Effect of Zirconia Content on the Oxidation Behavior of
Silicon Carbide/Zirconia/Mullite Composites
Cheng-Yuan Tsai and Chien-Cheng Lin
*
,†
Department of Materials Science and Engineering, National Chiao Tung University,
Hsinchu 300, Taiwan
Avigdor Zangvil
*
Materials Research Laboratory and Department of Materials Science and Engineering, University of Illinois,
Urbana, Illinois 61801
Ai-Kang Li
*
Materials Research Laboratory, ITRI, Chutung 310, Taiwan
The oxidation of hot-pressed SiC-particle (SiC
p
)/zirconia
(ZrO
2
)/mullite composites with various ZrO
2
contents, ex-
posed in air isothermally at 1000° and 1200°C for up to 500
h, was investigated; an emphasis was placed on the effects
of the ZrO
2
content on the oxidation behavior. A clear
critical volume fraction of ZrO
2
existed for exposures at
either 1000° or 1200°C: the oxidation rate increased dra-
matically at ZrO
2
contents of >20 vol%. The sharp transi-
tion in the oxidation rate due to the variation of ZrO
2
con-
tent could be explained by the percolation theory, when
applied to the oxygen diffusivity in a randomly distributed
two-phase medium. Morphologically, the composites with
ZrO
2
contents greater than the critical value showed a
large oxidation zone, whereas the composites with ZrO
2
contents less than the critical value revealed a much-
thinner oxidation zone. The results also indicated that the
formation of zircon (ZrSiO
4
) at 1200°C, through the reac-
tion between ZrO
2
and the oxide product, could reduce the
oxidation rate of the composite.
I. Introduction
C
ERAMICS are promising materials for making structural
components for use at elevated temperatures. Specifically,
mullite has been considered to be one of the best candidates,
because of its good high-temperature strength, high creep re-
sistance, good thermal-shock resistance, excellent chemical
stability, and low theoretical density. However, the low frac-
ture toughness (∼2 MPa⭈m
1/2
) has limited its use. One of the
remedies is the incorporation of reinforcements. Partially sta-
bilized zirconia (PSZ) and various forms of silicon carbide
(SiC), such as particles, whiskers, and fibers, have shown their
effectiveness in toughening mullite.
1–5
Because mullite-matrix
composites are considered for use in high-temperature envi-
ronments, their oxidation behavior, as well as the ensuing deg-
radation of their mechanical properties, is of great concern.
The oxidation of monolithic SiC has been subjected to in-
vestigation in the last few decades. Previous studies suggested
that the inward diffusion of oxygen through the growing oxide
layer was the rate-controlling step.
6–11
Under such a condition,
linear–parabolic oxidation kinetics were obeyed: the oxide
thickness was initially proportional to time and then to the
square root of time. The oxidation behavior of SiC-reinforced
ceramic-matrix composites has also been investigated.
12–24
Porter and Chokshi
19
examined the oxidation behavior of an
18-vol%-SiC-whisker-reinforced alumina (Al
2
O
3
) composite
at temperatures in the range of 1500°–1700°C. They reported
that mullite formed because of interaction between the matrix
and the oxidation product. In a study on the oxidation of SiC
p
/
Al
2
O
3
and SiC
p
/mullite composites at 1375°–1575°C, Luthra
and Park
16
found a parabolic rate behavior, and the parabolic
rate constant increased as the SiC content in the composites
increased. Two reaction products, i.e., mullite and an amor-
phous aluminosilicate, were observed in the oxidized SiC
p
/
Al
2
O
3
composite, and some partially oxidized SiC particles
were found between the reaction product layer and the unoxi-
dized zone. Kriven et al.
20
found that SiC whisker (SiC
w
)/
Al
2
O
3
composites had a faster oxidation rate than SiC
w
/mullite
composites, whereas Borom and co-workers
13,14
emphasized
that the oxidation behavior of SiC-reinforced alumina- or mul-
lite-matrix composites was affected by the reaction of the oxi-
dation product and matrix.
There were some studies on the oxidation behavior of zir-
conia-containing (ZrO
2
-containing) composites as well. Back-
haus-Ricoult
18
stated that the microstructure of SiC-reinforced
ZrO
2
/Al
2
O
3
composites after exposure in air could be charac-
terized by three typical subscales: (i) a glassy aluminosilicate in
the outermost surface; (ii) an intermediate white scale with
Al
2
O
3
, ZrO
2
, reaction products (mainly mullite), and large
pores; and (iii) an inner black scale that contained zircon
(ZrSiO
4
), partially ‘‘dissolved’’ SiC whiskers, and carbon. Liu
et al.
21
claimed that the oxidation rate of a SiC
w
/ZrO
2
/mullite
composite was faster than that of a SiC
w
/mullite composite,
because of the reduced crystallization rate in silica (SiO
2
)by
the existence of ZrO
2
. Lin
22
and Lin et al.
23,24
found that the
oxidation rate of some SiC
w
/ZrO
2
/mullite composites was
much faster than that of a SiC
w
/mullite composite, because of
higher oxygen diffusion in ZrO
2
, which provided a rapid trans-
port route of oxygen in the ZrO
2
-containing matrices. There-
fore, addition of PSZ particles into the mullite matrix, as an
effective toughening phase, could cause degradation of the oxi-
dation resistance of the SiC
w
/mullite composite. Oxidation ver-
sus depth behavior was studied for several composites, and
oxidation modes were defined and quantified based on plots of
N. S. Jacobson—contributing editor
Manuscript No. 191493. Received May 15, 1997; approved December 8, 1997.
Supported by the National Science Council of Taiwan under Grant No. NSC83-
0405-E-009-006. Author AZ was supported by the U.S. Dept. of Energy, Materials
Science Division, under Grant No. DE-FG02-91ER45439.
*
Member, American Ceramic Society.
†
Author to whom correspondence should be addressed.
J. Am. Ceram. Soc., 81 [9] 2413–20 (1998)
J
ournal
2413
oxide-layer thickness (around whiskers) versus depth.
22–24
However, they studied only composites with ∼30% ZrO
2
and
did not elucidate the effect of ZrO
2
content on the oxidation
behavior of SiC
w
/ZrO
2
/mullite composites.
In the present study, hot-pressed SiC
p
/ZrO
2
/mullite compos-
ites are exposed isothermally in air at 1000° and 1200°C. The
purpose is to investigate the effect of the matrix composition,
i.e., the variation of ZrO
2
content, on the oxidation behavior of
SiC
p
/ZrO
2
/mullite composites. The effect of the formation of
ZrSiO
4
, as a result of the interaction between ZrO
2
and the
oxidation product, on the oxidation rate is also investigated.
II. Experimental Procedure
Composites of SiC
p
/mullite and SiC
p
/ZrO
2
/mullite, all of
which contained 30 vol% of SiC particulates, were fabricated
by hot pressing. The starting materials were commercial mul-
lite powder (0.2 m average particle size, KM-mullite, Ky-
oritsu Ceramic Materials Co., Nagoya, Japan), 3 mol% Y
2
O
3
partially stabilized ZrO
2
(0.3 m average particle size, Product
No. TZ-3Y, Toyo Soda Mfg. Co., Tokyo, Japan), and SiC
powder (7 m average particle size, Cerac, Milwaukee, WI).
Powder mixtures of desired composition were dispersed and
homogenized by using an ultrasonic dispersing process (Model
XL-2020, SONICATOR威, Heat Systems, Farmingdale, NY)
with methanol as a medium. To form a stable suspension, the
pH of each slurry was adjusted to 10 with ammonium hydrox-
ide (NH
4
OH). The slurry was partially dried on a hot plate
under continuous stirring and oven-dried thereafter. The dried,
crushed, and sieved powder mixtures were uniaxially pre-
pressed into disks 60 mm in diameter. The green compact was
then coated with boron nitride (BN) and placed into a graphite
die that was lined with graphite foil. The composites that con-
tained various contents of ZrO
2
were hot-pressed at 1600°C in
an argon atmosphere under a pressure of 30 MPa for 45 min.
The designations, compositions, and hot-pressing conditions of
the composites are listed in Table I. For the sake of conve-
nience, the ‘‘zirconia + mullite’’ portion in each SiC
p
/ZrO
2
/
mullite composite is called the ‘‘matrix,’’ even though this
matrix itself is a composite. The matrix, either mullite or zir-
conia + mullite, has a content of 70 vol% in each composite,
whereas the ZrO
2
content is expressed as a volume percentage
of the matrix only, rather than a volume percentage of the
entire composite (see footnotes in Table I).
The as-hot-pressed composites were ground and then cut
into pieces with dimensions of 10 mm×6mm×3mm.All
specimens were ground and polished with diamond paste and
weighed before and after oxidation using a precision electronic
balance (Model R200D, Satorius AG, Goettingen, Germany);
the balance was accurate to 0.01 mg. Composite samples were
exposed to air in a box furnace (Model 51333, Lindberg, Wa-
tertown, WI) isothermally at 1000° and 1200°C for up to 500
h. Specimens were loaded into the furnace at room tempera-
ture, and then the furnace temperature was increased at a rate
of 20°C/min until the set temperature was attained. One speci-
men for each composition was drawn from the furnace at vari-
ous intervals to measure the weight change. Specimens drawn
from the furnace were not reloaded for further oxidation, to
avoid the formation of extended cracks due to thermal shock.
The extent of oxidation was expressed by the weight change
per unit surface area (called weight gain hereafter).
The major phases on the surfaces of the as-hot-pressed speci-
mens and the exposed specimens were identified via X-ray
diffractometry (XRD) (Model MXP18, Mac Science, Tokyo,
Japan). The as-polished cross sections of oxidized specimens
were observed by using polarized light microscopy (Model
BH-2, Olympus, Tokyo, Japan) as well as by using a scanning
electron microscopy (SEM) microscope (Model S-2500, Hita-
chi, Tokyo, Japan) that was equipped for energy-dispersive
X-ray spectroscopy (EDS) (Kevex Instruments, Valencia, CA).
III. Results
Figure 1 shows the plots of weight gain versus time for
various SiC
p
/ZrO
2
/mullite composites after exposure at 1000°
and 1200°C, respectively. The weight gain increased as time
increased for each composite at 1000° and 1200°C. The weight
gains of the composites that contained <20 vol% ZrO
2
were
significantly less than those of the composites that contained
ⱖ30 vol% ZrO
2
. Figure 2 shows the relationship between the
weight gain and the volume fraction of ZrO
2
for exposure at
1000°C for 500 h and 1200°C for 500 h. This figure indicates
that the weight gains for the composites that contained <20
vol% ZrO
2
were comparatively small, whereas the weight gain
increased rapidly when the composites contained >20 vol%
ZrO
2
. It appeared that there was a critical volume fraction of
ZrO
2
content required for rapid oxidation. In other words, the
oxidation rate would be accelerated abruptly beyond this criti-
cal ZrO
2
content, whereas the weight gain remained fairly low
below the critical value. It is also noted that the weight-gain
curve for 1200°C slightly declined as the ZrO
2
content in-
creased to >80 vol%.
Figure 3 shows the XRD spectrum of the surface of the
as-hot-pressed MZY50/SiC composite, as well as the XRD
spectra after exposure for 500 h at 1000° and 1200°C. The
as-hot-pressed sample consisted of four major phases: mullite,
monoclinic ZrO
2
(m-ZrO
2
), tetragonal ZrO
2
(t-ZrO
2
), and SiC
(Fig. 3(a)). After exposure at 1000°C for 500 h, the matrix was
rather stable; i.e., the interaction between the oxide product and
the matrix was insignificant (Fig. 3(b)). No new crystalline
phases were found. However, observations via transmission
electron microscopy (TEM) in a previous study
22
indicated that
there was an amorphous SiO
2
layer around each of SiC par-
ticles located near the surface of the composite. As shown in
Table I. Designation, Compositions, and Hot-Pressing Conditions of Composites
Designation Composition
†
Hot-pressing
conditions
‡
MZY0/SiC Mullite + 30 vol% SiC 1675°C/45 min
MZY5/SiC (Mullite + 5 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY10/SiC (Mullite + 10 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY15/SiC (Mullite + 15 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY20/SiC (Mullite + 20 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY25/SiC (Mullite + 25 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY30/SiC (Mullite + 30 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY40/SiC (Mullite + 40 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY50/SiC (Mullite + 50 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY60/SiC (Mullite + 60 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY80/SiC (Mullite + 80 vol% 3Y-PSZ) + 30 vol% SiC 1600°C/45 min
MZY100/SiC 3Y-PSZ + 30 vol% SiC 1600°C/45 min
†
The volume percentage of the matrix in parentheses is 70% for each composite; the balance is SiC
p
. The ZrO
2
content is based on the matrix itself; i.e., the amount (in vol%) of Y-PSZ ⳱ volume of Y-PSZ × 100/(volume of
mullite + volume of Y-PSZ).
‡
All hot pressing was performed under a pressure of 30 MPa and undera1atm
argon-gas atmosphere.
2414 Journal of the American Ceramic Society—Tsai et al. Vol. 81, No. 9
Fig. 3(c), after exposure at 1200°C for 500 h, some stronger
diffracted peaks of cristobalite were identified, whereas ZrSiO
4
was also found (due to the interaction between SiO
2
and ZrO
2
),
which led to the reduction of the ZrO
2
peak intensities. Figure
4 shows the XRD spectra of composites with various ZrO
2
contents after exposure at 1200°C for 500 h. It is apparent that
the relative intensity of the corresponding ZrSiO
4
peaks in-
creased as the ZrO
2
content increased.
Figure 5 shows polarized-light optical micrographs of the
MZY15/SiC composite (for 500 h), the MZY20/SiC composite
(for 500 h), and the MZY50/SiC composite (for 25 h), all after
exposure at 1200°C. The cross-sectional samples clearly ex-
hibited a distinct layered structure; this layering was due to the
different extent of oxidation at various depths, which caused a
change in composition. Figure 5(a) reveals the cross section of
the MZY15/SiC composite; it has two oxidized surfaces that
are glued together, the interfaces of which are located at the
center of the micrograph. The oxidation zone consists of an
outer white scale and an inner black scale, whereas the gray
region is the unoxidized substrate. Figure 5(b) demonstrates a
two-layer structure for the MZY20/SiC composite; both the
white and black scales are thicker than the corresponding scales
in the MZY15/SiC composite (Fig. 5(a)). Figure 5(c) shows the
MZY50/SiC composite after exposure at 1200°C for 25 h; this
figure indicates much-thicker white and black regions (the un-
oxidized region is not shown). Figure 6 shows an SEM micro-
graph of the cross section of the MZY20/SiC composite after
exposure at 1200°C for 500 h. The outermost surface of the
composite is shown on the left. The bright particles are ZrO
2
embedded in the mullite matrix (shown as the gray regions in
the figure). Angular SiC particles were encompassed by a dark
layer (indicated by the arrow). The dark layer was identified as
being the oxidation product, SiO
2
. The variation of the oxida-
tion of SiC particles with depth is also observed. There is a
sharp change in the thickness of the SiO
2
layer at a depth of
∼45 m. The outer region (‘‘A–B’’) corresponds to the white
scale shown in Fig. 5(b), whereas the inner region (‘‘B–C’’)
corresponds to the black scale in Fig. 5(b).
Figures 7 and 8 demonstrate the effect of ZrO
2
content on
the oxidation morphology of SiC
p
/ZrO
2
/mullite composites.
Figure 7 shows an SEM micrograph of the MZY15/SiC com-
posite after exposure at 1200°C for 500 h, indicating that SiC
particles at a depth beyond ∼40 m did not oxidize noticeably.
That is, there was a small oxidation zone in the MZY15/SiC
composite after exposure at 1200°C for 500 h. Figure 8 shows
a series of micrographs for the MZY30/SiC composite at vari-
ous depths after exposure at 1200°C for 25 h. The SiC particles
were slightly oxidized, even at a depth of >600 m, which
indicates a much-larger oxidized depth than that of the
MZY15/SiC composite exposed for 500 h.
IV. Discussion
For the sake of convenience in our discussion, two terms are
first defined. First, the silica layer of an individual SiC particle
in a SiC-containing composite means the layer of SiO
2
that is
formed as a result of the oxidation reaction occurring on the
surface of the SiC particle. When a SiC particle is partially
oxidized, a silica layer will encompass it. Second, the oxidation
zone of a SiC-containing composite after exposure in an oxi-
dizing environment is defined as the zone from the outermost
surface of the composite to the depth where no oxidation of the
incorporated SiC particles can be detected.
The inward diffusion of oxygen from the atmosphere causes
the oxidation of SiC particles within the oxidation zone, which
results in weight gain, because of the formation of SiO
2
, ac-
cording to the reaction
SiC(s)+
3
2
O
2
(g) → SiO
2
(s) + CO(g) (1)
where most of the gaseous product CO was believed to diffuse
all the way out to the surface. Thus, the atomic-weight differ-
ence between the SiC and SiO
2
was the reason for the weight
gain during oxidation, and the extent of oxidation could be
represented by the weight gain of each sample. In contrast, no
weight gain was detected in the oxidation test of mullite and
ZrO
2
/mullite composites,
22
which indicates that the inward dif-
fusion of oxygen was negligible in the composites that con-
tained no SiC.
Fig. 1. Curves of weight gain versus time at (a) 1000° and (b)
1200°C for samples listed in Table I.
Fig. 2. Curves of weight gain versus ZrO
2
content at 1000° and
1200°C for 500 h.
September 1998 Effect of Zirconia Content on the Oxidation Behavior of Silicon Carbide/Zirconia/Mullite Composites 2415
Fig. 3. XRD spectra of the MZY50/SiC composite (as hot-pressed (spectrum ‘‘(a)’’), after exposure at 1000°C for 500 h (spectrum ‘‘(b)’’), and
after exposure at 1200°C for 500 h (spectrum ‘‘(c)’’)). (M ⳱ mullite, Z ⳱ zircon, C ⳱ cristobalite, m ⳱ m-ZrO
2
,t⳱ t-ZrO
2
, and S ⳱ SiC.)
Fig. 4. XRD spectra of the MZY15/SiC (spectrum ‘‘(a)’’), MZY25/SiC (spectrum ‘‘(b)’’), and MZY50/SiC (spectrum ‘‘(c)’’) composites, all after
exposure at 1200°C for 500 h. (M ⳱ mullite, Z ⳱ zircon, C ⳱ cristobalite, m ⳱ m-ZrO
2
,t⳱ t-ZrO
2
, and S ⳱ SiC.)
2416 Journal of the American Ceramic Society—Tsai et al. Vol. 81, No. 9
As shown in Figs. 1 and 2, the weight gain abruptly in-
creased when the amount of incorporated ZrO
2
was above a
certain volume fraction. There appeared to be a threshold value
of ∼20 vol% ZrO
2
for the oxidation rate of SiC
p
/ZrO
2
/mullite
composites. A plausible explanation for this phenomenon can
be based on the percolation theory.
25–27
It applies to a wide
variety of transport phenomena, including diffusivity and ther-
mal or electric conductivity. In a two-phase system, when one
phase has a very high diffusivity of a species compared to that
of the other phase, there exists a critical volume fraction, or
percolation threshold ( f
c
), of the high-diffusivity phase. Per-
colation occurs beyond this threshold, and the diffusivity of the
particular species in the composite has the order of magnitude
of the high diffusivity. In other words, for a binary composite
with D
2
>> D
1
, percolation theory predicts that D
m
has the
approximate order of D
1
if f < f
c
and has the approximate order
of D
2
if f > f
c
(where D
1
and D
2
are the diffusivities of a certain
species in the two phases, D
m
is the diffusivity of that species
in the composite, and f is the volume fraction of the second
phase).
Because oxygen diffusivity in ZrO
2
is much higher than that
in mullite (Table II) and using mullite and ZrO
2
as the first and
second phases, respectively, the relationships of oxygen diffu-
sivity in ZrO
2
/mullite matrices would be
D
matrix
O
≈ order of D
zirconia
O
(if f > f
c
)(2a)
D
matrix
O
≈ order of D
mullite
O
(if f < f
c
)(2b)
where the superscript denotes the diffusing species (oxygen).
Note that the ZrO
2
/mullite assembly is called the ‘‘matrix,’’
because it serves as a matrix for the SiC
p
. From the results
shown in Figs. 1 and 2, f
c
≈ 20 vol% ZrO
2
. The oxygen dif-
fusivity in a matrix with >20 vol% ZrO
2
should be very close
to that in ZrO
2
, which leads to a rapid increase in the oxidation
rate of SiC particles. On the other hand, the diffusivities in
matrices with f < f
c
should be of the same order as that of
mullite (which is very low); thus, SiC particles were better
protected by these matrices.
In addition to the sharp transition at f
c
, the weight gain of
SiC
p
/ZrO
2
/mullite composites also increased as the ZrO
2
con-
tent increased, either when f < f
c
or when f > f
c
. For the oxi-
dation of an ‘‘individual’’ SiC particle in SiC
p
/ZrO
2
/mullite
composites, the inward diffusion of oxygen will proceed via
the following serial steps: (i) oxygen diffuses through the
ZrO
2
-containing matrix to the surface of SiC, where the oxi-
dation reaction occurs; and (ii) as the silica layer is formed,
oxygen must further diffuse through it to arrive at the interface
of the silica layer and the unoxidized SiC core. That is, the
inward diffusion of oxygen may be controlled by either diffu-
sion through the ZrO
2
-containing matrix or diffusion through
the silica layer, depending on which one is slower. It is easy to
compare the diffusivities of oxygen in SiO
2
, ZrO
2
, and mullite
(Table II). At 1000°C, the diffusivity in ZrO
2
is between 10
−9
and 10
−5
cm
2
/s; this value is dependent on crystal structure,
additives, and the stoichiometry.
28,29
At the same temperature,
the diffusivity in amorphous silica is much lower (10
−13
–10
−12
cm
2
/s).
30,31
The diffusivity in mullite is unknown. It was sug-
gested by Cherkasoy et al.
32
(quoted by Luthra and Park
16
)to
be similar to that in Al
2
O
3
, which, in turn, is between 10
−20
and
10
−18
cm
2
/s (based on reports quoted by Luthra and Park
16
).
Oxygen diffusion in Al
2
O
3
, and possibly in mullite, being ex-
tremely slow in the bulk, is mainly through grain boundaries
and, therefore, is strongly dependent on the microstructure.
However, it may be safe to generally state that oxygen diffu-
sivity in SiO
2
is very small compared to that in ZrO
2
but is very
large compared to that in mullite. Therefore, if f < f
c
(e.g., the
MZY15/SiC composite), the oxygen diffusivity in the matrix is
much lower than that in the silica layer and, thus, the oxidation
rate is likely to be controlled by the oxygen diffusion in the
matrix. More accurately, the controlling step is determined by
the lowest reciprocal diffusion impedance, or diffusivity (D)
divided by the corresponding thickness. Most of the oxygen
will be consumed continuously by the oxidizing SiC particles
at or near the surface until they are completely oxidized, which
results in a shallow oxidation zone. Because the oxygen diffu-
sivity slowly increases as the ZrO
2
content increases in the
range of f < f
c
, according to the effective medium theory,
33,34
Fig. 5. Polarized-light optical micrographs of the (a) MZY15/SiC
(500 h/1200°C), (b) MZY20/SiC (500 h/1200°C), and (c) MZY50/SiC
(25 h/1200°C) composites.
September 1998 Effect of Zirconia Content on the Oxidation Behavior of Silicon Carbide/Zirconia/Mullite Composites 2417