《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_Fracture mechanisms for SiC fibers and SiC/SiC composites under stress-rupture conditions at high temperatures.pdf_大学文库

J.A. DiCarlo et al. Appl Math. Comput. 152(2004)473-481 fiber-reinforced SiC matrix composites [1, 2]. Since these SiC/SiC composites are still in their infancy in terms of selecting and demonstrating the optimum fiber, interphase, and matrix constituents, there currently exists a strong need for studies that mechanistically analyze and predict the fracture- limited enve lope of thermostructual capability provided by currently available constituents Thus the objective of this paper is to present mechanistic models concerning the high-temperature stress-rupture behavior of SiC fibers and Sic/Sic com posites of current technical interest. Since the time-dependent fracture of these materials is controlled by creep-induced flaw growth, these models are based on key experimental observations made on the creep-rupture behavior for various SiC fiber types. These observations are important not only because the fibers are the primary structural constituents controlling ultimate composite rupture, but also because SiC fibers display microstructures and fracture be- havior representative of SiC matrices, which can also carry structural loads Sic/SiC composites 2. SiC fiber creep-rupture In terms of providing good thermomechanical reinforcement capability for SiC/SiC composites, small-diameter SiC fiber types with low oxygen content, such as the non-stoichiometric (C/Si>)Hi-Nicalon fiber from Nippon Car- bon and the stoichiometric(C/SiN I)Sylramic fiber from Dow Corning, have many desirable physical and chemical properties [3, 4]. For this reason, con- siderable creep-rupture property data exist for these two fiber types not only as single fibers [3-8], multifilament tows[9, 10], and woven fabric pieces [ll],but so as reinforcement for Sic/Sic composites with various types of Sic ma trices [12-15]. For example, Fig. I shows typical creep strain versus time curves ④F140c cREEPsTRAIN TIME. HRS Fig. 1. Typical creep curves for Sic fibers tested under high-temperature stress-rupture conditions

fiber-reinforced SiC matrix composites [1,2]. Since these SiC/SiC composites are still in their infancy in terms of selecting and demonstrating the optimum fiber, interphase, and matrix constituents, there currently exists a strong need for studies that mechanistically analyze and predict the fracture-limited envelope of thermostructual capability provided by currently available constituents. Thus the objective of this paper is to present mechanistic models concerning the high-temperature stress-rupture behavior of SiC fibers and SiC/SiC composites of current technical interest. Since the time-dependent fracture of these materials is controlled by creep-induced flaw growth, these models are based on key experimental observations made on the creep-rupture behavior for various SiC fiber types. These observations are important not only because the fibers are the primary structural constituents controlling ultimate composite rupture, but also because SiC fibers display microstructures and fracture behavior representative of SiC matrices, which can also carry structural loads in SiC/SiC composites. 2. SiC fiber creep-rupture In terms of providing good thermomechanical reinforcement capability for SiC/SiC composites, small-diameter SiC fiber types with low oxygen content, such as the non-stoichiometric (C/Si >1) Hi-Nicalon fiber from Nippon Carbon and the stoichiometric (C/Si 1) Sylramic fiber from Dow Corning, have many desirable physical and chemical properties [3,4]. For this reason, considerable creep-rupture property data exist for these two fiber types not only as single fibers [3–8], multifilament tows [9,10], and woven fabric pieces [11], but also as reinforcement for SiC/SiC composites with various types of SiC matrices [12–15]. For example, Fig. 1 shows typical creep strain versus time curves 0 10 20 30 40 50 0.0 0.5 1.0 1.5 2.0 2.5 TIME, HRS 1400C 270 MPa AIR : Sylramic : Hi-Nicalon. : Ultra SCS : Nicalon : SCS-6 C R E E P S T R A I N, % Fr R. Fr A D D B B E C A C E Fig. 1. Typical creep curves for SiC fibers tested under high-temperature stress-rupture conditions. 474 J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481

J.A. DiCarlo et al. /Appl. Math. Comput. 152(2004)473-481 for single Hi-Nicalon and Sylramic fibers as measured at 1400C in air under a constant stress of 270 MPa Rupture points are indicated by the symbol Fr Also shown are typical creep curves measured under the same conditions for the oxygen-containing Nicalon SiC fiber and for the oxygen-free SCS-6 and Ultra-SCS SiC fibers from Textron Specialty Materials [5, 16]. As will be dis- cussed, although the large diameters of the SCS fibers limit their formability for complex-shaped Sic/Sic components, their microstructures are similar those of SiC matrices formed by chemical vapor infiltration, a common matrix formation process for SiC/SiC composites One important observation from the Fig. I data is that all Sic fibers display a primary or transient creep stage that varies in magnitude depending on fiber type. This effect is typically observed in Si-based ceramics in which there exists an initial"" or recoverable creep component controlled by grain- boundary sliding [17]. Along with the anelastic component, there also exists a viscoelastic""or steady-state creep component, which is typically accompanied by creep-induced micro-crack formation, resulting in critical flaw growth and non-recoverable strain. Both components are enhanced by small grain size and by the high diffusion of impurity second phases in the grain boundaries. Thus the high creep behavior of the Nicalon fiber can be explained both by its ox- ygen content and its very fine grain size(5 nm); while the high creep of the H Nicalon fiber is primarily due to its fine grain size of 10 nm. The rather large transient stage for these fibers may also be caused in part by concurrent grain growth during the high-temperature stress-rupture test [7]. Impurity effects also appear to be responsible for the creep behavior of the coarse-grained (100 nm) Sylramic and SCS-6 fibers, which contain boron sintering phases and free silicon, respectively. This conclusion is supported by the observation that the Sylramic fiber transient stage decreases significantly by post-treatments that remove boron from the fiber bulk [8], and by the Fig. I creep data for the UItra-SCS fiber, which has the same grain size as the SCs-6 fiber, but no boron or excess silicon. Thus microstructure plays a strong role in the creep behavior of Sic fibers in particular and Sic-based ceramics in general Another point not shown in Fig. I is that creep rates and total creep strains for a given time period are typically increased for some fiber types by testing in an argon environment, with the effect being the largest for the Sylramic fiber [6, 8]. Although the exact mechanism responsible for this behavior is still un clear, the source appears to be related to decomposition effects. That is, testing in an argon or inert environments should enhance the gaseous volatilization of atomic species(such as silicon or its oxides) from the fiber surface and from within the fiber bulk, either along grain boundaries or through open porosity On the other hand, testing in air can form a silica coating on the fiber surface, which will effectively seal off gaseous volatilization from the bulk. As for possible volatizing species for the Sylramic fiber, this type does contain second phase impurities, such as boron and TiB. However, their removal should only

for single Hi-Nicalon and Sylramic fibers as measured at 1400 C in air under a constant stress of 270 MPa. Rupture points are indicated by the symbol Fr. Also shown are typical creep curves measured under the same conditions for the oxygen-containing Nicalon SiC fiber and for the oxygen-free SCS-6 and Ultra-SCS SiC fibers from Textron Specialty Materials [5,16]. As will be discussed, although the large diameters of the SCS fibers limit their formability for complex-shaped SiC/SiC components, their microstructures are similar to those of SiC matrices formed by chemical vapor infiltration, a common matrix formation process for SiC/SiC composites. One important observation from the Fig. 1 data is that all SiC fibers display a primary or transient creep stage that varies in magnitude depending on fiber type. This effect is typically observed in Si-based ceramics in which there exists an initial ‘‘anelastic’’ or recoverable creep component controlled by grainboundary sliding [17]. Along with the anelastic component, there also exists a ‘‘viscoelastic’’ or steady-state creep component, which is typically accompanied by creep-induced micro-crack formation, resulting in critical flaw growth and non-recoverable strain. Both components are enhanced by small grain size and by the high diffusion of impurity second phases in the grain boundaries. Thus the high creep behavior of the Nicalon fiber can be explained both by its oxygen content and its very fine grain size (5 nm); while the high creep of the HiNicalon fiber is primarily due to its fine grain size of 10 nm. The rather large transient stage for these fibers may also be caused in part by concurrent grain growth during the high-temperature stress-rupture test [7]. Impurity effects also appear to be responsible for the creep behavior of the coarse-grained (100 nm) Sylramic and SCS-6 fibers, which contain boron sintering phases and free silicon, respectively. This conclusion is supported by the observation that the Sylramic fiber transient stage decreases significantly by post-treatments that remove boron from the fiber bulk [8], and by the Fig. 1 creep data for the Ultra-SCS fiber, which has the same grain size as the SCS-6 fiber, but no boron or excess silicon. Thus microstructure plays a strong role in the creep behavior of SiC fibers in particular and SiC-based ceramics in general. Another point not shown in Fig. 1 is that creep rates and total creep strains for a given time period are typically increased for some fiber types by testing in an argon environment, with the effect being the largest for the Sylramic fiber [6,8]. Although the exact mechanism responsible for this behavior is still unclear, the source appears to be related to decomposition effects. That is, testing in an argon or inert environments should enhance the gaseous volatilization of atomic species (such as silicon or its oxides) from the fiber surface and from within the fiber bulk, either along grain boundaries or through open porosity. On the other hand, testing in air can form a silica coating on the fiber surface, which will effectively seal off gaseous volatilization from the bulk. As for possible volatizing species for the Sylramic fiber, this type does contain secondphase impurities, such as boron and TiB2. However, their removal should only J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481 475

476.A.DiCarloetal./Appl.Mat.Comput.152(2004)473-481 reduce rather than enhance creep. Therefore, at the present time, it would appear that the most likely species for volatilization is silicon from the Sic grains, which in turn can reduce grain size and enhance fiber creep rate. Indeed this effect should be especially observable in the Sylramic fiber, which has the smallest diameter(10 um)and therefore the largest surface to volume ratio of all the Fig. I fibers. Thus, from a technical point-of-view, one might expect differences in fiber creep behavior within a SiC/SiC composite, depending on whether the fiber surfaces are exposed to inert or to oxidizing environments However, for most high-temperature applications of current interest, the typ- ical service environments for the composites are oxidizing; so that even with porous interfacial zones, silica protective layers should eventually form on the Sic/SiC outer surfaces and effectively inhibit further volatilization of gaseous species from the fiber. It follows then that for mechanistic modeling of Sic/SiC creep-rupture behavior, the fiber in situ environment should be considered; but for most applications, the single fiber data in air should be appropriate A final point of interest for Fig. I is the rupture life behavior of the various SiC fiber types. In this regard, some important general observations have been made in previous studies [8, 18]. These include: (1) the activation energy con- trolling high-temperature rupture of the SiC fibers is nearly constant and equal to the activation energy controlling fiber creep;(2)fiber rupture strains for a given test temperature and test environment are fairly independent of the minimum creep rate at rupture, but are dependent on fiber type; (3)when the minimum creep rate is enhanced by argon testing, rupture times are shorter rupture strains are typically 100% higher in air than in argon; and(5)some of the fiber types rupture before clear attainment of steady-state creep behavior (see Fig. 1). The activation energy observation confirms that the most im- portant mechanism controlling the high-temperature fracture of SiC fibers(and monolithic ceramics)is creep-induced micro-crack growth. The observa of rupture strain independence on creep rate and shorter rupture times higher creep rates in argon indicate that for each fiber type, a certain amount of micro-crack growth is required to create the critical flaw size. Thus lower creep rates are conducive to longer fiber rupture lives. The observation of an approximate doubling of fiber rupture strain in air suggests that the con trolling micro-crack growth is primarily on the fiber surface, where silica for mation can blunt the stress concentration arising from this micro-crack growth. Finally, the observation of rupture before obvious steady-state creep behavior indicates that viscoelastic creep and micro-crack growth are on going The high-temperature creep curves of monolithic Si-based ceramics, such as pressureless sintered Si3N3[19], are very similar to those of the Sic fibers. For the monolithic materials, a convenient empirical approach for modeling rup- ture time is by use of Monkman-Grant(MG) diagrams that plot the log of average rupture time versus the log of minimum creep rate at various tem-

reduce rather than enhance creep. Therefore, at the present time, it would appear that the most likely species for volatilization is silicon from the SiC grains, which in turn can reduce grain size and enhance fiber creep rate. Indeed, this effect should be especially observable in the Sylramic fiber, which has the smallest diameter (10 lm) and therefore the largest surface to volume ratio of all the Fig. 1 fibers. Thus, from a technical point-of-view, one might expect differences in fiber creep behavior within a SiC/SiC composite, depending on whether the fiber surfaces are exposed to inert or to oxidizing environments. However, for most high-temperature applications of current interest, the typical service environments for the composites are oxidizing; so that even with porous interfacial zones, silica protective layers should eventually form on the SiC/SiC outer surfaces and effectively inhibit further volatilization of gaseous species from the fiber. It follows then that for mechanistic modeling of SiC/SiC creep-rupture behavior, the fiber in situ environment should be considered; but for most applications, the single fiber data in air should be appropriate. A final point of interest for Fig. 1 is the rupture life behavior of the various SiC fiber types. In this regard, some important general observations have been made in previous studies [8,18]. These include: (1) the activation energy controlling high-temperature rupture of the SiC fibers is nearly constant and equal to the activation energy controlling fiber creep; (2) fiber rupture strains for a given test temperature and test environment are fairly independent of the minimum creep rate at rupture, but are dependent on fiber type; (3) when the minimum creep rate is enhanced by argon testing, rupture times are shorter; (4) rupture strains are typically 100% higher in air than in argon; and (5) some of the fiber types rupture before clear attainment of steady-state creep behavior (see Fig. 1). The activation energy observation confirms that the most important mechanism controlling the high-temperature fracture of SiC fibers (and SiC monolithic ceramics) is creep-induced micro-crack growth. The observations of rupture strain independence on creep rate and shorter rupture times under higher creep rates in argon indicate that for each fiber type, a certain amount of micro-crack growth is required to create the critical flaw size. Thus lower creep rates are conducive to longer fiber rupture lives. The observation of an approximate doubling of fiber rupture strain in air suggests that the controlling micro-crack growth is primarily on the fiber surface, where silica formation can blunt the stress concentration arising from this micro-crack growth. Finally, the observation of rupture before obvious steady-state creep behavior indicates that viscoelastic creep and micro-crack growth are on going during the transient stage. The high-temperature creep curves of monolithic Si-based ceramics, such as pressureless sintered Si3N3 [19], are very similar to those of the SiC fibers. For the monolithic materials, a convenient empirical approach for modeling rupture time is by use of Monkman–Grant (MG) diagrams that plot the log of average rupture time versus the log of minimum creep rate at various tem- 476 J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481

peratures [20]. On MG diagrams, the log–log results at various temperatures typically fall on a set of parallel straight lines that allow average rupture time t to be described by the simple equation, tem ¼ C: ð1Þ Here e is the steady state (or minimum) creep rate, and m and C are empirically determined parameters. Using this approach, Fig. 2 shows that the rupture behavior for the five SiC fiber types of Fig. 1 can indeed be best-fit to straight MG lines using data measured at 1200 C in air and at 1400 C in air for the Hi-Nicalon fiber. For these lines, fiber creep rate is equivalent to minimum creep rate or to the instantaneous creep rate at fiber rupture since no tertiary stages were ever observed. Fig. 2 also shows that, except for the Nicalon and Hi-Nicalon fibers, the rupture lines of the different fiber types at 1200 C in air do not fall on one master line as has been observed for various alumina-based fibers [21]. This can probably be attributed in part to the measured creep rates being larger than the viscoelastic creep rates due to the presence of variable anelastic creep components at fiber rupture. However, in relation to Eq. (1), all the lines display approximately the same slope with an m exponent of 1.2. The fact that the m value is greater than unity indicates that the average fiber rupture strain from the viscoelastic component ( e t) increased with increasing rupture time or decreasing stress on the fibers. Also, by increasing test temperature from 1200 to 1400 C, all fiber lines shifted upward (>fivefold increase in rupture time for a given creep rate), like that of the Hi-Nicalon fiber in Fig. 2. However, the HiNicalon line at 1400 C displayed an m value less than unity, implying reduced 1E-10 1E-09 1E-08 1E-07 1E-06 1E-05 0.0001 0.01 0.1 1 10 100 1,000 RUPTURE TIME, hours CREEP RATE, sec -1 Sylramic (1200C) Ultra-SCS (1200C) Hi-Nicalon and Nicalon CVI SiC Matrix (1300C) (1200C) Hi-Nicalon (1400C) :: + + + D D SiC/SiC Composites + Zhu et al., 1997 (1300C) O Zhu et al., 1999 (1300C) D Hurst et al., 2001 (1315C) Fig. 2. Monkman–Grant lines measured in air for SiC fibers, SiC matrices, and SiC/SiC composites. J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481 477

J.A. DiCarlo et al. /Appl. Math. Comput. 152(2004)473-481 rupture strain with longer rupture time. This reduction in strain with time may be related to detrimental effects associated with long-term microstructural hanges or oxidation. Thus the constant C in Eq. (1)should be dependent be on SiC fiber microstructure and test temperature, but is fairly independent fiber stress or fiber cross-sectional area since the creep rate variation was ob- tained by changing the applied load on the fibers. It follows then from Fig. 2 that for a particular SiC/SiC application temperature, one cannot select fibe rupture time independently of fiber creep rate. For example, up to 1200C, the only approach for obtaining a 1000-h fiber lifetime is to assure that the posite application conditions do not create fiber creep rates more than 10 s for the creep-prone Nicalon and Hi-Nicalon fibers, and more than 10 s- for the more creep-resistant Sylramic fiber 3. SiC/SiC composite creep-rupture The single fiber results presented above indicate that many intrinsic extrinsic mechanistic factors control the high-temperature rupture of Sic types of current interest. As an initial step toward understanding how factors can in turn affect SiC/SiC composite behavior, the following discussion uses elementary composite theory and the single fiber data to develop simple mechanistic models for composite fracture at high temperatures under typical stress-rupture testing conditions. Common practice for these tests is to first raise the composites to the test temperature in ambient air, and then to increase load to a given level where creep strain is measured versus time until composite fracture. Typically the load is applied in a direction parallel to a set of fiber bundles that have an effective volume fraction of from 15% to 25% in the load direction. In this situation, there are two conditions of practical interest:(A) the maximum applied load is low enough to initially inhibit through-thickness cracking of the matrix, and(b)the maximum applied load is high enough to initially cause through-thickness cracking of the matrix, thereby leaving the fiber bundles bridging the matrix cracks and exposed to the test environment For this latter situation, the stress on the bridging fibers can be directly cal- culated from the composite stress; so that composite rupture can be predicted based on the stress dependence for creep-rupture of the reinforcing fibers. The details of one simple rupture model for condition(B)and its verification data are presented elsewhere [22, 23 For condition(A), a more desirable situation for long-term composite use the average stresses on the fibers and matrix are not constant; but change with time in a complicated manner due to differences in creep behavior between the two constituents. The initial creep rate of the uncracked composite will gen- erally be controlled by the more creep-prone constituent and the final creep rate by the more creep-resistant constituent. For modeling purposes, one can

rupture strain with longer rupture time. This reduction in strain with time may be related to detrimental effects associated with long-term microstructural changes or oxidation. Thus the constant C in Eq. (1) should be dependent both on SiC fiber microstructure and test temperature, but is fairly independent of fiber stress or fiber cross-sectional area since the creep rate variation was obtained by changing the applied load on the fibers. It follows then from Fig. 2 that for a particular SiC/SiC application temperature, one cannot select fiber rupture time independently of fiber creep rate. For example, up to 1200 C, the only approach for obtaining a 1000-h fiber lifetime is to assure that the composite application conditions do not create fiber creep rates more than 108 s1 for the creep-prone Nicalon and Hi-Nicalon fibers, and more than 109 s1 for the more creep-resistant Sylramic fiber. 3. SiC/SiC composite creep-rupture The single fiber results presented above indicate that many intrinsic and extrinsic mechanistic factors control the high-temperature rupture of SiC fiber types of current interest. As an initial step toward understanding how these factors can in turn affect SiC/SiC composite behavior, the following discussion uses elementary composite theory and the single fiber data to develop simple mechanistic models for composite fracture at high temperatures under typical stress-rupture testing conditions. Common practice for these tests is to first raise the composites to the test temperature in ambient air, and then to increase load to a given level where creep strain is measured versus time until composite fracture. Typically the load is applied in a direction parallel to a set of fiber bundles that have an effective volume fraction of from 15% to 25% in the load direction. In this situation, there are two conditions of practical interest: (A) the maximum applied load is low enough to initially inhibit through-thickness cracking of the matrix, and (B) the maximum applied load is high enough to initially cause through-thickness cracking of the matrix, thereby leaving the fiber bundles bridging the matrix cracks and exposed to the test environment. For this latter situation, the stress on the bridging fibers can be directly calculated from the composite stress; so that composite rupture can be predicted based on the stress dependence for creep-rupture of the reinforcing fibers. The details of one simple rupture model for condition (B) and its verification data are presented elsewhere [22,23]. For condition (A), a more desirable situation for long-term composite use, the average stresses on the fibers and matrix are not constant; but change with time in a complicated manner due to differences in creep behavior between the two constituents. The initial creep rate of the uncracked composite will generally be controlled by the more creep-prone constituent and the final creep rate by the more creep-resistant constituent. For modeling purposes, one can 478 J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481

J.A. DiCarlo et al. Appl Math. Comput. 152(2004)473-481 assume isostrain conditions within the composite, so that the measured creep ate of the composite is equal to that of the fiber and matrix. Assuming near steady-state conditions are reached during composite life it is proposed that the composite creep rate and the MG lines of Fig. 2 could then be used to predict which constituent will rupture first for a given SiC/SiC creep rate. To determine whether this model has any validity, it is interesting to note that most Sic/Sic composites fabricated today contain SiC matrices produced by hemical vapor infiltration(CVI). Assuming these matrices creep and rupture in a manner equivalent to that of the Ultra-SCS fiber, Fig. 2 suggests that at any composite creep rate at temperatures of 1200C or higher, the Cvi SiC matrix should crack before the Nicalon or Hi-Nicalon fibers rupture. This should be the case no matter what the fiber stress or volume fraction within the composite On the other hand, the Sylramic fibers should rupture before the CVI SiC matrix cracks Although there currently exists very little SiC/Sic data to validate the condition(A)model, one can examine two recent papers by Zhu et al. who measured the creep-rupture behavior of two types of SiC/SiC composites containing CVI SiC matrices manufactured by the same vendor For the first composite type with Nicalon fibers, data taken at 30, 45, and 60 MPa and at 1300C in air show low composite creep curves and relatively long lives until an abrupt and rapid increase in creep strain occurs, followed by composite rupture within a time period less than 5% of the original creep stage [12]. This behavior supports the prediction described above in which the first abrupt change should be caused by matrix cracking of the CVI SiC matrix, and the final failure by rupture of the fully loaded Nicalon fibers. Assuming this to be the case, one can use the composite creep rates and times at matrix cracking to plot the three [+ data points shown in Fig 2. For the second composite type with Hi-Nicalon fibers, the authors were actually able to measure CVI SiC matrix cracking from periodic hysteresis tests [13]. The two [o] data points in the Fig. 2 display these time results versus the composite creep rates at cracking. The observations that all points fall very well on the same straight line and that this line has an m value of 1. 2 strongly support the assumptions of the condition(A) model. In addition the matrix cracking line falls below that Ultra-SCS fiber at 1200C, suggesting that the CVi SiC matrix is not pure and stoichiometric like the Ultra-sCS fiber. Indeed recent raman studies CVI SiC matrices made by the same vendor have detected free silicon in the matrix bulk [24]. Thus the dashed line can be considered the empirically de termined MG line for CVI SiC matrices fabricated by this particular vendor Recently, Sic/Sic matrices have been fabricated with Sylramic fibers and with the same CVI SiC matrix from the same vendor as the major structural component of the matrix. Stress-rupture data taken in air at 1315C and composite stresses from 70 to 105 MPa show composite creep curves with a transient and steady state stage, but with a small tertiary stage before final

assume isostrain conditions within the composite, so that the measured creep rate of the composite is equal to that of the fiber and matrix. Assuming near steady-state conditions are reached during composite life, it is proposed that the composite creep rate and the MG lines of Fig. 2 could then be used to predict which constituent will rupture first for a given SiC/SiC creep rate. To determine whether this model has any validity, it is interesting to note that most SiC/SiC composites fabricated today contain SiC matrices produced by chemical vapor infiltration (CVI). Assuming these matrices creep and rupture in a manner equivalent to that of the Ultra-SCS fiber, Fig. 2 suggests that at any composite creep rate at temperatures of 1200 C or higher, the CVI SiC matrix should crack before the Nicalon or Hi-Nicalon fibers rupture. This should be the case no matter what the fiber stress or volume fraction within the composite. On the other hand, the Sylramic fibers should rupture before the CVI SiC matrix cracks. Although there currently exists very little SiC/SiC data to validate the condition (A) model, one can examine two recent papers by Zhu et al. who measured the creep-rupture behavior of two types of SiC/SiC composites containing CVI SiC matrices manufactured by the same vendor. For the first composite type with Nicalon fibers, data taken at 30, 45, and 60 MPa and at 1300 C in air show low composite creep curves and relatively long lives until an abrupt and rapid increase in creep strain occurs, followed by composite rupture within a time period less than 5% of the original creep stage [12]. This behavior supports the prediction described above in which the first abrupt change should be caused by matrix cracking of the CVI SiC matrix, and the final failure by rupture of the fully loaded Nicalon fibers. Assuming this to be the case, one can use the composite creep rates and times at matrix cracking to plot the three [+] data points shown in Fig 2. For the second composite type with Hi-Nicalon fibers, the authors were actually able to measure the times for CVI SiC matrix cracking from periodic hysteresis tests [13]. The two [O] data points in the Fig. 2 display these time results versus the composite creep rates at cracking. The observations that all points fall very well on the same straight line and that this line has an m value of 1.2 strongly support the assumptions of the condition (A) model. In addition, the matrix cracking line falls below that Ultra-SCS fiber at 1200 C, suggesting that the CVI SiC matrix is not pure and stoichiometric like the Ultra-SCS fiber. Indeed, recent Raman studies on CVI SiC matrices made by the same vendor have detected free silicon in the matrix bulk [24]. Thus the dashed line can be considered the empirically determined MG line for CVI SiC matrices fabricated by this particular vendor. Recently, SiC/SiC matrices have been fabricated with Sylramic fibers and with the same CVI SiC matrix from the same vendor as the major structural component of the matrix. Stress-rupture data taken in air at 1315 C and composite stresses from 70 to 105 MPa show composite creep curves with a transient and steady state stage, but with a small tertiary stage before final J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481 479

480A.DiCarloetal./Appl.Mat.Comput.152(2004)473 rupture [ 15]. Unlike the Nicalon composite of Shi et al. [12], there was no clear evidence that allowed an understanding of which constituent ruptured first or what mechanism controlled final rupture. However, if one plots minimum composite creep rates versus the times at initiation of the tertiary stage, the two data points in Fig. 2 show typical results. Since both the Sylramic fiber line at 1200C and the CvI Sic line at 1300C move up with a temperature in crease to 1315C, the fact that the Sylramic composite data points fall closer the estimated MG line for the CVI SiC matrix would suggest that the matrix rather than the Sylramic fiber was responsible for composite rupture. Clearly more studies are needed to validate the accuracy of this condition(A)model Nevertheless, the position of the Fig. 2 MG line for the Ultra-SCS fiber, even at 1200C, would suggest that the rupture lives of these Sic/Sic composites could be significantly improved at any creep rate (or composite stress or constituent olume fraction)if CVI SiC matrices could be developed that displayed a more stoichiometric composition than current versions 4. Summary and conclusions This paper has shown that the high-temperature fracture of Sic/Sic com- posites is primarily controlled by creep-induced flaw growth in the Sic fibers and matrices. Thus for composite applications requiring long service life, it is very important to reduce effects that enhance constituent creep both from within the microstructure(small grains, impurity phases) and from within the application(high stress, high temperature, inert environment). Despite these complicating factors, it is also shown that empirical rupture models, based on fiber and matrix Monkman-Grant diagrams, offer a simple approach for mechanistic analysis and prediction of creep-induced rupture for Sic/SiC composites and ceramic composites in general References [D. Brewer, Mater. Sci. Eng. A261(1999)284 [2T. Kameda, Y. Itoh, T. Hijikata, T. Okamura, in: Proceedings of the International Gas Turbine& Aeroengine Congress. The American Society of Mechanical Engineers. New York, NY,2000( Paper2000-GT-67) H Ichikawa, T Ishikawa. in: A. Kelly, C Zweben, T Chou(Eds ) Comprehensive Composite Materials, vol. 1, Elsevier Science Ltd, Oxford, England, 2000. Pp. 107-14 4R. E. Tressler, J.A. DiCarlo, in: R. Naslain, J. Lamon, D. Doumeingts(Eds ) Proceedings of HT-CMC-l Woodland Publishing Ltd, Cambridge, England, 1993 Proceedings of HT-CMC-2, Ceramic Transactions, vol. 57, 1993, p. 141, pp 33-49 [H.M. Yun, J.C. Goldsby, J.A. DiCarlo, in: Advances in Ceramic Matrix Composites Il. Ceramic Transactions, vol 46, 1994, p. 17 [6 H M. Yun, J.C. Goldsby, J.A. DiCarlo, in: Proceedings for HTCMC-2, Ceramic Transactions, o.57,1995,p.331

rupture [15]. Unlike the Nicalon composite of Shi et al. [12], there was no clear evidence that allowed an understanding of which constituent ruptured first or what mechanism controlled final rupture. However, if one plots minimum composite creep rates versus the times at initiation of the tertiary stage, the two [D] data points in Fig. 2 show typical results. Since both the Sylramic fiber line at 1200 C and the CVI SiC line at 1300 C move up with a temperature increase to 1315 C, the fact that the Sylramic composite data points fall closer to the estimated MG line for the CVI SiC matrix would suggest that the matrix rather than the Sylramic fiber was responsible for composite rupture. Clearly more studies are needed to validate the accuracy of this condition (A) model. Nevertheless, the position of the Fig. 2 MG line for the Ultra-SCS fiber, even at 1200 C, would suggest that the rupture lives of these SiC/SiC composites could be significantly improved at any creep rate (or composite stress or constituent volume fraction) if CVI SiC matrices could be developed that displayed a more stoichiometric composition than current versions. 4. Summary and conclusions This paper has shown that the high-temperature fracture of SiC/SiC composites is primarily controlled by creep-induced flaw growth in the SiC fibers and matrices. Thus for composite applications requiring long service life, it is very important to reduce effects that enhance constituent creep both from within the microstructure (small grains, impurity phases) and from within the application (high stress, high temperature, inert environment). Despite these complicating factors, it is also shown that empirical rupture models, based on fiber and matrix Monkman–Grant diagrams, offer a simple approach for mechanistic analysis and prediction of creep-induced rupture for SiC/SiC composites and ceramic composites in general. References [1] D. Brewer, Mater. Sci. Eng. A261 (1999) 284. [2] T. Kameda, Y. Itoh, T. Hijikata, T. Okamura, in: Proceedings of the International Gas Turbine & Aeroengine Congress, The American Society of Mechanical Engineers, New York, NY, 2000 (Paper 2000-GT-67). [3] H. Ichikawa, T. Ishikawa, in: A. Kelly, C. Zweben, T. Chou (Eds.), Comprehensive Composite Materials, vol. 1, Elsevier Science Ltd., Oxford, England, 2000, pp. 107–145. [4] R.E. Tressler, J.A. DiCarlo, in: R. Naslain, J. Lamon, D. Doumeingts (Eds.), Proceedings of HT-CMC-1, Woodland Publishing Ltd., Cambridge, England, 1993; Proceedings of HT-CMC-2, Ceramic Transactions, vol. 57, 1993, p. 141, pp. 33–49. [5] H.M. Yun, J.C. Goldsby, J.A. DiCarlo, in: Advances in Ceramic Matrix Composites II, Ceramic Transactions, vol. 46, 1994, p. 17. [6] H.M. Yun, J.C. Goldsby, J.A. DiCarlo, in: Proceedings for HTCMC-2, Ceramic Transactions, vol. 57, 1995, p. 331. 480 J.A. DiCarlo et al. / Appl. Math. Comput. 152 (2004) 473–481

《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_Fracture mechanisms for SiC fibers and SiC/SiC composites under stress-rupture conditions at high temperatures

《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_Fracture mechanisms for SiC fibers and SiC/SiC composites under stress-rupture conditions at high temperatures