Journal of Alloys and Compounds 790 (2019) 1119e1126

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journal homepage: http://www.elsevier.com/locate/jalcom

Designing oxidation resistant ultra-high temperature ceramics

through the development of an adherent native thermal barrier
Gaoyuan Ouyang a, b, *, Matthew F. Besser b, Matthew J. Kramer a, b, Mufit Akinc a, b,
Pratik K. Ray a, b
a
Department of Materials Science and Engineering, Iowa State University, Ames, IA, 50011, USA
b
Ames Laboratory, US-DOE, Ames, IA, 50011, USA

a r t i c l e i n f o a b s t r a c t

Article history: We present a design concept for developing ZrB2-SiC-AlN composites with enhanced oxidative stability
Received 1 January 2019 at ultra-high temperatures (~2000
C) and low pressures (100 Torr). The oxidative stability of these
Accepted 17 March 2019 materials arises from a protective silica based scale. However, active oxidation of SiC above 1700
C
Available online 20 March 2019
presents a challenge, which we circumvent through the in-situ growth of a zirconia layer that serves as a
thermal barrier, ensuring that the effective temperature at the zirconia/Si rich subscale is less than the
Keywords:
active oxidation temperature. The design concept is validated by a series of ultra-high temperature
Oxidation
oxidation experiments under static as well as cyclic conditions.
Ultra-high temperature ceramics (UHTCs)
Thermal cycling
© 2019 Elsevier B.V. All rights reserved.
Silica

1. Introduction operational temperatures of Ni-based superalloys are limited by the

formation of low melting topologically close-packed phases, as well
Thermal protection systems in hypersonic vehicles call for a as a loss of creep resistance above 900
C. As these alloys approach
unique combination of properties required for meeting the chal- their limits, the range of operational temperatures is expanded via
lenges of harsh environments. At extreme velocities (mach 7 or active cooling and thermal barrier coatings. We recognize that the
higher), the temperatures at the leading edges of hypersonic air- ZrB2-SiC composites have the potential for an in-situ natively grown
crafts can often exceed 1800
C as a result of aerodynamic heating thermal barrier, in the form of zirconia. Underneath this “native”
during the re-entry phase. At stratospheric conditions, the oxygen thermal barrier, the presence of a silica rich scale will provide the
pressures are extremely low, and such high enthalpy conditions oxidation resistance. Zirconia, however, has relatively poor thermal
results in the dissociation, forming air plasma. Such conditions shock resistance. Failure of the zirconia through thermal shock,
require that the material system be (1) resistant to high tempera- especially under cyclic conditions, is therefore identified as a key
tures; (2) have a high thermal conductivity for conducting away the factor that limits the reusability of the material system. Our concept
heat and (3) have adequate oxidation resistance under static and is based around mitigating the shock by ensuring the presence of a
cyclic conditions, with the latter being especially critical for reus- subscale pliable enough to accommodate the thermal expansion
ability of the thermal protection system. Recent efforts have iden- and contraction of the overlying zirconia, yet tenacious enough to
tified ZrB2-SiC system to be a potentially viable candidate system retain the zirconia grains. This is achieved by appropriately
for these applications, especially in light of their high melting adjusting the viscosity of the underlying subscale, and builds on our
temperatures, and formation of a silica scale [1e4]. Nonetheless, as previous work where we demonstrated addition of AlN can result
the temperature exceeds 1700
C, active oxidation of SiC prevents in significant modifications of the viscosity [8,9]. Past efforts have
the formation of a protective scale [5e7]. indicated that addition of AlN results in improved densification of
In this paper, we propose a design concept for future develop- ZrB2-SiC composites under hot pressing conditions. Additionally,
ment of ZrB2-SiC composites, similar to Ni-based superalloys. The AlN is a high melting ceramic that has excellent thermal conduc-
tivity e which is certainly desirable in a thermal protection system.
In the present work, we investigate the oxidation behavior of a
* Corresponding author. Department of Materials Science and Engineering, Iowa number of ZrB2-SiC-AlN composites at temperatures higher than
State University, Ames, IA, 50011, USA. 1900
C at a pressure of 100 Torr under cyclic and static conditions
E-mail address: gaoyuan@iastate.edu (G. Ouyang).

https://doi.org/10.1016/j.jallcom.2019.03.250
0925-8388/© 2019 Elsevier B.V. All rights reserved.
1120 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126

in order to showcase our design concept. The experimental results ~14 mm inside the hollow holder) for each test. The front surface
corroborate our proposition, with the ultra-high temperature and the 6 mm length side surface were included in the surface area
oxidation resulting in the formation of a multilayered scale calculation. All tests started at an identical standoff to maintain an
comprising of an external ZrO2 layer and a silica rich subscale. The identical initial heating rate. The standoff was adjusted manually
stability of subscale, especially under cyclic conditions, indicate during the test to maintain 2000 (±50)
C for 10 min. For the cyclic
that the active oxidation has been arrested, which lends credence tests, the samples were cooled in a vacuum for 30 min with the
to our hypothesis that the primary role of the polycrystalline ZrO2 plasma gun off before it was removed for measurement and in-
in the oxidation of these materials is to provide a thermal barrier spection, and then reloaded for the next cycle. The temperature was
layer leading to a drop in the temperature at the internal ZrO2/ recorded during testing via ModView software (Ircon Inc., Santa
subscale interface. Cruz, CA) at 1.5 s intervals. The temperature profile for ZS73 cyclic
testing is shown in Supplemental Information Fig. S2. All other tests
2. Experimental details maintained similar temperature profiles. The sample mass was
measured after each cycle. The sample surface was analyzed by
Commercially-available powders were used as raw materials: SEM (Quanta-250, FEI Company, Hillsboro, Oregon) and XRD (Phi-
ZrB2 (Grade B, ~2 mm particle size, H.C.Starck, Karlsruhe, Germany) lips PANalytical, Almelo, The Netherlands), the cross-section was
SiC (Grade UF-10, ~1 mm particle size, H.C.Starck, Karlsruhe, Ger- examined by stereo microscope (SZX12, Olympus Co., Tokyo, Japan)
many), and AlN (Grade C, ~1 mm particle size, H.C.Starck, Karlsruhe, with a digital camera (DP21, Olympus Co., Tokyo, Japan), SEM and
Germany). Nominal compositions of ZrB2-30 vol%SiC (ZS73), ZrB2- EDS (Oxford Aztec, Oxford Instruments, Abingdon, United
30 vol%SiC-10 vol%AlN (ZSA631), ZrB2-20 vol%SiC (ZS82), ZrB2- Kingdom). The XRD patterns were obtained using Cu Ka radiation
20 vol%SiC-10 vol%AlN (ZSA721) were prepared and studied in this in Bragg-Brentano reflection geometry.
paper. Samples were prepared using a method described in our
previous paper [8], in accordance with the approach adopted from
Zhang et al. [10]. Powders were mixed and milled in plastic jars
with methyl ethyl ketone using tungsten carbide as milling media. 3. Volatility diagram
Organic binder (QPAC40, polypropylene carbonate, Empower Ma-
terials Inc., New Castle, DE) was added in the milling procedure to To understand the thermal stability of each phase, a volatility
increase the density of the green body. Binder-coated powders diagram was constructed. The chemical reactions involved are
were harvested from the mixture using roto-evaporation followed tabulated in Table 1 where chemical reactions for boron and silicon
by grinding and sieving through 44 mm screen. Cylindrical com- were cited from the literature [4,5]. Thermodynamics data and
pacts with a diameter of ~12 mm and height of ~20 mm were uni- equilibrium constants were extracted from the NIST-JANAF tables
axially pressed followed by cold isostatic pressing at 310 MPa. [12]. The volatility diagram in Fig. 1a covers the species with the
Binder removal was completed at 600
C in flowing argon atmo- highest vapor pressure at each oxygen partial pressure PO2.
sphere prior to the sintering. A sintering temperature of 2000
C As shown in the volatility diagram, the significant vapor species
was used for 30 vol% SiC samples, while 2200
C was used for 20 vol in this system are B2O3, SiO, and Al2O. Therefore, a separate graph of
% SiC samples. Both sintering profiles included two 1-h isothermal vapor pressure containing these species as a function of tempera-
holds at 1250 and 1450
C in vacuum to remove the surface oxides. ture is calculated and plotted in Fig. 1b. B2O2 is also considered,
The sintering was done in a resistively heated furnace with graphite since it has comparable vapor pressure to B2O3 at the ZrB2-ZrO2
heating elements (3060-FP20, Thermal Technology Inc., Santa Rosa, interface and has a higher vapor pressure than B2O3 at the lower
CA). The sintered densities were measured according to ASTM oxygen partial pressure regime. ZrO is the most dominant vapor
standard B962-15 [11] using Archimedes principle. Relative density specie among the Zr species, so it is also plotted on the diagram.
was expressed as a fraction of calculated theoretical density for SiO2 is also considered, due to its high vapor pressure at high ox-
each composition. ygen partial pressures. Vapor pressures of B2O3 and SiO2 are in
A plasma gun (SG-100, Praxair Surface Technologies, Indian- equilibrium with their corresponding liquid and independent of
apolis, IN), with a power input of 40 kW was used for heating the PO2. Since SiO has the highest vapor pressure at Si-SiO2 interface
test samples to the target temperature (2000
C). Argon (flow rate: [13], PO2 at this interface was utilized. A similar approach was used
80 SCFH (standard cubic feet per hour)) was used as the plasma gas for ZrO, Al2O, and B2O2 to determine the PO2 needed. The calculated
and helium (flow rate: 44 SCFH) was used as the auxiliary gas. A vapor pressures are in good agreement with the work of Opeka
ZrO2 coated graphite tube with a hollow push rod was utilized to et al. [13].
hold the test specimen. A separate vacuum line was used against The volatility diagram suggests that a silica based scale would be
the rear surface of the sample through the push rod to hold the unstable at the gas/solid interface, due to the loss of Si as SiO.
sample in place. A manual sliding sample holder system was Similarly, the vapor pressure of B2O3, as well as prior research on
designed to change the sample standoff (distance from the sample's the volatilization of B2O3 implies that B2O3 cannot remain at the
anterior surface to the plasma gun). The gun and sample holder gas/solid interface [14]. Therefore, the only possible oxides that can
were placed inside a vacuum chamber. A pressure of 100 Torr was be stable in the external scale is Al2O3 and ZrO2. However, ZrO2 has
used for the test with continuous air supply. The oxygen content an extremely low thermal conductivity, as a result of which the
was measured as 21% by an oxygen sensor (MAX-250E, Maxtec Inc, subscale temperatures would be substantially lower leading to the
Murray, UT), whose probe was placed ~1 inch from the sample front presence of a silica rich scale deeper in the composite. This has in
surface. Sample surface temperature was measured by a two-color fact been observed by a number of researchers. Hence, the volatility
optical pyrometer (Modline 5R-3015, Ircon Inc., Santa Cruz, CA). The diagram suggests that the oxidation resistance of this material can
pyrometer was calibrated using type R/S thermal couple (Omega be improved if the ZrO2 layer in the oxide scale can be preserved.
Engineering Inc., Stamford, CT) and black body with the accuracy of While, under isothermal conditions, this would not be problematic,
the measurements as ± 10
C. The layout of the experimental setup cyclic conditions would induce thermal shocks. A thermal shock
is shown in Supplemental Information Fig. S1. damage mitigation mechanism is therefore essential to prevent
All compositions were subjected to 10-min one and five-cycles ZrO2 layers from spalling off, which is our focus in the remainder of
testing. The sample was place 6 mm above the holder (with the paper.
 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126 1121

Table 1
Reactions used in ZrB2-SiC-AlN system.

Reactions with ZrB2 as condensed phase Reactions with B2O3 as condensed phase

ZrB2 (cr) þ3O2 (g) / ZrO2 (cr)þ2BO2 (g) B2O3(l) / B2O3(g)

ZrB2 (cr) þ 2.5O2 (g) / ZrO2 (cr)þ B2O3 (g) B2O3(l) þ 0.5 O2(g) / 2BO2(g)
ZrB2 (cr)þ2O2 (g) / ZrO2 (cr)þB2O2 (g) B2O3(l) / B2O2 (g) þ 0.5O2 (g)
ZrB2 (cr)þ2O2 (g) / ZrO2 (cr)þ2BO (g) B2O3(l) / 2BO(g) þ 0.5O2 (g)
ZrB2 (cr) þ 1.5O2 (g) / ZrO2 (cr)þB2O (g) B2O3 (l) / B2O (g) þ O2 (g)
ZrB2 (cr)þO2 (g) / ZrO2 (cr)þB2 (g) B2O3 (l) / B2 (g) þ 1.5O2 (g)
ZrB2 (cr)þO2 (g) / ZrO2 (cr)þ2B (g) B2O3 (l) / 2B (g) þ 1.5 O2 (g)
Reactions with SiC as condensed phase Reactions with SiO2 as condensed phase
SiC (cr) þ 1.5O2 (g) / SiO2 (g) þ CO (g) SiO2 (l) / SiO2 (g)
SiC (cr) þ O2 (g) / SiO (g) þ CO (g) SiO2 (l) / SiO (g)þ 0.5O2 (g)
SiC (cr) þ 0.5O2 (g) / Si (g) þ CO (g) SiO2 (l) / Si (g)þ O2 (g)
Reaction with AlN as condensed phase Reaction with Al2O3 as condensed phase
AlN (cr) þO2 (g) /AlO (g) þ NO (g) Al2O3 (cr) / 2AlO (g) þ 0.5O2 (g)
AlN (cr) þ1.5O2 (g) /AlO2 (g) þ NO (g) Al2O3 (cr) þ0.5 O2 (g) / 2AlO2 (g)
AlN (cr) þ0.5O2 (g) /0.5Al2 (g) þ NO (g) Al2O3 (cr) / Al2 (g) þ 1.5O2 (g)
AlN (cr) þ0.75O2 (g) /0.5Al2O (g) þ NO (g) Al2O3 (cr) / Al2O (g) þ O2 (g)
AlN (cr) þO2 (g) /0.5Al2O2 (g) þ NO (g) Al2O3 (cr) / Al2O2 (g) þ 0.5O2 (g)

Fig. 1. a. ZrB2-SiC-AlN volatility diagram; b. Vapor pressure of relevant vapor species as a function of temperature.

4. Results and discussion surface after a 10-min oxidation at 2000
C. The bright phase rep-
resents ZrO2. The SEM observations are consistent with the XRD
All samples achieved a relative density of 93% or higher. Fig. 2 patterns of the surface, as only ZrO2 (confirmed by EDS) is seen on
shows the microstructures of the sintered samples. SiC or a com- the ZS73 sample surface, while ZrO2 coexists with a darker phase in
bination SiC and AlN (dark phase) was well distributed in the ZrB2 other compositions. Most the surface structures of ZS73 were
matrix (bright phase). It has been reported in the literature that comprised of discrete clusters of ZrO2. Large channels (or pores)
addition of SiC results in improved densification of ZrB2-SiC com- were present in between the ZrO2 clusters. ZSA631 shows a
posites [15]. Consequently, a higher sintering temperature (2200
C completely different microstructure compared to the other three
as opposed to 2000
C for the ZS73 and ZSA631) was required for samples, with the surface showing a fine lamellar morphology. EDS
samples with lower SiC content (ZS82 and ZSA721) in order to analyses indicated the presence of Al, Si, and oxygen in the surface
attain significant densification. Higher sintering temperature scale. Most the ZS82 surface reveals ZrO2 agglomerates and the Si-O
possibly accounts for a coarser grain structure of ZS82 and ZSA721 rich phase as shown in Fig. 4c. ZSA721 shows similar microstruc-
in comparison to ZS73 and ZSA631. Since AlN and SiC have very tures as compared to ZS82 with a ZrO2 (bright phase) and regions
similar Z contrast, it is not possible to distinguish these two phases rich in Al, Si, and O (dark phase). The porous ZrO2 scale formed in
with backscattered electron imaging. Both of the two phases ZS73 (Fig. 4a) cannot be expected to provide adequate oxidation
appeared dark in the backscattered images. Fig. 2(e) and (f) show resistance over longer time intervals, due to easy diffusion paths for
the elemental Al distribution acquired using EDS. Comparing EDS oxygen through the pores and boundaries. The ZSA631 shows a
maps in 2(e) and (f) between 2(b) and (c), Al is present in only a significant amount of silica on the surface. Active oxidation of silica
portion of the dark contrast phase features. The remaining of the above 1700
C would render such a scale unusable as well. ZSA82
dark phase has high Si content. AlN and SiC phases are thus and ZSA721, on the other hand, show a mix of ZrO2 and SiO2.
believed present in separate phases. The difference in size of the Furthermore, these samples seem to have a denser surface as
AlN features also form the same trend as ZrB2 phases. compared to ZS73.
Oxidation at ~2000
C for 10 min resulted in the development of The mass change and the oxide scale thickness data after the 10-
a yellowish oxide layer. XRD patterns from the sample surface min oxidation are presented in Table 2. ZS73 and ZSA631 lose sig-
showed the formation of a ZrO2 layer, with small amounts of SiO2 nificant amount of mass, while ZS82 gains mass. Mass change for
(crystoballite) as seen in Fig. 3. Fig. 4 shows the SEM images of the the ZSA721 sample is minimal. This agrees well with the analytical
1122 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126

Fig. 2. SEM micrographs of sintered samples: a. ZS73, b. ZSA631, c. ZSA721 d. ZS82; and Al elemental EDS maps: e. ZSA631, f. ZSA721. All six images share the same magnification.

in Fig. 5b (ZSA631) and 4d (ZSA721). The ZSA631 sample shows an

Al-rich top layer.
Thermodynamically-stable oxides for the ZrB2-SiC-AlN system
include Al2O3, SiO2, B2O3, and ZrO2. Destabilization of the stable
oxides is possible only at extremely low oxygen pressures. For
instance, the destabilizing of these oxides at 2027
C (2300 K) re-
quires an oxygen pressure below 1010 atm. The stability of the
oxides also depends on their relevant vapor species. For instance, as
shown in Fig. 1b, an increase in the temperature from 1527 to
2027
C (1800e2300 K) results in more than two orders of
magnitude increase in B2O3 vapor pressure. At the current test
temperature, B2O3 is expected to evaporate significantly, once it
forms. During the experiment, a green fluorescence was observed
at the initial stage of the plasma exposure, but diminished after a
few minutes. This green fluorescence emission corresponds to
volatilization of the boron species [18]. EDS further confirmed bo-
ron is largely absent from the surface. SiO2 has a sufficiently low
vapor pressure that it is not likely to vaporize as gas phase SiO2.
Fig. 3. XRD pattern of samples surfaces after 10 min oxidation in plasma. However, SiO2 is prone to volatilization during active oxidation
according to the following reaction [19]:

prediction from Parthasarathy et al. [16] that evaporation of SiO is SiO2 (l) / SiO (g) þ0.5O2 (g) (1)
significant above 1600
C, which leads to a large mass loss, as well
as a more severe mass loss in samples having a higher SiC content. This reaction is accompanied by a decrease in mass. The high
The sample, ZS73, shows the highest, ZS82 shows the minimum, vapor pressure of SiO is likely to result in a large depletion of the Si
while ZS631 and ZSA721 are comparable in overall scale thickness. species at high temperatures. The highest vapor pressure in Al
Savino et al. [17] reported scale thickness around 150 mm after species is more than one order of magnitude lower than Si species.
6.4 min of arc jet exposure around 1780 to 1810
Cdthe same order Therefore, Al should be less prone to depletion in comparison with
of magnitude compared to current samples. Given the higher Si. At the testing temperature (~2027
C), the Al2O3 vapor pressure
testing temperature and long exposure time as well as higher Si is of the order of 103 atm [20], while the oxygen partial pressure
content, it is reasonable the scale thickness of the current samples needed for formation of Al2O is less than 1011.8 atm. The oxygen
is higher. partial pressure at ZrB2-ZrO2 interface at 2027
C, based on Fig. 1a,
The cross-section SEM and corresponding EDS maps are shown is 1010.43 atm, higher than the oxygen partial pressure of Al2O.
in Fig. 5. These results are consistent with the results provided in Hence, thermodynamically, Al2O3 would form instead of Al2O (the
Table 2, with the maximum scale thickness observed in the ZS73 formation of some Al2O is possible, due to uncertainties in tem-
sample and the minimum observed in the ZS82 sample. In all perature and other variables). ZrO2 is unlikely to volatilize from the
samples, the scale primarily consists of ZrO2. Si completely depletes surface, since its vapor ZrO reaches only 105.58 atm at 2027
C
in ZS73, while still is present in ZSA631, ZS82, and ZSA721. ZSA631 (2300 K). Therefore, the thermodynamically favorable overall sur-
shows the highest Si amount. Elemental distribution of Al is shown face oxides would be ZrO2 and possibly some SiO2 and Al2O3. Not
 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126 1123

Fig. 4. Surface micrographs of 10-min tests samples. Micrographs a-d shows the representative features of the samples ZS73, ZSA631, ZS82, ZSA721 respectively. All four images
share the same magnification.

Table 2 surprising, both ZrO2 and SiO2 are observed in the XRD pattern and
Mass change data and oxygen scale thickness for samples after 10 min tests.
SEM micrographs (except for ZS73, which shows the presence of
Sample Mass change (mg/cm2) Scale thickness (mm) ZrO2 only). The absence of SiO2 in the diffraction pattern of ZS73 is
ZS73 11.3 498 ± 42 consistent with the surface micrographs of ZS73 (Fig. 4a), which
ZSA631 10.1 287 ± 55 shows a preponderance of ZrO2 clusters, but no visible SiO2 glassy
ZS82 10.7 262 ± 62 outer scale. The EDS map (Fig. 5a) suggests a large oxygen-affected
ZSA721 0.1 325 ± 82
region and indicates the presence of SiO2 at depths below 300 mm
from the surface; hence, no detection by x-rays. On the other hand,
ZS82 (Fig. 5c) has a thinner scale. As a result, the SiO2 is sufficiently
close to the surface and observed in the corresponding XRD

Fig. 5. Cross-section SEM and EDS maps of 10-min tested samples. (a) ZS73, (b) ZSA631, (c) ZS82, (d) ZSA721. The first column on the left shows the back scattered electron images.
The elements mapped in EDS maps are indicated below the EDS maps. The arrow shows the depth of oxygen affect region. All four images share the same magnification.
1124 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126

pattern. ZSA631 and ZSA721 (Fig. 5b and d, respectively) show the resistance. Oxidation occurs rapidly yet again, following the spall-
presence of Si and Al oxides at the surface. ation event, until a sufficiently thick external ZrO2 layer forms yet
All samples display a net mass loss after cyclic tests. The mass again. Therefore, the formation of multiple interfacial regions is
change is accompanied by oxide scale spallation, with the oxide likely due to the scale spallation occurring due to thermal shocks in
layer spalling completely in ZS82 (Supplemental Information different cycles.
Fig. S3), possibly because of thermal shock and high stresses, due The cross-section microstructure for ZSA631 (Fig. 6b) differs
to the subsonic plasma stream. Since the entire oxide scale for ZS82 significantly from that of ZS73. The outer layers of the scale for
spalled off after the second cycle, the sample was not subjected to ZSA631 are significantly more porous than the other samples. The
further cycling. Other samples show varying degrees of oxide scale inset shows a higher magnification image of the interfacial region
spallation. Nonetheless, none showed a complete spallation of the between the external oxide scale and the base alloy. EDS analyses
scale as observed in ZS82. Mass changes reflect the combined ef- (Fig. 7a) in this region confirmed the presence of Al-rich alumino-
fects of spallation and oxidation. Hence, cross-section microstruc- silicate (dark contrast) along with ZrO2 (bright contrast). ZSA631
tures, rather than mass measurements, provide a clearer picture of shows the maximum mass change (and recession) during testing
the oxidation process. with the sample progressively ablated. Unlike the ZS73 sample, the
The cross-section micrographs at the end of the cyclic oxidation external scale for ZSA631 is not comprised exclusively of zirconia
tests are presented in Fig. 6, with all the sections obtained from the with a fair amount of aluminosilicate with high aluminum content
surfaces exposed directly to plasma heating. For ZS73, Fig. 6 (a) present as well. Fig. 6c shows the cross-section of the ZS82 sample
shows the scale spalls off in some regions, while remaining intact in after two cycles, where the external ZrO2 layer is completely
other regions. More importantly, an interfacial zone is observed in missing. The inset shows the side face of the sample (not directly
the sub-surface, underneath the regions from where the scale exposed to the plasma stream during testing). Clearly, the scale
spalled off. A closer look at this microstructure shows the presence spalls off from the exposed face and from a small region on the side
of two interfacial regions. Interestingly, the second interfacial re- of the sample. The scale is still present in the regions exposed to
gion is present underneath the region where the external scale has relatively lower temperatures and the scale sub-surface shows a
spalled off. According to our design rationale, the external ZrO2 acts degree of similarity with ZS73, albeit the volume fraction of ZrO2 is
as a thermal barrier, resulting in an effective temperature drop. considerably higher and the SiO2 considerably lower in comparison
Once the temperature is lower than the active oxidation tempera- with ZS73. This is possibly due to the higher Zr content of the
ture, silica/aluminosilica provides an oxidation resistant layer. pristine alloy in ZS82 as opposed to ZS73. Fig. 6d shows the cross-
Formation of a Si rich protective layer occurs by outward diffusion section microstructure of ZSA721. This sample does not show any
of Si [5], which results in the development of a Si depleted region. obvious scale spallation, but the external ZrO2 region is inter-
Once the external scale spalls, the thermal barrier effect is lost, and spersed with aluminosilicate. Note, the aluminum content in the
the silica is exposed to temperatures that rise above the active scale for the ZSA721 sample is much lower than for the ZSA631
oxidation temperature, which adversely affects the oxidation sample as shown in Fig. 7a. The inset shows a high magnification

Fig. 6. Cross-section micrographs of samples a. ZS73, b. ZSA631, c. ZS82, d. ZSA721 after cyclic tests (cross-section micrographs pertaining to ZS82 were recorded after two test
cycles). All four images share the same magnification.
 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126 1125

Fig. 7. EDS spectrum of the samples, (a) EDS spectrum of the surface gray region shown in Fig. 5, (b) EDS spectrum of the interphase gray region shown in Fig. 5.

image for the interfacial region between the base alloy and the accommodate the transformation stresses arising during the
oxide scale. EDS analyses reveal the interfacial chemistry is a reversible transformation from high temperature tetragonal
combination of silicon, aluminum, and oxygen, likely consisting of structure to monoclinic structure in ZrO2 associated with a volume
aluminosilicate as shown in Fig. 7b. Again, the aluminum content in increase with the volume strain of 4.7% at room temperature [24].
this interphase region is lower than for ZSA631. Such a layered structure provides a rationale for designing the
The presence of an external ZrO2 layer is likely to modify the UHTC chemistry. The scale structure as shown in Fig. 8 is somewhat
temperature profile as a function of depth from the surface. The analogous to that for a superalloy with a thermal barrier coating.
temperature drop across the external ZrO2 layer can be approxi- The amount of ZrB2 should be high enough to form a near exclusive
mately estimated using the heat conduction equation: external ZrO2 scale. The external ZrO2 acts as the thermal barrier,
since ZrO2 has a relatively low thermal conductivity
Q кAðT1  T2 Þ (~2.2 Wm1K1 [25]). This thermal barrier protects the underlying
¼ ; (2) material from high temperature damage. The interfacial region
t d
between the base alloy and the external scale functions similar to
where Q is the heat flow, in this case the enthalpy of the plasma jet; the thermally grown oxide, imparting oxidative stability to the
t is the time; к is the thermal conductivity of ZrO2; T1, T2 are the underlying material. Furthermore, columnar zirconia grains are
temperature on the sample surface and below ZrO2; and d is the “glued” together, if the interfacial chemistry is appropriate. Pure
thickness of the ZrO2 layer. The heat flux is expected to be close to SiO2 crystallizes at ultra-high temperatures. If crystalline SiO2 is
the range of 250 W/cm2, based on the extrapolated value from the present at the base of the columnar ZrO2, the damage tolerance will
heat fluxes and surface temperatures plotted from the literature be relatively poor. On the other hand, a glassy layer of aluminosil-
[21] using 2000
C surface temperature. The barrier layer is icate with relatively low viscosity will better accommodate the
considered solely ZrO2. Under these conditions, using the measured stresses during thermal cycling. Hence, a proper balance of ZrB2,
thicknesses of ZrO2 layers from the microstructures, the tempera- SiC, and AlN is essential for designing a damage-tolerant, robust
ture at the inner face of the ZrO2 layer can be estimated to be below UHTC. ZS82 suffered during cyclic tests, due to the presence of
1800
Cdjust below the active oxidation regime for SiC. excessive ZrO2, and insufficient strain tolerant glassy interfacial
Based on these results, we can now elucidate the oxidation
mechanism for ZrB2-SiC-AlN composites at ~2000
C and 100 Torr
pressure. This material oxidizes to form ZrO2, B2O3 (due to the
oxidation of ZrB2), SiOx, COx (due to the oxidation of SiC), Al2O3, and
NOx (due to the oxidation of AlN). At the test temperature, 2000
C,
B2O3 volatilizes. Similarly, due to active oxidation of SiC, a stable
silicate layer does not form. Consequently, the external oxide
comprises of ZrO2, which co-exists with a Si-depleted subscale
because of active oxidation. Across the ZrO2 scale, the temperature
eventually drops to 1700
C or lower. Under these conditions, a
stable silicate layer forms, which is modified by the presence of
Al2O3 in the AlN containing samples. Al2O3 modified alloys result in
lower viscosity of the silicate, allowing for faster coverage in this
region. This layer affords oxidation resistance. As seen in earlier
work [9], the Al2O3 content is quite critical. If the viscosity reduces
drastically (high Al2O3 content), this subscale silicate does not
afford sufficient tenacity to the overlying ZrO2. If the viscosity
reduction is insignificant (low Al2O3 content), the stresses due to
the mismatch in the coefficients of thermal expansion
(6.8  106 K1) for ZrB2 [22] and 10.8  106 K1 for ZrO2 [23] of
ZrO2 and the base material cannot be accommodated, thus leading
to scale spallation. Similarly, a low viscosity subscale may also fail to Fig. 8. Schematic of the scale structure after ultra-high temperature oxidation.
1126 G. Ouyang et al. / Journal of Alloys and Compounds 790 (2019) 1119e1126

layer to accommodate the thermal stresses. To a lesser extent, temperature composites over a wide range of SiC content, J. Eur. Ceram.
Soc. 32 (14) (2012) 3875e3883.
similar issues occurred with ZS73 (relatively lower ZrO2 content
[5] W.G. Fahrenholtz, Thermodynamic analysis of ZrB2-SiC oxidation: formation
somewhat mitigated the susceptibility to cyclic damage with of a SiC-depleted region, J. Am. Ceram. Soc. 90 (1) (2007) 143e148.
respect to ZS82). ZSA631 had a significant aluminosilicate interface, [6] X.H. Zhang, P. Hu, J.C. Han, Structure evolution of ZrB2-SiC during the oxida-
but lacked a near exclusive ZrO2 scale. In the presence of an tion in air, J. Mater. Res. 23 (7) (2008) 1961e1972.
[7] S.N. Karlsdottir, J.W. Halloran, Rapid oxidation characterization of ultra-high
adequate “thermal barrier,” the oxide scale in ZSA721 performed temperature ceramics, J. Am. Ceram. Soc. 90 (10) (2007) 3233e3238.
better under cyclic conditions, since a relatively high ZrB2 content [8] G. Ouyang, P.K. Ray, M.J. Kramer, M. Akinc, High-temperature oxidation of
ensured a ZrO2-rich external scale, while a high Al:Si ratio with ZrB2eSiCeAlN composites at 1600
C, J. Am. Ceram. Soc. 99 (3) (2016)
808e813.
sufficient SiC ensured the presence of an aluminosilicate interlayer [9] G. Ouyang, P.K. Ray, M.J. Kramer, M. Akinc, Effect of AlN substitutions on the
anchoring the external columnar ZrO2 grains. oxidation behavior of ZrB2eSiC composites at 1600
C, J. Am. Ceram. Soc. 99
(10) (2016) 3389e3397.
[10] S.C. Zhang, G.E. Hilmas, W.G. Fahrenholtz, Pressureless sintering of ZrB2-SiC
5. Conclusions ceramics, J. Am. Ceram. Soc. 91 (1) (2008) 26e32.
[11] ASTM Standard B962 - 15 Standard Test Methods for Density of Compacted or
A new design concept involving growth of layered thermal Sintered Powder Metallurgy (PM) Products Using Archimedes' Principle, West
Conshohocken, PA, 2015.
barrier coating and its stabilization for enhanced oxidative stability [12] M.W. Chase, NIST-JANAF Thermochemical Tables, fourth ed., American Insti-
for ultra high temperature ceramic has been proposed. Such tute of Physics, Washington, DC, New York, 1998.
approach mirrors the theme of nickel base super alloy, where the [13] M.M. Opeka, I.G. Talmy, J.A. Zaykoski, Oxidation-based materials selection for
2000
C þ hypersonic aerosurfaces: theoretical considerations and historical
alloy must develop an outer scale with low thermal conductivity
experience, J. Mater. Sci. 39 (19) (2004) 5887e5904.
and thermal stability to protect the base alloy from high temper- [14] T. Karahan, G. Ouyang, P.K. Ray, M.J. Kramer, Mufit Akinc, Oxidation mecha-
ature damage, and an interphase with proper viscosity to alleviate nism of W substituted Mo-Si-B alloys, Intermetallics 87 (2017) 38e44.
the thermal stress of this thermal barrier scale. ZrB2-SiC-AlN sys- [15] X.H. Zhang, P. Hu, J.C. Han, S.H. Meng, Ablation behavior of ZrB2-SiC ultra high
temperature ceramics under simulated atmospheric re-entry conditions,
tem were found to be of great potential following this design Compos. Sci. Technol. 68 (7e8) (2008) 1718e1726.
concept, where ZrO2 grows as the outer thermal barrier layer, and [16] T.A. Parthasarathy, R.A. Rapp, M. Opeka, M.K. Cinibulk, Modeling oxidation
anchored by the aluminum silicate scale whose viscosity can be kinetics of SiC-containing refractory diborides, J. Am. Ceram. Soc. 95 (1) (2012)
338e349.
tuned by varying Al:Si ratio. [17] R. Savino, M.D. Fumo, D. Paterna, A. Di Maso, F. Monteverde, Arc-jet testing of
ultra-high-temperature-ceramics, Aero. Sci. Technol. 14 (3) (2010) 178e187.
Acknowledgements [18] M. Playez, D.G. Fletcher, J. Marschall, W.G. Fahrenholtz, G.E. Hilmas, S.M. Zhu,
Optical emission spectroscopy during plasmatron testing of ZrB2-SiC
ultrahigh-temperature ceramic composites, J. Thermophys. Heat Transf. 23 (2)
The authors would like to acknowledge the funding support (2009) 279e285.
from AFOSR HTAM under contract # FA9550-11-1-201. [19] A. Rezaie, W.G. Fahrenholtz, G.E. Hilmas, Oxidation of zirconium diboride-
silicon carbide at 1500
C at a low partial pressure of oxygen, J. Am. Ceram.
Soc. 89 (10) (2006) 3240e3245.
Appendix A. Supplementary data [20] W.M. Haynes, CRC Handbook of Chemistry and Physics, CRC press, 2014.
[21] T.H. Squire, J. Marschall, Material property requirements for analysis and
design of UHTC components in hypersonic applications, J. Eur. Ceram. Soc. 30
Supplementary data to this article can be found online at
(11) (2010) 2239e2251.
https://doi.org/10.1016/j.jallcom.2019.03.250. [22] N.L. Okamoto, M. Kusakari, K. Tanaka, H. Inui, M. Yamaguchi, S. Otani, Tem-
perature dependence of thermal expansion and elastic constants of single
References crystals of ZrB2 and the suitability of ZrB2 as a substrate for GaN film, J. Appl.
Phys. 93 (1) (2003) 88e93.
[23] T.I.A. Kosolapova, Handbook of High Temperature Compounds : Properties,
[1] W.C. Tripp, H.H. Davis, H.C. Granham, Effect of an SiC addition on the oxida- Production, Applications, Hemisphere Pub. Corp., New York, 1990.
tion of ZrB2, Am. Ceram. Soc. Bull. 52 (8) (1973) 612e616. [24] M. Ruhle, A.G. Evans, High toughness ceramics and ceramic composites, Prog.
[2] F. Monteverde, A. Bellosi, Oxidation of ZrB2-based ceramics in dry air, Mater. Sci. 33 (2) (1989) 85e167.
J. Electrochem. Soc. 150 (11) (2003) B552eB559. [25] G.V. Samsonov, The Oxide Handbook, second ed., Wiley-VCH Verlag GmbH &
[3] W.G. Fahrenholtz, G.E. Hilmas, Oxidation of ultra-high temperature transition Co. KGaA, New York and London, 1982.
metal diboride ceramics, Int. Mater. Rev. 57 (1) (2012) 61e72.
[4] P. Williams, R. Sakidja, J. Perepezko, P. Ritt, Oxidation of ZrB2-SiC ultra-high