Pittsburgh, PA). The alumina pow-
der had a Brunauer–Emmett–Teller analysis (BET) sur-
face area of 9.1 m2/g and a D50 particle size of 0.45 ␮m.
The alumina/TEOS ratio was calculated to achieve a stoi-
chiometric mullite ratio of 72 wt% alumina and 28 wt%
silica. The silica was precipitated on the suspended alu-
mina particles by adding ammoniated water and subse-
III. RESULTS AND DISCUSSION
A. Thermal and x-ray analysis
DTA plots for various powder mixtures involving
SCA powders alone and with partially stabilized zirconia
and with different amount of seeds are presented in Fig. 1.
The heating rate was 5 °C/min. The data for the pure
SCA powder [Fig. 1(a)] exhibit a broad peak of mulliti-
zation with a maximum at 1501 °C. This temperature
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
1359
IP address: 128.218.248.200
http://journals.cambridge.org

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
FIG. 1. DTA plots for (a) silica-coated alumina powder (SCA) and (b) SCA with 15 vol% PSZ, and SCA/PSZ composition with (c) 0.5%, and
(d) 2% seeds. Heating rate: 5 °C/min.
agrees well with that reported by Wang and Sacks for
similar silica-coated particles,16 despite the different av-
erage SCA particle size (0.46 versus 0.2 m2/g). The ad-
dition of partially stabilized zirconia particles to the
mixture reduces the mullitization temperature by 65 °C
(to 1436 °C). The mullitization peak also becomes nar-
row and sharp [Fig. 1(b)]. The addition of seeds to an
SCA/zirconia mixture does not further change the tem-
perature of mullite formation significantly. Without zir-
conia, however, a significant effect of seeds has been
observed.10
An important additional effect of zirconia inclusions is
shown on the DTA plot [Fig. 1(b)]. This involves the
appearance of a new sharp, strong exothermic peak at
about 1383 °C that precedes the formation of mullite. As
seeds are added, these two peaks broaden and eventually
merge. This is associated with a decrease of relative in-
tensity of the exothermic peak and an increase of that of
the endothermic peak. The effect is more pronounced
with increasing seed concentration.
XRD shows that the formation of transient zirconia
(ZrSiO4) is responsible for the formation of this exother-
mic peak (Fig. 2). The samples for XRD study were
prepared by heating the powder compacts to different
temperatures followed by a hold for 12 min. The short
hold was necessary in order to reveal the transient
phases, which tend to disappear over time.
Formation of zircon from the mixture of ZrO2 and
SCA is thermodynamically possible in the observed
temperature region with a heat of formation at 1427 °C
(1700 K) of about −30 kJ/mol. (calculated using data
from the JANAF Thermochemical Tables). Nevertheless,
in the presence of alumina the formation of mullite is
thermodynamically preferable (i.e., the reaction of zircon
with alumina to form mullite and zirconia leads to a
further decrease of the Gibbs free energy). The existence
of zircon in this temperature interval was not reported in
some previous studies dealing with similar (but finer)
starting materials,8 while in another study limited zircon
formation was observed for one mixture.17 There are two
possible explanations for this. One is that mullite forms
at a lower temperature, removing the driving force for
zircon formation. More likely however, zircon was not
found because it is a transient phase and those materials
were only observed after sintering for 2 or 8 h. The
tendency for removal of the zircon reaction by doping
with mullite seeds (demonstrated by the DTA data) is in
agreement with the results of XRD analysis.
For the SCA–PSZ mixture without seeds the formation
and growth of mullite proceeds in the presence of tran-
sient zircon. Significant x-ray reflections of zircon are
characteristic for this material in all temperature intervals
of the mullite reaction [Fig. 2(a)]. The reactions to form
zircon and mullite are not as well separated in tempera-
ture by the x-ray data as they are in the DTA plots.
Instead the onset of mullite and zircon formation seem to
be linked. The maximum zircon formation occurs be-
tween 1450 and 1475 °C before the decomposition of
zircon begins to dominate. Through the addition of a
sufficient number of mullite seeds the formation of tran-
sient zircon is eliminated.
The seeding effect is demonstrated in Figs. 2(b) and
2(c). For the materials containing 0.5% seeds the inten-
sity of zircon reflections is reduced, which indicates a
1360
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
http://journals.cambridge.org
IP address: 128.218.248.200

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
decrease in its amount, which is balanced by synthesis
and decomposition [Fig. 2(b)]. With 2% seeds only very
small quantities of zircon are detected at 1450 °C
[Fig. 2(c)]. The total temperature interval of the mulliti-
zation reaction for the material with 2% seeds is much
narrower (only the final mullite and zirconia phases are
detected here at 1475 °C). Without seeds the reaction is
still incomplete at 1525 °C (i.e., the reflections of initial
alumina and transient zircon are still detectable).
The influence of heating rate on the reactions was also
evaluated via DTA (Fig. 3). As the heating increases to
10 °C/min, all the peak temperatures shift higher as ex-
pected. Nevertheless, the changes to individual peaks are
different. For reactions that obey Avrami kinetics (crys-
tallization through nucleation and growth) this shift can
be used to calculate the activation energy of reaction. On
comparing Figs. 1 and 3, we see that, with increasing
heating rate, the mullitization peak for pure SCA powder
is better defined and the peak temperature increases to
1512 °C. This peak exhibits the lowest temperature shift
and, consequently, the highest activation energy. The
mullitization temperature decreases when seeds are in-
troduced into the SCA powder.
Such a decrease has also been observed by Hong
et al.18 for diphasic mullite gels seeded with mullite par-
ticles. They reported the largest decrease when the seed
concentration was increased to 2 wt%. A further increase
of concentration showed little effect. The existence of a
critical concentration close to 1 wt% of seeds was also
detected by dilatometry.15 It was suggested that this con-
centration marks the change in the mullite formation
mechanism from nucleation-controlled to interface reac-
tion-controlled.11,15
The comparison of Figs. 1 and 3 shows that the posi-
tion of the DTA peaks for zircon is the most sensitive to
the heating rate. The kinetics of zircon crystallization has
been shown previously to be characteristic of a nuclea-
tion and growth mechanism, in which the controlling step
is nucleation.19 We calculated the apparent activation
energy on the basis of the effect of heating rate on the
DTA peak temperatures, using the method developed by
Kissinger.20 The activation energy of zircon formation
estimated by this method is 325 kJ/mol. It is significantly
less than the activation energy of mullitization for dipha-
sic and SCA precursors, which is reported to be about
1050 kJ/mol.9
Comparison of the activation energy of formation
for these two compounds provides the possible reason for
the formation of transient zircon during sintering
SCA–ZrO2 composites. Without zirconia, formation of
mullite from diphasic precursor or SCA powders pro-
FIG. 2. XRD spectra for SCA/PSZ mixtures that demonstrate phase
development during reaction sintering. The compacts were heated to
the indicated temperatures and held for 12 min (a) without seeds, (b)
0.5% seeds, and (c) 2% seeds. Designation: A, alumina; C, crysto-
balite; Z, zircon; T, tetragonal zirconia; and M, mullite.
ceeds by nucleation and growth inside the siliceous
phase. The nucleation step controls this process with a
high activation energy (1050 kJ/mol). The formation of
zircon is also controlled by crystal nucleation19,21 but the
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
1361
IP address: 128.218.248.200
http://journals.cambridge.org

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
FIG. 3. DTA plot for (a) silica-coated alumina powder (SCA) and (b) composition of SCA with 15 vol% for PSZ, and SCA/PSZ composition with
(c) 0.5%, and (d) 1% seeds. Heating rate: 10 °C/min.
materials with 0, 0.5, and 1% seeds performed at a
activation energy is much lower (350 kJ/mol). Due to a
heating rate of 5 °C/min are presented in Fig. 4. The
shorter incubation period this reaction precedes the for-
introduction of unstabilized zirconia also leads to the
mation of mullite. In other words the two-step reaction
formation of transient zircon, although its peak tem-
is a means to decrease the high-energy barrier for mullite
perature is higher (now it is stable at 1450 °C). Also,
direct crystal nucleation by replacing it with two lower
without yttria, zirconia does not reduce the mullite for-
energy processes. The reduction of mullite crystallization
mation temperature nearly as much (only by 8 °C versus
temperature in the presence of zirconia is evidence of this
65 °C for PSZ). The reduction of the mullite formation
decrease.
temperature with the introduction of seeds is not sig-
With increasing heating rate the peaks of zircon and
nificant for this case. Thus the new positions of the
mullite get closer and begin to overlap. Since mullite is
DTA peaks for zircon and mullite are rather close for
the thermodynamically stable compound, mullite forma-
seeded samples (1445 and 1454 °C), although it always
tion takes precedence over zircon formation. This sub-
remains lower for the SCA–PSZ combination. This result
stitution is more pronounced at a heating rate of 10 °C/
is in agreement with our assertion that mullite formation
min than for 5 °C/min. The addition of 1% seeds in the
is the dominant process and that once mullitization is
first case practically eliminates the zircon peak. In the
moved to a sufficiently low temperature to overlap zircon
latter case it is detected with even 2% seeds. The experi-
formation, zircon formation does not occur [Figs. 4(b)
ment shows that an increase of heating rate as well as
and 4(c)].
seed concentration promotes the mullitization reaction
The crystal lattice characteristics for some mullite ma-
with less (if any) transient zircon.
terials after sintering for 4 h at 1550 °C are presented in
The yttria dopant in zirconia also influences the
Fig. 5. The data are compared with Cameron’s calibra-
reaction temperatures. For examples, Claussen and
Jahn22 included a hold at 1450 °C when sintering alu-
tion line showing the relation between the mullite lattice
parameters.23 The increase of a-cell edge is usually as-
mina–zircon powders to make mullite–zirconia com-
posites. After this hold specimens still consisted of
sociated with an increase of Al2O3 content in mullite,
while the formation of a metastable ZrO2 solid solution
only alumina and zircon, which converted to mullite
on mullite can be detected by the increase of the cell
only after heating above 1500 °C. In our case zircon is
volume above the trend line.24 No significant solubility
not stable at 1450 °C, presumably due to the influence
of ZrO2 in the bulk was detected for these materials. At
of yttria.
the same time there is a difference in a edge for materials
To assess the possible influence of Y2O3 on the reac-
sintered with and without seeds. The presence of seeds
tions just discussed we have repeated the experiments
slightly increases the a parameter of mullite. The lattice
using unstabilized zirconia. The DTA data for these
1362
J. Mater. Res., Vol. 15, No. 6, Jun 2000
http://journals.cambridge.org
Downloaded: 15 Mar 2015
IP address: 128.218.248.200

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
B. Dilatometry studies
parameters of seeds after crystallization at 1400 °C are
also shown on the plot. Seeds have also been crystallized
into alumina-enriched mullite, which could be the matrix
for growing crystals of the same stoichiometry.
The yttria-doped composites demonstrate a lower a
parameter. This shows the active role played by yttria in
the crystallization process. Nevertheless, its solubility in
the mullite matrix as a possible reason for this effect
should be further studied.
The phase evolution in the SCA–zirconia system has
also been studied by using high-temperature dilatometry.
The densification of SCA with and without PSZ is illus-
trated in Fig. 6. The various stages of phase evolution can
be detected on the dilatometric curves due to the volume
change associated with each reaction.
FIG. 5. Plot of the cell parameter a versus cell volume of mullite in
various composites with unstabilized and partially stabilized zirconia.
Solid symbols represent unseeded composites, while open symbols are
for materials with 2% DeGussa seeds. The parameter set derived from
crystallization of the DeGussa seeds themselves is also shown (marked
×). The dashed line represents the data of Cameron.24
FIG. 4. DTA plot for composition of SCA (a) with 15 vol% unstabi-
lized zirconia, and SCA–ZrO2 composition with (b) 0.5%, and (c) 1%
seeds. Heating rate: 5 °C/min.
FIG. 6. Densification curves for SCA powder (0.5% seeds) and the
same powder with 15 vol% PSZ.
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
1363
IP address: 128.218.248.200
http://journals.cambridge.org

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
Sintering by viscous flow of amorphous silica pro-
ceeds until it is blocked by the formation of a crystobalite
network.15 This is followed by mullite formation, which
is marked by a volume expansion on the densification
curves, since the density of mullite is slightly lower than
that of the starting alumina/silica mixture. The depth of
the depression on the density versus temperature plot
reflects the intensity of the reaction. The reaction tem-
peratures observed in these measurements are consistent
with those inferred from the x-ray measurements (com-
pare Figs. 2 and 6).
Moya and Osendi13 also attempted to use dilatometry
to study densification processes in ultrafine premullite
powders. However, the characteristic temperatures could
not be detected in this case due to the much lower tem-
perature of mullite crystallization, which resulted in
overlapping volume effects of sintering and reaction.
When PSZ is added to the SCA powder (Fig. 6), the
densification rate is increased both before and especially
after formation of mullite. Without zirconia the densifi-
cation of mullite during the high-temperature stage is
very slow. Zirconia also decreases the temperature for
the onset of mullitization. The latter result is in agree-
ment with our DTA data [Figs. 1(a) and 1(b)] and
with that derived by Shyu and Chen using the XRD
FIG. 7. Densification curves for SCA powder compact with 15 vol%
nonstabilized ZrO2, SCA with 15 vol% nonstabilized ZrO2 plus 0.5%
technique.16
The influence of yttria and seed additions on densifi-
cation is seen in Fig. 7. The addition of 0.5% seeds to a
SCA/(nonstabilized)zirconia composite decreases the
temperature of mullite formation while increasing the
densification rate of the mullite matrix after this forma-
tion. The use of stablized zirconia results in a further
lowering of the mullitization temperature (in agreement
with DTA data) and a higher densification rate after mul-
lite formation. Moya and Osendi14 suggested that the
presence of zirconia in a mullite matrix increases densi-
fication during sintering, due to the formation of a solid
solution at the grain boundary, thus enhancing the grain
boundary diffusion rate. The influence of seeds on den-
sification rate can be the result of a decreasing grain size.
The possible explanation of an yttria effect may be its
solubility in the bulk or at grain boundaries. However,
this would need to be confirmed by transmission electron
microscopy observations.
seeds, and SCA with partially stabilized zirconia plus 0.5% seeds.
gested that the presence of amorphous siliceous films at
the boundaries could be responsible for this. Such films
would decrease the interfacial energy and diminish the
driving force for material transport. Hence, thermal
grooving would be more limited.
For the seeded samples [Figs. 8(b) and 8(c)], the ther-
mal etching is distinct, which may reflect a reduction of
siliceous films on the boundaries. The microstructure
formed in the presence of 0.5% seeds exhibits mainly
equiaxed grains. Some PSZ inclusions are seen inside
mullite grains. They are much smaller than those on grain
boundaries, because of the low diffusion rate inside the
mullite crystals. The zirconia particles present on grain
boundaries undergo significant coalescence.
With increasing seed concentration the fraction of fine
grains increases, evidently due to an increased number of
nuclei for mullite crystallites. Also the formation of some
elongated mullite grains can be detected. Very few in-
tragranular zirconia inclusions can be seen in this struc-
ture [Fig. 8(c)].
The microstructure of the samples that were sintered
without seeds and where significant amounts of transient
zircon were detected have similar features to those ob-
served in materials derived from zircon–alumina mix-
tures. In accordance with a model proposed for such
C. Microstructure development
SEM images of the microstructure for materials with
different seed concentration are presented in Fig. 8. The
nonseeded sample is not responsive to thermal etching
[Fig. 8(a)] although different temperatures and durations
were tried. An analogous situation was observed by
Sacks et al. for unseeded microcomposite powders,10 in
which some grain boundaries appeared to be poorly
etched and others were only faintly visible, giving the
appearance of a subgrain structure. The authors sug-
1364
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
http://journals.cambridge.org
IP address: 128.218.248.200

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
FIG. 8. SEM micrographs of samples sintered at 1580 °C: (a,d) without seeds; (b) 0.5% seeds; (c,e) 2% seeds; (a–c) polished samples; (d,e)
fracture surfaces.
mixtures25,26 during heating above 1450 °C alumina re-
acts rapidly with zircon to form a “noncrystalline mul-
lite” matrix and ZrO2 particles. The ZrO2 particles have
a rounded morphology, since they are growing in a non-
crystalline phase without grain boundaries. These grains
could be observed in the Fig. 8(d). The inclusion of seeds
changes the fracture surface from more glassylike to a
mode characteristic of polycrystalline fracture [Figs. 8(d)
and 8(e)]. Once the mullite finally crystallizes, many
inclusions appear inside the mullite grains. As the recrys-
tallization of zirconia occurs primarily through grain
boundary diffusion, further growth of intragranular zir-
conia is limited. The inhomogeneity of the Al2O3 distri-
bution in material formed through the zircon reaction can
cause the formation of continuous thick glassy films in
the grain boundaries, as seen in Figs. 8(a) and 8(d).
Samples with this structure can exhibit good bending
strength and fracture toughness at room temperature al-
though the high-temperature mechanical properties will
be poor.27
on phase evolution and sintering behavior in silica-
coated alumina powders. The following conclusions can
be drawn:
(1) During the sintering of SCA/zirconia mixtures two
possible modes of mullite formation are found: (i)
through the formation of transient zircon, which de-
creases the potential barrier for mullite crystal nuclea-
tion; (ii) by direct synthesis from SCA powder, through
growth on mullite seeds.
(2) Seeding of the crystal nuclei promotes mullite for-
mation by direct synthesis.
(3) The yttria in partially stabilized zirconia is an ac-
tive sintering additive which decreases the potential bar-
rier of crystal nucleation for both zircon and mullite
formation. In the presence of seeds it promotes mulliti-
zation direct synthesis rather than by transient zircon
formation.
(4) There are significant differences in mullite micro-
structure formed through the above mentioned reactions.
The variation of seed concentration offers a method of
controlling the microstructure parameters of materials
over the wide variety of conditions.
The inclusion of seeds appears to reduce the formation
of the thick glassy films at the grain boundaries. The
formation of intragranular zirconia inclusions is not char-
acteristic of direct growth of mullite on crystal nuclei, as
dominates at the higher seed concentration [Fig. 8(c)].
ACKNOWLEDGMENTS
The authors are grateful for financial support of this
work from Caterpillar Inc., Peoria, IL. Several useful
discussions with K. Kelley and M. Long of Caterpillar
are also acknowledged.
IV. CONCLUSIONS
In this study we have observed the effect of zirconia
(with and without yttria stabilization) and mullite seeds
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
1365
IP address: 128.218.248.200
http://journals.cambridge.org

V. Yaroshenko et al.: Phase evolution during sintering of mullite/zirconia composites using silica-coated alumina powders
13. J.S. Moya and M.I. Osendi, J. Mater. Sci. 2909 (1984).
14. J.S. Moya and M.I. Osendi, J. Mater. Sci. Lett. 2, 559 (1983).
15. V. Yaroshenko, D.S. Wilkinson, J. Am. Ceram. Soc. (2000, in
press).
16. K. Wang, M.D. Sacks, J. Am. Ceram. Soc. 79, 12 (1996).
17. J.J. Shyu and Y.C. Chen, J. Mater. Res. 10, 63 (1995).
18. S-H. Hong, W. Cermignani, and G.L. Messing, J. Eur. Ceram.
Soc. 16, 133 (1996).
REFERENCES
1. I.A. Aksay, D.M. Dabbs, and M. Sarikaya, Am. Ceram. Soc. 74,
2343 (1991).
2. W.E. Lee and W.M. Rainforth, Ceramic Microstructures. Prop-
erty Control by Processing (Chapman & Hall, London, United
Kingdom, 1994).
3. Mullite and Mullite Based Composites, Ceramic Transaction Vol.
6, edited by S. Somiya, R.F. Davis, and J.A. Pask (American
Ceramic Society, Westerville, OH, 1990).
4. Mullite Processing Structure and Properties (topical issue),
J. Am. Ceram. Soc. 74, 2341 (1991).
5. Mullite ’94 (special issue), J. Eur. Ceram. Soc. 16, 99 (1996).
6. M.D. Sacks and H-W. Lee, in Mullite and Mullite Based Compos-
ites, Ceramic Transaction Vol. 6, edited by S. Somiya, R.F. Davis,
and J.A. Pask (American Ceramic Society, Westerville, OH,
1990), pp. 167–207.
7. M.D. Sacks, G.W. Scheifele, N. Bozkurt, and R. Raghunathan, in
Ceramic Powder Science IV, Ceramic Transaction Vol. 22, edited
by G.L. Messing, S.I. Hirano, and H.H. Hausner (American Ce-
ramic Society, Westerville, OH, 1991), pp. 437–455.
8. M.D. Sacks, N. Bozkurt, and G.W. Scheifele, J. Am. Ceram. Soc.
74, 2428 (1991).
19. P. Tartaj, C.J. Serna, J.S. Moya, J. Requena, M. Osana, S. De Asa,
and F. Guitian, J. Mater. Sci. 31, 6089 (1996).
20. H.E. Kissinger, J. Res. Natl. Bur. Stand. (U.S.) 50, 217 (1956).
21. P. Boch, T. Chartier, and J.P. Giry, in Mullite and Mullite Based
Composites, Ceramic Transaction Vol. 6, edited by S. Somiya,
R.F. Davis, and J.A. Pask (American Ceramic Society, Wester-
ville, OH, 1990), pp. 473–494.
22. N. Claussen and J. Jahn, J. Am. Ceram. Soc. 63, 228 (1980).
23. W.E. Cameron, Am. Ceram. Soc. Bull. 56, 1003 (1977).
24. M. Yoshimura, Y. Hanaue, and S. Somiya, in Mullite and Mullite
Based Composites, Ceramic Transaction Vol. 6, edited by S.
Somiya, R.F. Davis, and J.A. Pask (American Ceramic Society,
Westerville, OH, 1990), pp. 449–456.
25. S. Lathabai, D.G. Hay, F. Wagner, and N. Claussen, J. Am. Ce-
ram. Soc. 79, 248 (1996).
26. J.S. Wallace, G. Petzow, and N. Claussen, in Science and Tech-
nology of Zirconia, Advances in Ceramics Vol. 12, edited by N.
Claussen, M. Ruhle, and A.H. Heuer (American Ceramic Society,
Westerville, OH, 1984), pp. 436–442.
9. M.D. Sacks, K. Wang, G.W. Scheifele, and N. Bozkurt, J. Am.
Ceram. Soc. 79, 571 (1996).
10. M.D. Sacks, Y-J. Lin, G.W. Scheifele, K. Wang, and N. Bozkurt,
J. Am. Ceram. Soc. 78, 2897 (1995).
11. J.C. Huling and G.L. Messing, J. Am. Ceram. Soc. 74, 10, 2374
(1991).
27. T. Koyama, S. Hayashi, A. Yasumori, K. Okada, M. Schmucker,
and H. Schneider, J. Eur. Ceram. Soc. 16, 231 (1996).
12. J.C. Huling and G.L. Messing, in Better Ceramics Through Chem-
istry IV, edited by B.J.J. Zelinski, C.J. Brinker, D.E. Clark, and
D.R. Ulrich (Mater. Res. Soc. Symp. Proc. 180, Pittsburgh, PA,
1990), pp. 515–526.
1366
J. Mater. Res., Vol. 15, No. 6, Jun 2000
Downloaded: 15 Mar 2015
http://journals.cambridge.org
IP address: 128.218.248.200