Zirconia-mullite composites consolidated by spark plasma reaction sintering from zircon and alumina

Article abstract of DOI:10.1111/j.1551-2916.2005.00234.x

Mullite-ZrO₂ composites have been fabricated by attrition milling a powder mixture of zircon, alumina, and aluminum metal with MgO or TiO₂ as sintering additives, heating at 1100°C to oxidize the aluminum metal, and consolidation by spark plasma sintering (SPS). The influence of the SPS temperature on the formation of mullite, and the density and the mechanical properties of the resulting composites have been studied. For the mullite-zirconia composites without sintering additives, the mullite formation was accomplished at 1540°C. In contrast, for the composites having MgO and TiO₂, the formation temperature dropped to 1460°C. The composites without sintering additives were almost fully dense (99.9% relative density) and retained a larger amount of tetragonal zirconia. Those materials attained the best mechanical properties (E = 214 GPa and K_IC = 6 MPa · m^1/2). To highlight the advantages of using the SPS technique, the obtained results have been compared with the characteristics of a mullite-zirconia composite prepared by the conventional reaction-sintering process.

Full text of DOI:10.1111/j.1551-2916.2005.00234.x

J. Am. Ceram. Soc., 88 [5] 1150–1157 (2005)
DOI: 10.1111/j.1551-2916.2005.00234.x
ournal
J
Zirconia–Mullite Composites Consolidated by Spark Plasma Reaction
Sintering from Zircon and Alumina
Enrique Rocha-Rangel
Department of Materials, Metropolitan Autonomous University-Atzcapotzalco, Mexico, D.F., 02200 Mexico
Sebastian Dıaz-de-la-Torre
´ ´
CIITEC, National Polytechnic Institute, Mexico D.F., 07300, Mexico
Minoru Umemoto
Toyohashi University of Technology, Toyohashi, Japan
Hiroki Miyamoto
Technology Research Institute of Osaka Prefecture, Izumi, Osaka 594-1154, Japan
Heberto Balmori-Ramı
´
rezw
Department of Metallurgical and Materials Engineering, National Polytechnic Institute, Mexico, D.F., 07300 Mexico
Mullite–ZrO2 composites have been fabricated by attrition mill-
ing a powder mixture of zircon, alumina, and aluminum metal
with MgO or TiO2 as sintering additives, heating at 11001C to
oxidize the aluminum metal, and consolidation by spark plasma
sintering (SPS). The influence of the SPS temperature on the
formation of mullite, and the density and the mechanical prop-
erties of the resulting composites have been studied. For the
mullite–zirconia composites without sintering additives, the mull-
ite formation was accomplished at 15401C. In contrast, for the
composites having MgO and TiO2, the formation temperature
dropped to 14601C. The composites without sintering additives
were almost fully dense (99.9% relative density) and retained a
larger amount of tetragonal zirconia. Those materials attained
the best mechanical properties (E 5 214 GPa and KIC 5 6
Al2O3, and zircon, ZrSiO4;3–5 (ii) sintering of ‘‘premullite’’ pow-
der mixed with ZrO2 particles;6,7 (iii) sol–gel method;8 (iv) re-
action sintering of Al2O3 and SiO2 with additions of ZrO2;5,9
and (v) through a combination of processes such as reac-
tion bonding of aluminum oxide and reaction sintering with
ZrSiO4.10
Reaction sintering of alumina and zircon is an economic
and relatively easy processing route to obtain homogeneous
mullite–zirconia composite ceramics. This reaction is expressed
as follows:
3Al2O3 þ 2ZrSiO4 ! 3Al2O3 ꢀ 2SiO2 þ 2ZrO2
(1)
It proceeds at temperatures 415001C. The ZrO2 precipitates
appear as a dispersed phase in the mullite matrix. Nevertheless,
the processing route must be carefully designed to obtain a
dense, homogeneous microstructure and to limit the severity of
critical defects. Additionally, the final microstructure must be
able to accommodate transformation and microcracking. It has
been proposed that it is necessary to obtain a high level of den-
sification before the onset of the reaction (1) in order to obtain
dense ceramics, because the reaction occurs with a volume ex-
pansion that competes with the densification shrinkage.4,11 The
utilization of reactive, submicrometer size powders of alumina
and zircon is necessary to stimulate both the chemical reaction
and densification during sintering.12 Additions of MgO, CaO,
and TiO2 promote reaction (1) and the densification through the
formation of a transient liquid phase.13–15
MPa m1/2). To highlight the advantages of using the SPS tech-
.
nique, the obtained results have been compared with the char-
acteristics of a mullite–zirconia composite prepared by the
conventional reaction-sintering process.
I. Introduction
ULLITE ceramics feature excellent properties like low ther-
mal conductivity, low coefficient of thermal expansion,
M
high elevated temperature strength, high creep resistance, and
phase stability up to the melting point.1 Such features have
stimulated intensive research in the field of refractories and ce-
ramics for high-temperature structural applications. However,
monolithic mullite bodies exhibit both low strength and fracture
toughness at room temperature. Additionally, they are also dif-
ficult to sinter in the absence of a glassy phase.
Spark plasma sintering (SPS), a variant of the plasma-acti-
vated sintering (PAS) process16, is effective at creating homoge-
neous and rapid densification of ceramic materials, provided
that they have high dielectric constants.17–19 SPS is also known
as field activated sintering.20 The major advantages of plasma
sintering are the fast heating rates that can be obtained and the
short sintering times required to attain full density, usually less
than 10 min, an interval in which controlled high temperatures
can be developed while applying pressure.19 These are favorable
factors for restraining grain growth and retaining metastable
phases.
Reinforcing mullite with ZrO2 inclusions allows an increase in
its strength and toughness at low and intermediate tempera-
tures.2 Mullite–ZrO2 composites can be prepared by various
routes, such as: (i) reaction sintering of powders of alumina,
Z. A. Munir—contributing editor
The SPS technique consists of two steps.20,21 First, the pow-
der is loaded in a graphite die and pressed between two graphite
punches. The firing starts by applying a pulse current of short
Manuscript No. 10742. Received December 17, 2003; approved October 21, 2004.
wAuthor to whom correspondence should be addressed. e-mail: hbalmori@ipn.mx
1150

May 2005
Zirconia–Mullite Composites Consolidated by Spark Plasma Reaction
1151
duration that passes through the graphite dies and punches and
through the ceramic powder. In the second step, the pulse cur-
rent is switched off and a direct current (DC) is applied to induce
Joule or resistance heating of the graphite dies, the graphite
punches, and the powder. The DC is maintained until a preset
temperature is reached or when the shrinkage stops. The axial
load applied directly on the powder is maintained throughout all
the SPS-event duration to assist densification.
The underlying mechanisms during the SPS process are still a
subject of research.20 The pulse current applied during the first
step activates the particles and removes surface impurities. It
promotes mass transport to the particle necks through the for-
mation of vapor or of intermittent plasma.21 The diffusional
processes that are necessary to annihilate the porosity start when
the application of the DC starts. It has been proposed that the
DC intensifies the diffusional processes during sintering. The
applied pressure promotes plastic flow.
Table I. Characteristics of the Raw Materials
Zircon
(wt%)
Al2O3
(ppm)
Aluminum
(wt%)
ZrO2
SiO2
Al2O3
TiO2
Fe2O3
66.2
32.5
0.5
0.2
0.07 Mg
Si
Na
K
Fe
Ca
3
1
8
3
2
Si
0.2
0.4
0.3
0.016
Fe
Cu
Mg
10
o4 ppb
U
Th
o5 ppb
Particle
size
Particle
1 (mm) size
Taimei Chemical
(Japan)
Particle
0.20 (mm) size
o5 (mm)
Kreutz
(Germany)
Analytical
(Mexico)
The SPS production of mullite–ZrO2 composites by reaction
(1) has been attempted by Khor et al.22 They employed a plasma
spheroidized zircon/alumina powder that was composed of
alumina and partially dissociated zircon. The mixture was
completely transformed to mullite and zirconia at an SPS
temperature of 13001C for 10 min. However, rapid coarsening
of the microstructure took place and the zirconia transformed to
its monoclinic variant. Their composite ended up with numer-
ous microcracks. It had an indentation fracture toughness of
area (BET) of the powders was evaluated with an automatic
volumetric gas sorption analyzer (AUTOSORB-1, Quanta
Chrome, Boynton Beach, FL).
The SPS equipment used to perform the sintering experiments
(Dr. Sinter-SPS 1020, Sumitomo Coal Mining Co., Tokyo,
Japan) features a maximum sintering pressure of 10 tons and
a maximum DC pulse generator output of 2000 A. A detailed
description of the SPS setting can be found elsewhere.19
Small batches (3 g) of the three powders MZ, MZM, and
MZT that had been attrition milled, heated to 11001C, and
milled in the planetary mill to break the agglomerates were
packed in the cylindrical graphite die (20 mm inner diame-
ter ꢁ 40 mm long) of the SPS machine. The powder was sand-
wiched between two graphite punches. A uniaxial pressure of
40 MPa was applied directly over the powder throughout the
experiment. The SPS equipment was operated in vacuum so that
the maximum displayed voltage was 6.5 V, the current 2000 A,
and the frequency was 60 Hz. The current was applied in On/Off
pulses that were set so that the DC current passed through the
sample for 12 ms, followed by a step of 2 ms during which the
current was turned off (12/2 On/Off pulses). It has been sug-
gested that these conditions allow the activation and cleaning of
the powder surface, enhancing the material transfer throughout
necking of the ceramic powder particles.18,19 These condi-
tions were maintained until a preset sintering temperature was
reached, which was varied between 13501 and 16401C for the
MZ powders, and between 13501 and 14601C for the MZM and
MZT powders. The temperature was monitored with an optical
pyrometer focused at the surface of the graphite dies. Under
these conditions, the heating rate from room temperature to the
sintering temperature was approximately 5001C/min. Once the
set temperature was reached, the applied current was turned off,
the applied load was released, and the samples were allowed to
cool down resting inside the SPS machine.
11.2 MPa ꢀ m1/2
.
The present study focuses on the production of ZrO2-tough-
ened mullite ceramic composites by spark plasma reaction sin-
tering according to reaction (1). In order to promote the
formation of a reactive precursor powder, the zircon was mixed
with a mixture of a-Al2O3 and aluminum metal which, after
heating at 11001C in air, converts in situ to a highly reactive and
well-dispersed alumina.10,23,24 A combination of techniques was
used: (a) wet attrition milling to reduce the particle size and ho-
mogenize the powder, (b) heating to 11001C to fulfill the oxi-
dation of aluminum, and (c) SPS. Samples with sintering
additives (MgO and TiO2) were also prepared. Both processing
and microstructural aspects of the SPS samples are reported. A
batch of the same powder was processed by conventional
reaction sintering in order to compare with the SPS technique.
II. Experimental Procedure
Batches (100 g) of zircon (B1 mm, Helmut Kreutz, GmbH,
Langenbauch, Germany), aluminum (B5 mm, Analytical de
Mexico, Mexico City, Mexico), and a-alumina (TM-10, Taimei
Chemicals, Tokyo, Japan), mixed in a ratio of 64/18.5/17.5
wt%, were prepared. The sum of the alumina formed in situ af-
ter the oxidation of the aluminum and the added a-Al2O3 cor-
responds to the stoichiometric amount of alumina required by
reaction (1). Powder mixtures with 1.8 wt% MgO or 7 wt%
TiO2 were also prepared. These particular concentrations of
MgO and TiO2 were chosen on the basis of the work by Le-
riche,15 who proposed that these were the appropriate levels to
attain the complete mullitization of the mixtures and a high
density. The three samples were identified as MZ for the zircon–
alumina powder, MZM for the powder with MgO, and MZT
for the powder with TiO2. The characteristics of the zircon, alu-
mina, and aluminum used are summarized in Table I.
Samples of the powder MZ were conventionally sintered to
compare with the results of the SPS process. Pellets (20 mm di-
ameter ꢁ 5 mm thickness) were prepared by uniaxial pressing at
100 MPa and cold isostatic pressing at 400 MPa. They were
heated up at a rate of 10 K/min to different temperatures (13501,
14001, 14501, 15001, 15501, and 15801C) for 60 min and allowed
to cool inside the furnace.
The powders were attrition milled for 12 h at 400 rpm in a
Union Process mill device (Union Process, Inc., Akron, OH) filled
with 3 kg of Y-TZP balls of 3 mm diameter. They were dispersed
in isopropanol to slow down the metal oxidation during milling
and to avoid sticking of the powder and the milling media. The
attrition-milled powders were recovered by filtration through a
325-mesh sieve and dried at B701C. The dried powders were
heated up to 11001C at a rate of 11C/min in air to ensure the
oxidation of the aluminum metal. The calcined powder was
milled in a planetary mill for 30 min to break the agglomerates
that were formed during the calcination. The specific surface
Both density and porosity of the consolidated pellets were
evaluated by the Archimedes principle. The microstructure of
the sintered samples was analyzed by X-ray diffraction (XRD,
Siemens D-5000, Bruker AVS, Karlsruhe, Germany) and scan-
ning electron microscopy (SEM, Jeol-6300, Jeol, Ltd., Tokyo,
Japan) with samples polished with standard ceramographic tech-
niques. The amount of tetragonal zirconia of the sintered samples
was calculated using the method proposed by Boch et al.12
Young’s modulus (E) was measured by an impulse excitation
method (Grind-O-Sonic). Niihara’s indentation technique for KIC
determination was used,25 applying a load of 500 N.

1152
Journal of the American Ceramic Society—Rocha-Rangel et al.
III. Results and Discussion
Vol. 88, No. 5
There are also several small peaks visible that belong to mullite
(M). The lack of peaks of silica in this XRD pattern indicates
that this phase is either amorphous at 14201C, or that it is com-
pletely consumed in the reaction with alumina to yield mullite.
The intensities of the peaks of mullite and zirconia increase
continuously as the SPS temperature increases. Simultaneously,
the intensities of the peaks of zircon and a-Al2O3 decrease.
These two precursor phases have totally disappeared at 15601C.
With regard to the zirconia particles, they occur as a mixture of
the tetragonal and monoclinic polymorphs at all temperatures
415001C. The tetragonal type predominates for the composites
sintered at o15601C, but the monoclinic ZrO2 is the major
phase at higher temperatures. Thermodynamically, the mono-
clinic zirconia is the low-temperature stable phase and the te-
tragonal phase is stable at higher temperatures. However, if the
particles are smaller than a critical size, the tetragonal ZrO2-
monoclinic ZrO2 transformation does not occur upon cooling
so that metastable tetragonal ZrO2 particles can be retained
at room temperature. The critical particle size is in the order of
0.6–1 mm for mullite–zirconia composites.26 It is reasonable to
assume that the zirconia particles that precipitated from the
zircon at the lower SPS temperatures are smaller than the
critical size, but that they coarsened beyond this limit at
the highest SPS temperatures and transformed to monoclinic
ZrO2 upon cooling.
The XRD patterns for the MZM powder sintered by SPS at
13801, 14001, 14201, and 14601C are shown in Fig. 2. The XRD
pattern of the sample SPS-treated at 13801C indicates that the
formation of mullite is well advanced at this temperature, be-
cause it displays high intensities for mullite and zirconia peaks.
The peaks of zircon are also present, but they are less intense.
This phase disappears at 14201C. The tetragonal zirconia pre-
dominates over the monoclinic at 13801C. The opposite occurs
at 14001 and 14201C. However, the tetragonal phase is again the
major zirconia phase at 14601C. It is possible that some mag-
nesia diffused into the zirconia particles at higher temperatures.
Magnesia is a well-known stabilizer of tetragonal zirconia
utilized for the production of partially stabilized zirconia
In this section, the processing of mullite–zirconia composites by
the SPS method will be described. It starts with a brief descrip-
tion of the oxidation of the aluminum metal. The evolution of
the microstructure during the SPS process will be addressed
next, with an emphasis on phase changes, the mullitization
reaction, and densification. A discussion of the mechanical
properties obtained will follow. At the end, the results of the
mullitization, the density, and the mechanical properties of the
composites processed by SPS will be compared with the results
obtained by conventional reaction sintering in order to highlight
the advantages of the SPS process.
(1) Oxidation of the Aluminum Metal
The oxidation of aluminum to yield alumina has been analyzed
previously.23,24 Approximately 42% of the aluminum in the
mixture oxidizes during the attrition milling. The remaining al-
uminum oxidizes during the heating in air to 11001C. The XRD
pattern of the powder oxidized at 11001C is shown in Fig. 1. It
shows the diffraction peaks of zircon and a-Al2O3, as well as a
short peak at 2y 5 67.21 that was indexed as g-Al2O3 (Fig. 1).
There is no evidence of aluminum metal in this XRD pattern.
The specific surface area of the mixture of zircon, alumina,
and aluminum metal before milling was 2.96 m2/g. It increased
to 21.57 m2/g after the attrition milling and to 52.57 m2/g after
firing at 11001C and breaking the agglomerates by planetary
milling. The significant increase in the specific surface area can
be explained by the reduction of the particle size during the at-
trition milling and by the formation of very fine and reactive
particles of Al2O3 after the oxidation of the aluminum. This
process ensures a homogeneous distribution of the alumina par-
ticles which are formed in situ during the heating of the powder
at 11001C.
(2) Phase Changes During the SPS Process
Figure 1 summarizes the XRD patterns of the MZ powder fired
at different SPS temperatures. The first evidence for the decom-
position of zircon (Z) appears in the XRD pattern of the sample
that was fired at 14201C. It shows a small peak at 2y 5 30.41 that
corresponds to tetragonal zirconia (t-ZrO2). The presence of this
phase is taken as an indication for the beginning of the decom-
position of zircon according to the following reaction:
ZrSiO4 ! SiO2 þ ZrO2
(2)
Fig. 2. X-ray diffraction patterns of MZM powder sintered by
spark plasma sintering at different temperatures. Z, zircon (ZrSiO4);
M, mullite; Zm, monoclinic ZrO2; Zt, tetragonal ZrO2; Sp, spinel
(MgOꢀ Al2O3).
Fig. 1. X-ray diffraction patterns of MZ powder processed by spark
plasma sintering at different temperatures. Z, ZrSiO4; a, a-Al2O3; g,
g-Al2O3; M, mullite; Zm, monoclinic ZrO2; Zt, tetragonal ZrO2.

May 2005
Zirconia–Mullite Composites Consolidated by Spark Plasma Reaction
1153
1.1
MZ
0.9
MZT
0.7
MZM
0.5
Conventional
Reaction-Sintering
0.3
0.1
− 0.1
1325 1375 1425 1475 1525 1575 1625 1675
Temperature (°C)
Fig. 4. The mullitization transformation of MZ, MZM, and MZT
specimens plotted as a function of spark plasma sintering (SPS) tem-
perature. A comparison between the MZ powder data sintered by con-
ventional reaction sintering and SPS is shown.
data for the MZ powder that was conventionally sintered for 2 h
at different temperatures.
For the MZ powders, Fig. 4 confirms that the formation of
mullite starts at 14201C and seems to be completed at 15601C.
Although the mullitization reaction rate evidently slows down at
14801oTo15501C, it speeds up again at T415501C. One pos-
sible explanation for this observation might be because of the
formation of a high-temperature liquid. This could be derived
from the SiO2 that was expelled from the zircon. Although the
XRD patterns did not show peaks of silica, it is possible that this
compound was precipitated as an amorphous, glassy phase. In
fact, working the MZ specimens at 15501C brings them very
close to the eutectic point of the binary SiO2–Al2O3 system.
Figure 4 summarizes the XRD observations. It indicates that
the mullitization of powders with additives such as MgO and
TiO2 starts at lower temperatures proceeding faster. Actually,
this reaction is nearly over at the temperature of 14601C, a tem-
perature at which the decomposition of zircon has just started in
the powder MZ not containing sintering additives. Both MgO
and TiO2 assist mullitization through the appearance of a tran-
sient liquid at low temperatures.13,14
The mullitization reaction for the MZ powders occurs in a
temperature interval that is higher than that measured by Khor
et al.22 They report that mullitization of their powder completes
at 13001C. There are two basic differences between their exper-
iments and those reported here that explain these discrepancies:
(i) they employed a plasma spheroidized zircon/alumina powder
in which the zircon was partially decomposed and contained
some mullite from the beginning of the SPS, and (ii) their ex-
periments lasted a total of 10 or 30 min at each SPS temperature,
in contrast to the non-holding (annealing) time set in this work.
Khor’s results and those reported here also differ in the amount
of tetragonal zirconia that is retained when the decomposition of
zircon and the formation of mullite is completed. The zirconia
inclusions of Khor’s composites are of the monoclinic phase,
while those prepared in this work consist of 65% tetragonal and
35% monoclinic phases (Table II). The results presented in this
work and those of Khor et al.22 indicate that different micro-
structures of the consolidated zirconia-toughened mullite com-
posites can be obtained by combining different preconditioning
processes of the powders with the SPS process.
Fig. 3. X-ray diffraction patterns of MZT powder sintered by spark
plasma sintering (SPS) at different temperatures. Z, zircon (ZrSiO4); M,
mullite; Zm; monoclinic ZrO2; Zt, tetragonal ZrO2; T, Al2TiO5.
ceramics.26 Another possible explanation for the higher amount
of tetragonal ZrO2 found in the MZM samples processed at
14601C is that their Young’s modulus increased because they
have a higher relative density. This will be explained below, in
the section on mechanical properties. It has been established
that the critical particle size for the tetragonal zirconia-mono-
clinic zirconia transformation decreases for composites with a
high Young’s modulus.26 All the XRD patterns have two small
peaks at 2y 5 36.61 and 45.21, which were indexed as spinel
(MgAl2O4). This phase forms through a reaction between the
added MgO and some Al2O3.
The XRD patterns for the MZT powders sintered by SPS are
shown in Fig. 3. For this powder, the zircon decomposition and
the mullite formation are also well advanced at 13801C. At
14001C, a small amount of undissociated zircon still remains in
the microstructure. The dissociation and mullitization reactions
are completed at 14601C. Although the monoclinic zirconia
phase predominates at all the SPS temperatures, both forms
of zirconia precipitates are present. These XRD patterns show
three small peaks diffracted at 2y 5 39.11, 42.61, and 45.91,
which indicate the formation of Al2TiO5.
(3) Mullitization During SPS
An analysis of the formation of mullite as a function of the SPS
temperature for the MZ, MZM, and MZT powders was made
using the following equation:12
IMð210Þ
FM
¼
(3)
IMð210Þ þ IZð200Þ
(4) Densification
where FM is the intensity ratio of the diffraction peaks of mullite
and zircon. IM and IZ are the integrated intensities of the (210)
peak of mullite and the (200) peak of zircon, respectively. When
this ratio is equal to one, it means that the decomposition of the
zircon and the mullitization of the ceramic are completed. The
results of these calculations are plotted in Fig. 4. It also contains
The influence of the SPS temperature on the density of MZ,
MZM, and MZT composites is shown in Fig. 5. The MZM and
MZT composites achieved densities greater than 97% theoret-
ical density (TD) at SPS temperatures 413801C. The MZ com-
posites achieved greater than 95% TD at SPS temperatures
414201C. The MZM and MZT composites densified faster,

1154
Journal of the American Ceramic Society—Rocha-Rangel et al.
Vol. 88, No. 5
Table II. Microstructural Characteristics of the Samples
Sintered by SPS
Grain
size (mm)
SPS
temperature
(1C)
Relative
density porosity
(%) (%)
Open
w
z
ZrO2
t-ZrO2
(%)
Sample
Mullite ZrO2
(%)
MZ
1560
1460
1460
1550
4–5 0.5 38.93 65
1–2 1.0 39.01 38
99.91 0.07
99.90 0.06
99.2 0.10
92.07 1.36
MZM
MZT
MZy
3–4
1
37.63 12
29.85
7
wVolume fraction of ZrO2 in the sample. zFraction of ZrO2 inclusions that were
y
retained tetragonal at room temperature. Sample conventionally fired at 15501C
for 2 h. SPS, spark plasma sintering.
presumably because of the formation of a low-temperature tran-
sitory liquid phase that assists the sintering process.13–15 The
density of the MZ composites increased continuously at SPS
temperatures 414201C. A maximum relative density of 99.9%
was attained at 15401 and 15601C, the density slightly decreased
for the samples sintered at higher SPS temperatures. The occur-
rence of microcracking induced by the tetragonal-monoclinic
transformation of the ZrO2 particles upon cooling from the SPS
temperature can explain the reduction of the density for these
composites. A similar explanation can be given to understand
the lower densities of the MZT composites, which had the largest
amount of monoclinic zirconia of the three studied composites.
The high degree of densification reached when the decompo-
sition of zircon and the formation of mullite were incipient
processes is worth noting. Wallace et al.4 and Claussen and
Jahn11 have suggested that this condition is necessary in order to
obtain dense mullite–zirconia composites by reaction sintering.
In practice, it is a condition that is very difficult to attain by
conventional processing. Special powder forming methods, very
fine and reactive powders, and carefully designed firing sched-
ules are required. On the other hand, the SPS process is suc-
cessful in attaining those conditions because it promotes fast
sintering at low temperatures. The microstructure obtained is
fine and more homogeneous, allowing fast reaction between the
alumina particles and the silica expelled during decomposition
of the zircon.
(5) Microstructure
The microstructures of powder MZ sintered by SPS at 15601C
and those of the MZM and MZT samples sintered by SPS at
14601C are shown in Fig. 6. Table II summarizes several mi-
Fig. 6. Microstructures of (a) MZ specimen processed by spark plasma
sintering (SPS) at 15401C, (b) MZM, and (c) MZT specimens processed
by SPS at 14601C.
crostructural features observed in the obtained pellets. The three
SPS microstructures look dense and homogeneous. These are
composed of zirconia particles (white phase) dispersed in a mull-
ite matrix (dark phase). The zirconia particles of the MZ spec-
imen are finer than their counterparts in the other two MZM
and MZT specimens. The zirconia inclusions of the MZM sam-
ples are coarser and more spherical. They adopted an angular
shape in the MZT sample. The MZ samples retained the largest
fraction of tetragonal ZrO2 of all the three studied composites
because they had the finest ZrO2 inclusions.
(6) Mechanical Properties
The Vickers hardness (HV), Young’s modulus (E), and inden-
tation fracture toughness (KIC) for the three composites MZ,
MZM, and MZT sintered at different SPS temperatures are
summarized in Table III. The good properties of these compos-
ites are associated with the high consolidation level that was
achieved by SPS. The Young’s modulus of 207 GPa of the MZ
composite sintered by SPS at 15601C is very close to the mod-
ulus of 220 GPa calculated using the rule of mixtures for a
composite that contains 65% mullite and 35% ZrO2.
Fig. 5. Relative density of MZ, MZM, and MZT specimens plotted as
a function of spark plasma sintering (SPS) temperature.

May 2005
Zirconia–Mullite Composites Consolidated by Spark Plasma Reaction
1155
Table III. Mechanical Properties of MZ, MZM, and MZT Specimens as a Function of the SPS-Sintering Temperature
MZ
MZM
E (GPa)
MZT
E (GPa)
Sintering temperature (1C)
HV (GPa)
E (GPa)
KIC (MPaꢀ m1/2
)
HV (GPa)
KIC (MPaꢀ m1/2
)
HV (GPa)
KIC (MPa ꢀ m1/2
)
1380
1400
1420
1460
1480
1500
1520
1540
1560
1580
1620
1640
—
—
—
—
—
—
1229
1303
1262
1282
—
—
—
—
—
—
181
176
183
—
—
—
—
—
—
5.1
5.5
5.7
—
—
—
—
—
—
—
—
1152
1164
—
1173
—
—
—
—
—
—
161
154
—
173
—
—
—
—
—
4.4
4.5
—
4.9
—
—
—
—
—
—
—
—
1865
1530
1506
1515
1640
1183
1558
1353
746
226
193
183
207
174
214
207
187
141
156
3.2
2.9
3.0
3.2
3.4
5.4
6.0
5.9
5.2
5.1
—
—
—
—
—
—
—
—
—
—
—
880
SPS, spark plasma sintering.
The variation of the fracture toughness for the three com-
posites is shown in Fig. 7. The indentation cracks were found to
run in a straight manner without any significant evidence of
crack deflection, but occasional crack branching occurred. In
Fig. 7 the maximum values of KIC 5 6 and 5.7 MPaꢀ m1/2 of the
MZ and MZM composites that were consolidated at SPS tem-
peratures of 15601 and 14601C, respectively, are highlighted.
These values represent a substantial improvement over the frac-
ture toughness of mullite, which is commonly reported around
The fracture toughness of the MZ composites consolidated at
SPS temperatures 415601C decreases as the amount of tetrago-
nal zirconia decreases, so that less transformation toughening
was obtained. The spontaneous tetragonal ZrO2 to monoclinic
ZrO2 transformation that occurs upon cooling is accompanied
by a volume expansion that can induce microcracking of the
matrix.26 This can explain the reductions of the relative density
and the Young’s modulus of these samples.
The composites that were prepared with TiO2, i.e., the MZT
composites, had a lower fracture toughness because they were
less dense, and also had a much smaller amount of tetragonal
zirconia after SPS. They only had o7% of tetragonal zirconia,
so that the transformation toughening mechanism was unlikely
to occur. However, the microcrack toughening mechanism26
could have been activated in these composites, because their
fracture toughness is also higher than that of pure mullite.
The maximum fracture toughness values reported in this pa-
per are higher than those reported in the literature for mullite–
zirconia composites produced by conventional reaction sintering of
KIC 5 2 MPa ꢀ m1/2 1
.
The fabrication of mullite–zirconia composites by SPS result-
ed in a sufficient amount of tetragonal zirconia phase that al-
lowed the transformation toughening mechanism26 to occur in
both MZ and MZM composites produced. This can be observed
in Fig. 8, which shows an XRD pattern taken from the surface
of the composite MZ after SPS, and another pattern taken at the
fracture surface of a specimen that was fractured by three-point
bending. This calculation showed that the SPS-treated specimen
had 65% tetragonal zirconia, while the fractured surface had
only 29% tetragonal ZrO2. These results confirm that the stress-
induced tetragonal ZrO2-monoclinic ZrO2 transformation
toughening mechanism occurred in the SPS-treated mullite–zir-
conia composites. The fracture toughness improvement was cer-
tainly helped by the good densification and fine microstructure
obtained for these composites.
zircon and alumina, which range between 2 and 4 MPa ꢀ m1/2 27,28
.
The explanation for this difference lies in the better densification
and mullitization that was attained by the SPS-processed sam-
ples. However, the fracture toughnesses are smaller than the
value of KIC 5 11.2 MPa ꢀ m1/2 reported by Khor et al.22 The
explanation again lies in the different microstructures that were
obtained in the two methods. The microstructures obtained by
Fig. 8. X-ray diffraction patterns of MZ specimen processed by spark
plasma sintering (SPS) at 15601C. Analysis on the (a) SPS-treated sur-
face; (b) fracture surface. Zm, monoclinic ZrO2; Zt, tetragonal ZrO2; M,
mullite; Z, zircon.
Fig. 7. Indentation fracture toughness of MZ, MZM, and MZT spec-
imens plotted as a function of the spark plasma sintering temperature.

1156
Journal of the American Ceramic Society—Rocha-Rangel et al.
Vol. 88, No. 5
Khor et al. had many microcracks and all the zirconia particles
were monoclinic. A different toughening mechanism operated in
these samples. However, it should be mentioned that comparing
the fracture toughness of ceramics measured by the indentation
method should be made with caution, because of experimental
errors that arise when measuring the crack length.
It is worth noting the good mechanical properties and the
quality of the microstructures of the specimens that were pro-
duced by SPS. Twenty millimeter pellets were produced in this
investigation because the equipment used to conduct this work
was of laboratory size (Dr. Sinter 1020). The SPS process can be
scaled up to make larger samples. Other SPS system grades are
on sale for industrial purposes, including those convenient to
fabricate discs of up to 200 mm diameter and metallic pistons
for motorcycle engines. One limitation of current SPS machines
is that it is not possible to automate the equipment for serial
processing.
tures of the SPS process. It is a consequence of the surface
cleaning and activation of the particles and of the improved
diffusional processes acting when an electric current and a si-
multaneous pressure are applied to the powder.30
For the SPS powder processed in this work, two phenomena
are assumed to have taken place at the final stage of sintering,
i.e., the removal of the last remnants of porosity and completion
of mullitization. In this manner, fully dense mullite–zirconia
composites were fabricated. The question whether it was neces-
sary that the two processes occurred simultaneously during the
SPS was raised in a companion paper.32 For this work, pellets of
the MZ powder were processed by SPS at 14201C. They had a
relative density of 95.5% whereas the formation of mullite was
incipient. The pellets were later fired in a conventional furnace at
15501C. Under these processing conditions, the pellets only
reached a relative density of 98%. It took 2 h to complete the
mullitization reaction. These experiments proved that the simul-
taneous application of an electric current and pressure, as in
SPS, is necessary to obtain dense and fully reacted mullite–zir-
conia ceramics.
(C) Mechanical Properties: The fracture toughness of
the MZ composite fabricated by conventional processing was
only 2.2 MPa ꢀ m1/2 (Fig. 7). Such a low value is not only a con-
sequence of the high porosity of the conventional MZ compos-
ite, but also of the small amount of tetragonal zirconia that it
had, which was only 7%. These results confirm that the SPS
process yields mullite–zirconia composites with excellent me-
chanical properties.
(7) Comparison of Mullite–Zirconia Composites Processed
by SPS and by Conventional Reaction Sintering
The results described above show that the mixtures of zircon,
alumina, and aluminum processed by SPS yield high-quality
mullite–zirconia composites. Now, in order to further highlight
the advantages of this process, in this section the results for the
MZ composites processed by SPS will be compared to the re-
sults obtained by conventionally processing the same powders.
Three aspects will be addressed: (i) mullitization, (ii) densifica-
tion and microstructure, and (iii) mechanical properties. A more
complete discussion of the fabrication of mullite–zirconia com-
posites by conventional processing can be found elsewhere.29
(A) Mullitization: Figure 4 shows data for the mullite
formation as a function of temperature for the MZ powder after
firing it in a conventional furnace at the specified temperatures
for 2 h. Similar mullitization trend data are seen for these pellets
and for the counterparts processed in the SPS machine. Evi-
dently, the mullitization reaction starts earlier for the SPS-proc-
essed powder. It is not easy to compare the experimental results
for the SPS and the conventional processes, because the SPS
temperatures were measured by an infrared optical pyrometer
focused at the surface of the graphite die, which might not be
representative of the actual bulk powder’s core temperature. A
recent modeling of the temperature evolution during SPS has
shown that there might be a large temperature difference be-
tween the surface of the die set and the powder.30 It can be as
high as 1001C depending on a number of factors, such as the
thermal properties of the powders and graphite dies and punch-
es, the SPS heating parameters, and the thermal characteristics
and design of the SPS machine. Therefore, it is not possible to
conclude unambiguously whether the SPS process reduces the
initiation temperature of the reaction (1) on the basis of the
current observations. It is clear, however, that the SPS process
accelerates the reaction. This conclusion is based on the exper-
imental observation that it was not possible to complete the
mullitization reaction for the MZ powder after firing it for 3 h at
15501C or for 2 h at 15801C.31 These are longer durations than
that required in the SPS process. A possible explanation for this
behavior is that the shorter SPS-processing times produce finer
microstructures that are retained at all temperatures. This would
reduce the time required to complete the reaction (1), which is a
diffusion-driven process that is faster for finer grain sizes.12 It
has also been proposed that the DC current applied during the
second step of the SPS process promotes the diffusion processes
in some cases.
IV. Conclusions
The reaction synthesis of fully dense zirconia-toughened mullite
composites from a powder mixture of ZrSiO4, Al2O3, and al-
uminum was achieved by SPS. A combination of techniques was
used to increase the reactivity of the powders. They include at-
trition milling a mixture of zircon, alumina, and aluminum met-
al, its oxidation in air at 11001C to convert the aluminum to fine,
reactive particles of alumina and, finally, SPS in vacuum. The
decomposition of zircon and the formation of mullite start at an
SPS temperature of 14201 and finish at 15601C. These two tem-
peratures are considerably reduced when MgO and TiO2 are
used as sintering additives to promote the formation of a tran-
sitory low-temperature liquid phase, which in turn enhances
the reaction rate and densification. However, better mechanical
properties are obtained for the composite that was prepared
without sintering additives, because it retained a larger amount
of tetragonal zirconia particles that activated the transformation
toughening mechanism. A comparison of the characteristics of
the mullite–zirconia composites fabricated by SPS and conven-
tional reaction sintering highlights the advantages of the former
process.
References
1H. Schneider, K. Okada, and J. A. Pask, Mullite and Mullite Ceramics. John
Wiley & Sons, Inc., New York, 1994.
2T. Koyama, S. Hayashi, A. Yasumori, and K. Okada, ‘‘Preparation and Char-
acterization of Mullite–Zirconia Composites from Various Starting Materials,’’
J. Eur. Ceram. Soc., 14, 295–302 (1994).
3K. A. Khor and Y. Li, ‘‘Effects of Mechanical Alloying on the Reaction Sin-
tering of ZrSiO4 and Al2O3,’’ Mater. Sci. Eng., A256, 271–9 (1998).
4J. S. Wallace, G. Petzow, and N. Claussen, ‘‘Microstructure and Property
Development of In Situ Reacted Mullite–ZrO2 Composites;’’ pp. 436–442 in
Advanced Ceramics Vol. 12: Science and Technology of Zirconia, Edited by
N. Claussen, M. Ruhle, and A. H. Heuer. American Ceramic Society, Columbus,
OH, 1983.
5P. Boch, T. Chartier, and P. D. D. Rodrigo, ‘‘High-Purity Mullite Ceramics by
Reaction Sintering;’’ pp. 353–74 in Ceramic Transactions Vol. 6: Mullite and Mull-
ite Matrix Composites, Edited by S. Somiya, R. F. Davis, and J. A. Pask. Amer-
ican Ceramic Society, Westerville, OH, 1990.
(B) Densification and Microstructure: The relative den-
sity of the MZ composite fabricated by conventional reaction
sintering at 15501C for 2 h was only 92.1%. It also had 1.36%
open porosity. In comparison, a relative density 495% was
reached for all the MZ composites processed at SPS tempera-
tures 414201C.
6S. Prochazka, J. S. Wallace, and N. Claussen, ‘‘Microstructure of Sintered
Mullite–Zirconia Composites,’’ J. Am. Ceram. Soc., 63, C125–7 (1983).
7J. S. Moya and M. I. Osendi, ‘‘Microstructure and Mechanical Properties of
Mullite/ZrO2 Composites,’’ J. Mater. Sci., 19, 2909–14 (1984).
8K. Rundgren, P. Elfing, H. Tabata, S. Kansaki, and R. Pompe, ‘‘Microstruc-
ture and Mechanical Properties of Mullite–Zirconia Composites Made from
The fast sintering and high densities that can be obtained by
SPS at low temperatures are usually described as the main fea-

May 2005
Zirconia–Mullite Composites Consolidated by Spark Plasma Reaction
1157
Inorganic Sols and Salts;’’ pp. 553–66 in Ceramic Transactions Vol. 6: Mullite and
Mullite Matrix Composites, Edited by S. Somiya, R. F. Davis, and J. A. Pask.
American Ceramic Society, Westerville, OH, 1990.
20J. R. Groza and A. Zavaliangos, ‘‘Sintering Activation by External Electrical
Field,’’ Mater. Sci. Eng., A287, 171–7 (2000).
21J. R. Groza, M. Garcia, and J. A. Schneider, ‘‘Surface Effects in Field-Assisted
Sintering,’’ J. Mater. Res., 16 [1] 286–92 (2001).
9P. Boch, J. P. Giry, and P. D. D. Rodrigo, ‘‘Reaction Sintering of Mullite and
Zirconia–Mullite Ceramics;’’ pp. 307–31 in Ceramic Materials and Components for
Engines, Edited by W. Bunk and H. Hausner. Deutsche Keramische Geselleschaft,
Bad Honnef, FRG, 1986.
22K. A. Khor, L. G. Yu, Y. Li, Z. L. Dong, and Z. A. Munir, ‘‘Spark Plasma
Reaction Sintering of ZrO2–Mullite Composites from Plasma Spheroidized Zir-
con/Alumina Powders,’’ Mater. Sci. Eng., A339, 286–9 (2003).
10S. Lathabai, D. G. Hay, F. Wagner, and N. Claussen, ‘‘Reaction Bonded
Mullite/Zirconia Composites,’’ J. Am. Ceram. Soc., 79, 248–56 (1996).
11N. Claussen and J. Jahn, ‘‘Mechanical Properties of Sintered In Situ Reacted
Mullite Zirconia Composites,’’ J. Am. Ceram. Soc., 63, 228–9 (1980).
12P. Boch, T. Chartier, and J. Giry, ‘‘Zirconia-Toughened Mullite/The Role of
Zircon Dissociation;’’ pp. 473–94 in Ceramic Transactions Vol 6: Mullite and
Mullite Matrix Composites, Edited by S. Somiya, R. F. Davis, and J. A. Pask.
American Ceramic Society, Columbus, OH, 1990.
23H. Balmori-Ramı
rez, E. Rocha-Rangel, E. Refugio-Garcıa, and R. C. Bradt,
´ ´
‘‘Dense Mullite from Attrition-Milled Kyanite and Aluminum Metal,’’ J. Am.
Ceram. Soc., 87, 144–6 (2004).
24S. Wu, D. Hols, and N. Claussen, ‘‘Mechanisms and Kinetics of RBAO Ce-
ramics,’’ J. Am. Ceram. Soc., 76, 970–80 (1993).
25K. Niihara, R. Morena, and D. P. H. Hasselman, ‘‘Evaluation of KIC of Brittle
Solids by the Indentation Method with Low Crack-to-Indent Ratios,’’ J. Mater.
Sci. Lett., 1, 13–6 (1982).
13M. F. Melo, J. S. Moya, P. Pena, and S. De Aza, ‘‘Multicomponent Tough-
ened Ceramic Materials Obtained by Reaction Sintering: Part 2. ZrO2–Al2O3–
SiO2–MgO System,’’ J. Mater. Sci., 20, 2702–10 (1985).
26D. J. Green, R. H. J. Hannink, and M. V. Swain, Transformation Toughening
of Ceramics. CRC Press, Boca Raton, FL, 1989.
27T. Koyama, S. Hayashi, A. Yasumori, and K. Okada, ‘‘Contribution of
Microstructure to the Toughness of Mullite/Zirconia Composites;’’ pp. 695–700 in
Ceramic Transactions, Vol. 51, Edited by H. Hausner, G. L. Messing, and S-H.
Hirano. American Ceramic Society, Westerville, OH, 1995.
14P. Miranzo, P. Pena, J. S. Moya, and S. De Aza, ‘‘Multicomponent Tough-
ened Ceramic Materials Obtained by Reaction Sintering: Part 3. ZrO2–Al2O3–
SiO2–TiO2 System,’’ J. Mater. Sci., 20, 2711–8 (1985).
15A. Leriche, ‘‘Mechanical Properties and Microstructure of Mullite–Zirconia
Composites;’’ pp. 541–52 in Ceramic Transactions Vol. 6: Mullite and Mullite Ma-
trix Composites, Edited by S. Somiya, R. F. Davis, and J. A. Pask. American
Ceramic Society, Columbus, OH, 1990.
28T. Koyama, S. Hayashi, A. Yasumori, and K. Okada, ‘‘Microstructure Prop-
erties of Mullite/Zirconia Composites Prepared from Alumina and Zircon Under
Various Firing Conditions,’’ J. Eur. Ceram. Soc., 16, 231–7 (1996).
29E. Rocha, H. Balmori, and S. Dı
´
az, ‘‘Preparation of Mullite–Zirconia Ce-
16J. A. Schneider, R. S. Mishra, and A. K. Mukherjee, ‘‘Plasma Activated Sin-
tering of Ceramic Materials;’’ pp. 143–51 in Ceramic Transactions Vol. 79: Ad-
vanced Synthesis and Processing of Composites and Advanced Ceramics II, Edited
by K. V. Logan, Z. A. Munir, and R. M. Spriggs. American Ceramic Society,
Westerville, OH, 1996.
ramics by Milling and Oxidation-Sintering;’’ pp. 661–6 in Proceedings of the In-
ternational Symposium on Designing, Processing and Properties of Advanced
Engineering Materials, JSPS AEM 156 Committee, Japan, 1997.
30J. Zhang, A. Zavaliangos, and J. R. Groza, ‘‘A Model for Simulation of
Temperature Field Evolution in Field Activated Sintering;’’ in Proceedings of Sin-
tering 2003 (electronic publication), Edited by R. German, G. Messing, and
R. Cornwall. Penn State University, Pennsylvania, U.S.A., 2003.
17H. Risbud and C. H. Shan, ‘‘Fast Consolidation of Ceramic Powders,’’ Mater.
Sci. Eng., A204, 146–51 (1995).
18J. Hong, L. Gao, S. D. de-la-Torre, H. Miyamoto, and K. Miyamoto, ‘‘Spark
Plasma Sintering and Mechanical Properties of ZrO2(Y2O3)–Al2O3 Composites,’’
Mater. Lett., 43, 27–31 (2000).
31E. Rocha-Rangel, H. Balmori-Ramı
´
rez, and S. D. de-la-Torre, ‘‘Reaction-
Bonded Mullite–ZrO2 Composites with Oxide Additives;’’ pp. 231–6 in Ceramic
Transactions 96, Advances in Ceramic Matrix Composites IV, Edited by J. P. Singh,
and N. P. Bansal. The American Ceramic Society, Westerville, OH, 1999.
19S. D. de-la-Torre, H. Miyamoto, K. Miyamoto, J. Hong, L. Gao, L. Tinoco-
D, E. Rocha-Rangel, and H. Balmori-Ramirez, ‘‘Spark Plasma Sintering of Nano-
Composite Ceramics;’’ pp. 892–7 in Proceedings of the 6th International Sympo-
sium on Ceramic Materials and Components for Engineering, Arita, Japan, 1997.
32E. Rocha-Rangel, S. D. de-la-Torre, and H. Balmori-Ramı
´
rez, Ceram. Trans.,
&
115, 285–92 (2000).