COMPOSITES SCIENCE AND TECHNOLOGY ELSEⅤIER Composites Science and Technology 61(2001)1923-1930 www.elsevier.com/locate/compscitech Thermal-shock behavior of a Nicalon-fiber-reinforced hybrid glass-ceramic composite N. Chawla.*, K.K. Chawla, M. Koopman, B Patel, C. Coffin, J.I. Eldridge a Department of Chemical and Materials Engineering, Arizona State University, PO Box 876006, Tempe, AZ 85287-6006, USA Department of Materials and Mechanical Engineering, University of Alabama at Birmingham BEC 254, 1530 3rd Av. South, birmingham, AL 35294, US.A "NASA Glenn Research Center, MS 106-5, 21000 Brookpark Road, Cleveland, OH 44135, USA Received 1l January 2001; received sed form 5 June 2001; accepted 3 July 2001 Abstract A Nicalon-fiber-reinforced hybrid composite with a matrix of barium magnesium aluminosilicate(BMAs) glass with silicon carbide whiskers was subjected to thermal shock from elevated to ambient temperatures. The combination of Sic whisker and BMAS glass resulted in a hybrid matrix with a lower thermal expansion than that of the fibers, inducing tensile stresses in the fiber upon thermal shock. This stress state resulted in microstructural damage in the form of fiber cracking and cracking along the fiber/ matrix interface, as opposed to the conventional matrix cracking which is typically observed in ceramic-matrix composites. Sig- nificant damage in the composite was only observed after three thermal shock cycles. Flexural resonance measurements, used to evaluate thermal shock-induced changes in Youngs modulus, showed a reduction in modulus that correlated well with the onset of microstructural damage. Finally, fiber push-out tests, performed to evaluate changes in fiber/matrix interface strength after thermal cycling, indicated a slight decrease in interfacial strength, which was attributed to recession of the carbon-rich fiber surface during hermal shock. C 2001 Elsevier Science Ltd. All rights reserved Keywords: A Ceramic matrix composites; Whisker; Thermal shock; Damage; Fiber cracking 1. Introduction role in determining the toughness of the composite [1, 3- 7. While several CMC systems have very high strength Fiber-reinforced glass and glass-ceramic composites on account of load transfer and toughening mechanisms constitute a class of materials suitable for applications provided by a tailored fiber/matrix interface, many sys requiring a combination of lightweight, strength, and tems have a very low matrix cracking stress. This is not toughness at intermediate to elevated temperatures [1]. a desirable attribute since embrittlement in aggressive The glass or glass-ceramic matrix provides low density, environments may take place as a result of oxidation at while the fibers contribute to strength, stiffness, and the interface by oxygen ingression through the cracks toughness. Following the onset of matrix cracking in the Furthermore the higher the matrix crack stress, the composite, the presence of a weak fiber /matrix interface higher the allowable design stress for a given compo- at the tip of the growing crack leads to toughening nent. The incorporation of whiskers in the glass matrix through mechanisms such as crack blunting and deflec. can significantly increase the stress required for matrix tion, which are crucial in providing non-catastrophic cracking, and the fibers, with appropriate interface tai failure in the composite [1, 2]. Other fiber/matrix inter- loring, still provide high strength, work of fracture, and face properties, such as chemical composition, presence non-catastrophic failure [8,91 of voids, microcracking, and the microstructural stress Under thermal shock conditions the mismatch in the state near the interface, have also been shown to play a coefficient of thermal expansion of the matrix and fibers can contribute significantly to microcracking and indir Corresponding author. Tel: + 1-480-965-24 ectly to degradation in fiber strength through oxidation Consequently, the thermal history of the material, as ell as the ptibility of the material to a thermally 0266-3538/01/ S.see front matter C 2001 Elsevier Science Ltd. All rights reserved. PII:S0266-3538(01)00096
Thermal-shock behavior of a Nicalon-fiber-reinforced hybrid glass-ceramic composite N. Chawlaa,*, K.K. Chawlab, M. Koopmanb, B. Patelb, C. Coffinb, J.I. Eldridgec a Department of Chemical and Materials Engineering, Arizona State University, PO Box 876006, Tempe, AZ 85287-6006, USA bDepartment of Materials and Mechanical Engineering, University of Alabama at Birmingham, BEC 254, 1530 3rd Av. South, Birmingham, AL 35294, USA c NASA Glenn Research Center, MS 106-5, 21000 Brookpark Road, Cleveland, OH 44135, USA Received 11 January 2001; received in revised form 5 June 2001; accepted 3 July 2001 Abstract A Nicalon-fiber-reinforced hybrid composite with a matrix of barium magnesium aluminosilicate (BMAS) glass with silicon carbide whiskers was subjected to thermal shock from elevated to ambient temperatures. The combination of SiC whisker and BMAS glass resulted in a hybrid matrix with a lower thermal expansion than that of the fibers, inducing tensile stresses in the fiber upon thermal shock. This stress state resulted in microstructural damage in the form of fiber cracking and cracking along the fiber/ matrix interface, as opposed to the conventional matrix cracking which is typically observed in ceramic-matrix composites. Significant damage in the composite was only observed after three thermal shock cycles. Flexural resonance measurements, used to evaluate thermal shock-induced changes in Young’s modulus, showed a reduction in modulus that correlated well with the onset of microstructural damage. Finally, fiber push-out tests, performed to evaluate changes in fiber/matrix interface strength after thermal cycling, indicated a slight decrease in interfacial strength, which was attributed to recession of the carbon-rich fiber surface during thermal shock. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: A. Ceramic matrix composites; Whisker; Thermal shock; Damage; Fiber cracking 1. Introduction Fiber-reinforced glass and glass-ceramic composites constitute a class of materials suitable for applications requiring a combination of lightweight, strength, and toughness at intermediate to elevated temperatures [1]. The glass or glass-ceramic matrix provides low density, while the fibers contribute to strength, stiffness, and toughness. Following the onset of matrix cracking in the composite, the presence of a weak fiber/matrix interface at the tip of the growing crack leads to toughening through mechanisms such as crack blunting and deflection, which are crucial in providing non-catastrophic failure in the composite [1,2]. Other fiber/matrix interface properties, such as chemical composition, presence of voids, microcracking, and the microstructural stress state near the interface, have also been shown to play a role in determining the toughness of the composite [1,3– 7]. While several CMC systems have very high strength, on account of load transfer and toughening mechanisms provided by a tailored fiber/matrix interface, many systems have a very low matrix cracking stress. This is not a desirable attribute since embrittlement in aggressive environments may take place as a result of oxidation at the interface by oxygen ingression through the cracks. Furthermore, the higher the matrix crack stress, the higher the allowable design stress for a given component. The incorporation of whiskers in the glass matrix can significantly increase the stress required for matrix cracking, and the fibers, with appropriate interface tailoring, still provide high strength, work of fracture, and non-catastrophic failure [8,9]. Under thermal shock conditions, the mismatch in the coefficient of thermal expansion of the matrix and fibers can contribute significantly to microcracking and indirectly to degradation in fiber strength through oxidation. Consequently, the thermal history of the material, as well as the susceptibility of the material to a thermally 0266-3538/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0266-3538(01)00096-3 Composites Science and Technology 61 (2001) 1923–1930 www.elsevier.com/locate/compscitech * Corresponding author. Tel.: +1-480-965-2402; fax: +1-480-965- 0037. E-mail address: nchawla@asu.edu (N. Chawla)
1924 N. Chawla et al. Composites Science and Technology 61(2001)1923-1930 ynamic environment, play a pivotal role in determining where m is the mass of the specimen, ff is the resonant mechanical properties of the material. Thermal shock frequency of the bar under flexural loading, and b, L, studies of conventional glass and glass-ceramic matrix and t are the width, length, and thickness of the bar, composites have included Nicalon"M fibers within various respectively. Ti can be written as: matrices, including calcium aluminosilicate(CAS), lithium aluminosilicate (LAS), magnesium aluminosilicate T1=1+6.585(1+0.0752v+081092 (MAS), and borosilicate glasses [2, 3, 10-16]. For a given fiber, differences in matrix chemistry among the various composites can yield variations in interfacial properties 8340(1+0.2023+2173y2)(/L) and corresponding effects on thermal and mechanical 1.00+6338(1+0.1408v+1.536v2)t/L) properties. The present study is aimed at examining the thermal shock behavior of a novel Nicalon -fiber -rein- forced barium magnesium aluminosilicate(bMas)glass- where v is the Poissons ratio of the material [18] ceramic matrix containing SiC whiskers. The effects of Fiber push-out tests were conducted on polished cross thermal shock on the evolution of microstructural sections of approximately 2 mm thickness. Push-out damage, modulus, and interfacial shear strength of the was conducted on a micromechanical testing system, composite were investigated using a conical diamond indenter with a 10 um diameter flat region on the bottom of the indenter. Details of the fiber push-out test and the testing apparatus used in 2. Experimental procedure these experiments are given elsewhere [19, 20 Laminated unidirectional composites were pi by passing the fibers through a glass slurry containing 3. Results and discussion Sic whiskers(Corning Inc, Corning, NY). The indivi dual lamina were stacked and hot-pressed in a nitrogen The microstructure of the BMAS composite showed it atmosphere between 1400 and 1500C. The volume to be fully dense(Fig. 1). A relatively homogeneous dis- fraction of fibers and whiskers was measured to be 0.42 tribution of Nicalon fibers was observed. The glass-cera and 0.22, respectively. Table I shows selected material mic matrix contained cordierite as the primary glass properties for the composite constituents [1, 17]. Sam- phase, and a uniform distribution of SiC whiskers. Energy ples were sectioned from the as-received plate to speci- dispersive spectroscopy(EDS)(Fig. 2), revealed a homo- mens approximately 20 mm in length, 6 mm wide, and 2 geneous distribution of Si in the fibers and whiskers. mm thick, and polished prior to thermal shock, to iso- and a uniform distribution of Al, Ba, and Mg in the late polishing-induced fiber damage. Samples for ther- matrix of the composite. Areas of residual glassy phase mal shock were brought to temperature and allowed to in the matrix, perhaps unreacted during the hot pressing stabilize for 15 min prior to quenching in water at 25C. process, were also observed The Youngs modulus of each sample was determined Damage induced by thermal shock, in the form of y an impulse resonance technique GrindoSonic). cracking, was only observed after three thermal shocks After measuring the flexural resonant frequency(kHz) There was no damage observed after one or two cycles, of the sample, the Youngs modulus(GPa) was calcu- except for isolated fiber cracking on the surface. This can lated by using the relationship be attributed to the high thermal-induced tensile stresses in the near-surface region after thermal shock [14, 21] E=0.9465 (1) After three cycles, however, cracking was not confined to the surface region alone, but was also observed in the specimen interior. The fact that more than one thermal shock is needed to induce damage in the volume of the material indicates that a few thermal shocks are required Table I to induce incipient damage in the fibers, which results in Selected material properties for composite constituent [1, 17] well developed cracking after three cycles. Fig. 3 shows a region prior to and after thermal shock from 850C to oung s Coefficient Poisson room temperature. The pulled out fibers at the surface modulu raction are an artifact of polishing. The cracking seems toinitiate in the fibers, primarily transverse to the fiber axis. No longitudinal cracking in the fibers was observed. Fiber BMAS 0.25 cracking seems to be followed by propagation into the 0.22 Nicalon 0.42 matrix or along the fiber/matrix interface, see Figs. 4 and 5. Blissett et al. [11] also documented two types of
dynamic environment, play a pivotal role in determining mechanical properties of the material. Thermal shock studies of conventional glass and glass–ceramic matrix composites have included NicalonTM fibers within various matrices, including calcium aluminosilicate (CAS), lithium aluminosilicate (LAS), magnesium aluminosilicate (MAS), and borosilicate glasses [2,3,10–16]. For a given fiber, differences in matrix chemistry among the various composites can yield variations in interfacial properties and corresponding effects on thermal and mechanical properties. The present study is aimed at examining the thermal shock behavior of a novel Nicalon-fiber-reinforced barium magnesium aluminosilicate (BMAS) glass– ceramic matrix containing SiC whiskers. The effects of thermal shock on the evolution of microstructural damage, modulus, and interfacial shear strength of the composite were investigated. 2. Experimental procedure Laminated unidirectional composites were produced by passing the fibers through a glass slurry containing SiC whiskers (Corning Inc., Corning, NY). The individual lamina were stacked and hot-pressed in a nitrogen atmosphere between 1400 and 1500 C. The volume fraction of fibers and whiskers was measured to be 0.42 and 0.22, respectively. Table 1 shows selected material properties for the composite constituents [1,17]. Samples were sectioned from the as-received plate to specimens approximately 20 mm in length, 6 mm wide, and 2 mm thick, and polished prior to thermal shock, to isolate polishing-induced fiber damage. Samples for thermal shock were brought to temperature and allowed to stabilize for 15 min prior to quenching in water at 25 C. The Young’s modulus of each sample was determined by an impulse resonance technique (GrindoSonic). After measuring the flexural resonant frequency (kHz) of the sample, the Young’s modulus (GPa) was calculated by using the relationship: E ¼ 0:9465 mff 2 b L3 t3 T1 ð1Þ where m is the mass of the specimen, ff is the resonant frequency of the bar under flexural loading, and b, L, and t are the width, length, and thickness of the bar, respectively. T1 can be written as: T1 ¼ 1 þ 6:585 1 þ 0:0752 þ 081092 t L 4 8:340 1 þ 0:2023 þ 2:1732 ð Þ t=L 4 1:00 þ 6:338 1 þ 0:1408 þ 1:5362 ð Þð Þ t=L 2 " # ð2Þ where is the Poisson’s ratio of the material [18]. Fiber push-out tests were conducted on polished cross sections of approximately 2 mm thickness. Push-out was conducted on a micromechanical testing system, using a conical diamond indenter with a 10 mm diameter flat region on the bottom of the indenter. Details of the fiber push-out test and the testing apparatus used in these experiments are given elsewhere [19,20]. 3. Results and discussion The microstructure of the BMAS composite showed it to be fully dense (Fig. 1). A relatively homogeneous distribution of Nicalon fibers was observed. The glass–ceramic matrix contained cordierite as the primary glass phase, and a uniform distribution of SiC whiskers. Energy dispersive spectroscopy (EDS) (Fig. 2), revealed a homogeneous distribution of Si in the fibers and whiskers, and a uniform distribution of Al, Ba, and Mg in the matrix of the composite. Areas of residual glassy phase in the matrix, perhaps unreacted during the hot pressing process, were also observed. Damage induced by thermal shock, in the form of cracking, was only observed after three thermal shocks. There was no damage observed after one or two cycles, except for isolated fiber cracking on the surface. This can be attributed to the high thermal-induced tensile stresses in the near-surface region after thermal shock [14,21]. After three cycles, however, cracking was not confined to the surface region alone, but was also observed in the specimen interior. The fact that more than one thermal shock is needed to induce damage in the volume of the material indicates that a few thermal shocks are required to induce incipient damage in the fibers, which results in well developed cracking after three cycles. Fig. 3 shows a region prior to and after thermal shock from 850 C to room temperature. The pulled out fibers at the surface are an artifact of polishing. The cracking seems to initiate in the fibers, primarily transverse to the fiber axis. No longitudinal cracking in the fibers was observed. Fiber cracking seems to be followed by propagation into the matrix or along the fiber/matrix interface, see Figs. 4 and 5. Blissett et al. [11] also documented two types of Table 1 Selected material properties for composite constituent [1,17] Young’s modulus (GPa) Coefficient of thermal expansion (106 / C) Poisson’s ratio Volume fraction BMAS 120 2.5 0.25 0.36 SiC whisker 400 3.0 0.30 0.22 Nicalon 192 4.0 0.30 0.42 1924 N. Chawla et al. / Composites Science and Technology 61 (2001) 1923–1930
N Chawla et al. / Composites Science and Technology 61(2001)1923-1930 matrix fibe Fig. 1. Optical micrographs of BMAS hybrid composite: (a) transverse and(b)longitudinal Fig. 2. Electron microprobe images of BMAS hybrid composite. (a) Backscattered electron image. The Nicalon fiber and BMAs matrix are indi- cated by arrows. Other arrows point to the glassy phase regions. (b)SiKa X-ray map. Si is seen in the Nicalon fiber as well as in SiC whiskers in the matrix.(c)AFKa X-ray map. Al2O3 is well distributed in the glass-ceramic matrix (d) Ba La X-ray map. Ba is evenly distributed in the glass- ceramic matrIx
Fig. 2. Electron microprobe images of BMAS hybrid composite. (a) Backscattered electron image. The Nicalon fiber and BMAS matrix are indicated by arrows. Other arrows point to the glassy phase regions. (b) Si–K X-ray map. Si is seen in the Nicalon fiber as well as in SiC whiskers in the matrix. (c) Al–K X-ray map. Al2O3 is well distributed in the glass–ceramic matrix. (d) Ba L X-ray map. Ba is evenly distributed in the glass– ceramic matrix. Fig. 1. Optical micrographs of BMAS hybrid composite: (a) transverse and (b) longitudinal. N. Chawla et al. / Composites Science and Technology 61 (2001) 1923–1930 1925
1926 N. Chawla et al. Composites Science and Technology 61(2001)1923-1930 -oo um Fig. 3. Optical micrographs of BMAS hybrid composite:(a) prior to thermal shock and (b) following three thermal shock cycles. Notice significant fiber cracking after shock. 的20mm Fig 4. Optical micrographs of BMAs hybrid composite: (a) prior to thermal shock and (b)following three thermal shock cycles. Notice cracking along the fiber matrix interface. in addition to fiber cracking. interface track her crack 10m Fig. 5. Higher magnification of Fig. 4:(a) prior to thermal shock and(b) following three thermal shock cycles. Notice cracking along the fiber/ matrix interface, in addition to fiber cracking matrix cracking unidirectional reinforced Nicalon/ discussing this point in some detail because it is opposite CAS. Single cracks along the fiber/matrix interface and of the commonly encountered situation in CMCs. Since multiple cracks perpendicular to fibers were observed he matrix itself is a composite of Sic whiskers and The phenomenon of fiber cracking due to thermal BMAS matrix, the thermal expansion of the matrix of shock, rather than the typically observed matrix crack the composite can be treated as a composite. We use the ing, can be explained by examining the thermal expan- approach of Turner [22] in estimating the thermal sion of the matrix and that of the fiber. It is worth expansion of a two-phase homogeneous, particulate
matrix cracking in unidirectional reinforced Nicalon/ CAS. Single cracks along the fiber/matrix interface and multiple cracks perpendicular to fibers were observed. The phenomenon of fiber cracking due to thermal shock, rather than the typically observed matrix cracking, can be explained by examining the thermal expansion of the matrix and that of the fiber. It is worth discussing this point in some detail because it is opposite of the commonly encountered situation in CMCs. Since the matrix itself is a composite of SiC whiskers and BMAS matrix, the thermal expansion of the matrix of the composite can be treated as a composite. We use the approach of Turner [22] in estimating the thermal expansion of a two-phase homogeneous, particulate Fig. 3. Optical micrographs of BMAS hybrid composite: (a) prior to thermal shock and (b) following three thermal shock cycles. Notice significant fiber cracking after shock. Fig. 4. Optical micrographs of BMAS hybrid composite: (a) prior to thermal shock and (b) following three thermal shock cycles. Notice cracking along the fiber matrix interface, in addition to fiber cracking. Fig. 5. Higher magnification of Fig. 4: (a) prior to thermal shock and (b) following three thermal shock cycles. Notice cracking along the fiber/ matrix interface, in addition to fiber cracking. 1926 N. Chawla et al. / Composites Science and Technology 61 (2001) 1923–1930
N Chawla et al. / Composites Science and Technology 61(2001)1923-1930 composite, since the whiskers are randomly distributed The use of the Sic whiskers in the low expansion The thermal expansion of the composite hybrid matrix, BMAS matrix also results in a composite with reduced Chm, is given by [22] thermal expansion, vis a vis that of the fiber. Using the upper and lower bounds for elastic modulus of the hybrid awVwKw+aBMAS VBMAS KBMAS (3) matrix, upper and lower bounds of the thermal expansion WkW+VBMAS KBMAS of the composite in the longitudinal direction were calcu- lated, using the following expression by Schapery [24] where K 5 and aw, Vw, aBMAs, and vBmas, are he thermal expansion and volume fraction of the whis- o m Em/m +arErkr ker and BMAS matrix, respectively. The elastic mod Emim+ erve ulus of the hybrid matrix was then calculated using the approach by Hashin and Shtrikman [23], that prescribes The difference in upper and lower bound of the elastic upper and lower bounds for the elastic moduli of the modulus of the hybrid matrix did not significantly affect composite. The upper and lower bounds on the bulk the thermal expansion of the composite, which was calcu modulus of the hybrid matrix, Khm, are given by [23] lated as 3.35x10-b/C. This compares very well with the experimentally determined value of 3. 0x10-/C [25] The fiber damage was evaluated in terms of a decrease upper =Kw+ (4) in Young's modulus of the composite In the 850C sam- ple after three cycles, a measured decrease in Youngs KBMAS - Kw 3Kw+4G modulus from 155 to 85 GPa was observed. Thermal shock after one cycle resulted in no decrease in modulus Khm. lower KBMAS which once again, correlated well with microstructural observations. In thermal cycling of this composite, the microstructures were essentially crack-free[26]. The non- destructive nature of modulus measurement would seem to be more attractive than destructive testing, e.g., flex- Kw-kbmas 3KBMAS + 4GBMAS ural testing after thermal shock [27]. It is important to (5) note that flexural testing may not adequately quantify damage in thermal shocked materials. because of the where G is the shear modulus and is given by G=x(+v. combination of tensile, compressive, and shear stresses The computed Youngs modulus of the hybrid matrix involved. Thus, while a predominantly tensile mode and the coefficient of thermal expansion, in comparison may operate in the undamaged composite, the presence to that of Nicalon fiber, are given in Table 2 of microcracks in the damaged state may reduce the In a conventional glass-ceramic matrix (with no interlaminar shear strength and the composite may fail whiskers) reinforced with SiC fibers, am will be higher in a shear mode [14]. Furthermore, shifting of the neu than af, so under thermal shock conditions the matrix tral axis in continuous fiber ceramic composites results would be expected to be in tension, and cracking due to in a significant enhancement of the strength [28, 29 thermal stresses should initiate in the matrix, Table 3 It is interesting to note that microstructural damage indicates that despite the addition of Sic whiskers to the and decrease in modulus are also very much affected by matrix(the Sic whiskers have a similar CtE to that of the hold time prior to thermal shock. Blissett et al. [10] BMAS matrix, shown in Table 3), the hybrid matrix still determined the thermal shock resistance of unidirec has a lower Cte than that of the fiber, so on quenching, tional and cross-plied Nicalon fiber reinforced CAs the fiber is in tension, and cracks initiate in the fiber. It is glass-ceramic composites. The composite specimens were important to point out that, from a design standpoint, allowed to age for one hour prior to shock. Their results loading of the fibers in tension, rather than the matrix, indicated a strength decrease at intermediate tempera would seem to be a more desirable loading configuration tures followed by a perceived"recovery"in strength because of the higher fiber strength Table 3 Table 2 Debond and sliding stresses of Nicalon fibers in bmas. as measured Elastic and thermal properties of hybrid matrix and Nicalon fiber by fiber pushout testing Youngs Coefficient of Volume fraction Fiber diameter nodulus thermal expansion (GPa)(10-6C) AS- received 179±29 16.4士5.1 124±4.6 Hybrid matrix Shocked,500°C 23.2±19 8.7±1.9 Nicalon fiber Shocked,950°C 214±24 13.2±6.8 9.1士3.9
composite, since the whiskers are randomly distributed. The thermal expansion of the composite hybrid matrix, hm, is given by [22]: hm ¼ wVwKw þ BMASVBMASKBMAS VwKw þ VBMASKBMAS ð3Þ where K ¼ E 3 1ð Þ 2 , and w, Vw, BMAS, and VBMAS, are the thermal expansion and volume fraction of the whisker and BMAS matrix, respectively. The elastic modulus of the hybrid matrix was then calculated using the approach by Hashin and Shtrikman [23], that prescribes upper and lower bounds for the elastic moduli of the composite. The upper and lower bounds on the bulk modulus of the hybrid matrix, Khm, are given by [23]: Khm; upper ¼ Kw þ 1 Vw 1 KBMAS Kw þ 3Vw 3Kw þ 4Gw 2 6 6 4 3 7 7 5 ð4Þ Khm; lower ¼ KBMAS þ Vw 1 Kw KBMAS þ 3 1ð Þ Vw 3KBMAS þ 4GBMAS 2 6 6 4 3 7 7 5 ð5Þ where G is the shear modulus and is given by G ¼ E 2 1ð Þ þ . The computed Young’s modulus of the hybrid matrix and the coefficient of thermal expansion, in comparison to that of Nicalon fiber, are given in Table 2. In a conventional glass–ceramic matrix (with no whiskers) reinforced with SiC fibers, m will be higher than f, so under thermal shock conditions the matrix would be expected to be in tension, and cracking due to thermal stresses should initiate in the matrix. Table 3 indicates that despite the addition of SiC whiskers to the matrix (the SiC whiskers have a similar CTE to that of the BMAS matrix, shown in Table 3), the hybrid matrix still has a lower CTE than that of the fiber, so on quenching, the fiber is in tension, and cracks initiate in the fiber. It is important to point out that, from a design standpoint, loading of the fibers in tension, rather than the matrix, would seem to be a more desirable loading configuration because of the higher fiber strength. The use of the SiC whiskers in the low expansion BMAS matrix also results in a composite with reduced thermal expansion, vis a` vis that of the fiber. Using the upper and lower bounds for elastic modulus of the hybrid matrix, upper and lower bounds of the thermal expansion of the composite in the longitudinal direction were calculated, using the following expression by Schapery [24]: cl ¼ mEmVm þ fEfVf EmVm þ EfVf ð6Þ The difference in upper and lower bound of the elastic modulus of the hybrid matrix did not significantly affect the thermal expansion of the composite, which was calculated as 3.35106 / C. This compares very well with the experimentally determined value of 3.0106 / C [25]. The fiber damage was evaluated in terms of a decrease in Young’s modulus of the composite. In the 850 C sample after three cycles, a measured decrease in Young’s modulus from 155 to 85 GPa was observed. Thermal shock after one cycle resulted in no decrease in modulus, which once again, correlated well with microstructural observations. In thermal cycling of this composite, the microstructures were essentially crack-free [26]. The nondestructive nature of modulus measurement would seem to be more attractive than destructive testing, e.g., flexural testing after thermal shock [27]. It is important to note that flexural testing may not adequately quantify damage in thermal shocked materials, because of the combination of tensile, compressive, and shear stresses involved. Thus, while a predominantly tensile mode may operate in the undamaged composite, the presence of microcracks in the damaged state may reduce the interlaminar shear strength and the composite may fail in a shear mode [14]. Furthermore, shifting of the neutral axis in continuous fiber ceramic composites results in a significant enhancement of the strength [28,29]. It is interesting to note that microstructural damage and decrease in modulus are also very much affected by hold time prior to thermal shock. Blissett et al. [10] determined the thermal shock resistance of unidirectional and cross-plied Nicalon fiber reinforced CAS glass-ceramic composites. The composite specimens were allowed to age for one hour prior to shock. Their results indicated a strength decrease at intermediate temperatures followed by a perceived ‘‘recovery’’ in strength at Table 2 Elastic and thermal properties of hybrid matrix and Nicalon fiber Young’s modulus (GPa) Coefficient of thermal expansion (106 / C) Volume fraction Hybrid matrix 173–195 2.86 0.58 Nicalon fiber 192 4.0 0.42 Table 3 Debond and sliding stresses of Nicalon fibers in BMAS, as measured by fiber pushout testing Fiber diameter (mm) debond (MPa) sliding (MPa) As-received 17.92.9 16.4 5.1 12.44.6 Shocked, 500 C 23.21.9 12.72.6 8.71.9 Shocked, 950 C 21.42.4 13.26.8 9.13.9 N. Chawla et al. / Composites Science and Technology 61 (2001) 1923–1930 1927