Internationa Journal ELSEVIER International Journal of Fatigue 24(2002)241-248 www.elsevier.com/locate/ijfatigue Observations of fatigue damage process in SiC/SiC composites at room and elevated temperatures Y Miyashita a,, K Kanda,S. Zhu,Y. Mutoh a, M. Mizuno A.J. McEvily Department of Mechanical Engineering, Nagaoka University of Technology, 1603-1 Kamitomioka, Nagaoka, Niigata 940-2188, Japar niversity of Electro-Communications, Chofu, Tokyo, Japan Japan Fine Ceramics Center, Atsuta-ku, Nagoya, Aichi, Japan IMS, University of Connecticut, Storrs, CT, USA Abstract Scanning Electron Microscope(SEM)in-situ observation of fatigue damage process of SiC/SiC composites was carried out at room and elevated temperatures. Although single edge V-notched specimens were used, fatigue crack initiated not only at the notch root but also at large pores far from the notch. The widespread damage means that the parameter of linear elastic fracture mechanics cannot be used to describe fatigue crack growth behavior of SiC/SiC materials. Instead, modulus of rigidity(MOR)may be used as a parameter to estimate fatigue damage process. The crack behavior was divided into two types of damage modes depending on the orientation of fiber bundles in front of a notch. No matter which type of damage mode, the final failure occurred at about 85%of the maximum value of mOR. c 2002 Elsevier Science Ltd. All rights reserved Keywords: SiC/SiC composite; Fatigue; Crack growth, Damage process; Elevated temperature 1. Introduction fatigue damage was correlated with the change in rigid- ity of the specimens The SiC fiber reinforced SiC (SiC/SiC) composite is one of the most important ceramic matrix composites for high temperature applications. It has been reported that fatigue limit of SiC/SiC composites is about 80% of the 2. Materials ultimate tensile strength at room temperature, but the fatigue limit decreases at 1000oC, a temperature lower The SiC (Nicalon) fibers were approximately 15 um than that for fiber creep [1-5]. One part of the fatigue in diameter. These fibers were assembled in flattened damage mechanism in SiC/SiC is debonding as a result bundles with about 500 fibers per bundle. The bundles of wear of the interface between the matrix and the fibers were cross-woven to form a layer approximately 0.3 mm [1, 2, 4], as observed by in-situ observations [6, 7).A com- thick. This pre-pleg layer was then coated with carbon lete evaluation of the fatigue damage process in by a Cvd( Chemical Vapor Deposition)technique to SiC/SiC has not been established provide a weak interface between fibers and the Sic In this work, fatigue tests of two types of SiC/Sic matrix which in the next step was formed by a chemical composites were carried out in an SEM(Scanning Elec- vapor infiltration(CVI) procedure. This infiltration pro- tron Microscope)chamber at room temperature and at cedure resulted in a significant amount of porosity. Two 800oC. The fatigue crack initiation and growth processes types of matrices were used. One matrix was simply were observed in detail in the sem. the evolution of standard sic. the second matrix consisted of sic to which boron had been added The boron reacts with oxy- gen to form glassy particulates which seal the matrix phase and inhibit oxidation of the carbon-interfacial Corresponding author. Tel. +81-258-47-9750; fax: +81-258-47 layer [8, 9]. This type of matrix is referred to as an enhanced matrix. The individual layers were then sin E-mailaddress: miyayuki@mech. nagaokaut ac jp(Y Miyashita). tered together to form a 3 mm thick panel 0142-1 123/02/. see front matter e 2002 Elsevier Science Ltd. All rights reserved P:S0142-1123(01)00078-0
International Journal of Fatigue 24 (2002) 241–248 www.elsevier.com/locate/ijfatigue Observations of fatigue damage process in SiC/SiC composites at room and elevated temperatures Y. Miyashita a,*, K. Kanda a , S. Zhu b , Y. Mutoh a , M. Mizuno c , A.J. McEvily d a Department of Mechanical Engineering, Nagaoka University of Technology, 1603-1 Kamitomioka, Nagaoka, Niigata 940-2188, Japan b University of Electro-Communications, Chofu, Tokyo, Japan c Japan Fine Ceramics Center, Atsuta-ku, Nagoya, Aichi, Japan d IMS, University of Connecticut, Storrs, CT, USA Abstract Scanning Electron Microscope (SEM) in-situ observation of fatigue damage process of SiC/SiC composites was carried out at room and elevated temperatures. Although single edge V-notched specimens were used, fatigue crack initiated not only at the notch root but also at large pores far from the notch. The widespread damage means that the parameter of linear elastic fracture mechanics cannot be used to describe fatigue crack growth behavior of SiC/SiC materials. Instead, modulus of rigidity (MOR) may be used as a parameter to estimate fatigue damage process. The crack behavior was divided into two types of damage modes depending on the orientation of fiber bundles in front of a notch. No matter which type of damage mode, the final failure occurred at about 85% of the maximum value of MOR. 2002 Elsevier Science Ltd. All rights reserved. Keywords: SiC/SiC composite; Fatigue; Crack growth; Damage process; Elevated temperature 1. Introduction The SiC fiber reinforced SiC (SiC/SiC) composite is one of the most important ceramic matrix composites for high temperature applications. It has been reported that fatigue limit of SiC/SiC composites is about 80% of the ultimate tensile strength at room temperature, but the fatigue limit decreases at 1000°C, a temperature lower than that for fiber creep [1–5]. One part of the fatigue damage mechanism in SiC/SiC is debonding as a result of wear of the interface between the matrix and the fibers [1,2,4], as observed by in-situ observations [6,7]. A complete evaluation of the fatigue damage process in SiC/SiC has not been established. In this work, fatigue tests of two types of SiC/SiC composites were carried out in an SEM (Scanning Electron Microscope) chamber at room temperature and at 800°C. The fatigue crack initiation and growth processes were observed in detail in the SEM. The evolution of * Corresponding author. Tel.: +81-258-47-9750; fax: +81-258-47- 9770. E-mail address: miyayuki@mech.nagaokaut.ac.jp (Y. Miyashita). 0142-1123/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S0142-1123(01)00078-0 fatigue damage was correlated with the change in rigidity of the specimens. 2. Materials The SiC (Nicalon) fibers were approximately 15 µm in diameter. These fibers were assembled in flattened bundles with about 500 fibers per bundle. The bundles were cross-woven to form a layer approximately 0.3 mm thick. This pre-pleg layer was then coated with carbon by a CVD (Chemical Vapor Deposition) technique to provide a weak interface between fibers and the SiC matrix which in the next step was formed by a chemical vapor infiltration (CVI) procedure. This infiltration procedure resulted in a significant amount of porosity. Two types of matrices were used. One matrix was simply standard SiC. The second matrix consisted of SiC to which boron had been added. The boron reacts with oxygen to form glassy particulates which seal the matrix phase and inhibit oxidation of the carbon-interfacial layer [8,9]. This type of matrix is referred to as an enhanced matrix. The individual layers were then sintered together to form a 3 mm thick panel
Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 o'bundle by using a specimen equipped with a strain gauge at room temperature and developing a strain-actuator dis- rain gauge placement relationship. A tungsten wire heater attached R10 to the specimen gauge section was used to heat the speci mens in the 800%C fatigue tests In order to stabilize the test conditions, the specimen was held at 800oC for 30 min prior to testing The fatigue tests were carried out under axial loading using a sinusoidal wave form at a stress ratio of 0. 1. The test frequency at both temperatures either 1 or 10 Hz, which was reduced to 0. 1 Hz when obtaining stress-strain data. In the fatigue tests the specimens were initially cycled at a maximum stress of 45 MPa, a level which was 15 MPa less than the critical stress for matrix cracking. If cracks were not observed at this level after 5000 cycles at 1.0 Hz or 20,000 cycles at 10 Hz, the Fig. 1. Specimen geometry(in mm) maximum stress was increased by 5 MPa. The specimen was then cycled for an additional 5000 or 20,000 cycles If no cracking occurred the incremental loading pro- 3. Specimens and tests cedure was repeated until a stress range was reached at which cracks were observed. The area under observation The specimen geometry for the fatigue tests is shown was within 2 mm of the crack tip in Fig. 1. A machined notch facilitated the study of the crack initiation and propagation processes. This notch was V-shaped with a notch depth of I mm and a notch 4. Results and discussion root radius of 15 um A servo-hydraulic fatigue machine designed for test- 4. 1. Fatigue life ing of the specimens in the chamber of the SEM was utilized for the in-situ observations and stress-strain The relationship between the maximum stress and hysteresis loops were obtained periodically during number of cycles to failure for both the standard SiC/SiC fatigue tests. In the room temperature tests strains were and enhanced SiC/SiC composites is shown in Fig. 2 measured by using a strain gauge of 1 mm gauge length The large scatter in results can be attributed to wide vari- mounted in front of the notch, as shown in Fig. 1. In ations in both crack initiation and growth behavior the tests at 800C, the displacement of the actuator was Cracks initiated not only in front of a notch tip but also utilized to estimate strains. A calibration was established at large pores remote from the notch tip. Fatigue life at Smooth sp Stress-Number of cycles curve of standard SiC/SiC at RT and 10Hz Stress-Number of cycles curve of standard SiC/SiC at 1000"C and 20Hz 200 The present notched specimen ● Standard SiC/SiC, R A Enhanced SiC/SiC, 800C. Load frequency of data unmarked is 10Hz. 7 Number of cycles to failure Fig. 2. Results of fatigue tests at room and elevated temperatures
242 Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 Fig. 1. Specimen geometry (in mm). 3. Specimens and tests The specimen geometry for the fatigue tests is shown in Fig. 1. A machined notch facilitated the study of the crack initiation and propagation processes. This notch was V-shaped with a notch depth of 1 mm and a notch root radius of 15 µm. A servo-hydraulic fatigue machine designed for testing of the specimens in the chamber of the SEM was utilized for the in-situ observations, and stress–strain hysteresis loops were obtained periodically during the fatigue tests. In the room temperature tests strains were measured by using a strain gauge of 1 mm gauge length mounted in front of the notch, as shown in Fig. 1. In the tests at 800°C, the displacement of the actuator was utilized to estimate strains. A calibration was established Fig. 2. Results of fatigue tests at room and elevated temperatures. by using a specimen equipped with a strain gauge at room temperature and developing a strain-actuator displacement relationship. A tungsten wire heater attached to the specimen gauge section was used to heat the specimens in the 800°C fatigue tests. In order to stabilize the test conditions, the specimen was held at 800°C for 30 min prior to testing. The fatigue tests were carried out under axial loading using a sinusoidal wave form at a stress ratio of 0.1. The test frequency at both temperatures was either 1 or 10 Hz, which was reduced to 0.1 Hz when obtaining stress–strain data. In the fatigue tests the specimens were initially cycled at a maximum stress of 45 MPa, a level which was 15 MPa less than the critical stress for matrix cracking. If cracks were not observed at this level after 5000 cycles at 1.0 Hz or 20,000 cycles at 10 Hz, the maximum stress was increased by 5 MPa. The specimen was then cycled for an additional 5000 or 20,000 cycles. If no cracking occurred the incremental loading procedure was repeated until a stress range was reached at which cracks were observed. The area under observation was within 2 mm of the crack tip. 4. Results and discussion 4.1. Fatigue life The relationship between the maximum stress and number of cycles to failure for both the standard SiC/SiC and enhanced SiC/SiC composites is shown in Fig. 2. The large scatter in results can be attributed to wide variations in both crack initiation and growth behavior. Cracks initiated not only in front of a notch tip but also at large pores remote from the notch tip. Fatigue life at
Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 Loading direction Loading direction notch#+- 20 um Loading direction notch s To0um Observations of fatigue crack growth path in standard SiC/SiC with frequency of 1 Hz at room temperature: (a) at maximum stress of 60 MPa after 5000 cycles;(b) at maximum stress of 80 MPa after 1 cycle; (c)final fracture at maximum stress of 100 MPa after I cycle room temperature was higher than that at high tempera- 4.2. Crack initiation and propagation processes ture. Comparable test results from the literature which were obtained at room temperature and at 1000C for Fig 3 shows examples of the crack growth path in unnotched specimens [4 are also shown in Fig. 2. Not standard SiC/SiC at room temperature. In this instance rprisingly, the fatigue resistance of the notched speci- the notch was situated in a longitudinal fiber bundle mens was less than that of the smooth specimens. The which was oriented in the same direction as the loading shorter fatigue lives in notched specimens resulted not axis. A crack initiated at the interface between fiber and only from the stress concentration at the notch tip but matrix near the notch tip (indicated by an arrow in Fig also from differences in loading history. The applied 3(a)). As this crack was growing along the interface stress was increased stepwise during fatigue test in the fiber-cracking occurred( Fig. 3(b). The crack then con- present study, and fatigue damage may have developed tinued to grow by a combination of interfacial cracking at each applied stress level. The fatigue strength at and fiber-cracking. However, the cracks initiated not 800C is seen to be greater than at room temperature. It only at the notch, some cracks initiated at pores and is also noted that although the difference in fatigue life where fiber bundles of different orientation met between standard and enhanced SiC/SiC was in general (indicated by(A)in Fig. 3(c). Nevertheless, the final not significant, the standard material did have a slightly fracture process always involved considerable crack longer fatigue life than the enhanced material at room branching and a linking up with a crack initiated at the temperature and at 800oC notch( Fig. 3(c)
Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 243 Fig. 3. Observations of fatigue crack growth path in standard SiC/SiC with frequency of 1 Hz at room temperature: (a) at maximum stress of 60 MPa after 5000 cycles; (b) at maximum stress of 80 MPa after 1 cycle; (c) final fracture at maximum stress of 100 MPa after 1 cycle. room temperature was higher than that at high temperature. Comparable test results from the literature which were obtained at room temperature and at 1000°C for unnotched specimens [4] are also shown in Fig. 2. Not surprisingly, the fatigue resistance of the notched specimens was less than that of the smooth specimens. The shorter fatigue lives in notched specimens resulted not only from the stress concentration at the notch tip but also from differences in loading history. The applied stress was increased stepwise during fatigue test in the present study, and fatigue damage may have developed at each applied stress level. The fatigue strength at 800°C is seen to be greater than at room temperature. It is also noted that although the difference in fatigue life between standard and enhanced SiC/SiC was in general not significant, the standard material did have a slightly longer fatigue life than the enhanced material at room temperature and at 800°C. 4.2. Crack initiation and propagation processes Fig. 3 shows examples of the crack growth path in standard SiC/SiC at room temperature. In this instance the notch was situated in a longitudinal fiber bundle which was oriented in the same direction as the loading axis. A crack initiated at the interface between fiber and matrix near the notch tip (indicated by an arrow in Fig. 3(a)). As this crack was growing along the interface fiber-cracking occurred (Fig. 3(b)). The crack then continued to grow by a combination of interfacial cracking and fiber-cracking. However, the cracks initiated not only at the notch, some cracks initiated at pores and where fiber bundles of different orientation met (indicated by (A) in Fig. 3(c)). Nevertheless, the final fracture process always involved considerable crack branching and a linking up with a crack initiated at the notch (Fig. 3(c))
244 Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 Fig. 4 provides examples of the crack paths in the notch root, but this crack also did not propagate until enhanced SiC/SiC tested at room temperature.At a maximum stress of 130 MPa was reached. The second maximum stress of 60 MPa a crack initiated at one side crack then propagated and merged with a third crack of the notch root in the first loading cycle(Fig. 4(a)). initiated at a pore away from the notch(Fig. 4(c)),and the applied stress to 90 MPa, a new crack was observed third crack had been initiated at a maximum stress of fter 5000 loading cycles at a pore on the other side of 45 MPa early in the fatigue test but had not propagated until the 130 MPa maximum stress level was reached Fatigue crack growth paths in standard SiC/SIC at Loading direction 800C are shown in Fig. 5. A crack initiated from the notch and propagated along the interface between the fibers and the matrix, as indicated by arrows in Fig. 5(a) A second crack initiated from a pore far from the notch was observed, as indicated by arrows in Fig. 5(b). In this specimen, the crack which initiated from the pore played a dominant role in the final fracture process. As with the enhanced SiC/SiC. no significant difference in crack initiation and propagation behavior was found between room temperature and 800oC When the fiber orientation in front of the notch was notch normal to the loading direction as shown in Fig. 6,a crack initiated at the notch tip(Fig. 6(a)). This crack hen propagated rapidly along the interface between the fibers and the matrix as shown in Fig. 6(b) oading airection notch pore notch 20um Loading direction Loading direction Fig. 4. Observations of fatigue crack growth path in enhanced SiC/SiC with frequency of 10 Hz at room temperature: (a)at maximum Fig. 5. Observations of fatigue crack growth path in standard SiC/SiC ress of 60 MPa after 1 cycle;( b)at maximum stress of 90 MPa after with frequency of 10 Hz at 800C:(a)at maximum stress of 70 MPa 5000 cycles;(c)at maximum stress of 45 MPa after I cycle after 1000 cycles; ( b)at maximum stress of 71 MPa after I cycle
244 Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 Fig. 4 provides examples of the crack paths in enhanced SiC/SiC tested at room temperature. At a maximum stress of 60 MPa a crack initiated at one side of the notch root in the first loading cycle (Fig. 4(a)). However, the crack did not propagate. After increasing the applied stress to 90 MPa, a new crack was observed after 5000 loading cycles at a pore on the other side of Fig. 4. Observations of fatigue crack growth path in enhanced SiC/SiC with frequency of 10 Hz at room temperature: (a) at maximum stress of 60 MPa after 1 cycle; (b) at maximum stress of 90 MPa after 5000 cycles; (c) at maximum stress of 45 MPa after 1 cycle. the notch root, but this crack also did not propagate until a maximum stress of 130 MPa was reached. The second crack then propagated and merged with a third crack initiated at a pore away from the notch (Fig. 4(c)), and this linking up resulted in failure of the specimen. This third crack had been initiated at a maximum stress of 45 MPa early in the fatigue test but had not propagated until the 130 MPa maximum stress level was reached. Fatigue crack growth paths in standard SiC/SiC at 800°C are shown in Fig. 5. A crack initiated from the notch and propagated along the interface between the fibers and the matrix, as indicated by arrows in Fig. 5(a). A second crack initiated from a pore far from the notch was observed, as indicated by arrows in Fig. 5(b). In this specimen, the crack which initiated from the pore played a dominant role in the final fracture process. As with the enhanced SiC/SiC, no significant difference in crack initiation and propagation behavior was found between room temperature and 800°C. When the fiber orientation in front of the notch was normal to the loading direction as shown in Fig. 6, a crack initiated at the notch tip (Fig. 6(a)). This crack then propagated rapidly along the interface between the fibers and the matrix as shown in Fig. 6(b). Fig. 5. Observations of fatigue crack growth path in standard SiC/SiC with frequency of 10 Hz at 800°C: (a) at maximum stress of 70 MPa after 1000 cycles; (b) at maximum stress of 71 MPa after 1 cycle
Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 245 Loading direction Fig. 6. Observations of fatigue crack growth path in standard SiC/SiC with frequency of 10 Hz at 800C: (a)at maximum stress of 90 MPa after 5000 cycles;(b)at maximum stress of 130 MPa after 50000 cycles The crack initiation and propagation processes in stan 9(a). On comparing Figs. 8(b)and 9(b), it is apparent dard SiC/SiC were similar to those in enhanced SiC/Sic that fiber pull-out is much more extensive at the higher both at room temperature and at 800oC temperature. Pull-out length of fiber at high temperature Based on the foregoing observations, the crack propa- is about 100 um, while that at room temperature is negli- gation processes were classified into two types as sche ble in average. This is the result of a decrease in bond- matically shown in Fig. 7 ing strength between the fibers and the matrix at elevated In Type la, the fibers ahead of the notch are parallel temperatures. From these fractographic observations to the loading direction, and final failure is caused by there is no significant difference of fracture morphology growth of a crack initiated from a notch with the crack between near-surface region and inside of the specimen propagating by a combination of fiber-breaking and Therefore, it may be speculated that the fracture mech- branching anisms observed on the surface coincide with those In Type Ib, the fibers ahead of the notch are perpen- within the bulk. However, further experiments for get- licular to the loading direction and final failure is caused ting direct evidence will be needed to make the fracture by the growth of a crack initiated from a notch with the mechanisms inside the specimen clear crack propagating along the interface between the fiber and the matrix 4. 4. Modulus of rigidity(MOR) In Type 2, a crack is initiated at the notch but does not grow until a second crack initiated at a pore grows In metals it is customary to treat the growth of a domi and joins with the first crack nant fatigue crack in terms of linear elastic fracture mech- anics with the rate of fatigue crack growth being a func 4.3. Fracture surface appearance tion of the stress intensity factor. However, in the SiC/SiC system such an approach is not feasible because ig. 8 shows fracture surface of the standard SiC/Sic of the complexity of the cracking process as described specimen tested at room temperature. Fig. 9 shows frac- above. Therefore, we seek for a parameter other than the ture surface of the standard SiC/Sic tested at stress intensity factor for estimating fatigue damage in 800C. There are three or four bundle thickness composite materials of this type. In order to characterize direction of the specimen as shown in Figs. 8(a)and the extent of fatigue damage in SiC/SiC composite
Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 245 Fig. 6. Observations of fatigue crack growth path in standard SiC/SiC with frequency of 10 Hz at 800°C: (a) at maximum stress of 90 MPa after 5000 cycles; (b) at maximum stress of 130 MPa after 50000 cycles. The crack initiation and propagation processes in standard SiC/SiC were similar to those in enhanced SiC/SiC both at room temperature and at 800°C. Based on the foregoing observations, the crack propagation processes were classified into two types as schematically shown in Fig. 7. In Type 1a, the fibers ahead of the notch are parallel to the loading direction, and final failure is caused by growth of a crack initiated from a notch with the crack propagating by a combination of fiber-breaking and branching. In Type 1b, the fibers ahead of the notch are perpendicular to the loading direction and final failure is caused by the growth of a crack initiated from a notch with the crack propagating along the interface between the fibers and the matrix. In Type 2, a crack is initiated at the notch but does not grow until a second crack initiated at a pore grows and joins with the first crack. 4.3. Fracture surface appearance Fig. 8 shows fracture surface of the standard SiC/SiC specimen tested at room temperature. Fig. 9 shows fracture surface of the standard SiC/SiC specimen tested at 800°C. There are three or four bundles to the thickness direction of the specimen as shown in Figs. 8(a) and 9(a). On comparing Figs. 8(b) and 9(b), it is apparent that fiber pull-out is much more extensive at the higher temperature. Pull-out length of fiber at high temperature is about 100 µm, while that at room temperature is negligible in average. This is the result of a decrease in bonding strength between the fibers and the matrix at elevated temperatures. From these fractographic observations, there is no significant difference of fracture morphology between near-surface region and inside of the specimen. Therefore, it may be speculated that the fracture mechanisms observed on the surface coincide with those within the bulk. However, further experiments for getting direct evidence will be needed to make the fracture mechanisms inside the specimen clear. 4.4. Modulus of rigidity (MOR) In metals it is customary to treat the growth of a dominant fatigue crack in terms of linear elastic fracture mechanics with the rate of fatigue crack growth being a function of the stress intensity factor. However, in the SiC/SiC system such an approach is not feasible because of the complexity of the cracking process as described above. Therefore, we seek for a parameter other than the stress intensity factor for estimating fatigue damage in composite materials of this type. In order to characterize the extent of fatigue damage in SiC/SiC composite
Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 (c)Type 2 Fig. 7. Schematic illustrations of crack growth processes. materials, the variation in MOR, as has been previously figure that the MOR reduction decreases with increase in suggested [10-12], was used in this study. The mOR was the total crack length. It is also noted that in some cases calculated from stress-strain hysteresis loops, which were the MOR increased with crack growth at low stress levels ecorded during fatigue tests. The MOR was taken to be(Fig. 10(a)). This anomalous behavior was observed for the slope of the unloading curve between the maximum all types of crack growth processes. Initial MOR, final stress and 10% of the maximum stress. the variation of mor and maximum mor are shown in fig. 11. values the mor reduction and of the total crack length with of the maximum mor are larger than the initial mOR increasing number of cycles is shown in Fig. 10, where for all specimens tested. In some cases, the final MOR MOR reduction is the mOr divided by the maximum is larger than the initial MOR. The increase in MOR has MOR and total crack length is the sum of all crack been attributed to an increase in the interfacial shear lengths within 2 mm of the notch tip. It is seen from the stress during long-duration cyclic loading [ 12]
246 Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 Fig. 7. Schematic illustrations of crack growth processes. materials, the variation in MOR, as has been previously suggested [10–12], was used in this study. The MOR was calculated from stress–strain hysteresis loops, which were recorded during fatigue tests. The MOR was taken to be the slope of the unloading curve between the maximum stress and 10% of the maximum stress. The variation of the MOR reduction and of the total crack length with increasing number of cycles is shown in Fig. 10, where MOR reduction is the MOR divided by the maximum MOR and total crack length is the sum of all crack lengths within 2 mm of the notch tip. It is seen from the figure that the MOR reduction decreases with increase in the total crack length. It is also noted that in some cases, the MOR increased with crack growth at low stress levels (Fig. 10(a)). This anomalous behavior was observed for all types of crack growth processes. Initial MOR, final MOR and maximum MOR are shown in Fig. 11. Values of the maximum MOR are larger than the initial MOR for all specimens tested. In some cases, the final MOR is larger than the initial MOR. The increase in MOR has been attributed to an increase in the interfacial shear stress during long-duration cyclic loading [12]
Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 8m图 Number of cycles×10 Fig.8. Observation of fracture surface for the specimen of standard 5 90 SiC/SiC with frequency of 1 Hz at room temperature. 8 100000 200000 Number of cycles Fig 10. Relationship between MOR reduction, total crack length and number of cycles: (a) Type 1;(b) Type 2. 00 ig. 12 shows the MOR reduction at the point of specimen failure, the final MOR and the fracture stress The MOR reduction is calculated as the (final MOR/maximum MOR X100. As shown in Fig. 10, the MOR has a complicated dependence on the number of loading cycles. However, at failure the value of the MOR is reduced to about 85% of the maximum mOR. The MOR reduction merits further consideration as a para- meter for estimating fatigue damage of SiC/SiC com- 5. Conclusions (a)An SEM in-situ observation of fatigue damage Observation of fracture surface for the sp process in SiC/SiC composites was carried out at room with frequency of 10 Hz at 800oC. temperature and at 800C. Single edge V-notched speci- mens were used, and fatigue cracks initiated not only at
Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 247 Fig. 8. Observation of fracture surface for the specimen of standard SiC/SiC with frequency of 1 Hz at room temperature. Fig. 9. Observation of fracture surface for the specimen of standard SiC/SiC with frequency of 10 Hz at 800°C. Fig. 10. Relationship between MOR reduction, total crack length and number of cycles: (a) Type 1; (b) Type 2. Fig. 12 shows the MOR reduction at the point of specimen failure, the final MOR and the fracture stress. The MOR reduction is calculated as the (final MOR/maximum MOR)×100. As shown in Fig. 10, the MOR has a complicated dependence on the number of loading cycles. However, at failure the value of the MOR is reduced to about 85% of the maximum MOR. The MOR reduction merits further consideration as a parameter for estimating fatigue damage of SiC/SiC composite materials. 5. Conclusions (a) An SEM in-situ observation of fatigue damage process in SiC/SiC composites was carried out at room temperature and at 800°C. Single edge V-notched specimens were used, and fatigue cracks initiated not only at
48 Y. Miyashita et al. /International Journal of Fatigue 24(2002)241-248 300 damage mode whIc depended on the orientation of fiber bundles in of a notch and the extent of 0 (b)Independent of the type of damage mode, final failures occurred at an mOR which was about 85% of 只 the value of the mor 200 150 Shijie zhu is grateful for the Grant-in-Aid for Encour- Ministry of Education, Science, Sports and Culture, 100 Jar Q Initial MOR Maximum mor References Type 1 Type 2 [] Rouby D, Reynaud P. Fatigue behavior related to interface modi- fication during load cy y in ceramic-matrix fiber composites Fig. 11. Initial MOR, final MOR and maximum MOR for the speci- Compos Sci Technol 1993: 48: 109-18 2] Evans AG, Zok FW, McMeeking RM. Fatigue of ceramic matrix composites. Acta Metall Mater 1995: 43(3): 859-75 3] Reynaud P, Rouby D Fantozzi G, Abbe F and Peres P. Cyclic fatigue at high tempe Ires of ceramic-matrix composites. High performance. In Evans AG and Naslain R, editors. Ceramic 0 Transactions, vol. 57. Westerville(OH, USA): American Ceramic 这 14 Mizuno M, Zhu S, Nagano Y, Sakaida Y, Kagawa Y, Kaya H. Cyclic fatigue behavior of SiC/SiC composite at room and high temperatures. J Am Ceram Soc 1996, 79(12): 3065-77. 弓 [5]DiCarlo JA. Creep limitations of current polycrystalline ceramic fibers. Compos Sci Technol 1994: 51: 213-22 16] Morris WL, Cox BN, Marshall DB, Inman RV, James MR. Fatigue mechanisms in graphite/SiC composites at room and high [7 Zhu S, Kagawa Y, Mizuno M, Guo SQ, Nagano Y, Kaya H. In situ observation of cyclic fatigue crack propagation of SiC fiber/SiC composite at room temperature. Mater Sci Eng, A MOR -o-Fracture stres [8] Fox DS. Oxidation kinetics of enhanced SiC/SiC. Ceram Eng Sci Proc1995;16:877-84 [9 Zhu S, Mizuno M, Kagawa Y, Cao J, Nagano Y, Kaya H Creep Fig. 12. MOR reduction and fracture stress for the specimens tested and fatigue behavior of enhanced SiC/SiC composite at high tem- peratures. J Am Ceram Soc 1998: 81: 2269-77 [10 Karandikar PG, Chou T-W. Damage development and reductions in Nicalon-calcium aluminosilicate composite the notch tip but also at large pores remote from the static fatigue and cyclic fatigue. J Am Ceram notch. The observed widespread damage indicates that 1993:76:1720- a linear elastic fracture mechanics parameter cannot be [11] Zawada LP, Butkus LM, Hartman GA. Tensile and fatigue applied to describe the fatigue crack growth behavior of behavior of silicon carbide fiber-reinforced aluminosilicate glass SiC/SiC materials. Instead, the MOR may be used as a J Am Ceram Soc 1991: 74: 2851-8 parameter for estimating the fatigue damage proces [12] Shuler SF, Holmes JW, Wu X. Influence of loading frequency on the room-temperature fatigue of a carbon-fiber/SiC-matrix The crack behavior was divided into several types of composite. J Am Ceram Soc 1993, 76(9): 2327-36
248 Y. Miyashita et al. / International Journal of Fatigue 24 (2002) 241–248 Fig. 11. Initial MOR, final MOR and maximum MOR for the specimens tested. Fig. 12. MOR reduction and fracture stress for the specimens tested. the notch tip but also at large pores remote from the notch. The observed widespread damage indicates that a linear elastic fracture mechanics parameter cannot be applied to describe the fatigue crack growth behavior of SiC/SiC materials. Instead, the MOR may be used as a parameter for estimating the fatigue damage process. The crack behavior was divided into several types of damage modes which depended on the orientation of fiber bundles in front of a notch and the extent of matrix porosity. (b) Independent of the type of damage mode, final failures occurred at an MOR which was about 85% of the maximum value of the MOR. Acknowledgements Shijie Zhu is grateful for the Grant-in-Aid for Encouragement of Young Scientists (No. 09750105) by the Ministry of Education, Science, Sports and Culture, Japan. References [1] Rouby D, Reynaud P. Fatigue behavior related to interface modi- fication during load cycling in ceramic-matrix fiber composites. Compos Sci Technol 1993;48:109–18. [2] Evans AG, Zok FW, McMeeking RM. Fatigue of ceramic matrix composites. Acta Metall Mater 1995;43(3):859–75. [3] Reynaud P, Rouby D Fantozzi G, Abbe F and Peres P. Cyclic fatigue at high temperatures of ceramic-matrix composites. Hightemperature ceramic-matrix composites I: design, durability and performance. In Evans AG and Naslain R, editors. Ceramic Transactions, vol. 57. Westerville (OH, USA): American Ceramic Society, 1995:85–94. [4] Mizuno M, Zhu S, Nagano Y, Sakaida Y, Kagawa Y, Kaya H. Cyclic fatigue behavior of SiC/SiC composite at room and high temperatures. J Am Ceram Soc 1996;79(12):3065–77. [5] DiCarlo JA. Creep limitations of current polycrystalline ceramic fibers. Compos Sci Technol 1994;51:213–22. [6] Morris WL, Cox BN, Marshall DB, Inman RV, James MR. Fatigue mechanisms in graphite/SiC composites at room and high temperature. J Am Ceram Soc 1994;77(3):792–800. [7] Zhu S, Kagawa Y, Mizuno M, Guo SQ, Nagano Y, Kaya H. In situ observation of cyclic fatigue crack propagation of SiC- fiber/SiC composite at room temperature. Mater Sci Eng, A 1996;220:100–8. [8] Fox DS. Oxidation kinetics of enhanced SiC/SiC. Ceram Eng Sci Proc 1995;16:877–84. [9] Zhu S, Mizuno M, Kagawa Y, Cao J, Nagano Y, Kaya H. Creep and fatigue behavior of enhanced SiC/SiC composite at high temperatures. J Am Ceram Soc 1998;81:2269–77. [10] Karandikar PG, Chou T-W. Damage development and moduli reductions in Nicalon-calcium aluminosilicate composites under static fatigue and cyclic fatigue. J Am Ceram Soc 1993;76:1720–8. [11] Zawada LP, Butkus LM, Hartman GA. Tensile and fatigue behavior of silicon carbide fiber-reinforced aluminosilicate glass. J Am Ceram Soc 1991;74:2851–8. [12] Shuler SF, Holmes JW, Wu X. Influence of loading frequency on the room-temperature fatigue of a carbon-fiber/SiC-matrix composite. J Am Ceram Soc 1993;76(9):2327–36