《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_j.1551-2916.2007.01669.x[1]

J.Am. Ceram.Soc,902135-2142(2007) DOl:10.161551-2916.2007.01669x c 2007 The American Ceramic Society urna Real-Time Monitoring of Thermal Cycling damage in Ceramic Matrix Composites Under a Constant Stress Hui Mei, Laifei Cheng, Litong Zhang, Peng Fang, Zhixin Meng, and Chidong Liu National Key Laboratory of Thermostructure Composite Materials, Northwestern Polytechnical University, Xian haanxi 710072. China Under a constant stress of 50 MPa. a thermal strain with a and partial debonding stress were investigated quantita ange of 0. 2% was measured on a carbon-fiber-reinforced Sic by morscher and colleagues. 24-27Kishi and Enoki, and matrix composite(C/siC) subjected to thermal cycling between Yu et al. also successfully established a detection system of 700 and 1200 C. Acoustic emission(AE) technology was im- microcracks to estimate matrix cracking behavior in the CMCs. plemented to assist in monitoring the occurrence of damag Moreover, some analytical approaches/algorithms for AE data during testing. The monitored AE signals, together with the lave been proposed to predict the loss of mechanical propertie measured strain, were shown to have a significant dependence on and identify the different damage mechanisms in the process of repetitive temperature. In a single cycle the cycled specimens Nevertheless, only limited work has been published concern- emitted fewer acoustic emissions during heating, but as the cool- ing fiber-reinforced CMCs in the direction of the ae character ing stage approached, the emission rate increased dramatically ization of thermal shock or thermal cycling damage,35-37 As the cycle proceeded, the AE energy increased stepwise, although under similar conditions, monolithic ceramic38-4 whereas this stepwise increment per cycle continuously de- and metal matrix composites(MMCs) have received con- creased until finally it nearly disappeared at 15 cycles, after siderable attention. Unfortunately, the ae technology in these which no further increase in thermal cycle creep strain was few published papers was mainly used to monitor damage de- observed with a rate of zero, and the measured coating crack velopment or microstructural changes of the thermally shocked density reached a stable value of about 5.0 CMCs during the monotonic tensile tests for the different ther mal shock temperature and or after different cycle numbers. A comprehensive understanding of the real-time damage proces L Introduction of the CMC materials subjected to thermal cycling on the basis of AE methodology has not yet been reported in a detailed FIBER-R EINEOiRC D e ramt ic madex dm Ites(CMCs) have manner. Efforts were made in this investigation to characterize of a typical CMC applications in a variety of technological fields. The wide ran (i.e, C/SiC) by the subtle AE response, and to correlate repetitive temperatures of applications for these composites are likely to include rocket strain responses, and AE readings with the dynamic damage The monitored AE signals were mainly conducted to better understand the underlying damage mechanisms derived from these structural components are like temperature thermal cycling processes changes combined with significant mechanical loading. Even ture changes may produce thermal stresses that exceed the matrix yield stress, leading to microstructural chang es, and nonreversible damage in the composites such as matrix Il. Experimental Procedure cracking. fiber-matrix debonding and fiber fracture. The acous- (1) Materials tic emission(AE) technique provides a means to monitor the Three-dimensional (3D) preforms were braided by a four-step in situ damage evolution in the material without interrupting the method using I K carbon fibers(T-300", Toray, Tokyo, Japan) test procedure Low-pressure Isothermal-CVI was used to deposit a pyrolytic Fracture process and mechanics of the fiber-reinforced CMCs carbon interphase on the fibers and the silicon carbide matrix at during mechanical loading have been widely investigated by 1000C. The volume fractions of fibers were about 40%, and the many previous researchers, mainly based on continuous AE braiding le was about 20. Finally, test specimens were cut tative discussion of the raw data. The AE studies of damage SiC by I-CVI under the same conditions until a thicknes on monitoring of the loading procedure and qualitative or quanti from the fabricated composite plates and further coated w evolutions and failure mechanisms in the CMCs have been c50 um Fiber architectures and Pyc interphase m xperimentally conducted under monotonic tension, hyster- he fabricated unloa stepwise inc spectively. The dimensions of the dog-bone-shat pression,bending, cyclic fatigue, and even creep stress are 185 mm x3 mm x 3 mm. The properties of the as-received vated C/SiC composite specimens are listed in Table I of Mei et al basis of data analysis of acoustic emission, interfacial mechan- cal parameters between fiber and matrix such as interfacial shear strength, interfacial debonding length, interfacial sliding (2) Thermal Cycling Test As shown in Fig. 2 of reference Mei et al. thermal cycling ex E. Lara-Curzo-contributing editor periments were conducted on an integrated system, which included an induction heating furnace controlled by a pro- grammable microprocessor and a servo-hydraulic machine Model 8801, Instron Ltd, High Wycombe, UK). The micro- Manuscript No. 22247. Received September processor was set to run thermal cycles between T,=700C and National Yo No 50425208). T2=1200C with a period of 210 s, temperature difference △T≈500°C. Only the middle part of the specimens was kept 213

Real-Time Monitoring of Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress Hui Mei,w Laifei Cheng, Litong Zhang, Peng Fang, Zhixin Meng, and Chidong Liu National Key Laboratory of Thermostructure Composite Materials, Northwestern Polytechnical University, Xi’an Shaanxi 710072, China Under a constant stress of 50 MPa, a thermal strain with a range of 0.2% was measured on a carbon-fiber-reinforced SiC￾matrix composite (C/SiC) subjected to thermal cycling between 7001 and 12001C. Acoustic emission (AE) technology was im￾plemented to assist in monitoring the occurrence of damage during testing. The monitored AE signals, together with the measured strain, were shown to have a significant dependence on temperature in a single cycle and to change periodically with repetitive temperature. In a single cycle, the cycled specimens emitted fewer acoustic emissions during heating, but as the cool￾ing stage approached, the emission rate increased dramatically. As the cycle proceeded, the AE energy increased stepwise, whereas this stepwise increment per cycle continuously de￾creased until finally it nearly disappeared at 15 cycles, after which no further increase in thermal cycle creep strain was observed with a rate of zero, and the measured coating crack density reached a stable value of about 5.0 mm-1. I. Introduction FIBER-REINFORCED ceramic matrix composites (CMCs) have found, during the last two decades, numerous industrial applications in a variety of technological fields. The wide range of applications for these composites are likely to include rocket engine nozzles, thrusters, combustion chambers, divergent/con￾vergent flaps, turbomachinery, and aircraft brakes,1–3 where these structural components are likely to undergo temperature changes combined with significant mechanical loading. Even slight temperature changes may produce thermal stresses that exceed the matrix yield stress, leading to microstructural chang￾es, and nonreversible damage in the composites such as matrix cracking, fiber-matrix debonding, and fiber fracture. The acous￾tic emission (AE) technique provides a means to monitor the in situ damage evolution in the material without interrupting the test procedure. Fracture process and mechanics of the fiber-reinforced CMCs during mechanical loading have been widely investigated by many previous researchers, mainly based on continuous AE monitoring of the loading procedure and qualitative or quanti￾tative discussion of the raw data. The AE studies of damage evolutions and failure mechanisms in the CMCs have been experimentally conducted under monotonic tension,4–6 hystere￾sis reloading/unloading,7–9 stepwise incremental loading,10 com￾pression,11 bending,12,13 cyclic fatigue,14–20 and even creep stress rupture both at room and elevated temperatures.21–23 On the basis of data analysis of acoustic emission, interfacial mechan￾ical parameters between fiber and matrix such as interfacial shear strength, interfacial debonding length, interfacial sliding stress, and partial debonding stress were investigated quantita￾tively by Morscher and colleagues.24–27 Kishi and Enoki,28 and Yu et al. 29 also successfully established a detection system of microcracks to estimate matrix cracking behavior in the CMCs. Moreover, some analytical approaches/algorithms for AE data have been proposed to predict the loss of mechanical properties and identify the different damage mechanisms in the process of mechanical degradation of CMCs.15,30–34 Nevertheless, only limited work has been published concern￾ing fiber-reinforced CMCs in the direction of the AE character￾ization of thermal shock or thermal cycling damage,35–37 although under similar conditions, monolithic ceramic38–41 and metal matrix composites (MMCs)42–45 have received con￾siderable attention. Unfortunately, the AE technology in these few published papers was mainly used to monitor damage de￾velopment or microstructural changes of the thermally shocked CMCs during the monotonic tensile tests for the different ther￾mal shock temperature and/or after different cycle numbers. A comprehensive understanding of the real-time damage process of the CMC materials subjected to thermal cycling on the basis of AE methodology has not yet been reported in a detailed manner. Efforts were made in this investigation to characterize thermal cycling damage of a typical CMC (i.e., C/SiC) by the subtle AE response, and to correlate repetitive temperatures, strain responses, and AE readings with the dynamic damage. The monitored AE signals were mainly conducted to better understand the underlying damage mechanisms derived from thermal cycling processes. II. Experimental Procedure (1) Materials Three-dimensional (3D) preforms were braided by a four-step method using 1 K carbon fibers (T-300t, Toray, Tokyo, Japan). Low-pressure Isothermal-CVI was used to deposit a pyrolytic carbon interphase on the fibers and the silicon carbide matrix at 10001C. The volume fractions of fibers were about 40%, and the braiding angle was about 201. Finally, test specimens were cut from the fabricated composite plates and further coated with SiC by I-CVI under the same conditions until a thickness of
50 um. Fiber architectures and PyC interphase morphology of the fabricated composites are shown in Figs. 1(a) and (b), respectively. The dimensions of the dog-bone-shaped specimens are 185 mm 3 mm 3 mm. The properties of the as-received C/SiC composite specimens are listed in Table 1 of Mei et al. 46 (2) Thermal Cycling Test As shown in Fig. 2 of reference Mei et al. 46 thermal cycling ex￾periments were conducted on an integrated system, which included an induction heating furnace controlled by a pro￾grammable microprocessor and a servo-hydraulic machine (Model 8801, Instron Ltd., High Wycombe, UK). The micro￾processor was set to run thermal cycles between T1 5 7001C and T2 5 12001C with a period of 210 s, temperature difference DT
5001C. Only the middle part of the specimens was kept in E. Lara-Curzio—contributing editor This work was supported by the Natural Science Foundation of China (Contract No. 90405015) and National Young Elitists Foundation (Contract No. 50425208). w Author to whom correspondence should be addressed. e-mail: phdhuimei@yahoo.com Manuscript No. 22247. Received September 14, 2006; approved February 23, 2007. Journal J. Am. Ceram. Soc., 90 [7] 2135–2142 (2007) DOI: 10.1111/j.1551-2916.2007.01669.x r 2007 The American Ceramic Society 2135

2136 Journal of the American Ceramic Society-Mei et al Vol. 90. No. 7 b Matrix Fiber Fig 1. SEM micrographs showing(a)fiber architectures and(b) a pyrolytic carbon interphase in the as-prepared 3D C/SiC composites the hot zone and inert atmosphere (i.e, 99.99% argon). Surface (3) AE Monitoring temperatures along the length of the symmetric test specimen Thermal cycling tests were monitored by the ae technique. The (185 mm x3 mm x3 mm)at the upper(1200"C)and lower AE signals were detected by two highly sensitive transducers (700C)limits of thermal cycle were measured, and then plotted (Model MICRO-80D, Physical Acoustic Corp., Princeton Junc- Fig 2 as a function of distance from the specimen symmetric tion, NJ), which were directly clamped at the ends of the spec- center. It was apparent that the top constant temperature of imen with a silicone compound. AE signals were frequency 1200oC almost covered 24.5 mm of the middle part of the spec- filtered between 20 kHz and 1 MHz, pre-amplified by 40 dB imen With cooling back to 700C, the hot constant temperature and then evaluated at a threshold level of 40 db to obtain ae domain on the specimen was extended to about 51.4 mm and hits. AEWin software was used to acquire, store, and analy the temperatures on both ends of the specimen increased the signals htly owing to heat transfer from the center to the end A slight stress of 50 MPa(smatrix cracking stress) was se- lected to apply on both ends of the cycled specimen. Th Il. Results and discussion even a slight temperature gradient might produce thermal stress- es that exceeded the matrix yield stress, resulting in microstruc- (1) Monotonic Tensile behavion tural changes of the composites to produce irreversible The static tensile stress-strain behavior of the 3D C/SiC com- strain. More importantly, in the presence of the stress, the posite material was measured to rupture on an Instron tester constraint thermal strain could be directly obtained, togethe Model 1196, Instron Ltd )at a loading rate of 0.001 mm/s at with the time-dependent creep strain by a contact Instron room temperature. Figure 3 gives a typical stress-strain curve extensometer during cycles. Thermal and mechanical loading with evolution of the corresponding ae signals during tensile is applied as follows testing. The energy of ae below a proportional stress of 50 MPa (1) Heat the test specimen to the upper temperature and (i.e, matrix cracking stress) was small, and that above the pr portional limit obviously became large, indicating that the (2) Apply the constant load and balance strain to zero and onset of significant matrix cracking correlated closely to the n, start thermal cycles, and simultaneously record the strain produced ducted to identify the first-matrix cracking stress in many Finally. th iblished works. After an initiation period, the accumulated he morphologies of the thermally cycled specimens AE energy increased gradually and the AE counts increased vere observed with a scanning electron -4700, Tokyo, Japan). ith increasing inelastic strain at a stress higher than 50 MPa Multiple matrix cracking and debonding of the interfaces resulted in a macroscopic nonlinear mechanical response lead ing to this Ae activity. It is interesting to note that near the top of the loading curve, at the point where a saturated matrix 12001°0~00D. cracking state was believed to have been reached and no more matrix cracks and interfacial debonding were believed to form 1000 △T=500°C 100200300400500600700800 己800 4.5x10 4.0x10° 600 3.5x10° E400 3525x10 200 252150 10x105 0 50 35 0.00.102030405060.70809 Relation curve between temperature and distance from the en center during cooling from the upper (1200.C Fig 3. Typical tensile stress-strain curve of the 3D C/SiC composites lower(700oC) limit of the thermal cycl with real-time acoustic emission signals

the hot zone and inert atmosphere (i.e., 99.99% argon). Surface temperatures along the length of the symmetric test specimen (185 mm 3 mm 3 mm) at the upper (12001C) and lower (7001C) limits of thermal cycle were measured, and then plotted in Fig. 2 as a function of distance from the specimen symmetric center. It was apparent that the top constant temperature of 12001C almost covered 24.5 mm of the middle part of the spec￾imen. With cooling back to 7001C, the hot constant temperature domain on the specimen was extended to about 51.4 mm and the temperatures on both ends of the specimen increased slightly owing to heat transfer from the center to the ends during cooling. A slight stress of 50 MPa (
matrix cracking stress) was se￾lected to apply on both ends of the cycled specimen. Thus, even a slight temperature gradient might produce thermal stress￾es that exceeded the matrix yield stress, resulting in microstruc￾tural changes of the composites to produce irreversible strain. More importantly, in the presence of the stress, the constraint thermal strain could be directly obtained, together with the time-dependent creep strain by a contact Instron extensometer during cycles. Thermal and mechanical loading is applied as follows: (1) Heat the test specimen to the upper temperature and hold the temperature; (2) Apply the constant load and balance strain to zero; and (3) Bring the temperature down, start thermal cycles, and simultaneously record the strain produced. Finally, the morphologies of the thermally cycled specimens were observed with a scanning electron microscope (Hitachi S-4700, Tokyo, Japan). (3) AE Monitoring Thermal cycling tests were monitored by the AE technique. The AE signals were detected by two highly sensitive transducers (Model MICRO-80D, Physical Acoustic Corp., Princeton Junc￾tion, NJ), which were directly clamped at the ends of the spec￾imen with a silicone compound. AE signals were frequency filtered between 20 kHz and 1 MHz, pre-amplified by 40 dB, and then evaluated at a threshold level of 40 dB to obtain AE hits. AEWin software was used to acquire, store, and analyze the signals. III. Results and Discussion (1) Monotonic Tensile Behavior The static tensile stress–strain behavior of the 3D C/SiC com￾posite material was measured to rupture on an Instron tester (Model 1196, Instron Ltd.) at a loading rate of 0.001 mm/s at room temperature. Figure 3 gives a typical stress–strain curve with evolution of the corresponding AE signals during tensile testing. The energy of AE below a proportional stress of 50 MPa (i.e., matrix cracking stress) was small, and that above the pro￾portional limit obviously became large, indicating that the onset of significant matrix cracking correlated closely to the proportional limit stress. AE technology has been widely con￾ducted to identify the first-matrix cracking stress in many published works.4–8 After an initiation period, the accumulated AE energy increased gradually and the AE counts increased with increasing inelastic strain at a stress higher than 50 MPa. Multiple matrix cracking and debonding of the interfaces resulted in a macroscopic nonlinear mechanical response lead￾ing to this AE activity. It is interesting to note that near the top of the loading curve, at the point where a saturated matrix cracking state was believed to have been reached and no more matrix cracks and interfacial debonding were believed to form, a b Matrix Fiber Interphase 500 um 400 nm Fig. 1. SEM micrographs showing (a) fiber architectures and (b) a pyrolytic carbon interphase in the as-prepared 3D C/SiC composites. Fig. 2. Relation curve between temperature and distance from the tested C/SiC specimen center during cooling from the upper (12001C) to the lower (7001C) limit of the thermal cycle. Fig. 3. Typical tensile stress–strain curve of the 3D C/SiC composites with real-time acoustic emission signals. 2136 Journal of the American Ceramic Society—Mei et al. Vol. 90, No. 7

July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2137 0.129 (2) Strain Response During Thermal Cycles a typical strain versus thermal cycle number N curve of the C/ Sic composite under a constant stress of 50 MPa is shown in 0.043 Fig 4. It is apparent that the measured strain is cycled with the lic temperature on a progressively increasing baseline. It wa believed that the increasing strain baseline(also referred to as resulted from damage of the C/SiC composites in terms of ma trix cracking during thermal cycles. As we know. the main -0129 ethodology to increase fracture toughness and inelastic strain of CMCs is derived from the viewpoint of microcracking mech- anisms because no plastic deformation can be expected in such naterials.28.47 As shown in Figs. 6(a)and(b), the damage pro- cess of the CMCs under thermal cycles is the multiplication of coating/matrix cracks that propagate perpendicular to the tensile direction. It has been widely accepted that in case of Thermal cycling Number, N fiber-reinforced ceramic matrix composites, the inelastic strain versus thermal cycle number N curve of the 3D represents a large part of the total strain and they have their C/SiC composites during thermal cycles at a constant stress of 50 MPa urce in the sum of the crack opening displacement of the transverse crack system. The inelastic damage strain is related to the crack density p as Thermal cycle number N △ L nUc (1) 50×10 where L is the extensometer length, n the number of cracks, and Ucop crack opening displacements. Thus, the inelastic strain is mply the density of transverse matrix microcracks p multiplied Tcc strain y their aspect ratio 8( dimension parameter of a crack). During thermal cycles, as also illustrated in Fig. 6. the dam- 0.086 10-4 age mechanisms of the CMCs favored the generation of micro- 0.129 racks oriented normal to the tensile axis and the microcracks 20x10 tended to propagate rapidly inside the brittle matrix across the entire width of fiber spacing. Therefore, according to Eq .(1), anges in the inelastic strains depend significantly on the den- 3.0x10 10502100315042005250 ty of transverse matrix microcracks B, which is associated with the extent of thermal cycling damage. It is commonly recognized Time(s) that the crack density increases with increasing thermal cycle Fig.5. Change in thermal cycle creep(TCC)strain and the correspond- number N and there exists a critical Ne beyond whic ing strain rate of the 3D C/SiC composites subjected to thermal cycles and a constant stress of 50 MPa ing to a stable inelastic strain of the composite. Figure 5 shows that the TCC strain increases gradually from the initial.215% with an increase in thermal cycle number n higher modulus fibers and bundles split manne and then maintains a constant value of.121% with a strain leading to an apparent""(marked with arrow in rate of zero(right y-axis in Fig. 5)after 15 cycles(3150 s), in- on the stress-strain curve. The corresponding cumula- dicating no initiation of the new cracks and no propagation of tive ae energy increased dramatically and coincided extremely the previous cracks once the crack density becomes saturate well with the"stiffening" point on the pic mechanical bviously, the contribution of the inelastic strain to the total esponse. It is therefore feasible to correlate the observed strain was rather large within the initial 15 cycles, and the ther AE events to the actual damage phenomena occurring in the mal cycling damage mainly took place before the critical number nvestigated materia Nc of 15. Therefore, TCC strain is considered to be damage 200um the grc graphs showing(a) multiplication of the coating cracks and(b ) matrix cracks penetrating across the fiber-free ceramic matrix regions in round and polished C/SiC composites after 25 thermal cycles

higher modulus fibers and bundles split in a sudden manner, leading to an apparent ‘‘stiffening’’ (marked with arrow in Fig. 3) on the stress–strain curve. The corresponding cumula￾tive AE energy increased dramatically and coincided extremely well with the ‘‘stiffening’’ point on the macroscopic mechanical response. It is therefore feasible to correlate the observed AE events to the actual damage phenomena occurring in the investigated material. (2) Strain Response During Thermal Cycles A typical strain versus thermal cycle number N curve of the C/ SiC composite under a constant stress of 50 MPa is shown in Fig. 4. It is apparent that the measured strain is cycled with the cyclic temperature on a progressively increasing baseline. It was believed that the increasing strain baseline (also referred to as thermal cycle creep (TCC) strain46; see the left y-axis in Fig. 5) resulted from damage of the C/SiC composites in terms of ma￾trix cracking during thermal cycles. As we know, the main methodology to increase fracture toughness and inelastic strain of CMCs is derived from the viewpoint of microcracking mech￾anisms because no plastic deformation can be expected in such materials.28,47 As shown in Figs. 6(a) and (b), the damage pro￾cess of the CMCs under thermal cycles is the multiplication of coating/matrix cracks that propagate perpendicular to the tensile direction. It has been widely accepted that in case of fiber-reinforced ceramic matrix composites, the inelastic strain represents a large part of the total strain and they have their source in the sum of the crack opening displacement of the transverse crack system. The inelastic damage strain is related to the crack density b as47 ein ¼ DLin L ¼ nUCOD L ¼ db (1) where L is the extensometer length, n the number of cracks, and UCOD crack opening displacements. Thus, the inelastic strain is simply the density of transverse matrix microcracks b multiplied by their aspect ratio d (dimension parameter of a crack). During thermal cycles, as also illustrated in Fig. 6, the dam￾age mechanisms of the CMCs favored the generation of micro￾cracks oriented normal to the tensile axis and the microcracks tended to propagate rapidly inside the brittle matrix across the entire width of fiber spacing. Therefore, according to Eq. (1), changes in the inelastic strains depend significantly on the den￾sity of transverse matrix microcracks b, which is associated with the extent of thermal cycling damage. It is commonly recognized that the crack density increases with increasing thermal cycle number N and there exists a critical Nc beyond which the crack density can reach saturated with damage saturation,37,48–50 lead￾ing to a stable inelastic strain of the composite. Figure 5 shows that the TCC strain increases gradually from the initial 0.215% with an increase in thermal cycle number N, and then maintains a constant value of 0.121% with a strain rate of zero (right y-axis in Fig. 5) after 15 cycles (3150 s), in￾dicating no initiation of the new cracks and no propagation of the previous cracks once the crack density becomes saturated. Obviously, the contribution of the inelastic strain to the total strain was rather large within the initial 15 cycles, and the ther￾mal cycling damage mainly took place before the critical number Nc of 15. Therefore, TCC strain is considered to be damage Fig. 4. Typical strain versus thermal cycle number N curve of the 3D C/SiC composites during thermal cycles at a constant stress of 50 MPa. Fig. 5. Change in thermal cycle creep (TCC) strain and the correspond￾ing strain rate of the 3D C/SiC composites subjected to thermal cycles and a constant stress of 50 MPa. a b 200 um 200 um Fig. 6. Micrographs showing (a) multiplication of the coating cracks and (b) matrix cracks penetrating across the fiber-free ceramic matrix regions in the ground and polished C/SiC composites after 25 thermal cycles. July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2137

2138 Journal of the American Ceramic Society-Mei et al v20 1200 -005 8 -0.0 Heating 6-0.10 -0.10 630660690720750780810840 ime (s) ustic emission activities, specimen strain, and temperature 700800 00100011001200 14000 Fig. 7. Effect of temperature on the strain of the specimen in a single 0.00 12000 controlled and referred to as an indicator of the extent of s-0.05 10000 damage of the composites during thermal cycles, whereas cycl thermal strain is recoverable upon removal of thermal loading 0.10 12发 Thus, the progression tenden 8000 curve d led significantly on the dan controlle aseline strain. The more severe the thermal cycling damage to the composites, the wider the crack oper 630735840945105011551260 displacement and the larger the contribution of the t strain to the total strain because the actual thermal strain is cycled on the basis of the TCC strain. When the crack density 9. Rela cimen strain ic emission(AE)activ reached saturation with increasing N, thermal cycling damage nergy with the thermal became stable and no further increase in tcc strain was expected Figure 7 shows that the measured strain also has a significant C/Sic is heated up to the matrix crack closure temperature (i.e 900C), the tensile stress acting on the matrix is gradually dependence on temperature. In a certain thermal cycle, as shown relaxed to zero and on further heating, compressive stresses in Fig. 7, the strain of the composite specimen gradual increases from the initial.165% at the start point I upon build up. Hence, above this temperature, further thermal heating from 700C, reaching a peak of 0.0474% at the peak xpansion of the Sic matrix on both sides of a crack is con- point 3 of 1200oC through the point 2 of 900oC, and ther trained by the reinforcing fibers under tension, and this xpansion procedure is also suppressed by the applied tensile decreases to the final.155% at the finish point 4 with cooling stress. Thus, it becomes more and more difficult to further in- back to 700C. Evidently, a substantial strain hysteresis and a crease the strain of the specimen, leading to a lower strain rate 900"C can be explained as follows: as a consequence of the fab- 900C b inflexion of the thermal stress redistribution beyond distinct knee at the point 2 on the heating curve are observed The likely rationale for the occurrence of the distinct knee at after the ntrast,the decrease of the specimen strain is barely rication at elevated temperatures(about 1000C), all CMCs restricted during cooling. prepared by CVI technology exhibit internal residual thermal As shown in Figs. 8 and 9, temperature-dependent strain stresses at room temperature and the residual thermal stress repeats periodically with the cyclic temperature and the period may be experimentally determined by a repetitive reloading/un- of each strain cycle is also equal to 210 s completely. Figures 4 loading method or analytically calculated by the following and 9 indicate that the strain range between the upper and lower relationship nperatures in each cycle is always identical in magnitude and the average range is about 0. 2%, although the TCC strain of the specimen at the lower temperature is continuou The temperature dependence of the strain during cycling results OM- 1-v2)2 in the formation of a hysteresis loop. Hence, thermal strain range Ae can be simply estimated =OM(xm-x)△TF (2 CAT where OM is related to the material parameter, and am and ar where a is the mean linear Cte of the specimen betweer refer to the linear coefficient of thermal expansion(CTE)of the selected temperatures and AT is the temperature gradient matrix and fiber, respectively. The ATF is the temperature hermal cycling ference between the processing temperature(Tp)and the tem- between 700 and 1200 C 46 The temperature difference perature of operation (To)(i.e, ATF=Tp-To). Generally peaking, below the processing temperature, the matrix is un- approximate to 0.1946%, which is in good accordance with der tensile stresses, whereas the reinforcement is under compres- the observed experimental value of 0.2% sive stresses as the Sic matrix normally has a greater Cte that the longitudinal carbon fiber. As a result, a pre-cracked as-received condition due to the Sic matrix crack e clastic and recoverable. Upon heating from room temperature, ""matrix cracking tem ing once cooled down from the processing temperature. When a eratureturns to"matrix crack closure temperature

controlled and referred to as an indicator of the extent of damage of the composites during thermal cycles, whereas cyclic thermal strain is recoverable upon removal of thermal loading. Thus, the progression tendency of the entire cyclic thermal strain response curve depended significantly on the damage￾controlled baseline strain. The more severe the thermal cycling damage to the composites, the wider the crack opening displacement and the larger the contribution of the TCC strain to the total strain because the actual thermal strain is cycled on the basis of the TCC strain. When the crack density reached saturation with increasing N, thermal cycling damage became stable and no further increase in TCC strain was expected. Figure 7 shows that the measured strain also has a significant dependence on temperature. In a certain thermal cycle, as shown in Fig. 7, the strain of the composite specimen gradually increases from the initial 0.165% at the start point 1 upon heating from 7001C, reaching a peak of 0.0474% at the peak point 3 of 12001C through the point 2 of 9001C, and then decreases to the final 0.155% at the finish point 4 with cooling back to 7001C. Evidently, a substantial strain hysteresis and a distinct knee at the point 2 on the heating curve are observed. The likely rationale for the occurrence of the distinct knee at 9001C can be explained as follows: as a consequence of the fab￾rication at elevated temperatures (about 10001C), all CMCs prepared by CVI technology exhibit internal residual thermal stresses at room temperature and the residual thermal stress may be experimentally determined by a repetitive reloading/un￾loading method51 or analytically calculated by the following relationship52 sRES M ¼ 1 þ E1 Ef
EmEfVfðam afÞ 2 1 ð12n12Þ 2ð1n12Þ 1 E1 Ef h i
E1ð1 n12Þ DTF ¼ YMðam afÞDTF (2) where YM is related to the material parameter, and am and af refer to the linear coefficient of thermal expansion (CTE) of the matrix and fiber, respectively. The DTF is the temperature dif￾ference between the processing temperature (Tp) and the tem￾perature of operation (To) (i.e., DTF 5 TpTo). Generally speaking, below the processing temperature, the matrix is un￾der tensile stresses, whereas the reinforcement is under compres￾sive stresses as the SiC matrix normally has a greater CTE than the longitudinal carbon fiber.53 As a result, C/SiC materials have a pre-cracked as-received condition due to the SiC matrix crack￾ing once cooled down from the processing temperature. When a C/SiC is heated up to the matrix crack closure temperaturez (i.e., 9001C),54 the tensile stress acting on the matrix is gradually relaxed to zero and on further heating, compressive stresses build up. Hence, above this temperature, further thermal expansion of the SiC matrix on both sides of a crack is con￾strained by the reinforcing fibers under tension, and this expansion procedure is also suppressed by the applied tensile stress. Thus, it becomes more and more difficult to further in￾crease the strain of the specimen, leading to a lower strain rate after the inflexion of the thermal stress redistribution beyond 9001C. In contrast, the decrease of the specimen strain is barely restricted during cooling. As shown in Figs. 8 and 9, temperature-dependent strain repeats periodically with the cyclic temperature and the period of each strain cycle is also equal to 210 s completely. Figures 4 and 9 indicate that the strain range between the upper and lower temperatures in each cycle is always identical in magnitude and the average range is about 0.2%, although the TCC strain of the specimen at the lower temperature is continuously increasing. The temperature dependence of the strain during cycling results in the formation of a hysteresis loop. Hence, thermal strain range De can be simply estimated as Derange ¼ a D
T (3) where a
is the mean linear CTE of the specimen between two selected temperatures and DT is the temperature gradient during thermal cycling. In the present work, the mean CTE is about 3.8911 106 /1C between 7001 and 12001C.46 The temperature difference DT
5001C. From Eq. (3), the thermal strain range should approximate to 0.1946%, which is in good accordance with the observed experimental value of 0.2%. Fig. 7. Effect of temperature on the strain of the specimen in a single thermal cycle. Fig. 8. Acoustic emission activities, specimen strain, and temperature as measured during the fourth thermal cycle. Fig. 9. Relationship of specimen strain, acoustic emission (AE) activ￾ities, and accumulated AE energy with the thermal cycle. z SiC matrix can crack when cooled down to the ‘‘matrix cracking temperature,’’ above which and below the processing temperature the deformation of the matrix is considered to be elastic and recoverable. Upon heating from room temperature, ‘‘matrix cracking tem￾perature’’ turns to ‘‘matrix crack closure temperature.’’ 2138 Journal of the American Ceramic Society—Mei et al. Vol. 90, No. 7

July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2139 (3) AE Characterization of Damage Initiation and Evolution the first cycle, although the driving force will decrease due to During Thermal Cycle reduction in material stiffness E after cracking, it is impossibly Acoustic emission. i.e. transient elastic waves generated within a below the matrix cracking stress ome immediately and the re- material due to sudden irreversible structural change. has been sidual thermal stress between fibers and matrix is impossibly bserved and recorded during thermal cycling of the C/SiC com- relaxed completely. As a result, the matrix continues to crack posite. Figure 8 presents AE activities, s en strain. and tem- with repetitive thermal cycles until the decreasing driving force ure measured during the fourth cycle as a typical example becomes smaller than omc, where the crack density reaches As can be seen, the specimen strain reveals an absolute depen discussed in Section Ill (2). It is coml dence on temperature. Namely, the strain increases upon heat that the modulus of the composites de ch cycle. It is gene cycles due to continuous matrix cracking. 374ases with thermal that AE signals are produced -by collective microstructural dam- Besides, in a single cycle, the resultant driving stress oM act ing on the matrix changes with temperature(see Fig 9): reach- testing emit fewer acoustic emissions at the stage of heat he lower ing, but as the cooling stage approaches the emission rate in temperature limit, creases rapidly. This indicates that a large damage accompanied by multiplication of microcracking in composite materials mainly AT1>0 occurs once cooled. When a C/Sic is heated in a certain cycle, the increasing temperature renders the wide crack opening displace 50 MP ment in the matrix more homogeneously by its thermal expan sion, and then above the crack closure temperature(. e, 900C), and. declining to the minimum value odr on heating to the tensile stress acting on the fibers emerges but does not pro- upper temperature limit, duce extensive fiber fracture. A reasonable interpretation is that the high strength of the fibers, the softness of the Pyc interphase (Fig. 1(b), and the lower top temperature (only 1200C)are F=0<0.=q8Bm 0. oAC= 50MPa fracture. Thus, detectable AE hardly appears before the top cycle temperature is reached. When the C/SiC specimen is cooled This driving stress just reveals an opposite tendency with ob- down from the upper cycle temperature, the compressive stress- served thermal strain: the higher the temperature the larger the es diminish rapidly and tensile stresses in the matrix build up. aE thermal strain and the smaller the resultant driving stress oM should take place due to relaxation of the tensile stresses in the matrix by the deformation mechanism of matrix cracking, and and vice versa. Thus. a ic stress between gdr and odr is eated in the matrix over the same period as the temperature should persist down to the lower cycle temperature. During cooling, thermally induced stresses in the matrix may Generally, the appearance of significant AE at a certain reload comprise, - apart from the residual thermal stresses as men- ing level below the previous maximum-applied stress level is tioned previously in Eq (2). thermal shock-induced stress due to referred to as the felicity effect. The ratio of stress at the onset temperature gradient between the exterior and interior of the of emission to the previous peak stress, known as the Felicit sample. Thus, the driving force behind the cracking of the ratio, is an indication of the severity and extent of damage matrix can be a sum of the thermally induced stresses ar As shown in Fig 9, AE responses exhibit the apparent Felicit he applied stress, as effect as AE in each cycle is detected at stresses below the pre- vious peak stress. Assuming that emissions within the n+I cycle (4) will occur as long as the driving force exceeds the matrix crack ing stress ome, the Felicity ratio, RF, may be simply expressed as The subscript M denotes the matrix. om refers to the applied R constant stress. oM is the thermal shock-induced stress at the surface of the material, and can be classically written as' Evidently AEα△Ti activity occurred because the resultant driving stress was much igher than the stress required to propagate the cracks mce ndicating a quite low Felicity ratio associated with the appear- where modulus E, CtE a, and poissons ratio v are either matrix ance of significant damage. Subsequently, the Felicity ratio in- or volume-averaged properties. The parameter"A"is termed creased as the driving stress od decreased after cracking and the"stress reduction factor. ATk is the temperature difference the damage of each cycle gradually diminished. Fi ween the interior(Tin)and exterior(Tex) temperature of the for the applied stress, the driving stress of the matrix cracking material during thermal shock (i.e, ATk= Tin-Tex). Thus, disappeared basically because the thermally induced stress multiple matrix cracking perpendicular to the fibers occurs when was released completely with the saturation both in the crack the resultant driving stress obr along the fiber direction be- density and in the damage when Gpr comes higher than the stress required to cause matrix fracture damage-saturated composite did not exhibit any response to (referred to as matrix strength omu, which is usually low for ce- the destructive energy of thermal cycling because macro-cracks ramics and approximately equal to ome). According to Eqs. (2), of the matrix and the weakened interphase had enough spac approximately obtained on cooling from the matrix cracking Felicity ratio approached one. The notion that thermae, the 4)and(5), a tensile driving stress of at least 286 MPa can be to tolerate micro-thermal expansion. As a consequen temperature to 700c by simply substituting the material damage decreased with cycles will be further confirmed by the parameter OM in Eq. 2 using the composite modulus E. There- accumulated AE energy as later shown in Fig. 10 fore, in the first cycle, the driving force of matrix cracking may igure 9 also depicts the relationship of the thermal strain, be high enough to initiate and develop the matrix cracks. After the AE activities, and the accumulated aE energy with cycles. Hs=14285GPa×48-(-0.1×10-6/C×(900 It is interesting that the stepwise increasing AE energy is stress 6x14285GPa x 3 enhanced only at the cooling stage of each cycle, and all these Abov e the 6 micra lothe data ratd ue bo frthe selected physical parameters(i.e, thermal strain, AE hits, and 210 s as well as their excitation source of temperature. It is also

(3) AE Characterization of Damage Initiation and Evolution During Thermal Cycles Acoustic emission, i.e., transient elastic waves generated within a material due to sudden irreversible structural change, has been observed and recorded during thermal cycling of the C/SiC com￾posite. Figure 8 presents AE activities, specimen strain, and tem￾perature measured during the fourth cycle as a typical example. As can be seen, the specimen strain reveals an absolute depen￾dence on temperature. Namely, the strain increases upon heating and decreases with cooling at each cycle. It is generally accepted that AE signals are produced by collective microstructural dam￾age generation and motion.42–45 The thermally cycled specimens during testing emit fewer acoustic emissions at the stage of heat￾ing, but as the cooling stage approaches the emission rate in￾creases rapidly. This indicates that a large damage accompanied by multiplication of microcracking in composite materials mainly occurs once cooled. When a C/SiC is heated in a certain cycle, the increasing temperature renders the wide crack opening displace￾ment in the matrix more homogeneously by its thermal expan￾sion, and then above the crack closure temperature (i.e., 9001C), tensile stress acting on the fibers emerges but does not pro￾duce extensive fiber fracture. A reasonable interpretation is that the high strength of the fibers, the softness of the PyC interphase (Fig. 1(b)), and the lower top temperature (only 12001C) are advantageous in preventing the fibers from undergoing a large fracture. Thus, detectable AE hardly appears before the top cycle temperature is reached. When the C/SiC specimen is cooled down from the upper cycle temperature, the compressive stress￾es diminish rapidly and tensile stresses in the matrix build up. AE should take place due to relaxation of the tensile stresses in the matrix by the deformation mechanism of matrix cracking, and should persist down to the lower cycle temperature. During cooling, thermally induced stresses in the matrix may comprise,52 apart from the residual thermal stresses as men￾tioned previously in Eq. (2), thermal shock-induced stress due to a temperature gradient between the exterior and interior of the sample. Thus, the driving force behind the cracking of the matrix can be a sum of the thermally induced stresses and the applied stress, as sDR M ¼ sRES M þ sTS M þ sAC M (4) The subscript M denotes the matrix. sAC M refers to the applied constant stress. sTS M is the thermal shock-induced stress at the surface of the material, and can be classically written as55 sTS M ¼ AEaDTk 1 n (5) where modulus E, CTE a, and Poisson’s ratio n are either matrix or volume-averaged properties. The parameter ‘‘A’’ is termed the ‘‘stress reduction factor.’’ DTk is the temperature difference between the interior (Tin) and exterior (Tex) temperature of the material during thermal shock (i.e., DTk ¼ Tin Tex). Thus, multiple matrix cracking perpendicular to the fibers occurs when the resultant driving stress sDR M along the fiber direction be￾comes higher than the stress required to cause matrix fracture (referred to as matrix strength smu, which is usually low for ce￾ramics and approximately equal to smc). According to Eqs. (2), (4) and (5), a tensile driving stress of at least 286 MPa can be approximatily obtainedy on cooling from the matrix cracking temperature to 7001C by simply substituting the material parameter YM in Eq. 2 using the composite modulus E. There￾fore, in the first cycle, the driving force of matrix cracking may be high enough to initiate and develop the matrix cracks. After the first cycle, although the driving force will decrease due to reduction in material stiffness E after cracking, it is impossibly below the matrix cracking stress smc immediately and the re￾sidual thermal stress between fibers and matrix is impossibly relaxed completely. As a result, the matrix continues to crack with repetitive thermal cycles until the decreasing driving force becomes smaller than smc, where the crack density reaches sat￾uration as discussed in Section III (2). It is commonly accepted that the modulus of the composites decreases with thermal cycles due to continuous matrix cracking.37,49–50 Besides, in a single cycle, the resultant driving stress sDR M act￾ing on the matrix changes with temperature (see Fig. 9): reach￾ing the maximum value sDR max on cooling to the lower temperature limit, sRES M ¼ sRES max To¼700C DTF > 0 > 0;sTS M ¼ sTS max Tex¼700C DTk > 0 > 0;sAC M ¼ 50 MPa (6) and, declining to the minimum value sDR min on heating to the upper temperature limit, sRES M ¼ sRES min To¼1200C DTF < 0 < 0;sTS M ¼ sTS min Tex¼1200C DTk < 0 < 0;sAC M ¼ 50 MPa (7) This driving stress just reveals an opposite tendency with ob￾served thermal strain: the higher the temperature, the larger the thermal strain and the smaller the resultant driving stress sDR M , and vice versa. Thus, a cyclic stress between sDR min and sDR max is repeated in the matrix over the same period as the temperature and its amplitude gradually decreases with the continued cycle. Generally, the appearance of significant AE at a certain reload￾ing level below the previous maximum-applied stress level is referred to as the Felicity effect. The ratio of stress at the onset of emission to the previous peak stress, known as the Felicity ratio, is an indication of the severity and extent of damage. As shown in Fig. 9, AE responses exhibit the apparent Felicity effect as AE in each cycle is detected at stresses below the pre￾vious peak stress. Assuming that emissions within the n11 cycle will occur as long as the driving force exceeds the matrix crack￾ing stress smc, the Felicity ratio, RF, may be simply expressed as RF ¼ smcðn þ 1Þ sDR maxðnÞ (8) Evidently, during the initial several cycles, major acoustic activity occurred because the resultant driving stress was much higher than the stress required to propagate the cracks smc, indicating a quite low Felicity ratio associated with the appear￾ance of significant damage. Subsequently, the Felicity ratio in￾creased as the driving stress sDR M decreased after cracking and the damage of each cycle gradually diminished. Finally, except for the applied stress, the driving stress of the matrix cracking disappeared basically because the thermally induced stress was released completely with the saturation both in the crack density and in the damage when sDR M ¼ smc. At that time, the damage-saturated composite did not exhibit any response to the destructive energy of thermal cycling because macro-cracks of the matrix and the weakened interphase had enough space to tolerate micro-thermal expansion.56 As a consequence, the Felicity ratio approached one. The notion that thermal cycling damage decreased with cycles will be further confirmed by the accumulated AE energy as later shown in Fig. 10. Figure 9 also depicts the relationship of the thermal strain, the AE activities, and the accumulated AE energy with cycles. It is interesting that the stepwise increasing AE energy is enhanced only at the cooling stage of each cycle, and all these selected physical parameters (i.e., thermal strain, AE hits, and AE energy) follow a strict periodicity over the same period of 210 s as well as their excitation source of temperature. It is also y Residual thermal stress sRES M ¼ 142:85 GPa ½4:8 ð0:1Þ 106= C ð900 700ÞC ¼ 140 MPa; Thermal shock-induced stress sTS M ¼ 0:6 142:85 GPa 3:6 106=C ð900 700Þ  C=ð1 0:36Þ ¼ 96MPa. All the data used can be found in Mei and colleagues46,52–54. Above the matrix crack closure temperature, due to further expansion the residual thermal stress in the matrix transforms into compression, leading to decrease in the driving stress. July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2139

2140 Journal of the American Ceramic Society-Mei et al Vol. 90. No. 7 decreases with increasing cycles. The decreasing rate of AE energy reveals that the tested composite is certain to attain a steady state during which the initiation of the new cracks and/or the propagation of the previous cracks will be terminated tran- sitorily for that specific applied stress and given temperature 24000 gradient conditions. If the applied stress or temperature gradient AT is increased, more cracks would be formed and propagate. and the dynamic equilibrium between the internal microstruc ures and the external applied conditions would be disturbed until reaching a new equilibrium. Additionally, noise is unavoid- 9600 able for AE measurement and cannot be removed completely during testing. Thus, the accumulated AE energy still exhibits a 4800 ing trend after 15 cycles, although its stepwise form becomes considerably weak In addition to the above qualitative evidence, it is interesting to discuss quantitatively the damage evolution during thermal Thermal cycling Number, N cycles. An energy percentage in a certain domain between N, and n, is define Fig. 10. umulated acoustic emission energy versus thermal cycle number N curve for the C/SiC composites during the thermal cycles. En noticeable that these results imply that the main changes of the matrix structure(matrix cracking) should appear in the temper- ture intervals of the ae events. The greater the microstructural changes, the stronger the ae events and the more severe the Where En, Etotal refer to the AE energy at cycle n and the total Figure 10 presents an entire accumulated AE energy versus accumulated energy. Thus, the energy percentages per five cy thermal cycle number N curve for the composite specimen dur les’ interval are31.5%N∈[0,527.6%N∈[5,10]22.9% ing testing. The stepwise increasing AE energy is found to hav ∈[10.15]10.2%N∈[5,20].and7.8%N∈[20,25], respec a decay rate as N proceeds, and increases rather modestly ively. The AE energy before 15 cycles (i.e, 60% of the maxi- cycles. More importantly, the incremental amount of mum 25)has reached 82% of the nergy nergy per one cycle decreases with increasing N and the that the major damage takes place the initial cycles form of the AE energy almost disappears after Furthermore, the decreasing energy percentage implies that 15 cycles. This is in good agreement with the critical cycle num- the newly created damage pre-cycle diminishes gradually with ber Ne, whose TCC strain rate is zero as determined in Fig. 5. N, and eventually becomes saturated when the critical value Thus, the ae data also confirm the notion that matrix damage arrive b Fig. 11. Micrographs showing initiation and evolution of the coating cracks, as well as the matrix cracks, of the C/SiC composites with thermal cycles. be determined as: (a)1. 8 mm- after five cycles, (b)3.5 mm- after 10 cycles. (c)4 I after 15 cycles, and (d)5.1 mm-Iafter

noticeable that these results imply that the main changes of the matrix structure (matrix cracking) should appear in the temper￾ature intervals of the AE events. The greater the microstructural changes, the stronger the AE events and the more severe the damage induced. Figure 10 presents an entire accumulated AE energy versus thermal cycle number N curve for the composite specimen dur￾ing testing. The stepwise increasing AE energy is found to have a decay rate as N proceeds, and increases rather modestly after 15 cycles. More importantly, the incremental amount of the AE energy per one cycle decreases with increasing N and the stepwise form of the AE energy almost disappears after 15 cycles. This is in good agreement with the critical cycle num￾ber Nc, whose TCC strain rate is zero as determined in Fig. 5. Thus, the AE data also confirm the notion that matrix damage decreases with increasing cycles. The decreasing rate of AE energy reveals that the tested composite is certain to attain a steady state during which the initiation of the new cracks and/or the propagation of the previous cracks will be terminated tran￾sitorily for that specific applied stress and given temperature gradient conditions. If the applied stress or temperature gradient DT is increased, more cracks would be formed and propagate, and the dynamic equilibrium between the internal microstruc￾tures and the external applied conditions would be disturbed until reaching a new equilibrium. Additionally, noise is unavoid￾able for AE measurement and cannot be removed completely during testing. Thus, the accumulated AE energy still exhibits a slow increasing trend after 15 cycles, although its stepwise form becomes considerably weak. In addition to the above qualitative evidence, it is interesting to discuss quantitatively the damage evolution during thermal cycles. An energy percentage in a certain domain between N1 and N2 is defined as PE ½N1;N2 ¼ PN2 n¼N1 En Etotal % (9) Where En, Etotal refer to the AE energy at cycle n and the total accumulated energy. Thus, the energy percentages per five cy￾cles’ interval are 31.5% NA[0, 5], 27.6% NA[5, 10], 22.9% NA[10, 15], 10.2% NA[15, 20], and 7.8% NA[20, 25], respec￾tively. The AE energy before 15 cycles (i.e., 60% of the maxi￾mum 25) has reached 82% of the total energy, indicating that the major damage takes place within the initial cycles. Furthermore, the decreasing energy percentage implies that the newly created damage pre-cycle diminishes gradually with N, and eventually becomes saturated when the critical value Nc arrives. Fig. 10. Accumulated acoustic emission energy versus thermal cycle number N curve for the C/SiC composites during the thermal cycles. Fig. 11. Micrographs showing initiation and evolution of the coating cracks, as well as the matrix cracks, of the C/SiC composites with thermal cycles. The crack density can be determined as: (a) 1.8 mm1 after five cycles, (b) 3.5 mm1 after 10 cycles, (c) 4.9 mm1 after 15 cycles, and (d) 5.1 mm1 after 25 cycles, respectively. 2140 Journal of the American Ceramic Society—Mei et al. Vol. 90, No. 7

July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2141 It can also be observed from Fig. 1l that thermal cycling can M. C. Halbig. D. N. Brewer, and A J. Eckel, " Degradation of Continuo result simultaneously in the initiation of the new cracks and nposites Under Constant Load Conditions": NASA ultiplication of the preexisting cracks n the co expected, the coating cracks of the thermally cycled C/Sic com- Mechanisms in Ceramic Matrix Composite Under Longitudinal Tensile Loading. posites, as well as the matrix cracks(e. g, Figs. 6(a) and(b), ater.29,1946-6l(19 could initiate and propagate with increasing thermal cycles, ar can eventually attain stability for a given applied condition. In Emission During Tensile Testing of SiC-Fibre-Rein forced BMAS Glass- Ceramic Composites, " Comp. Part. A. 28A. 473-80(1997) Fig. ll, the crack density can be determined approximately: (a) M. E. Walter and G. Ravichandran. "" Experimental Investigation of Damage 1.8 ter five cycles, (b)3.5 mm- after 10 cycles, (c)4.9 ASME n 7. 101-8 995, Matrix Composite,". Eng. Mater. Technol. Trans tively. After 15 cycles, the crack density almost reaches a con G Morscher and A L Gyekenyesi, ""The Velocity and Attenuation of Acous- Emission Waves in SiC/SiC Composites Loaded in Tension, Compos. Sci. stant value of about 5.0 mm. Obviously, the macroscopic Techno., 62, 1171-80(2002) TCC strain and the real-time ae data correspond extremely wel "Modal Acoustic Emission of d with the microstructural changes. These results suggest that K. Sato, Y. Kagawa, H Iba, SO.Guo, H. Ka thermal cycling damage to the composites is limited and there Acoustic Emission and fracture Behavior of sic Fibe Composite Fabricated by PIP Process, "Ceram. Eng. Sci. Proc. 21 [3]407-14 tend toward a negligible level for a given applied condition R. Dharani. J. E. Goethe. S B Haug. w. P. Cai and M.A. Lankiord, Com (go Hybrid Ceramic Matrix Composites, "Ceram. Eng Under a thermal cycling between 700 and 1200C with a con- stant stress of 50 MPa, cyclic thermal strains of a 3D C/SiC Harrigan, J. Strife and A. K Dhingra. Warrendale, PA, 1985 composite with a range of 0. 2 have been measured on a pro- of the Failure Modes in Hybrid Ceramic Matrix Composites. ". Compos. Tech- gressively increasing baseline strain. The temperature dependence nol.Res,18[3]194201(1996) M. Enoki. S. Ohtake and T. Kishi. "Classification of microfracture Process of the strain in a single cycle resulted in a substantial strain hys- Type in Glass Matrix Composites by Quantitative Acoustic Emission Method, teresis. a distinct knee occurred on the strain hysteresis loop during heating, which was in large part ascribed to thermal stress M. K each, and R. D. Rawlings. "Acoustic Emission Studies (e, 900.C). No further increase in TCC strain was observe re redistribution in composites above the crack closure temperatur Fatigue-Loaded Sic Platelet-reinforced Y-TZP, J. Mater. Sci. Lett. 13, 211- the crack density became saturated after 15 cycles M. Drissi-Habti and D. Rouby, "Assessment of the Static Fatigue Beha uring testing, the monitored AE signals(hits and energy),as of the SiC Fibre- Reinforced Lithium Magnesium Aluminosilicate Glass -Cerami well as the measured strain, were found to follow a strict per dicity over the same period as their excitation temperature. AE aque and M. Rahman. ""Durability and Damage Development in w appears in certain temperature ranges, being much less during heating than during cooling. A stepwise increasing AE energy is evated Temperatures,"J.Eng Mater. Techimol Trans ASME,122,394-401(2000). aya. F. Kaya and H. Mor, "Damage Assessment of Alumina Fibre- ature cycle. moreover, the incremental amount of ae energy Ambient and Elevated Tem)2出题 per cycle gradually decreased with increasing cycles until it V Kostopoulos, Y. Z Pappas, and Y. P. Markopoulos, "Fatigue Damage tion in 3-Dimensional SiC/iC Composites. " J. Euro. Ceram. Soc.. finally nearly disappeared after 15 cycles. AE responses also 207-15( 99/E. G Henneke, and K L Reifsnider, "Damage Chara exhibited the apparent Felicity effect, and the Felicity ratio increased as the thermally induced stress was gradually relaxed a Cross-Ply SiC/CAS-Il Ceramic Composite Under Fatigue Loading Using during repetitive cooling by the deformation mechanism of ma- Real- Time Acousto-Ultrasonic NDE Technique,"J. Compos. Technol Res, 17, trix cracking. The decreasing AE energy per cycle and increasing Felicity ratio indicated that thermal cycling damage gradually Kaya, F. Kaya, and H. Mori, ""Non-Destructive Damage Evaluation of diminished with cycles until final saturation. As expected, the Using Forced Resonance and Acoustic Emission Techniques, . Mater. Sci. Lett coating and matrix cracks were observed to initiate and prop- agate with cycles, and could eventually attain stability beyond N. Morscher and A Room Temperature Creep of SiC/Sic the critical number of 15. Thus, the measured macroscopic TCC strain and the monitored real-time Ae data correspond extreme. Rupturcora sve 81- Llon BN- nterphase, SiC-Matrix Composite in Air imply that thermal cycling damage to the composite, in the form N. Morscher and J. Hurst, "Stress-Rupture and Stress-Relaxation of SiC/ SiC Composites at Intermediate Temperature, Ceram. Eng. Sci. Proc., 22. 539-46 of significant matrix cracking, mainly takes place during the(2001 cooling stage in each single cycle and before 15 cycles for a given G.N.Morscher,J. Martinez-Fernandez, and M. J. Purdy, "Determination hermal and mechanical applied condition. AE methodology nterfacial Properties Using a Single-Fiber Microcomposite Test. J Am. Ceram. that gives in situ information can be advantageously used to M. Park, E M. Chong. D. J. Yoon, and J H. Lee. "Interfacial Properti understand comprehensively the damage process of the ther- of Two SiC Fiber-Reinforced Polycarbonate Composites Using the Fragmenta- mally cycled ceramic matrix composites on Test and Acoustic Emission, Polym. Compos., 19, 747-58(1998). 2T. Kishi a.Eng,A143.103-10(199 27Y. Yamade. Y. Kawaguchi, N. Takeda, and T Kishi. "Interfacial Debondin Acknowledgments Behavior of Mullite/ SiC Continuous Fiber Composite, "J. Am. Ceran. Soc. The Program for Changjiang Scholars and Innovative Research Team in 320416(1995) university(PCSIRT) is appreciated greatly of Microcracks, "Key Eng. Mater. 127-131, 63-72(1997) 2Y. H. Yu, J. H Choi, J. H. Kweon, and D. H. Kim, "A Study on the Failure References os and y Ke Fabrication and Application of C/C, C/SiC and SiC/Sic Analysis of 2D Carbon/Carbon Using Acoustic Emission Monitoring. " NDT&E emperature Ceramic Naslain, and H. Schneider. Wiley-VCH Press, m13131576 haracterization and Acoustic Emission Monitoring of a 2-D Carbon/Carbon Composite, "Eng Fract. Mechan. R Advanced Ceramic Matrix Composite Materials for Current and Future T. Kishi and J. H. Koo,"Non-Destructive Evaluation of Engineering Propulsion Technology Applications, Acta Astronautica, 55, 409-20(2004) Ceramics, Key Eng. Mater. 2. 587-92(1999)

It can also be observed from Fig. 11 that thermal cycling can result simultaneously in the initiation of the new cracks and multiplication of the preexisting cracks in the composites. As expected, the coating cracks of the thermally cycled C/SiC com￾posites, as well as the matrix cracks (e.g., Figs. 6(a) and (b)), could initiate and propagate with increasing thermal cycles, and can eventually attain stability for a given applied condition. In Fig. 11, the crack density can be determined approximatily: (a) 1.8 mm1 after five cycles, (b) 3.5 mm1 after 10 cycles, (c) 4.9 mm1 after 15 cycles, and (d) 5.1 mm1 after 25 cycles, respec￾tively. After 15 cycles, the crack density almost reaches a con￾stant value of about 5.0 mm1 . Obviously, the macroscopic TCC strain and the real-time AE data correspond extremely well with the microstructural changes. These results suggest that thermal cycling damage to the composites is limited and there exists a critical cycle number Nc beyond which the damage will tend toward a negligible level for a given applied condition. IV. Conclusions Under a thermal cycling between 7001 and 12001C with a con￾stant stress of 50 MPa, cyclic thermal strains of a 3D C/SiC composite with a range of 0.2% have been measured on a pro￾gressively increasing baseline strain. The temperature dependence of the strain in a single cycle resulted in a substantial strain hys￾teresis. A distinct knee occurred on the strain hysteresis loop during heating, which was in large part ascribed to thermal stress redistribution in composites above the crack closure temperature (i.e., 9001C). No further increase in TCC strain was observed as the crack density became saturated after 15 cycles. During testing, the monitored AE signals (hits and energy), as well as the measured strain, were found to follow a strict peri￾odicity over the same period as their excitation temperature. AE appears in certain temperature ranges, being much less during heating than during cooling. A stepwise increasing AE energy is found to be only enhanced at the cooling stage of each temper￾ature cycle. Moreover, the incremental amount of AE energy per cycle gradually decreased with increasing cycles until it finally nearly disappeared after 15 cycles. AE responses also exhibited the apparent Felicity effect, and the Felicity ratio increased as the thermally induced stress was gradually relaxed during repetitive cooling by the deformation mechanism of ma￾trix cracking. The decreasing AE energy per cycle and increasing Felicity ratio indicated that thermal cycling damage gradually diminished with cycles until final saturation. As expected, the coating and matrix cracks were observed to initiate and prop￾agate with cycles, and could eventually attain stability beyond the critical number of 15. Thus, the measured macroscopic TCC strain and the monitored real-time AE data correspond extreme￾ly well with the observed microstructural changes. These results imply that thermal cycling damage to the composite, in the form of significant matrix cracking, mainly takes place during the cooling stage in each single cycle and before 15 cycles for a given thermal and mechanical applied condition. AE methodology that gives in situ information can be advantageously used to understand comprehensively the damage process of the ther￾mally cycled ceramic matrix composites. Acknowledgments The Program for Changjiang Scholars and Innovative Research Team in university (PCSIRT) is appreciated greatly. References 1 F. Christin, ‘‘Design, Fabrication and Application of C/C, C/SiC and SiC/SiC Composites’’; pp. 731–43 in High Temperature Ceramic Matrix Composites, Vol. 4, Edited by W. Krenkel, R. Naslain, and H. Schneider. Wiley-VCH Press, Weinheim, 2001. 2 S. Schmidt, S. Beyer, H. Knabe, H. Immich, R. Meistring, and A. Gessler, ‘‘Advanced Ceramic Matrix Composite Materials for Current and Future Propulsion Technology Applications,’’ Acta Astronautica, 55, 409–20 (2004). 3 M. C. Halbig, D. N. Brewer, and A. J. Eckel, ‘‘Degradation of Continuous Fiber Ceramic Matrix Composites Under Constant Load Conditions’’; NASA/ TM-209681, January, 2000. 4 J. J. Luo, S. C. Wooh, and I. M. Daniel, ‘‘Acoustic Emission Study of Failure Mechanisms in Ceramic Matrix Composite Under Longitudinal Tensile Loading,’’ J. Comp. Mater., 29, 1946–61 (1995). 5 M. Surgeon, E. Vanswijgenhoven, M. Wevers, and O. Van Oer Biest, ‘‘Acous￾tic Emission During Tensile Testing of SiC-Fibre-Reinforced BMAS Glass￾Ceramic Composites,’’ Comp. Part. A., 28A, 473–80 (1997). 6 M. E. Walter and G. Ravichandran, ‘‘Experimental Investigation of Damage Evolution in a Ceramic Matrix Composite,’’ J. Eng. Mater. Technol. Trans. ASME, 117, 101–8 (1995). 7 G. Morscher and A. L. Gyekenyesi, ‘‘The Velocity and Attenuation of Acous￾tic Emission Waves in SiC/SiC Composites Loaded in Tension,’’ Compos. Sci. Technol., 62, 1171–80 (2002). 8 G. N. Morscher, ‘‘Modal Acoustic Emission of Damage Accumulation in a Woven SiC/SiC Composite,’’ Compos. Sci. Technol., 59, 687–97 (1999). 9 K. Sato, Y. Kagawa, H. Iba, S. O. Guo, H. Kakisawa, and M. Mizuno, ‘‘Acoustic Emission and Fracture Behavior of SiC Fiber-Reinforced Si–N–C Ma￾trix Composite Fabricated by PIP Process,’’ Ceram. Eng. Sci. Proc., 21 [3] 407–14 (2000). 10D. R. Carroll, L. R. Dharani, J. E. Goethe, S. B. Haug, W. P. Cai, and M. A. Hall, ‘‘Damage Evolution in Hybrid Ceramic Matrix Composites,’’ Ceram. Eng. Sci. Proc., 16 [5] 949–56 (1995). 11J. Lankford, ‘‘Compressive Strength and Damage Mechanisms in a SiC-Fiber Reinforced Glass–Ceramic Matrix Composite,’’ pp. 587–602 in Proceedings of the 5th International Conference on Composite Materials, ICCM-V. Edited by W. C. Harrigan, J. Strife and A. K. Dhingra. Warrendale, PA, 1985. 12J. E. Goethe, L. R. Dharani, and D. R. Carroll, ‘‘Experimental Investigation of the Failure Modes in Hybrid Ceramic Matrix Composites,’’ J. Compos. Tech￾nol. Res., 18 [3] 194–201 (1996). 13M. Enoki, S. Ohtake, and T. Kishi, ‘‘Classification of Microfracture Process Type in Glass Matrix Composites by Quantitative Acoustic Emission Method,’’ Mater. Trans., 42 [1] 108–13 (2001). 14P. M. Kelly, C. A. Leach, and R. D. Rawlings, ‘‘Acoustic Emission Studies of Fatigue-Loaded SiC Platelet-reinforced Y–TZP,’’ J. Mater. Sci. Lett., 13, 211–2 (1994). 15M. Drissi-Habti and D. Rouby, ‘‘Assessment of the Static Fatigue Behaviour of the SiC Fibre-Reinforced Lithium Magnesium Aluminosilicate Glass–Ceramic Matrix Composite Tested Under Uniaxiale Tensile Loading,’’ Key Eng. Mat., 132– 136, 1910–3 (1997). 16A. Haque and M. Rahman, ‘‘Durability and Damage Development in Woven Ceramic Matrix Composites Under Tensile and Fatigue Loading at Room and El￾evated Temperatures,’’ J. Eng. Mater. Technol. Trans. ASME., 122, 394–401 (2000). 17C. Kaya, F. Kaya, and H. Mori, ‘‘Damage Assessment of Alumina Fibre￾Reinforced Mullite Ceramic Matrix Composites Subjected to Cyclic Fatigue at Ambient and Elevated Temperatures,’’ J. Eur. Ceram. Soc., 22, 447–52 (2002). 18V. Kostopoulos, Y. Z. Pappas, and Y. P. Markopoulos, ‘‘Fatigue Damage Accumulation in 3-Dimensional SiC/SiC Composites,’’ J. Euro. Ceram. Soc., 19, 207–15 (1999). 19A. Tiwari, E. G. Henneke, and K. L. Reifsnider, ‘‘Damage Characterization of a Cross-Ply SiC/CAS-II Ceramic Composite Under Fatigue Loading Using a Real-Time Acousto-Ultrasonic NDE Technique,’’ J. Compos. Technol. Res., 17, 221–7 (1995). 20C. Kaya, F. Kaya, and H. Mori, ‘‘Non-Destructive Damage Evaluation of Cyclic-Fatigued Alumina Fiber-Reinforced Mullite Ceramic Matrix Composites Using Forced Resonance and Acoustic Emission Techniques,’’ J. Mater. Sci. Lett., 21, 1333–5 (2002). 21G. N. Morscher and A. Gyekenyesi, ‘‘Room Temperature Creep of SiC/SiC Composites,’’ Ceram. Eng. Sci. Proc., 22, 547–52 (2001). 22G. N. Morscher, J. Hurst, and D. Brewer, ‘‘Intermediate-Temperature Stress Rupture of a Woven Hi-Nicalon, BN-Interphase, SiC-Matrix Composite in Air,’’ J. Am. Ceram. Soc., 83, 1441–9 (2000). 23G. N. Morscher and J. Hurst, ‘‘Stress-Rupture and Stress-Relaxation of SiC/ SiC Composites at Intermediate Temperature,’’ Ceram. Eng. Sci. Proc., 22, 539–46 (2001). 24G. N. Morscher, J. Martinez-Fernandez, and M. J. Purdy, ‘‘Determination of Interfacial Properties Using a Single-Fiber Microcomposite Test,’’ J. Am. Ceram. Soc., 79, 1083–91 (1996). 25J. M. Park, E. M. Chong, D. J. Yoon, and J. H. Lee, ‘‘Interfacial Properties of Two SiC Fiber-Reinforced Polycarbonate Composites Using the Fragmenta￾tion Test and Acoustic Emission,’’ Polym. Compos., 19, 747–58 (1998). 26T. Kishi, M. Enoki, and H. Tsuda, ‘‘Interface and Strength in Ceramic Matrix Composites,’’ Mater. Sci. Eng., A143, 103–10 (1991). 27Y. Yamade, Y. Kawaguchi, N. Takeda, and T. Kishi, ‘‘Interfacial Debonding Behavior of Mullite/SiC Continuous Fiber Composite,’’ J. Am. Ceram. Soc., 78, 3204–16 (1995). 28T. Kishi and M. Enoki, ‘‘Development of Ceramic Matrix Composites and the Meaning of Microcracks,’’ Key Eng. Mater., 127–131, 63–72 (1997). 29Y. H. Yu, J. H. Choi, J. H. Kweon, and D. H. Kim, ‘‘A Study on the Failure Detection of Composite Materials Using an Acoustic Emission,’’ Comp. Struct., 75, 163–9 (2006). 30Y. Z. Pappas, Y. P. Markopoulos, and V. Kostopoulos, ‘‘Failure Mechanisms Analysis of 2D Carbon/Carbon Using Acoustic Emission Monitoring,’’ NDT&E Int., 31 [3] 157–63 (1998). 31Y. Z. Pappas and V. Kostopoulos, ‘‘Toughness Characterization and Acoustic Emission Monitoring of a 2-D Carbon/Carbon Composite,’’ Eng. Fract. Mechan., 14, 1557–73 (2001). 32T. Kishi and J. H. Koo, ‘‘Non-Destructive Evaluation of Engineering Ceramics,’’ Key Eng. Mater., 2, 587–92 (1999). July 2007 Thermal Cycling Damage in Ceramic Matrix Composites Under a Constant Stress 2141

2142 Journal of the American Ceramic Society-Mei et al Vol. 90. No. 7 F. Ziegler, "Acoustic Emission from Strain-Determined Sources, "Int J. Sol- a tes Submitted to Thermal Cycling. " Mater. Sci. Eng. A234-23, 774-7 hmelik. P. Lukac. J. Kiehn. L. B. Mordike. K. U. Kainer and G. T. Z-Fiber Types. Characteristics of Thermal Cycling in a Magnesi lloy Composite, Am. Ceram.Soc,88146-5302005). Shocked Discontinuous Fid K.M.Prew."Fracture Process of Thermally P. Lukac. Z. Trojanova, F. Chmelik, and J. Kiehn. "Thermal Stress Relax Sci,32,6l53-62 2-5(goes Studied by Non-Destructive Methods, "Mater. Res.Adv. S. Jeelani, "Thermal Shock 4H. Mei. L. F. Cheng. L. T. Zhang, and Y. D. Xu, "Effect of Temperature Behavior of Nicalon Fiber Sicn Ceramic Symp.Exb,49,39546602004 llekeen, J N. Amoako, H. Mahfuz, and S. Jeelani, ""Mechanical Property egradation of a Nicalon Fiber Reinforced SiNC Ceramic Matrix Composite Technol,61,2285-97(200 F Mignard, C. Olagnon, and G Fantoni -A 78. 47-85:(2007 H. Mei, L. F. Cheng. and L. T. Zhang. "Damage Mechanisms of Cl Composites Subjected to Constant Load and Thermal Cycling in Oxidizing A umics Submitted to Thermal Shock, " J. Eur. Panda. T.S. Kannan, J. Dubois. C. Olagnon, and G. Fantozzi. Ther- H. Mei. L. E cheng, and L. T. Zhang. "Thermal Cycling Damage Mecha- mal Shock and Thermal Fatigue Study of Ceramic Materials on a Newly Devel- 中mCmM,.闭到 Constraint and Oxidizing Atme oed Ascending Thermal Shock Test Equipment, "Sci. Tecno/. Adv. Mater. X. W. Yin, L. F. Cheng. L. T. Zhang, and Y. D. Xu. "Thermal Shock Behavior of 3-Dimensional C/SiC Composite " Carbon, 40. 905-10(2002) on. B. proceedings e ter. Sci. Eng, A250[ astritseas P. A. Smith and J. A. Yeomans. "Thermal Shock Fracture in Unidirectional Fibre-Reinforced Ceramic-Matrix Composites, Comp. Sci. Te Engines. Edited by w. Bunk and H. Hausner. Deutsche Keramische, Gesellschaft. Halbig. "The Oxidation Kinetics of Continuous Carbon Fibers in a 4P. K Panda. T.S. Kannan, J. Dubois. C. Olagmon and G. Fantozzi. ""Ther mal Shock and Thermal Fatigue Study of Alumina, " J. Eur. Ceram. Soc. 22. L. F. Cheng. Y D Xu. L. T. Zhang, and Q. Zhang. "Effect of Heat Treatment on the Thermal Expansion of 2D and 3D C/SiC Composites from Room Tem- F. Chmelik J Kiehn, P Lukac K. U. Kainer. and B. L Mordike.""Acoustic perature to 1400"C, Carbon 411, 1645-87(2002) Emission and Dilatometry for Non-Destructive Characterisation of Microstruc- Sw. D. Kingery, "Factors Affecting Thermal Stress Resistance of Ceramic tural Changes in Mg Based Metal Matrix Composites Submitted to Thermal Materials, J. Am. Ceran. Soc. 38 3-15(195. H. Mei, L. F. Cheng. L. T. Zhang. X. G. Luan, and J. Zhang. ""Behavior of hmelik, Z. Trojanova, J. Kiehn, P. Lukac, and K. U. Kainer, "Non-De- structive Characterization of microstructure Evolution in Mg Based metal matrix

33F. Ziegler, ‘‘Acoustic Emission from Strain-Determined Sources,’’ Int. J. Sol￾ids Struct., 39, 5465–79 (2002). 34G. N. Morscher, H. M. Yun, and J. A. DiCarlo, ‘‘Matrix Cracking in 3D Orthogonal Melt-Infiltrated SiC/SiC Composites With Various Z-Fiber Types,’’ J. Am. Ceram. Soc., 88 [1] 146–53 (2005). 35K. Ogi, N. Takeda, and K. M. Prewo, ‘‘Fracture Process of Thermally Shocked Discontinuous Fibre-Reinforced Glass Matrix Composites under Tensile Loading,’’ J. Mater. Sci., 32, 6153–62 (1997). 36S. Salekeen, J. N. Amoako, H. Mahfuz, and S. Jeelani, ‘‘Thermal Shock Behavior of Nicalon Fiber SiCN Ceramic Matrix Composites,’’ Int. SAMPE Symp. Exhib., 49, 3954–66 (2004). 37S. Salekeen, J. N. Amoako, H. Mahfuz, and S. Jeelani, ‘‘Mechanical Property Degradation of a Nicalon Fiber Reinforced SiNC Ceramic Matrix Composite Under Thermal Shock Loading,’’ Comp. Struct., 78, 477–85 (2007). 38F. Mignard, C. Olagnon, and G. Fantozzi, ‘‘Acoustic Emission Monitoring of Damage Evaluation in Ceramics Submitted to Thermal Shock,’’ J. Eur. Ceram. Soc., 15, 651–3 (1995). 39P. K. Panda, T. S. Kannan, J. Dubois, C. Olagnon, and G. Fantozzi, ‘‘Ther￾mal Shock and Thermal Fatigue Study of Ceramic Materials on a Newly Devel￾oped Ascending Thermal Shock Test Equipment,’’ Sci. Technol. Adv. Mater., 3, 327–34 (2002). 40M. Busawon, B. Augustyniak, G. Fantozzi, and D. Rouby, ‘‘Relation Between Acoustic Emission and Damage Caused by Thermal Fatigue and Thermal Shocks in Structural Ceramics,’’ pp. 799–806 in Proceedings of the 2nd International Symposium on Ceramic Materials & Components for Engines. Edited by W. Bunk and H. Hausner. Deutsche Keramische, Gesellschaft, 1986. 41P. K. Panda, T. S. Kannan, J. Dubois, C. Olagnon, and G. Fantozzi, ‘‘Ther￾mal Shock and Thermal Fatigue Study of Alumina,’’ J. Eur. Ceram. Soc., 22, 2187–96 (2002). 42F. Chmelik, J. Kiehn, P. Lukac, K. U. Kainer, and B. L. Mordike, ‘‘Acoustic Emission and Dilatometry for Non-Destructive Characterisation of Microstruc￾tural Changes in Mg Based Metal Matrix Composites Submitted to Thermal Cycling,’’ Scr. Mater., 38, 81–7 (1997). 43F. Chmelik, Z. Trojanova, J. Kiehn, P. Lukac, and K. U. Kainer, ‘‘Non-De￾structive Characterization of Microstructure Evolution in Mg Based Metal Matrix Composites Submitted to Thermal Cycling,’’ Mater. Sci. Eng., A234-23, 774–7 (1997). 44F. Chmelik, P. Lukac, J. Kiehn, L. B. Mordike, K. U. Kainer, and G. T. Langdon, ‘‘Characteristics of Thermal Cycling in a Magnesium Alloy Composite,’’ Mater. Sci. Eng., A325, 320–3 (2002). 45P. Lukac, Z. Trojanova, F. Chmelik, and J. Kiehn, ‘‘Thermal Stress Relax￾ation in Mg Composites Studied by Non-Destructive Methods,’’ Mater. Res. Adv. Tech., 90, 892–5 (1999). 46H. Mei, L. F. Cheng, L. T. Zhang, and Y. D. Xu, ‘‘Effect of Temperature Gradients and Stress Levels on Damage of C/SiC Composites in Oxidizing Atmosphere,’’ Mater. Sci. Eng., A430, 314–9 (2006). 47S. Baste, ‘‘Inelastic Behaviour of Ceramic-Matrix Composites,’’ Comp. Sci. Technol., 61, 2285–97 (2001). 48H. Mei, L. F. Cheng, and L. T. Zhang, ‘‘Damage Mechanisms of C/SiC Composites Subjected to Constant Load and Thermal Cycling in Oxidizing At￾mosphere,’’ Scr. Mater., 54, 163–8 (2006). 49H. Mei, L. F. Cheng, and L. T. Zhang, ‘‘Thermal Cycling Damage Mecha￾nisms of C/SiC Composites in Displacement Constraint and Oxidizing Atmo￾sphere,’’ J. Am. Ceram. Soc., 89 [7] 2330–4 (2006). 50X. W. Yin, L. F. Cheng, L. T. Zhang, and Y. D. Xu, ‘‘Thermal Shock Behavior of 3-Dimensional C/SiC Composite,’’ Carbon, 40, 905–10 (2002). 51M. Steen, ‘‘Tensile Mastercurve of Ceramic Matrix Composites: Significance and Implications for Modeling,’’ Mater. Sci. Eng., A250 [2] 241–8 (1998). 52C. Kastritseas, P. A. Smith, and J. A. Yeomans, ‘‘Thermal Shock Fracture in Unidirectional Fibre-Reinforced Ceramic-Matrix Composites,’’ Comp. Sci. Tech￾nol., 65, 1880–90 (2005). 53M. C. Halbig, ‘‘The Oxidation Kinetics of Continuous Carbon Fibers in a Cracked Ceramic Matrix Composite’’; NASA/TM-2001-210520, June 2001. 54L. F. Cheng, Y.D Xu, L. T. Zhang, and Q. Zhang, ‘‘Effect of Heat Treatment on the Thermal Expansion of 2D and 3D C/SiC Composites from Room Tem￾perature to 14001C,’’ Carbon, 411, 1645–87 (2002). 55W. D. Kingery, ‘‘Factors Affecting Thermal Stress Resistance of Ceramic Materials,’’ J. Am. Ceram. Soc., 38, 3–15 (1955). 56H. Mei, L. F. Cheng, L. T. Zhang, X. G. Luan, and J. Zhang, ‘‘Behavior of Two-dimensional C/SiC Composites subjected to Thermal Cycling in Controlled Environments,’’ Carbon, 44, 121–7 (2006). & 2142 Journal of the American Ceramic Society—Mei et al. Vol. 90, No. 7