com
Received 6 June 2007; received in revised form 28 September 2007; accepted 3 November 2007
Abstract
In this paper, tensile strength and mode I fracture toughness of a low cost carbon–carbon composite (C/C) and liquid metal infiltrated
composites Si–C/C and Cu–C/C are investigated based on tensile testing and compact tension testing. Si–C/C and Cu–C/C are manu-
factured by infiltrating liquid silicon and copper into the low cost C/C, respectively, to make the densification of porous C/C, and to
improve the oxidation resistance and the frictional and wearing properties of C/C for various potential applications under high temper-
ature and frictional environments. Experimental results and analysis of microstructures of three composites reveal that the infiltration of
liquid metal into C/C significantly influences the tensile strength and fracture toughness of the composite. The tensile strength and frac-
ture toughness of Si–C/C are much lower than original C/C. On the other hand, infiltration of copper into C/C significantly improves the
stiffness and strength of Cu–C/C.
Ó 2007 Elsevier Ltd. All rights reserved.
1359-835X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compositesa.2007.11.004
232 W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242
materials at 106 cycles was above 70% of each ultimate ten- infiltration on the microstructures, tensile properties and
sile strength. On the other hand, the tensile strength and fracture toughness of these composites. Microstructures
fracture toughness of these composites are also two impor- were analyzed using optical microscope, scanning electron
tant material properties to the design of brakes. Cracking microscope (SEM) and electron probe microanalyzer
from the corner of slot area of a disk is one of the major (EPMA).
damages. However, the study on the effects of the infiltra-
tion of metals into C/C on the tensile strength and fracture
toughness of Si–C/C and Cu–C/C is not sufficient up to
now.
In the present paper, analysis of microstructures, tensile
testing and compact tension (CT) testing were conducted to
investigate the tensile properties and fracture toughness of
low cost C/C and metal infiltrated composites Si–C/C and Ai: (Pi,δi,ai)
Load, P
Cu–C/C. Attention was focused on the effects of metal Ai+1: (Pi+1,δi+1,ai+δa)
Table 1
Material properties of C/C, Si–C/C and Cu–C/C composites
Material Density Porosity Coefficient of Interlaminar shear
(g/cm3) (%) friction strength (MPa)
(P = 10 MPa)a
O δir δ(i+1)r δi δi+1
C/C 1.7 20 0.25 4 ± 0.3
Cu–C/C 2.68 9 0.23 14.1 ± 1.2 Displacement, δ
Si–C/C 2.05 5 0.22 23.3 ± 1.1
Fig. 2. Determination of critical energy release rate from load–displace-
a
The coefficient of friction varies with contact pressure P. ment curve.
20
Al
4
60
L
r =10
0.3
1.0
48
1
13
40
(mm)
50
low cost C/C usually contains about 20% pores. To reduce laminate to form Si–C/C or Cu–C/C composite laminate
the porosity and to improve the oxidation resistance, fric- [6]. The infiltration of Si into C/C is performed under the
tional properties, and wearing resistance, liquid silicon or conditions of about 1600 °C, 1 hPa low pressure, and Ar
copper was infiltrated into the pores of the low cost C/C gas atmosphere for about 1 h, while the infiltration of Cu
Fig. 4. SEM images of impregnated Si and Cu, and EPMA mapping of carbon of Si–C/C.
W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242 235
into C/C is performed under the conditions of about mens for mechanical tests are described in Fig. 1.
1200 °C, 10 MPa pressure, and Ar gas atmosphere for Specimens for tensile testing were rectangular, as shown
about 20 min. Details of these infiltration processes can in Fig. 1a. The length of tensile specimens of C/C and
be found in Ref. [6]. It is noted that a potential application Si–C/C was 140 mm and the length of tensile specimens
of Cu–C/C composite to the brakes of trucks and train cars of Cu–C/C was 130 mm due to the limitation of original
is specially expected since the working temperature of these laminate of Cu–C/C. The gauge length of all specimens
brakes may be lower than the melting point of Cu. Table 1 of three composites was the same as 60 mm. Aluminum
gives several material properties of these composites. It is tabs of 1 mm thickness were bonded to the two ends of
seen that the porosity was reduced to 5% for Si–C/C and specimens to protect the grip regions. A strain gauge was
9% for Cu–C/C. On the other hand, the density increased located at the center of gauge length of each specimen to
by about 20% for Si–C/C and 57% for Cu–C/C. The inter- measure the strain. Fig. 1b shows the geometry of speci-
laminar shear strength (ILSS) was obtained from double mens for compact tension testing. The thickness of all spec-
notched compressive shear testing [11]. imens of three composites was 4 mm. The width of
Specimens used in the present experiments were cut specimens is denoted by W and was 40 mm which is equal
from the laminates of C/C, Si–C/C and Cu–C/C manufac- to the distance from the loading line to left edge. The length
tured by above mentioned methods. Geometries of speci- of pre-crack is denoted by a which is equal to the distance
Fig. 5. In situ CCD camera video images of three composites at tensile failure, and fractured specimens.
236 W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242
from loading line to crack tip. Pre-crack was made firstly a cyclic loading. The critical energy release rate is expressed
using a cutting machine with a 0.3 mm thick blade and then by
using a razor blade to obtain a relatively sharp crack tip. P iC DC i di dir
GC ¼ ; DC i ¼ C iþ1 C i ; Ci ¼ ð1Þ
2BDai P iC
2.2. Testing procedures
where Ci represents the compliance at loading Pi during a
Analysis of microstructures of three composites was per- cyclic loading, PiC denotes the critical load, Dai denotes
formed using optical microscope, scanning electron micro- the effective crack extension, and di and dir represent the
scope (SEM) and electron probe microanalyzer (EPMA) displacement and residual displacement, respectively. Effec-
before mechanical tests to investigate the effects of metal tive crack length ai was determined from the compliance Ci
infiltration on the original C/C. and compliance calibration curves as mentioned above.
Tensile testing was performed using a MTS 810 material
testing system at room temperature. Six specimens were
tested for C/C and five specimens were tested for each of
Si–C/C and Cu–C/C. Loading rate was 0.5 mm/min. Strain 300
was obtained from the strain gauge located at the center of C/C
250
each specimen. One side edge of each specimen was pol-
ished. Optical observation of side edges of specimens was
Stress (MPa)
200
carried out before and after testing to identify the damage
evolution. CCD video camera system was also used to 150
observe the damage evolution during loading.
Compact tension testing for three composites was con- 100
ducted also using MTS 810 material testing system at room
50
temperature to evaluate the mode I fracture toughness.
Loading was applied through a pin-and-shackle system 0
and loading rate was also 0.5 mm/min. It is well known 0 0.1 0.2 0.3 0.4 0.5 0.6
that damages, such as fiber breaking, matrix cracking, fiber Strain (%)
pull-out, debonding, and delamination, might occur
mixedly around the crack tip when a crack grows in C/C 350
composites. Thus it is difficult to clearly define the crack Cu-C/C
300
length during testing. In the present work, an effective
crack length approach [12] using the compliance calibra- 250
Stress (MPa)
40
fact proved that the low cost C/C is a very tough material
comparing other two composites. A theory of linear elastic 30
fracture mechanics applied to composites [13] was adopted
20
to approximately evaluate the fracture toughness of present
three composites, although the crack growth process 10
involving with multi-damages was quite complicated.
Fig. 2 schematically describes the determination of critical 0
0 0.05 0.1 0.15 0.2
energy release rate from the load–displacement curve. The
shaded area represents the energy contributed to the crack Strain (%)
growth, whereas the white area is considered to be con- Fig. 6. Typical stress–strain curves of C/C, Cu–C/C and Si–C/C
sumed by other damages and hysteretic energy loss during composites.
W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242 237
3. Experimental results and discussion in Fig. 3. To see the detail morphology of infiltrated metals,
microscopic analysis of Si–C/C and Cu–C/C was conducted
3.1. Microstructures using SEM and EPMA. SEM images of infiltrated silicon
and copper are shown in Fig. 4a and EPMA mapping of car-
Optical microscope images presented in Fig. 3 show the bon of Si–C/C is shown in Fig. 4b. It is seen that a silicon
polished side surfaces of C/C, Cu–C/C, and Si–C/C lami- carbide (SiC) interlayer exists between silicon and surround-
nates. Global images through the laminate thickness are ing C/C due to the reaction between silicon and carbon at
given in Fig. 3a and enlarged local images are given in high temperature during the infiltration process. This fact
Fig. 3b. Dark spots and crack-like strips represent pores is proved by the EPMA mapping of carbon of Si–C/C, as
and voids. Many pores and voids are observed in the case shown in the left one of Fig. 4b. In contrast, there is no
of C/C laminate, whereas a few pores and voids can be seen any reacted interlayer between copper and carbon in Cu–
in the cases of Cu–C/C and Si–C/C. These facts reveal that C/C. The formation of SiC interlayer between silicon and
metal infiltration indeed is an effective method for the den- surrounding C/C leads to a high ILSS as given in Table 1
sification of porous C/C composite. On the other hand, it is since the infiltrated silicon formed some networks in C/C
worth noting that new cracks appeared between every two from the view point of three-dimension and the SiC inter-
original cracks after infiltration process in the case of Si– layer induced a strong combination of carbon and silicon.
C/C, though the original cracks were infiltrated with silicon. However, high ILSS is not always an advantage for com-
This kind of new cracks can not be observed in the case of posites and it may lead to low strength and fracture tough-
Cu–C/C. One of the reasons for the difference of microstruc- ness of composites, which can be realized from following
ture between Si–C/C and Cu–C/C is due to the influence of experimental results. Furthermore, as shown in Fig. 4c,
infiltration process as mentioned in the above section. some of carbon fibers near to the infiltrated Si seem to be
Besides, it is noted that pores and voids equally existed in eroded by the chemical reaction between silicon and carbon
0° layers of laminates of three composites, though the voids fiber. This degradation of carbon fibers may reduce the
in 90° layers seem to be more obvious than those in 0° layers mechanical properties of carbon fibers.
350
Averaged values of tensile strength and
200
150
100
54.1±10 62.5±14 55.2±7 57.8±10
50
0
C/C Cu-C/C Si-C/C
Fig. 9. Averaged stiffness and strength of C/C, Cu–C/C and Si–C/C composites.
W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242 239
3.2. Tensile testing results before (left) and after (right) tensile testing are presented
in Fig. 7. A delamination and a coalition of two cracks
Typical in situ video images of specimens at tensile fail- can be seen in C/C specimen, and a debonding between
ure are presented in Fig. 5a, and fractured specimens are infiltrated copper and surrounding C/C is observed in
shown in Fig. 5b. Three composites appeared quite differ- Cu–C/C specimen. Fig. 8 presents typical images of Si–C/
ent failure behaviors. Fracture of 90° layers and large scale C specimens before (left) and after (right) tensile testing.
delamination are observed in the case of C/C specimens, Comparing with the specimens of C/C and Cu–C/C, many
relatively small scale delamination together with fracture cracks are observed. These cracks certainly lead to the
of 90° layers is observed in the case of Cu–C/C specimens, increase of strain of the specimen during tensile testing,
whereas only fracture of 90° layers can be observed in the which finally results in a nonlinear stress–strain curve for
case of Si–C/C specimens before the final fracture of 0° lay- Si–C/C specimen. The lower and right one in Fig. 8 also
ers. In other words, no obvious delamination can be
observed from the Si–C/C specimens. In addition, plastic
deformation due to the effect of infiltrated copper can be 1400
observed from the image of Cu–C/C specimen. These fea- C/C
1200
tures also appear in the typical stress–strain curves of spec-
imens of three composites, as shown in Fig. 6. Stress–strain 1000
Load P (N)
curves of specimens of C/C and Cu–C/C are linear or
800
slightly nonlinear until their critical loads, respectively.
After critical load and before final fracture, the decrease 600
of stress in the C/C specimen is induced by the large delam-
400
ination occurring and partially breaking of 0° layers,
whereas a slight decrease of stress in the Cu–C/C specimen 200
reveals a pseudo-plastic behavior since Cu is ductile and
0
the interface between Cu and carbon has very weak bond- 0 1 2 3
ing. On the other hand, the stress–strain curve of Si–C/C Displacement (mm)
specimen becomes obviously nonlinear as the load increas-
ing, and the final facture occurs almost at critical load. The 1400
critical strain of Si–C/C is also much smaller than those of Si-C/C
1200
C/C and Cu–C/C specimens. The nonlinear behavior is
considered to be induced by the cracking in the regions 1000
Load P (N)
0
0 0.5 1 1.5 2
2.0E-06 Displacement (mm)
C/C
Si-C/C 1800
1.5E-06 Cu-C/C 1600 Cu-C/C
Compliance (m/N)
1400
1200
Load P (N)
1.0E-06
1000
800
600
5.0E-07
400
200
0.0E+00 0
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5 2
a/W Displacement (mm)
Fig. 10. Compliance calibration curves of C/C, Si–C/C and Cu–C/C Fig. 11. Typical loading–unloading curves of C/C, Si–C/C and Cu–C/C
specimens. specimens.
240 W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242
shows the partial fractured location of a Si–C/C specimen. sile strength and stiffness and Si–C/C shows the lowest ten-
It is noted that the images before testing were taken at sile strength. Comparing with C/C, infiltration of copper
selected locations of each specimen. However, it was diffi- into C/C leads to 17% increase in tensile strength and
cult to predict the final fracture location of the specimen 15% increase in tensile stiffness, whereas infiltration of sili-
before testing. Hence, only very few images before testing con results in very low tensile strength and 7% increase in
could catch the fracture location. In addition to the differ- tensile stiffness. The stiffness and strength of Cu–C/C are
ence of failure behaviors of three composites, the tensile almost in consistence with the prediction using mixed rule.
strength of Si–C/C is also very different from those of C/ However, the mixed rule is not available for Si–C/C. Refer-
C and Cu–C/C. Averaged values and scatter ranges of ten- ring to the microstructure analysis it is considered that the
sile strength and tensile stiffness of three composites are major reasons for low strength and slight improvement in
given in Fig. 9. Cu–C/C composite shows the highest ten- stiffness in the case of Si–C/C are the occurrence of new
Fig. 12. Typical fractured specimens of C/C, Si–C/C and Cu–C/C composites.
W.-X. Wang et al. / Composites: Part A 39 (2008) 231–242 241
(kJ/m2)
of delamination which plays an important role in the relax- 10
ation of stress concentration in the 0° layer ahead the crack 8
tip. Consequently the high stress level leads to the fracture
6
of 0° layers of Si–C/C specimens at a low loading level.
4
Thus restraining the crack occurrence, fiber degradation
2
and the formulation of SiC interlayer during infiltration
0
process is necessary to obtain high strength Si–C/C 0 0.2 0.4 0.6 0.8 1
composite. Normalized equivalent crack length a/W
3.3. Compact tension testing results Fig. 13. Critical energy release rate of C/C, Si–C/C and Cu–C/C
composites.