Morphological evolution and weak interface development within chemical-vapor-deposited zirconia coating deposited on Hi-Nicalon fiber

Article abstract of DOI:10.1111/j.1151-2916.2002.tb00312.x

The phase contents and morphology of a ZrO₂ fiber coating deposited at 1050°C on Hi-Nicalon by chemical vapor deposition (CVD) were examined as a function of deposition time from 5-120 min. The morphological evolution in the ZrO₂ coating was correlated to the development of delamination within the ZrO₂ coating. The delamination appears to occur as a result of: (i) continuous formation of tetragonal ZrO₂ nuclei on the deposition surface; (ii) martensitic transformation of the tetragonal phase to a monoclinic phase on reaching a critical grain size; and (iii) development of significant compressive hoop stresses because of the volume dilation associated with the transformation. Our observations suggest that it will be of critical importance to further understand and eventually control the nucleation and grain growth behavior of CVD ZrO₂ and its phase transformation behavior for its potential applications for composites.

Full text of DOI:10.1111/j.1151-2916.2002.tb00312.x

J. Am. Ceram. Soc., 85 [6] 1561–68 (2002)
journal
Morphological Evolution and Weak Interface Development within
Chemical-Vapor-Deposited Zirconia Coating Deposited on
Hi-Nicalon™ Fiber
Hao Li, Jinil Lee, Matthew R. Libera, and Woo Y. Lee*,†
Department of Chemical, Biochemical and Materials Engineering, Stevens Institute of Technology,
Hoboken, New Jersey 07030
Anteneh Kebbede*
General Electric Corporate Research and Development, Niskayuna, New York 12309
Michael J. Lance*
Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6068
Hongyu Wang*
General Electric Power System, Schenectady, New York 12345
Gregory N. Morscher*
Ohio Aerospace Institute, c/o NASA Glenn Research Center, Cleveland, Ohio 44135
The phase contents and morphology of a ZrO2 fiber coating
deposited at 1050°C on Hi-NicalonTM by chemical vapor
deposition (CVD) were examined as a function of deposition
time from 5–120 min. The morphological evolution in the ZrO2
coating was correlated to the development of delamination
within the ZrO2 coating. The delamination appears to occur as
a result of: (i) continuous formation of tetragonal ZrO2 nuclei
on the deposition surface; (ii) martensitic transformation of
the tetragonal phase to a monoclinic phase on reaching a
critical grain size; and (iii) development of significant com-
pressive hoop stresses because of the volume dilation associ-
ated with the transformation. Our observations suggest that it
will be of critical importance to further understand and
eventually control the nucleation and grain growth behavior of
CVD ZrO2 and its phase transformation behavior for its
potential applications for composites.
ϳ700°–1000°C. This oxidation problem, which is commonly
referred to as “pesting,” is considered to be one of the major
problems remaining to be solved before SiC/SiC composites
can be reliably introduced into next-generation aircraft engines,
power generation gas turbines, space vehicles, and other indus-
trial applications.1
Recently, Lee et al.2 reported that SiC/SiC minicomposite
specimens containing a SiO2/ZrO2/SiO2 multilayer coating pre-
pared by chemical vapor deposition (CVD) exhibited graceful
tensile failure behavior and extensive crack deflection within the
multilayer interface region.2 The minicomposite possessed a rea-
sonable degree of load-carrying capability and retained much of
the composite characteristics after short-term exposure to pesting
conditions. The inner SiO2/ZrO2 interface was found to be the
weakest interface in the multilayer coating region. They speculated
that the significant coefficient of thermal expansion (CTE) mis-
match might result in the development of extreme residual strains
and consequent “weakening” of the interfaces of the multilayer
coating.
I. Introduction
NLIKE traditional brittle ceramic materials, SiC/SiC composites
A possible role of ZrO2 phase transformation on the apparent
weak interface behavior was not included in the discussion by Lee
et al.2 of the experimental observations. It is well known that
various applications of ZrO2 stem from its polymorphism and
martensitic phase transformation behavior. The martensitic phase
transformation of ZrO2 from t (tetragonal) to m (monoclinic) is
accompanied by significant volume dilation and shear.3 If the
t-to-m transformation occurred in the multilayer coating during the
deposition process or subsequent cooling, the volume change
associated with the transformation would result in the development
of significant hoop compressive stresses in the ZrO2 coating. This
might cause the ZrO2 layer to delaminate from the SiO2 layer. The
objective of this study was to carefully examine the phase content
and morphology of the CVD–ZrO2 coating as a function of
deposition time (i.e., coating thickness) by transmission electron
microscopy (TEM) and Raman microscopy to assess the role of the
phase transformation on the apparent weak interface observed in
the SiO2/ZrO2/SiO2 multilayer coating.
U
are made to be tough and damage tolerant by incorporating a
weak fiber coating (e.g., boron nitride (BN)), which can deflect
cracks and promote debonding at the fiber/matrix interface region.
However, the BN fiber coating, whereas it provides superior weak
interface behavior, has proved to be susceptible to oxidation from
N. S. Jacobson—contributing editor
Manuscript No. 187464. Received September 10, 2001; approved February 12,
2002.
Supported by NSF (DMR-GOALI 9971623) with co-funding from Air Force
Research Laboratory (AFRL). Author MJL was supported in part by the Division of
Materials Sciences and Engineering, Office of Basic Energy Sciences, Dept. of
Energy.
*Member, American Ceramic Society.
†Author to whom correspondence should be addressed.
1561

1562
Journal of the American Ceramic Society—Li et al.
Vol. 85, No. 6
II. Experimental Procedure
polished using a tripod-polishing technique,4 and some of them
were also ion-milled with an ion-milling machine (DuoMillTM
600, Gatan, Inc., Pleasanton, CA).
Six CVD-ZrO2 coating specimens were prepared at different
deposition times: 5, 10, 20, 30, 60, and 120 min. A hot-wall CVD
reactor was used for ZrO2 deposition. As schematically shown in
Fig. 1, the reactor chamber was made of a fused SiO2 tube (3.4
cm outside diameter and 34 cm length). A long as-received
Hi-NicalonTM fiber tow (Nippon Carbon Co., Tokyo, Japan)
was wrapped around a fused SiO2 holder to make approxi-
mately nine fiber tow specimens (ϳ25 cm long) for each
deposition experiment.
The reactor was heated using a resistance furnace with an
effective heating zone of ϳ30 cm. The temperature at the center
region of the reactor was measured and controlled using a K-type
thermocouple interfaced to a temperature controller. The pressure
inside the reactor chamber was controlled using a mechanical
vacuum pump and an exhaust valve (Type 253, MKS, Waltham,
MA) interfaced to a pressure controller (Type 252A, MKS) and a
capacitance manometer (Baratron, MKS). The reactor was oper-
ated using ZrCl4, CO2, and H2 as precursors at a temperature of
1050°C and a pressure of 4 kPa for the CVD experiments.
A 500 cm3 stainless steel vaporizer was used to hold ZrCl4,
which is a solid at room temperature. The vaporizer was kept at a
constant temperature of 190°C using heating tapes to maintain the
vapor pressure of ZrCl4. An argon stream was used to carry the
ZrCl4 vapor from the vaporizer to the reactor chamber. The flow
rates of argon, H2, and CO2 were controlled by mass flow meters
(Type 258, MKS, Andover, MA): 40 cm3/min for argon, 120
cm3/min for H2, and 120 cm3/min for CO2 at standard temperature
and pressure (STP).
Tensile tests were conducted at room temperature. The length of
fiber tow specimens was 20 cm. The entire length of each fiber tow
specimen was glued to a plastic strip (20 cm ϫ 2 cm ϫ 0.1 mm)
using a 5-min epoxy, except for the 2.5 cm gauge length. At least
five tows were tested for each condition. For comparison, five
flame-desized fiber tows were also tested. Flame-desizing was
achieved by slowly moving the fiber tow through a torch flame.
A field-emission scanning electron microscope (FEG-SEM;
Model DSM 982, LEO Electron Microscopy, Inc., Thornwood,
NY) and a LaB6 transmission electron microscope (Model CM30,
Philips, Eindhoven, Netherlands) were used for coating morphol-
ogy characterization. The phases in the ZrO2 coating were char-
acterized by selected-area electron diffraction (SAD) and
convergent-beam electron diffraction (CBED). The diffraction
patterns were indexed using electron microscopy simulation
(EMS) on-line (http://cimesg1.epfl.ch/CIOLS/crystal1.pl). Crys-
tallographic data from a recent review paper3 on the crystal
structure of ZrO2 were used in analyzing the data. TEM specimens
were prepared by placing fiber tows of interest between two single
crystal silicon wafer pieces using an epoxy adhesive (M-Bond 610,
Measurement Group, Raleigh, NC). The “sandwich” specimens
were then cured for 2 h at 170°C in air and then cross-sectioned to
small coupons of 0.5 mm ϫ 2 mm ϫ 5 mm. The coupons were
A Dilor XY 800 triple stage Raman microprobe (JY, Inc.,
Edison, NJ) and an Innova 308C argon ion laser (Coherent, Inc.,
Santa Clara, CA) operating at 514.5 nm with a 100 mW output
power were used to detect the phase content of the CVD-zirconia
coating at Oak Ridge National Laboratory. The laser was focused
on areas of interest with an optical objective providing a spatial
resolution of 2 ␮m. The method for calculating the phase content
was based on comparing the Raman peak heights of monoclinic
5
and tetragonal zirconia and is described in detail in Lance et al.
III. Results
Based on initial SEM results, three specimens were selected for
in-depth TEM investigation to study the morphologic evolution of
CVD-ZrO2 on the surface of Hi-Nicalon fiber as a function of
deposition time (5, 20, and 120 min). As shown in Figs. 2(a) and
(b), two distinct regions were observed for the ZrO2 coating
deposited for 5 min. In some regions, the ZrO2 coating was
continuous with a thickness of ϳ100 nm (Fig. 2(a)). The “contin-
uous” regions were characterized by SAD. The rings in the
diffraction pattern in Fig. 2(a) were indexed to c (cubic) and/or t
nanocrystallites, whereas the spots were assigned to larger c and/or
t particles. However, no m phase was observed in the continuous
regions.
Note that the c and t phases have very similar structures, and
therefore it is generally difficult to distinguish the c and t phases
based on diffraction patterns. To complement TEM analysis,
Raman microscopy was also used for phase determination. Even
though the CVD-ZrO2 coating after the 5 min deposition was too
thin to be detected by Raman microscopy, Raman data from other
CVD-ZrO2 coatings indicated that t was the only phase detected.
So we suppose that the phase indexed to be c/t by electron
diffraction patterns is mainly t phase. For the rest of the paper, we
will use the t notation to distinguish this phase indexed by TEM
from the m phase.
In other regions, the ZrO2 coating was discontinuous with
individual grains directly attached to the fiber surface. As shown in
Fig. 2(b), the grains were ϳ50 nm thick and ϳ50–100 nm long
along the fiber surface. Some of the grains were t, whereas the
others were m. Figs. 2(c) and (d) are high-resolution images of the
៮
m particles with all the fringes corresponding to the (111) plane of
៮
the m phase. Note that the (111) plane has the lowest surface free
energy for the m phase.6
Because the t phase is a high-temperature phase for ZrO2, the t
grains could not have been transformed from the m phase during
the CVD process at 1050°C or during cooling from the deposition
temperature. This argument implies that the t grains should have
been t when they were deposited at 1050°C. On the other hand, the
m grains observed by TEM could either be m phase when
deposited, or they could have transformed from the t phase during
the CVD process and/or the cooling step.
Figure 3 shows TEM images of the CVD–ZrO2 coating after the
20 min deposition. As shown in Fig. 3(a), delamination occurred
within the ZrO2 coating close to the fiber surface. This observation
indicates that a weak interface developed during this deposition
time period. In contrast, no debonding was observed between the
ZrO2 coating and the fiber or within ZrO2 coating after the 5 min
deposition. High-resolution images (Figs. 3(b) and (c)) show that
the inner ZrO2 layer retained on the fiber side is ϳ50–100 nm
thick and contains both t and m phases. It is interesting to note that
៮
the t (111) plane and m (111) planes are almost normal to the
coating growth direction, suggesting the development of through-
thickness preferred orientations. Note that these planes represent
the lowest energy surfaces for their crystal structures.6 The
interface between the ZrO2 layer and the fiber surface was sharp,
indicating the absence of any major reaction there.
Based on the analysis of diffraction patterns collected at various
coating locations, the amount of the m phase present in the ZrO2
Fig. 1. Schematic diagram of the CVD apparatus used for ZrO2
deposition.

June 2002
Morphological Evolution and Weak Interface Development within CVD-ZrO2-Coated Hi-Nicalon™ Fiber
1563
Fig. 2. TEM images of CVD-ZrO2 deposited for 5 min: (a) continuous area, (b) discontinuous area, (c) and (d) high resolution images of the discontinuous
៮
area. All lattice fringes shown in (c) and (d) corresponded to the m(111) planes with a d-spacing of 0.316 nm.
layer after the 20 min deposition was observed to be higher than
that observed after the 5 min deposition, particularly in the outer
layer that was detached from the fiber. This observation suggests
that some of the t phase in the outer region of the coating might
have been transformed to the m phase during the additional CVD
process and/or the cooling step.
Cross-sectional SEM and TEM images of the CVD-ZrO2
coating after the 120 min deposition are shown in Fig. 4.
Similar to the 20 min sample, delamination was observed within
the ZrO2 coating at a distance of ϳ50–100 nm from the fiber
surface. The surface SEM images shown in Fig. 5 provide
additional evidence that the delamination occurred within the
ZrO2 coating. It is evident from the SEM images that the inner
layer is composed of very small particles (ϳ50–100 nm)
deposited on the fiber surface. Above the inner layer, larger
ZrO2 grains were observed and were identified to be mostly m
by SAD. However, at the very top surface of the ZrO2 coating,
very small t and m particles less than ϳ50–100 nm were again
observed, as shown in the SEM pictures in Fig. 5 and the TEM
images in Fig. 4(c). The observations suggest that ZrO2 nuclei
were continuously formed on the deposition surface throughout
the 120 min period.

1564
Journal of the American Ceramic Society—Li et al.
Vol. 85, No. 6
Fig. 3. TEM images of CVD-ZrO2 deposited for 20 min.
The Raman spectra of both the surface and the cross section
Figure 7 also compares the failure loads of coated fiber tows to
those of flame-desized fiber tows. All the coated samples
showed similar or higher failure loads than the desized fiber
tows. This demonstrated that the ZrO2 coatings did not de-
grade fiber strength. In contrast, a CVD-SiO2 coating severely
degraded the strength of Hi-Nicalon when its thickness ex-
ceeded ϳ80 nm.7 It was also interesting to observe that for
two of the five 120 min specimens, the fiber tow did not fail
totally within the gauge length, as some fibers showed a
fiber pull-out behavior in the “matrix” formed by the epoxy.
The epoxy here is used to bond the fiber tow to its plastic
support.
of the CVD-ZrO2 coating after the 120 min deposition are
shown in Fig. 6. Only the m phase was detected from the cross
section of this specimen, but the presence of the t phase (ϳ13
vol%) was detected from the surface of this specimen. How-
ever, the c phase was not detected in either case. The difference
between the two spectra is consistent with our TEM observa-
tions that the t phase was at the outer surface of the ZrO2
coating and the m phase was mainly present in the middle
region of this coating.
Figure 7 shows the failure loads of Hi-Nicalon fiber tows
coated with the CVD-ZrO2 as a function of deposition time.

June 2002
Morphological Evolution and Weak Interface Development within CVD-ZrO2-Coated Hi-Nicalon™ Fiber
1565
Fig. 4. SEM (a) and TEM images (b)–(d) of CVD-ZrO2 deposited for 120 min.
IV. Discussion
transformation was also reported to be accompanied by a 4.9%
volume dilation at room temperature and 3.5% volume dilation at
950°C.8 On the other hand, there is almost no volume change for
the c-to-t transformation.3
The first major question encountered in this study is why the
high temperature t phase of ZrO2 was present in our CVD-ZrO2
coating at room temperature. A brief review of phase transforma-
tions in ZrO2 is needed before discussion of the experimental
observation. The ambient-pressure phases of ZrO2 are a mono-
clinic phase (P21/c), which is stable up to 1205°C, a tetragonal
phase (P42/nmc) from 1205°–2377°C, and a cubic fluorite struc-
It has been previously reported that the high-temperature t and
c phases can be retained to room temperature without the use of
stabilizers.9,10 A few mechanisms have been postulated to explain
the stabilization of the high-temperature phases of ZrO2, especially
for the t phase. Garvie11,12 first proposed that the lower surface
energy of the t-ZrO2 is responsible for the t phase being present in
nanocrystalline forms at or even below room temperature. He
predicted that ZrO2 particles Ͻ10 nm in diameter could be
៮
ture (Fm3m) from 2377°C to the melting point at 2710°C. The
t-to-m transformation is diffusionless and martensitic and is
accompanied by ϳ9o shear and 4.5% volume dilation.3 The t-to-m

1566
Journal of the American Ceramic Society—Li et al.
Vol. 85, No. 6
Fig. 5. SEM images of CVD-ZrO2 surfaces for a 120 min deposition time.
stabilized in the tetragonal form. Those above this critical size
would be subjected to the t-to-m transformation.
According to a simple thermodynamic approach recently pre-
sented by Chraska,13 the Gibbs free energy change because of the
t-to-m phase transformation of an unconstrained spherical particle,
may be expressed as:
where ⌬H is the enthalpy difference between the t and m phase
(Ϫ2.82 ϫ 108 J/m3) at the equilibrium bulk transformation
temperature, and Tb is the t-m transformation temperature for bulk
ZrO2 (1250°C).12,3 This relationship suggests that the critical size
increases with increasing temperature, as shown in Fig. 8. This
phenomenon is often referred as the “size effect” in the literature.11
This simple model suggests that, at our typical deposition
temperature of 1050°C, ZrO2 grains with a critical radius Ͻ35 nm
(or ϳ70 nm in grain size) could be stable as tetragonal if the grains
were nucleated directly as the tetragonal phase. However, this
model should be used carefully for direct comparison to our
experimental observations because: (i) constraints, i.e., in the
matrix or by the fiber substrate, are ignored,16 (ii) grain size
distribution is not considered, and (iii) the surface free energy
changes at elevated temperature are not incorporated. As shown in
Fig. 2(b), in the CVD-ZrO2 coating after the 5 min deposition,
some grains with size of ϳ50 nm, which is much larger than the
calculated critical grain size 10 nm at room temperature, were still
observed with TEM. Also, t-ZrO2 grains with sizes Ͼ10 nm were
also found in the inner CVD-ZrO2 coating retained on fiber (Fig.
3(c)). It is possible that the transformation of these t-ZrO2 grains
could be locally “constrained” by the fiber substrate.
4
⌬G͑r͒ ϭ ␲r3⌬Gv ϩ 4␲r2͑␥m Ϫ ␥t͒
(1)
3
where r is the radius of a particle, ⌬Gv is the chemical energy
difference for the phase transformation per unit volume of an
infinite crystal, and ␥t and ␥m are the average surface free energy
for the t and m phases, respectively. ␥t and ␥m have been measured
by calorimetry for bulk polycrystalline specimens to be 0.77 J/m2
and 1.13 J/m2, respectively. Also, the surface free energy will not
change much with temperature, and the d␥t/dT is only approxi-
mately Ϫ10Ϫ4 J/(m2⅐°C).14,15 A critical value (rc), below which
the t phase can be stabilized, can then be determined as a function
of temperature (T) by setting ⌬G(rc) ϭ 0. Writing ⌬Gv as a linear
function of ⌬H, one finds:
Ϫ3
rc ϭ
͑␥t Ϫ ␥m͒
(2)
In addition to the size and constraint effect, there are also other
explanations for stabilization of the t and c phases at lower
temperature, including water vapor, domain boundaries, anionic
⌬H͑1 Ϫ T/Tb͒
Fig. 6. Raman spectra of CVD-ZrO2 deposited for 120 min: (a) cross
section and (b) surface. Tetragonal bands in spectrum (b) are labeled with
a “t.”
Fig. 7. Tensile failure load of fiber tows coated with CVD-ZrO2 as a
function of deposition time in comparison to that of flame-desized fiber
tows. The error bars represent standard deviation.

June 2002
Morphological Evolution and Weak Interface Development within CVD-ZrO2-Coated Hi-Nicalon™ Fiber
1567
vacancies favorable for t-nucleation, and topotactic crystalliza-
tion.13 For example, Tomaszewski et al.17 observed that a lower
oxygen partial pressure was responsible for the metastability of
undoped c-ZrO2 and t-ZrO2 particles dispersed in an alumina
matrix.
Based on our results to date, a possible mechanism for the
observed weak interface can be considered. Our observations
correlate the morphological evolution of the CVD-ZrO2 to the
development of the weak interface behavior, as schematically
illustrated in Fig. 9.
At initiation of the CVD process (Fig. 9(a)), ZrO2 nucleates on
the amorphous SiC fiber surface. Note that it takes several adatoms
to form critical nuclei because of the tremendous supersaturation
in typical CVD conditions.18 Once formed, some nuclei will
coalesce and grow via the arrival of adatoms produced by
heterogeneous chemical reactions at the deposition surface. Be-
cause of the size effect and fiber surface constraints, the initial
coating layer primarily consists of t nanocrystals, but also possibly
contains some m particles of ϳ50–100 nm in some areas.
Meanwhile, the formation of new ZrO2 nuclei is continuing at the
coating surface. We speculate that the ZrO2 deposited for the 5 min
deposition might have been at this stage.
Fig. 9. Schematic representation of the relationship between morpholog-
ical evolution and weak interface development observed within the
CVD-ZrO2, with increasing the deposition time from (a) to (f).
As the CVD deposition proceeds, normal grain growth takes
place within the ZrO2 layer, as illustrated in Fig. 9(b). First, t
nanocrystals coalesce to form larger grains, and some of them
grow to the critical grain size (ϳ70 nm from Eq. (2)). This grain
growth step is expected to be diffusion controlled. Above this
critical size, most t grains martensitically transform to the m phase.
Because the t-to-m transformation is accompanied by significant
volume dilation, the ZrO2 layer becomes compressively stressed.
Once a certain degree of the transformation process is accumu-
lated, the overall compressive stress developed in the ZrO2 layer
may be high enough to delaminate the transformed ZrO2 region
from the fiber surface during CVD and/or cooling.
The m grains in the delaminated side of the ZrO2 layer continue
to grow by consuming other neighboring grains in the lateral
direction. At the same time, new nuclei are being formed on the
deposition surface. The CVD-ZrO2 layer produced after 20 min
might have been at this stage. With increasing deposition time, the
average gain size of the m phase in the delaminated layer increases.
After 120 min, the coating morphology shown in Fig. 9(f)
emerges. It is evident from the TEM results that delamination
consistently occurs at a distance of ϳ50–100 nm from the fiber
surface and within 20 min deposition. It is interesting that the
delamination occurs within the ZrO2 coating rather than the
interface between the ZrO2 and the fiber. This can be attributed to
the transformation lag in the inner ZrO2 layer, where the t-to-m
phase transformation is constrained by fiber substrate. Such a
constraint may lead to a larger critical size for transformation of
the grains on the fiber surface. Consequently, the grains on the
fiber surface transform late or only partially, as shown in Fig. 3(c).
Once the delamination takes place, the ZrO2 grains in the inner
layer exhibit no further significant grain growth (Figs. 3 and 4).
Also, the grains remaining on the fiber surface are preferentially
oriented after 20 min of the deposition step and appear to reach a
stagnant columnar microstructure.19 Therefore, for these reasons,
most of the t-ZrO2 grains in the inner ZrO2 layer do not grow much
and consequently do not transform and delaminate even after 120
min. Meanwhile, the less physical contact between the ZrO2 grains
close to the delaminated regions also impedes the grain growth rate
of those grains. Those grains are thus smaller than those deposited
afterward.
Based on the above discussion, the weak interface behavior
associated with the ZrO2 fiber coating is primarily attributed to the
t-to-m phase transformation within the ZrO2 coating during the
CVD process (and/or possibly during cooling), which leads to
delamination within the coating layer. So the composite behavior
previously observed2 can be explained by the delamination within
ZrO2 coating, serving as the mechanism for the apparent weak
interface behavior. The CTE mismatch, which must have been
involved in the evolution of the CVD-ZrO2 coating, was also
considered in this study. However, it seems that the CTE mismatch
couldn’t be the dominant effect controlling the weak interface
behavior. If CTE mismatch is dominant, the ZrO2 layer should
shrink much more than the Hi-Nicalon fiber substrate during
cooling, because ZrO2 has the biggest CTE in the system. This
shrinkage would eventually lead to the development of hoop
tensile stresses in the ZrO2 coating. Failure would then be by
microcraking rather than by delamination.
To serve as a robust fiber coating for SiC/SiC ceramic matrix
composites, there are other issues to be concerned with besides
the weak interface and retention of fiber strength. For example,
the integrity and uniformity of the fiber coating are important in
terms of protecting the fiber surface from matrix infiltration by
chemical vapor infiltration or melt infiltration. In this regard, an
undelaminated but compressively stressed coating might be
preferred before the matrix infiltration. We are currently
studying how to proactively control the nucleation behavior of
CVD ZrO2 and, consequently, the extent of the phase transfor-
mation in the coating layer for optimizing various interfacial
properties including delamination and sliding resistance of
SiC/SiC minicomposites.
Fig. 8. In accordance with
a thermodynamic model proposed by
Chraska,13 the critical particle radius, below which the t phase can be
stabilized because of the lower surface energy of the t phase over that of the
m phase, increases with temperature.

1568
Journal of the American Ceramic Society—Li et al.
Vol. 85, No. 6
2W. Y. Lee, E. Lara-Curzio, and K. L. More, “Multilayered Oxide Interphase
Concept for Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 81 [3] 717–20 (1998).
3E. H. Kisi and C. J. Howard, “Crystal Structure of Zirconia Phases and Their
Inter-Relation,” Key Eng. Mater., 153–154, 1–36 (1998).
V. Conclusions
Our results suggest that the tetragonal-to-monoclinic (t-to-m)
phase transformation within the CVD-ZrO2 layer is most likely the
key mechanism responsible for the apparently weak interface
behavior previously observed with SiC/SiC minicomposite speci-
mens containing a SiO2/ZrO2/SiO2 multilayer coating. A simple
thermodynamic model, which describes the size effect in the t-to-m
phase transformation, was used to understand morphologic evolu-
tion observed for the ZrO2 layer as a function of deposition time.
The morphological evolution was related to the development of a
delamination within the ZrO2 layer. The delamination location was
within the ZrO2 layer, but ϳ50–100 nm from the fiber surface.
The delamination appears to result from the following chain of
events: (i) continuous formation of t-ZrO2 nuclei on the deposition
surface; (ii) diffusion-controlled grain growth; (iii) the tetragonal-
to-monoclinic martensitic transformation of grains on reaching
their critical size, above which the tetragonal phase is no longer
stable; and (iv) development of significant compressive hoop
stresses because the volume dilation associated with the transfor-
mation. Our observations suggest that it will be of critical
importance to further understand and eventually control the nucle-
ation and grain growth behavior of CVD ZrO2 and its phase
transformation behavior for effectively assessing the potential of
CVD-ZrO2 as replacement for the boron nitride fiber coating for
SiC/SiC composites.
4S. J. Klepeis, J. P. Benedic, and R. M. Anderson, “A Grinding/Polishing Tool for
TEM Sample Preparation,” Mater. Res. Soc. Symp. Proc., 115, 179–84 (1988).
5M. J. Lance, J. A. Haynes, M. K. Ferber, and W. R. Cannon, “Monoclinic Zirconia
Distributions in Plasma-Sprayed Thermal Barrier Coatings,” J. Therm. Spray Tech-
nol., 9 [1] 68–72 (2000).
6A. Christensen, and EA Carter, “First Principles Study of the Surfaces of
Zirconia,” Phys. Rev. B, 58, 8050 (1998).
7J. I. Lee, H. Li, M. Libera, W. Y. Lee, H. Wang, and G. N. Morscher, “Role of
First SiO2 Layer in a Multilayer SiO2/ZrO2/SiO2 Fiber Coating for SiC/SiC Ceramic
Composites”; pp.15–26 in Advances in Ceramic Matrix Composites VI, Ceramic
Transactions, 124. Edited by J. P. Singh, N. O. Bansal, and E. Ustundag, 102nd
Annual Meeting of The American Ceramic Society, St. Louis, Missouri, 2001.
8W. M. Kriven, “Possible Alternative Transformation Tougheners to Zirconia:
Crystallographic Aspects,” J. Am. Ceram. Soc., 71 [12] 1021–30 (1988).
9J. Minet, F. Langlais, and R. Naslain, “On The Chemical Vapour Deposition of
Zirconia from ZrCl4–H2–Ar Gas Mixture: II, An Experimental Approach,” J.
Less-Common Met., 132, 273–87 (1987).
10M. Morstein, I. Pozsgai, and N. D. Spencer, “Composition and Microstructure of
Zirconia Films Obtained by MOCVD with a New Liquid, Mixed Acetylacetonato-
Alcoholato Precursor,” Chem. Vap. Deposition, 5, [4] 151–58 (1999).
11R. C. Garvie, “The Occurrence of Metastable Tetragonal Zirconia as a Crystallite
Size Effect,” J. Phys. Chem. Solids, 69, 1238–43 (1965).
12R. C. Garvie, “Stabilization of the Tetragonal Structure in Zirconia Microcrys-
tals,” J. Phys. Chem. Solids, 82, 218–24 (1978).
13T. Chraska, A. H. King, and C. C. Berndt, “On the Size-Dependent Phase
Transformation in Nanoparticulate Zirconia,” Mater. Sci. and Eng. A, A286, 169–78
(2000).
14H. F. Holmes, E. L. Fuller Jr., and R. B. Gammage, “Heats of Immersion in the
Zirconium Oxide-Water System,” J. Phys. Chem., 76, 1497–502 (1972).
15D. T. Livey and P. Murray, “Surface Energies of Solid Oxides and Carbides,”
J. Am. Ceram. Soc., 29, 363–72 (1956).
Acknowledgments
16F. F. Lange, “Transformation Toughening: Part 1, Size Effects Associated with
the Thermodynamics of Constrained Transformation,” J. Mater. Sci., 17, 225–34
(1982).
We are grateful to Dr. B. Greenberg at Stevens Institute of Technology for his
discussion on crystallography for the EMS simulations.
17H. Tomaszewski and K. Godwod, “Influence of Oxygen Partial Pressure on the
Metastability of Undoped Zirconia Dispersed in Alumina Matrix,” J. Eur. Ceram.
Soc., 15, 17–23 (1995).
References
18J. Bloem, “Nucleation and Growth of Silicon by CVD,” J. Cryst. Growth, 50,
581–604 (1980).
1D. W. Johnson, A. G. Evans, R. W. Goettler, M. P. Harmer, J. Lipowitz, K. L.
Luthra, P. D. Palmer, K. M. Prewo, R. E. Tressler, and D. Wilson, Ceramic Fibers
and Coatings: Advanced Materials for the Twenty-First Century, National Materials
Advisory Board, National Academy Press, Washington, DC, 1998.
19C. V. Thompson, “Grain Growth in Thin Films,” Annu. Rev. Mater. Sci., 20,
245–68 (1990).
Ⅺ