Materials Transactions, Vol. 46, No. 3 (2005) pp.

543 to 551
Special Issue on Fusion Blanket Structural Materials R&D in Japan
#2005 The Japan Institute of Metals

Evaluation of Tensile Properties of SiC/SiC Composites

with Miniaturized Specimens
Takashi Nozawa1; * , Yutai Katoh2 and Akira Kohyama1
1
Institute of Advanced Energy, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan
2
Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

Mechanical testing after neutron irradiation is a critical research tool for evaluating materials for fusion systems, such as silicon carbide
fiber silicon carbide matrix (SiC/SiC) composites. However, single-axis tensile testing, which is required to build a fundamental database,
requires large specimens. Therefore miniaturization of tensile test specimens has long been pursued as a method to reduce the irradiation volume
to fit the capsule size limitation. The objective of this study is to identify specimen size effects on tensile properties of SiC/SiC composites from
the viewpoints of the influences of fabric architecture and tensile loading axis, with a final goal to establish a small specimen test technique for
tensile testing of the composites. The axial fiber volume fraction plays an important role in achieving good tensile properties. However the size
dependent change of the axial fiber volume fraction gives specimen size effect. The composites with much fiber volume content tended to have
superior tensile strength, elastic modulus and proportional limit stress. Contrarily, the tensile properties of the composites with the same axial
fiber volume fraction were almost independent of the specimen size. This type of size effect is generally common in any types of architecture.
The size-relevant fracture mode in off-axis tension: detachment in shorter widths vs. in-plane shear at larger widths, also gives specimen size
effect on tensile properties, resulting in strict limitation of miniaturization of the tensile specimen. Finally we proposed a miniature tensile
specimen for the composites.

(Received September 21, 2004; Accepted February 15, 2005)

Keywords: ceramic matrix composites, small specimen test technique, size effect, tensile testing

1. Introduction Composite materials are composed of the constituent fiber,

SiC/SiC composites are candidate materials for fusion
blankets due to elevated temperature capability, low radio-
activity, and inherently good neutron irradiation tolerance.1,2)
Mechanical testing after neutron irradiation is a major
requirement in the research for fusion-grade SiC/SiC
composites. The flexural test has long been employed for
evaluation of neutron irradiated materials. However, the
complex fracture mode in flexure: mixture of tension,
compression and shear, limits this test to screening purposes
only. Simple fracture mode tests: loading in tension and
shear, are therefore receiving more recent attention. Single-
axis tensile test is a useful and fundamental test methodology
with a simple fracture mode. However, there is no established
tensile test technique for irradiation research on ceramic
matrix composites (CMCs). The strict requirement of the
comparatively large specimens in conventional tensile testing
prevents use in neutron irradiation experiments because of
the irradiation volume limitations. For this purpose, a small
specimen test technique has been developed as a simple and
useful testing method in the series of the international
standardization activities.3,4) Miniaturization is an effective
means to evaluate irradiated materials because of the reduced
radiological hazard potential. Small size also allows an
increase in the test numbers, beneficial to statistical reli-
ability. In addition, the small specimen test technique is
required to give maximum use of the lab-scale products with
very limited material volumes.
The ASTM standard C1275, familiar as a tensile testing
methodology of the composites, provides various types of
standard specimens with differed sizes and geometries.
*Present address: Metals and Ceramics Division, Oak Ridge National
Laboratory, Oak Ridge, TN37831, USA
matrix and the fiber/matrix (F/M) interphase, each with
finite scale. Therefore the specimen size of the composites is
often determined by the constituent such as a fiber bundle
composite. The ASTM standard specimens contain almost
48–120 bundles in the gauge section, considering the fiber
bundle with 1.5 mm width and 0.25 mm thickness. A tensile
specimen with <10 fiber bundles in the gauge section is our
recent goal to meet irradiation capsule requirements. How-
ever miniaturization of the specimen must not spoil the
inherent feature obtained from the characteristic composite
architecture. This is because the composite strength is
strongly dependent on the submicron-order constituents,
giving rise to specimen size effects.
Specimen size effects of the composites have been
investigated by both experimental and analytical ap-
proaches.5–16) Specifically the recent attention is paid on the
following factors: (1) inside sub-micron flaws, (2) micro-
structure, (3) stress gradient and (4) test considerations. This
technique cannot be established understanding of the influ-
ence of theses factors.
Sub-micron flaws induced in processing and machining are
primary factors that govern specimen size effects in ceramics
and quasi-brittle materials. The most familiar relationship
between the defects and their strength is the weakest link
theory. The failure strength of a uniformly stressed volume of
the brittle material is determined by the weakest defect,
among defects randomly distributed in materials. Specifically
the failure strength of the larger specimen tends to decrease
because the probability of flaws increases with volume. This
relationship of failure strength to specimen volume is well
described by Weibull statistics.17) The global load sharing
theory, based on the Weibull statistics, is often mentioned as
a good analytical model to predict composite strength
including the specimen scaling issue.5–7) This theory hy-
544 T. Nozawa, Y. Katoh and A. Kohyama
pothesizes that the redistribution of load is shared equally for
all intact fibers when fibers progressively fail. In this
hypothesis no stress concentration can occur. Under this
hypothesis, the composite failure strength can be propor-
tional to the axial fiber volume fraction. In contrast, the local
load sharing theory assumes a non-uniform redistribution of
load on the intact fibers.8,9) Under this condition, the stress
concentration around the broken fibers varies with the
specimen size, resulting in another size effect. It is reported
that failure strength decreases as laminate thickness decreas-
es, due to increasing stress concentration.8,9)
The microstructure of the composites is also important in
determining the defects that may give rise to a size effect. The
constituent fibers and matrices themselves are subject to
damage during processing and machining, producing the
defects responsible for ultimate composite failure. Specifi-
cally, degradation of the reinforcing fibers becomes a critical
issue to achieving high fracture strength because the fibers
must transfer almost all the applied load beyond first matrix
cracking. Brittle fibers such as carbon and glass exhibit a
strength size effect due to flaws in the microstructure of the
fibers; tensile strength increases with reduced fiber diameter
and gauge length, respectively.10,11) Fabric architecture is
also important to discuss mechanical properties of the
composites. The multi-dimensional weave can easily create
pockets where voids can be trapped or cracks can initiate.
Stitched composites have another set of defects built into
their microstructures. The stitching damages fibers and the
stitches themselves form defects where cracks can initiate.
The increasing defects raise the probability of composite
failure. In particular, these origins of the failure initiation are
dependent on the specimen size, because the architecture is
strictly restricted by the constituents with a finite size. The
fabric architecture also provides anisotropy of the compo-
sites. Silicon carbide fiber reinforced composites exhibit
anisotropic fracture behavior due to their discontinuous
fabric structure. (The SiC fiber itself never shows anisotropy
of fracture strength since the fibers have an isotropic
microstructure.) In general, tensile strength tends to decrease
with an increase in the angle between the fiber longitudinal
direction and the loading direction, because the resulting
reduction of the axial fiber volume fraction adds the shear
failure at the F/M interface.
A stress gradient is another factor that governs the
specimen size effect. In many tests, the stress field is not
constant. For example, in flexure there is a stress gradient
through the thickness. It has been shown theoretically from
the Weibull theory that the strength in flexure should be
higher than the strength in tension.12) The redistribution of
load that occurs as fibers progressively fail may produce a
higher strength in flexure than in pure tension.13) The effect of
the stress gradient on failure has also been used to explain the
variation in tensile strength of the notched composites as a
function of the notch size.14) Moreover, the stress gradient
will affect the stress field during compressive loading. The
compressive failure is in general caused by shear instability,
often referred to as microbuckling. The stress gradient will
affect the stress field where instability occurs.15,16)
Testing considerations are important technical issues in
discussion of the specimen size effect issue. As mentioned
previously, different test methods can give rise to other size
effects. Even if the same test method is employed, differences
in the thickness and the aspect ratio of the specimens can
provide premature failures due to unexpected bending
moment produced by misalignment. Moreover early failure
due to stress concentration near the grips may mask a size
effect.
In order to discuss mechanical properties of the compo-
sites, the influence of the tensile loading axis is specifically
recognized as an important issue because of anisotropic
features of the composites. The objective of this study is to
evaluate specimen size effect on the tensile properties of the
various architecture types of SiC/SiC composites by varied-
axis tensile tests.
2. Experimental
2.1 Materials
Specimen size effects on tensile properties were evaluated
using SiC/SiC composites fabricated by the polymer
impregnation and pyrolysis (PIP) process. Three types of
composites were produced: a plane-weave (P/W) composite,
an eight-harness satin-weave (S/W) composite and an
orthogonally woven three-dimensional (3D) composite. Both
P/W and S/W SiC/SiC composites had Z-directional fiber
stitching. The ratio of the fabric configuration of 3D SiC/SiC
composite was X:Y:Z ¼ 1:0:1:0:0:2. Here X and Y mean the
fiber longitudinal direction, while Z means the through-
thickness direction. Typically these composites had relative-
ly high porosity above 20%. They exhibited several large
pores inside the structure, primarily along the Z-stitches. The
Tyranno
-LoxM (Si–Ti–C–O) fibers used as reinforcements
had undergone a surface modification before weaving in
order to optimize the fiber-matrix interface.18) Due to this
treatment, an interphase with excess carbon was formed close
to the fiber surfaces. Details of materials are described
elsewhere.19,20)
2.2 Tensile Test
Room-temperature tensile tests were carried out using a
face-loaded rectangular tensile specimen (Fig. 1) following
the method of ASTM C1275. The main reason to use straight-
Y irectio
n
dinal d
longitu
Fiber
θ =0, 15, 30, 45°
W
X (loading axis)
L
Thickness: T
L+60 (Units: mm)
Valuables Constants
L=5.0, 15.0, 30.0, 45.0
3.0W x3.0T (P/W & S/W)
2.0W x2.0T, 3.0Wx2.0T, 3.0Wx3.0T (3D)
15.0Lx3.0T (P/W & S/W)
W=1.5, 3.0, 6.0, 9.0
15.0Lx2.0T, 15.0Lx3.0T (3D)
T=1.0, 1.5, 2.0, 2.5, 3.0 15.0Lx3.0W (All)
Fig. 1 Schematic illustrations of tensile specimens for specimen size effect
study.
 Evaluation of Tensile Properties of SiC/SiC Composites with Miniaturized Specimens 545

sided geometry is to prevent from the accumulation of 400

undesired machining flaws and also to avoid stress concen-

Tensile Stress, σ / MPa

tration at the root of these flaws. All tensile specimens were
machined using a diamond cutting device and machined
surfaces were polished to a surface roughness of
5 mm to
reduce machining flaws. The fiber direction was parallel to
the tensile loading axis for the axial tensile test or rotated off
15 , 30 and 45 for the off-axis tensile test. For simplicity,
we denote them as [0 /90 ], [15 /75 ], [30 /60 ] and [45 /
45 ], respectively. The gauge size of each specimen was
designed by the number of fiber bundles included in the
cross-section. At least one bundle (
1:5 mm/bundle) in
width and four bundles (
0:25 mm/bundle) in thickness
were included. Specimen gauge length was determined on the
basis of total specimen length, 15–45 mm (10–30 bundles),
which is expected for neutron irradiation tests. The straight-
sided ASTM standard specimen with the gauge volume of
L
30  W 6  T 3 mm3 (total size: L 100  W 6  T 3 mm3 ) was
used for comparison in the fiber longitudinal axis test. No
standard specimen was used in the off-axis tests due to very
limited volume of the supplied composites. All tensile
specimens were clamped by wedge-typed grips, and to avoid
gripping damage, 1.0 mm thick aluminum end tabs were
bonded on both sides of the specimen. Crosshead displace-
ment rate was 0.5 mm/min for all tests. The strain of the
composites was measured by averaging two readings of strain
gauges bonded to both faces of the center gauge section.
After tensile tests, fracture surfaces were analyzed by optical
and scanning electron microscopy to investigate the fracture
mechanism.
3. Results
3.1 Tensile Properties in Fiber Longitudinal Axis
Figure 2 shows typical tensile behaviors of P/W, S/W and
300 3D [0°/90°]
S/W [0°/90°]
200
P/W [0°/90°]
100 P/W [45°/45°]
3D [45°/45°]
0
0 0.5 1 1.5 2
Strain, ε / %
Fig. 2 Typical tensile stress–strain curves of P/W, S/W and 3D SiC/SiC
composites.
3D SiC/SiC composites. All types of SiC/SiC composites
exhibited non-linear segment beyond proportional limit due
to progressive accumulation of matrix cracks. Less elonga-
tion for P/W composites (0.4–0.7) and S/W composites
(0.4–0.6) was attributed to the structural restrictions caused
by weaving fiber bundles. Fracture surfaces for P/W SiC/SiC
composites showed fibrous appearances although they failed
with shorter fiber pullout lengths (Fig. 3). In contrast, 3D
SiC/SiC composites exhibited much longer fiber pullout
because of less restriction to the total elongation due to
straightly aligned fibers parallel to the loading direction.
Specifically fiber bundle pullout was characteristic in 3D
SiC/SiC composites. These failure patterns are almost kept
in entire tested specimen size range. For example, no
significant difference in tensile fracture appearances was
observed for 3D SiC/SiC composites with varied widths

(a) 2D (P/W) (b) 3D

500µm 500µm

50µm 50µm

Fig. 3 Typical tensile fracture surfaces of (a) 2D and (b) 3D SiC/SiC composites.
546 T. Nozawa, Y. Katoh and A. Kohyama

5.0 mm fibers perpendicular to the loading direction do not transfer

load once detachment occurs at the F/M interface.
Figure 5 exhibits specimen size effects on tensile strength,
elastic modulus and proportional limit stress (PLS). Tensile
results with percent bending larger than 5% were not
included in this figure. Upper and lower error bars in the
plots indicate the maximum and minimum values, respec-
tively. Tensile strength of both P/W and S/W composites
was nearly independent of the specimen size in the entire
1.5 mm in Width 3.0 mm in Width tested range, while the tensile strength of the 3D composites
was significantly dependent on the specimen length, width
and thickness: increased tensile strength for the composites
with the gauge size of L 15  W 3  T 1:5, L 15  W 3  T 2 and
L
15  W 3  T 3 mm3 . In contrast, elastic moduli of both 2D
and 3D composites were nearly independent of the specimen
size. The only exception was obtained for the 5.0-mm-long
3D composites; where the tensile modulus decreased
30%.
The 2D composites did not exhibit significant size effects on
tensile modulus. Similarly, the PLS was apparently inde-
6.0 mm in Width 9.0 mm in Width pendent of specimen gauge length, width and thickness for
(Length: 15.0mm, Thickness: 2.0mm) both 2D and 3D SiC/SiC composites. However, there was
significantly large scatter in the smaller gauge volume,
Fig. 4 Tensile fracture appearances of 3D SiC/SiC composites with varied especially for small gauge width (<3:0 mm) 3D composites.
gauge widths.
Tensile properties obtained from the standard specimen were
almost followed by these trends.

(Fig. 4). The only exception of the stress concentration
around stitching fibers is important in the 1.5-mm-wide case
for 3D SiC/SiC composites. A large stress concentration
should arise along the detached Z-stitches, resulting in
premature failure of the composites because the stitching
3.2 Tensile Properties in Varied Loading Axis
Figure 2 also shows typical off-axis tensile behaviors of P/
W and 3D SiC/SiC composites. Stress–strain curves for all
materials were mostly non-linear except for a very short
linear region in the initial stage of the curves. Proportional

3D (3W x3T) P/W (3W x3T) 3D (15Lx3T) P/W (15Lx3T) 3D (15Lx3W ) S/W (15Lx3W )
3D (3W x2T) P/W-std. (6W x3T) 3D (15Lx2T) P/W-std. (30Lx3T) 3D-std. (30Lx6W ) S/W-std. (30Lx6W)
3D (2W x2T) S/W (3W x3T) 3D-std (30Lx3T) S/W (15Lx3T) P/W (15Lx3W )
3D-std. (6W x3T) S/W-std. (6W x3T) S/W-std. (30Lx3T) P/W-std. (30Lx6W)
600
Tensile Strength,

500
400
σs / MPa

300
200
100
0
Elastic Modulus,

100
E / GPa

50

0
Proportional Limit
Stress, σmc / MPa

80
60
40
20
0
0 10 20 30 40 0 2 4 6 8 0 1 2 3 4
Gauge Length, L / mm Gauge Width, W / mm Gauge Thickness, T / mm

Fig. 5 Effects of the specimen size on tensile properties of P/W, S/W and 3D PIP SiC/SiC composites. Upper and lower error bars in the
plots indicate the maximum and minimum values, respectively.
 Evaluation of Tensile Properties of SiC/SiC Composites with Miniaturized Specimens
(a) W=1.5mm
1.0mm
(c) W=6.0mm
1.0mm
(L=15.0mm, T=3.0mm)
Fig. 6 Typical tensile fracture surfaces of P/W SiC/SiC composites with varied gauge widths in the 45 off-axis loading test.
limit stress and elastic modulus were very low due to
anisotropy of the composites. Beyond proportional limit,
accumulation of the strain was quite different between [0 /
90 ] and [45 /45 ] SiC/SiC composites. The latter trans-
ferred the load primarily by interfacial shear stress. Low
stiffness of interfacial deformation resulted in progressive
strain accumulation at the lower stress level. Tensile fracture
behavior of the [15 /75 ] and [30 /60 ] composites were
very similar to that of the [45 /45 ] composites.
Figure 6 shows typical fracture surfaces of the off-axis
tensile specimens for P/W SiC/SiC composites. In general,
all specimens exhibited a V-shaped fracture plane. Specif-
ically most fibers of the specimens with shorter gauge widths
were remained as intact. The detachment of the F/M
interface was characteristic in these composites. In contrast,
the specimens with larger widths (>6:0 mm) incorporated
with fiber failure and interfacial detachment. The 3D
composites also exhibited the similar trend. Contrarily, there
is no significant difference in fracture behaviors of the
specimens with differed lengths or thicknesses (with constant
width) for both P/W and 3D composites.
Figure 7 shows the effects of the specimen size on tensile
properties of P/W and 3D PIP SiC/SiC composites for
various directional loading tests. Tensile strength and PLS
decreased with increasing specimen widths for 2D and 3D
composites, although they never show significant length or
thickness effects. The elastic modulus was hardly dependent
on the specimen length, width and thickness. However severe
degradation was obtained for the 1.5 mm wide specimen in
the [15 /75 ] loading test of 3D composites.
4. Discussion
4.1 Specimen Size Effect on Tensile Properties in Fiber
Longitudinal Axis
In the ideal composite system, the tensile strength in the
547
(b) W=3.0mm
1.0mm
(d) W=9.0mm
X (Length)
Z (Thickness)
1.0mm
Y (Width)
fiber longitudinal direction is proportional to the axial fiber
volume fraction.5) This relation should hold true for 2D and
3D composite systems because the longitudinal fibers acting
alone can transfer the entire load following first matrix
cracking. We obtained a good relationship between the axial
fiber volume fraction and the tensile strength regardless of the
specimen size (Fig. 8). The large uncertainity of the axial
fiber volume fraction in a 3D structure can be easily obtained
by imaging its structural unit cell18) (Fig. 9). The presence of
the Z-stitches can separate a fiber rich region or a matrix rich
region using specimens smaller than a structural unit width of

2:0 mm, resulting in the large scatter of axial fiber content.
Similarly, the axial fiber volume fraction is also subject to
variation within the specimen thickness. However, there
might be a minor effect on the tensile properties in cases
where the specimen thickness is large enough to cover many
fiber bundles. The axial fiber volume fraction was nearly
same through entire thickness range. In contrast, the tightly
woven P/W and S/W composites can cause a small change
of the axial fiber volume fraction with respect to the specimen
width and thickness.
Here we defined a new parameter: normalized tensile
strength, which is the maximum tensile stress divided by the
axial fiber volume fraction. Normalized tensile strength
means the stress where the entire load is applied only to the
axial fibers. This new parameter will be good to evaluate the
behavior after the first matrix cracking of the composites.
Figure 10 plots the specimen width effect on normalized
tensile strength. This revealed no width effect for widths
>3:0 mm and no thickness effect for >1:5 mm, although a
slight reduction (
30%) of normalized tensile strength was
seen in shorter gauge widths or thicknesses. There is also no
apparent length dependence in the range tested. A slight
decrease in the tensile strength of 1.5-mm-wide and 1.0-mm-
thick specimens might be explained by the influence of
machining damage on the bare fibers near the specimen
548 T. Nozawa, Y. Katoh and A. Kohyama

200

Tensile Strength,
3D [15°/75°] 3D [45°/45°]
3D [30°/60°] P/W [45°/45°]
150
σ s / MPa 100

50

0
Elastic Modulus,

40
E / GPa

20

0
Proportional Limit
Stress, σ mc / MPa

30

20

10

0
0 10 20 30 40 0 2 4 6 8 0 1 2 3 4
Gauge Length, L / mm Gauge Width, W / mm Gauge Thickness, T / mm
(W=3mm, T=3mm) (L=15mm, T=3mm) (L=15mm, W=3mm)

Fig. 7 Off-axis tensile properties of P/W and 3D SiC/SiC composites with varied specimen width. Upper and lower error bars in the plots
indicate the maximum and minimum values, respectively.

800
15Lx1.5Wx2T
15Lx9Wx2T
15Lx3Wx2T
15Lx3Wx1T
15Lx6Wx2T
15Lx3Wx1.5T
III
15Lx3Wx2.5T 15Lx3Wx3T 5Lx2Wx2T
Tensile Strength, σs / MPa

600 30Lx3Wx2T 45Lx3Wx2T h1

h2
II
400

200 I e

0
b

10 15 20 25 30
x
Axial Fiber Volume Fraction, f x / vol% z
Fig. 8 Effect of the axial fiber volume fraction on tensile strength of 3D c
d

PIP SiC/SiC composites.

surfaces. Discontinuous fibers are ineffective to transfer the
applied load. Actual volume fraction of the intact fibers was
probably smaller than that used in normalization. Otherwise
this might be explained by the local load sharing analysis,
although further investigation is required. Recently exper-
imental and numerical analyses have been developed to
evaluate this phenomenon.8,9)
Tensile elastic modulus of the 3D composites is also
estimated by the simple analytical model based on the rule of
mixtures.18) This indicates that the tensile elastic modulus
exhibits monotonic increase with the axial fiber volume
fraction increasing.19,20) As similar to tensile strength, tensile
a y
Fig. 9 Schematic illustration of the structural unit cell in an orthogonal 3D
composite system.
elastic modulus exhibited large scatter in the gauge width of
1.5 mm. Figure 11 plots measured and predicted axial elastic
moduli relevant to the axial fiber volume fraction for 3D PIP
SiC/SiC composites. The size-related change of the axial
fiber volume fraction will possibly affect elastic modulus,
although this effect might be small. The primary mechanism
 Evaluation of Tensile Properties of SiC/SiC Composites with Miniaturized Specimens 549

(a) 150
3500 15Lx1.5Wx2T 15Lx3Wx2T 15Lx6Wx2T
Strength, σ s /fx / MPa

3D (3W x3T) P/W (3 W x3T)

3000 3D (3W x2T) P/W-std. (6 W x3T)
Normalized Tensile

15Lx9Wx2T 15Llx3Wx1T 15Lx3Wx1.5T

3D (2W x2T) S/W (3 W x3T)

Elastic Modulus, E / GPa

2500 3D-std. (6W x3T) S/W-std. (6 W x3T) 15Lx3Wx2.5T 15Lx3Wx3T 5Lx2Wx2T

2000 30Lx3Wx2T 45Lx3Wx2T

100
1500
1000
Em=30
500
Em=20
0 50
0 10 20 30 40 50 Em=10
Gauge Length, L / mm
(b) 3500
Strength, σs /fx / MPa

3D (15Lx3T) P/W (15 Lx3T)

Normalized Tensile

3000 3D (15Lx2T) P/W-std. (30 Lx3T) 0

3D-std (30 Lx3T) S/W (15 Lx3T)
2500 S/W-std. (30 Lx3T) 10 15 20 25 30
2000 Axial Fiber Volume Fraction, f x / vol%
1500
Fig. 11 Effect of the axial fiber volume fraction on tensile elastic modulus
1000
of 3D PIP SiC/SiC composites. Fitting curves were estimated by eq. (7) in
500 ref. 19 using characteristic lengths, a ¼ b ¼ 1:25 mm, c ¼ 0:75 mm,
0 d ¼ e ¼ 0:37 mm, h1 ¼ h2 ¼ 0:25 mm, fiber volume fraction of a fiber
0 2 4 6 8 10 bundle composite, f ¼ 0:7, and elastic moduli of the fiber and the matrix,
Gauge Width, W / mm Ef ¼ 187 GPa and Em ¼ 10{30 GPa, respectively.
(c) 3500
3D (15Lx3W ) S/W (15 Lx3W )
Strength, σs /fx / MPa

3000 3D-std. (30Lx6W ) S/W-std. (30 Lx6W )

Normalized Tensile

P/W (15 Lx3W ) 120

2500 P/W-std. (30 Lx6W )
15Lx1.5Wx2T 15Lx3Wx2T 15Lx6Wx2T
Proportional Limit Stress, σmc / MPa

15Lx9Wx2T 15Lx3Wx1T 15Lx3Wx1.5T

2000 100 15Lx3Wx2.5T 15Lx3Wx3T 5Lx2Wx2T
1500 30Lx3Wx2T 45Lx3Wx2T
1000 80
500
0 60
0 1 2 3 4
Gauge Thickness, T / mm 40

Fig. 10 Effects of specimen (a) length, (b) width and (c) thickness on
normalized tensile strength, defined as tensile strength divided by the axial
fiber volume fraction. Upper and lower error bars in the plots indicate the
maximum and minimum values, respectively.
that causes the reduced tensile modulus for the 5.0-mm-long
specimen might be due to width effect typical for the short
width (
2:0 mm) specimen.
Proportional limit stress also shows a function of the axial
fiber volume fraction.21) According to the analytical model
prediction, PLS tends to monotonically increase with
increasing fiber content. However the quantitative analysis
indicates that the average PLS was constant regardless of the
axial fiber volume fraction (Fig. 12), but with very large
scatter of individual data. The difference comes from the
large scatter of other parameters, which might mask the
specimen size effect on PLS.
4.2 Specimen Size Effect on Tensile Properties in Varied
Loading Axis
Off-axis tensile tests of the composite materials resulted in
low fracture stress and high elongation compared to the axial
tensile tests. In this case, the composite failure was
significantly dependent on the detachment strength between
fiber and matrix rather than only on the tensile strength of
fibers. This is apparently true according to fracture surfaces.
The V-shaped fracture appearance was characteristic of each
20
0
10 15 20 25 30
Axial Fiber Volume Fraction, f x / vol%
Fig. 12 Effect of the axial fiber volume fraction on proportional limit stress
of 3D PIP SiC/SiC composites.
surface. In-plane shear strength calculated from the off-axis
tensile data converged to a constant with increasing specimen
gauge widths (Fig. 13). In contrast, the in-plane shear
strength obtained by Iosipescu shear test shows no size
dependency.22) Iosipescu shear test is one of the testing
methods to evaluate the in-plane shear properties of ceramics
and composites. This test provides pure shear stress field
between two notches curved on both sides of the specimen.
These results indicate that the key fracture mechanism in off-
axis tension had a different dependence on specimen width
because of the presence of several simultaneous failure
modes. In general, three kinds of stresses; tensile stress in the
longitudinal fiber direction, transverse tensile stress perpen-
dicular to the longitudinal fiber direction and in-plane shear
parallel to the fibers, are active during off-axis tensile
loading. In shorter widths, the off-axis tensile strength might
be determined by the weak detachment strength at the F/M
interface rather than by fiber tension and in-plane shear. This
550
100
In-Plane Shear Strength, τ IPSS / MPa
Iosipescu shear:
[45°/45°] tension:
80
60
40
20
0 the axial tensile properties depended significantly on the
0 5
Gauge Width, W / mm
Fig. 13 Specimen width effect on in-plane shear strength of P/W and 3D
SiC/SiC composites. Upper and lower error bars in the plots indicate the
maximum and minimum values, respectively.
is because the lack of cross-weave enables the F/M interface
to detach easily. In contrast, in larger widths (>6:0 mm), the
composites strength should be determined by in-plane shear
mode fracture, although the interfacial detachment mode may
still operate. Moreover, in [15 /75 ] loading tests, a tensile
failure mode as well as the in-plane shear and detachment
failure modes might also operate in the larger width range.
Elastic modulus of the off-axis tensile test is determined by
considering tensile elastic deformation of the fibers and the
matrix in both fiber longitudinal and radial directions, and in-
plane shear deformation at the F/M interface. Fiber volume
content in the gauge section becomes a constant in the off-
axis loading test regardless of the gauge size for all composite
types. In-plane shear modulus is also independent of the
specimen size because of no microscopic change at the F/M
interface. Therefore the meaning size effect was hardly
obtained. The primary reason to the reduced elastic modulus
for the 1.5 mm-wide specimen of [15 /75 ] 3D composites is
probably due to mechanical damage during machining but
still unclear.
As discussed, the off-axis tensile specimens were failed
primarily by detachment and/or in-plane shear fracture
mode. The only detachment at the F/M interface is a primary
fracture mode for the narrow specimen. At the crack
initiation, most cracks propagated along the F/M interface.
It is reported that push-out test results indicate that the
interfacial debonding shear strength first increases and
converges into a constant with the increasing axial lengths
(specimen thickness).23) This can explain the slight reduction
of PLS for short width specimens. Another possible explan-
ation for the reduced PLS in shorter widths is machining
damage on the F/M interface. Detachment strength is usually
very weak comparing fiber strength and/or in-plane shear
strength of the woven composites. The weak detachment
strength provides significantly lower off-axis tensile strength.
Under the single fracture mode tests (length and thickness
effect tests), no size effect was detected.
4.3 Generality of Specimen Size Effect
The effects of the axial fiber volume fraction on tensile
T. Nozawa, Y. Katoh and A. Kohyama
properties are common in most types of composites. A recent
P/W, 3D study revealed the width effect on tensile strength of various
P/W, 3D
types of SiC/SiC composites.24) In that report, a three
directionally woven Tyranno
-SA fiber reinforced compo-
site made by the modified PIP process25) and a cross-plied
Iosipescu shear
Tyranno
-SA fiber reinforced laminate composite made by
the nano-infiltration transient eutectic phase (NITE) proc-
ess26,27) were evaluated. Both composites featured a highly
crystalline and near stoichiometric SiC matrix. Specifically
the NITE SiC/SiC composite provides a robust structure and
dense matrix (
3:1 g/cm3 ). In 3D advanced PIP SiC/SiC,
Off-axis tension
10 15 cross-sectional fiber volume content similar to the case of 3D
PIP SiC/SiC composites. In contrast, the cross-plied laminate
NITE SiC/SiC composites had no width effect on tensile
properties because the axial fiber content is constant in any
specimen widths under the same thickness due to the lay-up
structure of the unidirectional fiber bundle composites.
4.4 Miniature Specimen
Not all CMCs follow the Weibull theory, due to the
presence of multiple fracture modes, even though the
constituent materials are brittle ceramics. We identified
significant width effect on tensile strength, tensile elastic
modulus and PLS. They are closely dependent on the axial
fiber volume fraction and differed in specimen widths. Data
scatter tended to increase in shorter gauge widths. In general,
these data are adequately normalized by simple model
prediction using the fiber volume content for any size of the
composites. However, strength reduction by the stitching
effect needs to be avoided in shorter widths. In conclusion,
specimen width is recommended to be 3.0–4.0 mm to include
a couple of fiber bundles. In contrast, off-axis tensile strength
is decreased in narrower gauge widths because of the size-
related change of fracture mode from the mixed mode of in-
plane shear to fiber detachment. Therefore specimen gauge
width should be 6.0–10.0 mm or larger for off-axis tensile
testing. Specimen thickness has a minor effect on tensile
strength in the range of 1.0–3.0 mm. However, thinner
specimens (<1:0 mm) tend to decrease tensile strength,
resulting in difficult testing in normal operation. Porous
composites such as chemical vapor infiltrated (CVI) SiC/SiC
composites contain many submicron inter-laminar pores; for
those we do not recommend using test specimens thinner than
1.0 mm. Specimen thickness of 1.5–2.0 mm is recommended
to reduce the irradiation volume. No meaningful length
effect, which was found in tensile strength in 30.0–50.0 mm-
long specimens with 15.0–20.0 mm gauge length under the
constant axial fiber volume fraction, is a great advantage to
design the miniature specimen.
No significant effect of the specimen size resulted in
unique tensile strength, elastic modulus and PLS, which were
well agreed with those obtained from standard size specimen.
In short, the miniature specimens used in this study can
predict the mechanical properties of the bulk composites. The
systematic size effects on the off-axis tensile properties can
also utilize to evaluate the bulk properties of the composites,
although no standard data were obtained. Specifically the key
mechanism of the specimen size effect might be remained for
the composites tailored by small fiber bundles; the tensile
 Evaluation of Tensile Properties of SiC/SiC Composites with Miniaturized Specimens
properties are easily predicted by considering the minimum
structural unit, in many cases fiber bundles, and its config-
uration. Therefore, much smaller specimen can be adopted to
utilize. However, machining damage needs to be avoided to
keep a good mechanical performance. Their tensile proper-
ties are more subject to these defects.
5. Summary
Development of mechanical testing methodologies has
been carried out. In order to establish a small specimen test
technique for tensile testing of ceramic matrix composites,
specimen size effects on tensile properties were investigated
from the viewpoint of architecture and varied loading
directions. The main conclusions are summarized as follows.
(1) The axial fiber volume fraction plays an important role
in achieving good tensile properties, but the size
dependent change of the axial fiber volume fraction
gives specimen size effect on them. The composites
with much fiber volume fraction tended to have greater
tensile strength, elastic modulus and PLS. This type of
size effect is generally common in any types of
architecture.
(2) Both 2D and 3D SiC/SiC composites with the same
axial fiber volume content hardly exhibited meaningful
size effect on the tensile properties in the fiber
longitudinal direction.
(3) The size-relevant fracture mode in off-axis tension:
detachment in shorter widths vs. in-plane shear at larger
widths, gives specimen size effect on the off-axis tensile
properties, resulting in strict limitation of miniaturiza-
tion of the tensile specimen.
(4) This study provides a promising approach to tensile
testing using the miniature specimen for ceramic matrix
composites.
Acknowledgements
The authors would like to thank Dr. Edgar Lara-Curzio at
Oak Ridge National Laboratory for useful discussion. This
work was performed as a part of ‘R&D of Composite
Materials for Advanced Energy Systems’ research project
supported by Core Research for Evolutional Science and
Technology (CREST) and ‘Fundamental R&D on Advanced
Material System for (High Efficiency Environment-
551
conscious) very High Temperature Gas-cooled Fast Reactor
Core Structures’ by IVNET Program of MEXT, Japan.
REFERENCES
1) A. R. Raffray, R. Jones, G. Aiello, M. Billone, L. Giancarli, H. Golfier,
A. Hasegawa, Y. Katoh, A. Kohyama, S. Nishio, B. Riccardi and
M. S. Tillack: Fusion Eng. Des. 55 (2001) 55–95.
2) B. Riccardi, L. Giancarli, A. Hasegawa, Y. Katoh, A. Kohyama, R. H.
Jones and L. L. Snead: J. Nucl. Mater. 329–333 (2004) 56–65.
3) W. R. Corwin, S. T. Rosinski and E. V. Walle: Small Specimen Test
Techniques, ASTM STP 1329 (American Society for Testing and
Materials, PA, 1998).
4) M. A. Sokolov, J. D. Landes and G. E. Lucas: Small Specimen Test
Techniques: Fourth Volume, ASTM STP 1418 (ASTM International,
West Conshohocken, PA, 2002).
5) W. A. Curtin: J. Am. Ceram. Soc. 74 (1991) 2837–2845.
6) M. Sutcu: Acta Metall. Mater. 37 (1989) 651–661.
7) W. A. Curtin: J. Am. Ceram. Soc. 77 (1994) 1072–1074.
8) M. Ibnabdeljalil and W. A. Curtin: Int. J. Solids Struct. 34 (1997) 2649–
2668.
9) W. A. Curtin: Compos. Sci. Technol. 57 (1997) 471–481.
10) A. A. Griffith: Phillos. Trans. R. Soc. London, Ser. A 221 (1920) 163–
198.
11) J. W. Hitchon and D. C. Phillips: Fiber Sci. Technol. 12 (1979) 217–
233.
12) J. C. McNulty and F. W. Zok: J. Am. Ceram. Soc. 80 (1997) 1535–
1543.
13) M. R. Wisnom: J. Compos. Mater. 26 (1992) 1173–1180.
14) R. T. Potter: Phillos. Trans. R. Soc. London, Ser. A 361 (1978) 325–
341.
15) M. R. Wisnom: J. Compos. Mater. 28 (1994) 66–76.
16) M. R. Wisnom and J. A. Atkinson: Composite A 28 (1997) 959–964.
17) W. Weibull: J. Appl. Mech. 18 (1951) 293–297.
18) T. Ishikawa, K. Bansaku, N. Watanabe, Y. Nomura, M. Shibuya and
T. Hirokawa: Compos. Sci. Technol. 58 (1998) 51–63.
19) T. Nozawa, T. Hinoki, Y. Katoh, A. Kohyama, E. Lara-Curzio and M.
Sato: Ceram. Trans. 124 (2001) 351–362.
20) T. Nozawa, T. Hinoki, Y. Katoh, A. Kohyama and E. Lara-Curzio:
ASTM STP 1418 (2002) 294–305.
21) A. G. Evans and F. W. Zok: J. Mater. Sci. 29 (1994) 3857–3896.
22) T. Nozawa, E. Lara-Curzio, Y. Katoh and A. Kohyama: Ceram. Trans.
139 (2002) 127–138.
23) C.-H. Hsueh: Acta Metall. Mater. 38 (1990) 403–409.
24) T. Nozawa, Y. Katho, A. Kohyama and E. Lara-Curzio: Ceram. Eng.
Sci. Proc. 24 (2003) 415–420.
25) T. Nozawa, K. Ozawa, Y. Katoh and A. Kohyama: Mater. Trans. 45
(2004) 307–310.
26) Y. Katoh, S. M. Dong and A. Kohyama: Ceram. Trans. 144 (2002) 77–
86.
27) Y. Katoh, S. M. Dong and A. Kohyama: Fusion Eng. Des. 61–62 (2002)
723–731.