ELSEVIER Materials Science and Engineering A209( 1996)251-259 A microstructural investigation of the mechanisms of tensile creep deformation in an Al,O3/SiCw composite C. O'Mearaa, T. Suihkonena. T. Hansson,, R. Warren Department of Physics, Chalmers University of Technology, Goteborg S-412 96, Sweden Department of Mechanical Engineering, Nagaoka University of Technology, Nagaoka, Japan Department of Materials Science and Production Technology, Luled University of Technology, Luled, Sweden Abstract The tensile creep behaviour of an SiCw (25%)reinforced alumina composite was investigated using scanning electron microscopy(SEM) and transmission electron microscopy and automatic image analysis in SEM. The creep tests were carried out in air in the ranges 1100-1300C and 11-67 MPa Each creep test was performed at a constant temperature. The material had a stress exponent of about three for all temperatures and an approximate activation energy of 650 kJ mol, The creep resistance of this composite is poorer than that of similar composites studied earlier. Microstructural examination revealed the microstruc- ture to be extremely inhomogeneous consisting of spherical whisker-rich clusters(20-100 um)surrounded/separated by AlO, rich rims(10 um). The secondary is dominated by a damage accumulation process namely cavitation and crack growth in both the Sic clusters and the Al,O, rims. Final fracture seems to occur through the alumina rich regions. The lower creep resistance of this composite compared to that of similar composites is attributed primarily to the inhomogeneity of the as-received Keywords: Tensile creep deformation; Alumina composites; Microstructure 1. Introduction whiskers into alumina is also expected to improve the creep resistance and this has largely been confirmed in Monolithic alumina exhibits only moderate strength bend and compression tests [5-12 but not in tension and creep resistance and, like most monolithic ceram- [13] ics, is extremely brittle. SiC whisker reinforcement of Observed creep mechanisms in monolithic alumina alumina(SICw/Al,O3) has been employed primarily include basal slip, diffusional creep and grain boundary and successfully to improve fracture toughness [1-4]. sliding(GBS)[14, 15. Grain boundary cavitation has various toughening mechanisms such as whisker bridg- also been observed in association with GBS [16, 17]. The ng and pullout, microcracking and crack deflection are stress exponent has generally been found to vary be- operative depending on microstructural factors and ex- tween I and 2 and the activation energy between 400 erimental conditions [1-4]. The improvements ob- 650 kJ mol". The creep resistance has been found to ained in the composite in both fracture toughness and increase with grain size and there is a general agreement strength as compared to monolithic alumina has led to that aluminas with"clean"grain boundaries exhibit s application as, for example, cutting tool inserts and higher creep resistance than aluminas with an amor extrusion valves, and give it potential for use in struc- phous grain boundary phase [14-24 tural applications at high temperatures. Ho Reinforcement of alumina with SiC-whiskers is ex- use of ceramic materials in high temperature structural pected to improve the creep resistance primarily by applications is inevitably dependent on their time interlocking/pinning of grains which then limits grain eep and boundary sliding [8]. However, several factors can be oxidation resistance. The incorporation of SiC- identified which will affect the mechanical response of 0921-509396S1500c 1996- Elsevier Science S.A. All rights reserved SSD09215093(95)101020
A E L S E V I E R Materials Science and Engineering A209 (1996) 251 259 A microstructural investigation of the mechanisms of tensile creep deformation in an AI203/SiC w composite C. O'Meara a, T. Suihkonen a, T. Hansson b, R. Warren c aDepartment of Physics, Chalmers University of Technology, G6teborg S-412 96, Sweden bDepartment of Mechanical Engineering, Nagaoka University of Technology, Nagaoka, Japan ~Department of Materials Science and Production Technology, Lule~ University of Technology, Lule~, Sweden Abstract The tensile creep behaviour of an SiCw (25%) reinforced alumina composite was investigated using scanning electron microscopy (SEM) and transmission electron microscopy and automatic image analysis in SEM. The creep tests were carried out in air in the ranges 1100-1300 °C and 11-67 MPa. Each creep test was performed at a constant temperature. The material had a stress exponent of about three for all temperatures and an approximate activation energy of 650 kJ mol 1. The creep resistance of this composite is poorer than that of similar composites studied earlier. Microstructural examination revealed the microstructure to be extremely inhomogeneous consisting of spherical whisker-rich clusters (20-100/~ m) surrounded/separated by A1203 rich rims (10 ~tm). The secondary creep rate is dominated by a damage accumulation process namely cavitation and crack growth in both the SiC clusters and the A1203 rims. Final fracture seems to occur through the alumina rich regions. The lower creep resistance of this composite compared to that of similar composites is attributed primarily to the inhomogeneity of the as-received material. Keywords: Tensile creep deformation; Alumina composites; Microstructure 1. Introduction Monolithic alumina exhibits only moderate strength and creep resistance and, like most monolithic ceramics, is extremely brittle. SiC whisker reinforcement of alumina (SiCw/A1203) has been employed primarily and successfully to improve fracture toughness [1-4]. Various toughening mechanisms such as whisker bridging and pullout, microcracking and crack deflection are operative depending on microstructural factors and experimental conditions [1-4]. The improvements obtained in the composite in both fracture toughness and strength as compared to monolithic alumina has led to its application as, for example, cutting tool inserts and extrusion valves, and give it potential for use in structural applications at high temperatures. However the use of ceramic materials in high temperature structural applications is inevitably dependent on their time dependent mechanical properties such as creep and oxidation resistance. The incorporation of SiC- 0921-5093/96/$15.00 © 1996 - Elsevier Science S.A. All rights reserved SSDI 0921-5093(95)10102-0 whiskers into alumina is also expected to improve the creep resistance and this has largely been confirmed in bend and compression tests [5-12] but not in tension [13]. Observed creep mechanisms in monolithic alumina include basal slip, diffusional creep and grain boundary sliding (GBS) [14,15]. Grain boundary cavitation has also been observed in association with GBS [16,17]. The stress exponent has generally been found to vary between 1 and 2 and the activation energy between 400- 650 kJ mol-1. The creep resistance has been found to increase with grain size and there is a general agreement that aluminas with "clean" grain boundaries exhibit higher creep resistance than aluminas with an amorphous grain boundary phase [14-24]. Reinforcement of alumina with SiC-whiskers is expected to improve the creep resistance primarily by interlocking/pinning of grains which then limits grain boundary sliding [8]. However, several factors can be identified which will affect the mechanical response of
252 C O Meara et al. Materials Science and Engineering A209(1996)251-259 the composite under creep conditions: the volume frac- The mixture was cold pressed to 20 mm diameter rods tion of whiskers; the strength of the interfacial bond and then HIPped (1600C, 160 MPa, I h). Cylindrical between the fibre and the matrix; the grain size of the test specimens were produced by precision machining matrix grains; the volume of intergranular amorphous from the rods. Each specimen had a total length of 150 phase and; the oxidation susceptibilty of the material mm and a diameter of 10 mm reducing to 4 mm over a which causes the formation of glass at the whisker 20 mm long gauge length matrix interface. These factors will vary from material to material and will complicate both the interpretation 2.2. Creep testing of the creep behaviour and the comparison of different The creep equipment was specially designed for the Studies on bending and compression of the testing of brittle materials and details of the test system composite indicate that there exists a transition stress e described in Ref. [13]. The creep tests we below which the creep is dominated by diffusion ac commodated mechanisms and above which a damage Each creep test was performed at a constant stress and accumulation process involving cavitation and microc- temperature. Two specimens were pre-heat treated in racking become increasingly important. In the low air at 1300oC prior to creep testing. The high tempera- stress regime the stress exponent is 1-2 while above the ture heat treatment was used to investigate the effect of energies similar to those in monolithic alumina are non heat treated specimens subjected to he aring with transition it increases to values of 5[6-10. Activation oxidation on the creep behaviour by cor same creep bserved. No direct evidence of dislocation activity has conditions een found In general whisker reinforcement increases the creep resistance of alumina, however in the high 2.3. Microstructural examination not provide further improvement and may even de crease the creep resistance [8]. Grain boundary and he microstructure of the as-received and crept mate rials were studied using both scanning and transmission interfacial amorphous phases are detrimental to creep electron microscopy (SEM/TEM)and quantitative esistance and may promote transition to a damage accumulation process [9]. This is consistent with the SEM using automatic image analysis(AIA) observation that creep resistance is lower in air than in inert atmospheres since the composite is sensitive to 3. TEM xidation Thin sections for TEM analysis were taken from the Work by the authors on tensile creep of the com- centre of the gauge section directly above the fracture posite has shown that in tension even in the low stress surface and were cut in the longitudinal direction regime damage accumulation was the dominant creep parallel to the stress axis. Thinned sections were dimple mechanism and a stress exponent of three was obtained ground followed by ion-beam thinning to perforation or all temperatures and stresses [13]. Previous investi TEM examination was carried out using a JEOL gations on ceramic materials tested in tension and 2000FX TEM/STEM instrument equipped with a Link flexure have shown stress exponents of three to arise Systems AN 10 000 EDX spectrometer from creep cavitation [25]. However for composite and multiphase ceramics because of the complex interaction 2. 3.2. SEM between the microstructural constituents, creep defor- SEM specimens were cut from the gauge section in mation mechanisms cannot be reliably deduced from the longitudinal direction from the fracture surface to a creep data alone [9] but require direct observation of distance of about 0. 5 cm along the gauge length. The the deformed microstructures specimens were mounted in transoptic plastic, ground This work presents a microstructural examination of and polished down to 0. 25 um using a Struers semiau the tensile creep behaviour of a SiCw(25%)reinforced tomatic polishing apparatus. The specimens were exam- alumina composite. Electron microscopy was used to ined in a CAM Scan S-4 80DV instrument equipped obtain information on the possible creep mechanisms. with a Link Systems AN 10000 EDX spectrometer The backscattered electron mode 2.E 2. Material (AIA)was used for cavity, alumina grain size and phase volume fraction estimation. A Jeol JXA/8600 The composite was produced from a powder mixture Electron Probe Micro Analyser(EPMA) was used with of alumina and whiskers without sintering additives. Kantron software
252 C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251-259 the composite under creep conditions: the volume fraction of whiskers; the strength of the interfacial bond between the fibre and the matrix; the grain size of the matrix grains; the volume of intergranular amorphous phase and; the oxidation susceptibilty of the material which causes the formation of glass at the whisker matrix interface. These factors will vary from material to material and will complicate both the interpretation of the creep behaviour and the comparison of different works. Studies on bending and compression creep of the composite indicate that there exists a transition stress below which the creep is dominated by diffusion accommodated mechanisms and above which a damage accumulation process involving cavitation and microcracking become increasingly important. In the low stress regime the stress exponent is 1-2 while above the transition it increases to values of 5 [6-10]. Activation energies similar to those in monolithic alumina are observed. No direct evidence of dislocation activity has been found. In general whisker reinforcement increases the creep resistance of alumina, however in the high stress regime whisker volume fractions above 20% do not provide further improvement and may even decrease the creep resistance [8]. Grain boundary and interfacial amorphous phases are detrimental to creep resistance and may promote transition to a damage accumulation process [9]. This is consistent with the observation that creep resistance is lower in air than in inert atmospheres since the composite is sensitive to oxidation. Work by the authors on tensile creep of the composite has shown that in tension even in the low stress regime damage accumulation was the dominant creep mechanism and a stress exponent of three was obtained for all temperatures and stresses [13]. Previous investigations on ceramic materials tested in tension and flexure have shown stress exponents of three to arise from creep cavitation [25]. However for composite and multiphase ceramics because of the complex interaction between the microstructural constituents, creep deformation mechanisms cannot be reliably deduced from creep data alone [9] but require direct observation of the deformed microstructures. This work presents a microstructural examination of the tensile creep behaviour of a SiCw (25%) reinforced alumina composite. Electron microscopy was used to obtain information on the possible creep mechanisms. 2. Experimental 2.1. Material The composite was produced from a powder mixture of alumina and whiskers without sintering additives. The mixture was cold pressed to 20 mm diameter rods and then HIPped (1600 °C, 160 MPa, 1 h). Cylindrical test specimens were produced by precision machining from the rods. Each specimen had a total length of 150 mm and a diameter of 10 mm reducing to 4 mm over a 20 mm long gauge length. 2.2. Creep testing The creep equipment was specially designed for the testing of brittle materials and details of the test system are described in Ref. [13]. The creep tests were carried out in air in the ranges 1100-1300 °C and 11 67 MPa. Each creep test was performed at a constant stress and temperature. Two specimens were pre-heat treated in air at 1300 °C prior to creep testing. The high temperature heat treatment was used to investigate the effect of oxidation on the creep behaviour by comparing with non heat treated specimens subjected to the same creep conditions. 2.3. Microstructural examination The microstructure of the as-received and crept materials were studied using both scanning and transmission electron microscopy (SEM/TEM) and quantitative SEM using automatic image analysis (AIA). 2.3.1. TEM Thin sections for TEM analysis were taken from the centre of the gauge section directly above the fracture surface and were cut in the longitudinal direction, parallel to the stress axis. Thinned sections were dimple ground followed by ion-beam thinning to perforation. TEM examination was carried out using a JEOL 2000FX TEM/STEM instrument equipped with a Link Systems AN 10 000 EDX spectrometer. 2.3.2. SEM SEM specimens were cut from the gauge section in the longitudinal direction from the fracture surface to a distance of about 0.5 cm along the gauge length. The specimens were mounted in transoptic plastic, ground and polished down to 0.25/zm using a Struers semiautomatic polishing apparatus. The specimens were examined in a CAM Scan S-4 80DV instrument equipped with a Link Systems AN 10000 EDX spectrometer. The specimens were examined in secondary and backscattered electron mode. 2.3.3. Quantitative microscopy (AIA) was used for cavity, alumina grain size and phase volume fraction estimation. A Jeol JXA/8600 Electron Probe Micro Analyser (EPMA) was used with Kantron software
C. O'Meara et al. Materials Science and Engineering A209(1996)251-259 253 Tensile creep test conditions and results Sample Temp Stress Time to fracture Strain to fracture Secondary creep Creep exponent Preheated at 1300C (%) (s-) 3.4 9×1 3.25 11001-35728 3.7×10-10 24×10 8×10-6 2.94 1300 252 20×10-9 .7 1300 19 4.1×10 !30 0.17 2.3×10 1275 28×10 1300 1.3 6.6×10-7 0 i Load increased during the test. ted before failur Defect in the sample 2.3.3.1. Cavity analysis. Two sets of measurements were 3. Results made for each fractured specimen, one at the fracture surface and one 3 mm further into the bulk. At a 3. 1. Creep results magnification of 5000 x, 50 fields were examined in ach measurement with a total frame area of 10 x 5 A summary of the creep results is given in Table 1 327. 12 um2=16356 um2. Cavities/pores in the size The composite exhibited limited regions of primary range 0. 1-10 um were measured by AIA. The parame- creep, relatively well defined secondary creep and en ter measured was the area of the pore. This was con- tered a tertiary creep region just before fracture for all erted to the equivalent diameter of that area by conditions of stress and temperature. The material had DCIRCLE-2/4 a stress exponent of approximately three for all temper (1) atures and an approximate activation energy of 650 kJ mol-. The creep resistance of this composite in ten Where dCirClE is the average diameter of the pore sion is poorer than that of similar composites studie and a is its measured area earlier in bend or compression. However, the creep resistance improved significantly following high temper 2.3.3. 2. Grain size analysis. For the alumina grain size ature pre-heat treatment analysis the specimens were first etched in argon at 1300C for 15 min For each specimen 10 micrographs t a time were taken in a line close to and parallel with the fracture surface. At least 500 alumina grains were analysed in each specimen. The parameter measured in Ala was the area of the grain which was converted to average diameter by Eq (1) 2.3.3.3. Volume fraction of phases. As will be discussed in the results, considerable inhomogeneity in the matrix was observed dividing the microstructure into whisker /rich”and“ alumina rich” areas. A qualitative AIA analysis was carried out simply by marking he whisker/rich"areas in one colour and the" alu- mina rich"areas in another in order to get an indica- tion of the extent of inhomogeneity. In addition ar estimation of the whisker fraction in the whisker /rich and"alumina rich"areas was carried out using EDX ing the inhomogeneity in the pherical whisker- rich clusters surrounded/ separated by Al, O, rich
C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251 259 253 Table 1 Tensile creep test conditions and results Sample Temp Stress Time to fracture Strain to fracture Secondary creep Creep exponent Preheated at 1300 °C (°C) (MPa) (h) (%) rate (s- i ) (h) 1 1200 t9 42.1 3.4 1.9 x 10 7 3.25 2 ~ 1100 11-35 728 2.3 3.7x I0 m 3.25 1.6 x 10 8 3 1200 35 13.9 1.7 2.5 x 10 7 3.28 - 4 1200 67 1.2 1.2 2.4x 10 -6 3.28 - 5 1300 19 1.7 2.5 3.8 X 10 -6 2.94 - 6 b 1300 11 252 3.4 2.0 X 10 9 3.7 - 7 1300 19 13.7 3.0 4.1 × 10 7 3.08 61 8 1300 35 0.17 1.5 2.3× 10 -5 2.94 .- 9 1275 35 2.2 2.5 2.8 × 10 -6 3.08 72 10 ~ 1300 1 t 4.9 1.3 6.6 x 10 -v 3.08 72 Load increased during the test. b Test aborted before failure. Defect in the sample. 2.3.3. I. Cavity analysis. Two sets of measurements were made for each fractured specimen, one at the fracture surface and one 3 mm further into the bulk. At a magnification of 5000 x, 50 fields were examined in each measurement with a total frame area of 10 x 5 x 327.12 pm2= 16356 pm 2. Cavities/pores in the size range 0.1-10 pm were measured by AIA. The parameter measured was the area of the pore. This was converted to the equivalent diameter of that area by ?.-,- DCIRCLE = 2 k/A (1) Where DCIRCLE is the average diameter of the pore and A is its measured area. 2.3.3.2. Grain size analysis. For the alumina grain size analysis the specimens were first etched in argon at 1300 °C for 15 min. For each specimen 10 micrographs at a time were taken in a line close to and parallel with the fracture surface. At least 500 alumina grains were analysed in each specimen. The parameter measured in AIA was the area of the grain which was converted to average diameter by Eq. (1). 2.3.3.3. Volume fraction of phases. As will be discussed in the results, considerable inhomogeneity in the matrix was observed dividing the microstructure into "whisker/rich" and "alumina rich" areas. A qualitative AIA analysis was carried out simply by marking the "whisker/rich" areas in one colour and the "alumina rich" areas in another in order to get an indication of the extent of inhomogeneity. In addition an estimation of the whisker fraction in the "whisker/rich" and "alumina rich" areas was carried out using EDX analysis. 3. Results 3.1. Creep results A summary of the creep results is given in Table 1. The composite exhibited limited regions of primary creep, relatively well defined secondary creep and entered a tertiary creep region just before fracture for all conditions of stress and temperature. The material had a stress exponent of approximately three for all temperatures and an approximate activation energy of 650 kJ mol-t. The creep resistance of this composite in tension is poorer than that of similar composites studied earlier in bend or compression. However, the creep resistance improved significantly following high temperature pre-heat treatment. Fig. 1. SEM image showing the inhomogeneity in the microstructure, spherical whisker-rich clusters surrounded/separated by AI203 rich rims
C. O'Meara et al. Materials Science and Engineering A209(1996)251-259 The cavity size measurements of cavities between 0. 1 and 10 um Mean cavity avity nucleation Area fraction of location cavities(1 m diameter (um) 50 115 163 161 041 0.45 1214.9 0 stands for as-received sample 3. 2. Microstructure AlO3 SiC and SiC/SiC contacts. The extent of cavita tion decreased somewhat with distance from the frac- 3. 2. 1. As-received material ture surface but was evident the entire length of th As is shown in Fig. I microstructural examination specimens (a 5 um). Cavitation between the AlO revealed that the as-received material had an inhomoge grains was observed to be very frequent and to occur neous microstructure consisting of spherical whisker two grain junctions. Ca Al2O3 rich rims( 10 um in length (i) The cavities/pores were observed at Al2 O3/ Al2O3 Al O3 grains following creep testing at 35 MPae cavity between two Fig. 2. TEM bright field image showing a lat
254 C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251-259 Table 2 The cavity size measurements of cavities between 0.1 and 10/~m Sample and Area density of Median cavity Mean cavity Cavity nucleation Area fraction of location cavities (l/ram -~) diameter (jzm) diameter (/zm) rate (l/h) cavities (%) 0 22 0.42 0.50 0.42 1 115 0.50 0.60 45 3.26 2 170 0.52 0.60 4.0 5.0 3 178 0.40 0.50 210 3.10 4 96 0.30 0.38 1306 1.1 5 200 0.51 0.60 1980 5.6 6 163 0.50 0.56 10.57 4.0 7 126 0.5I 0.61 150 4.0 8 198 0.53 0.57 19317 5.1 9 161 0.41 0.45 1214.9 2.6 0 stands for as-received sample. 3.2. Microstructure 3.2.1. As-received material As is shown in Fig. 1 microstructural examination revealed that the as-received material had an inhomogeneous microstructure consisting of spherical whiskerrich clusters (~ 20-100 /~m) surrounded/separated by A1203 rich rims ( ~ 10/lm). The inhomogeneity is probably inherited from spray drying used during processing of the composite powders which resulted in a clustering of the SiC phase. The results of AIA and EDX analysis indicate that approximately 40% of the microstructure has a SiC content of less than 15 vol.% while about 60% has a SiC content of ,-~ 30 vol.%. The microstructure is in effect a composite within a composite. In addition the alumina grain size was small (sub-micron) and bi-modal in distribution with the larger grains in the rims and very small grains in the whisker clusters. In TEM very little ( _- 10 pm in length. (i) The cavities/pores were observed at AI203/A1203, A1203/SiC and SiC/SiC contacts. The extent of cavitation decreased somewhat with distance from the fracture surface but was evident the entire length of the specimens (~ 5 /zm). Cavitation between the A1203 grains was observed to be very frequent and to occur both at triple and two grain junctions. Cavitation occurred most commonly at two grain contacts and here the grain facets were separated the entire length of the grains forming lath like cavities (Fig. 2). The nucleation of smaller cavities along two-grain facets was also frequently observed in TEM. These cavities were approximately 100 nm in length and lenticular in shape and encroached on both grains as is shown in Fig. 3. No grain growth was detected following creep even for the preheated samples and few dislocations were observed. Indication of possible GBS was found both in TEM and in SEM examination of etched specimens (see Fig. 4). Cavities also formed at AI203/SiC and SiC/SiC contacts. As is shown in Fig. 5, SiC debonding and pullout occurred frequently. When these debonded contacts lay close together a small isolated intergranular cavity formed. Fracture of long SiC whiskers was also observed. Fig. 6 shows the presence of an amorFig. 2. TEM bright field image showing a lath like cavity between two AI203 grains following creep testing at 35 MPa and 1200 °C
C. O'Meara et al./ Materials Science and Engineering A209(1996)251-259 tested at 1300oC and 35 MPa (TEM), lacets in a specimen creep Fig. 3. Cavities along twograin Al phous coating(10 um) was observed: (i) through the Al,O3 in the rims around the clusters originating at the edge of the specimen and;(ii)in the interior of the clusters associated predominantly with the SiC whiskers. In both cases the cracking is mainly perpen Fig 4. Grain boundary sliding of alumina grains observed in SEM. dicular to the tensile direction and is intergranular. The
C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251-259 IIIII 255 Fig. 3. Cavities along two-grain A1203 facets in a specimen creep tested at 1300 °C and 35 MPa (TEM). phous coating (~2 nm) observed in TEM on the surface of some SiC whiskers in cavitated areas. This phenomena was observed in all specimens examined indicating that these interfaces were oxidised during creep exposure. It is pointed out that the TEM foils were taken from the centre of the specimens where it is assumed that the effects of oxidation are at a minimum; nonetheless, some qualification must be made on these observations: (1) the amorphous film was not observed at all AI203/SiC and SiC/SiC cavitated contacts in the same specimen; (2) no significant increase could be detected in the volume of intergranular amorphous phase in non-cavitated regions of the specimen or at the A1203/A1203 cavities as compared to as-received material; (3) Fig. 6, a typical oxidised cavity shows a cleanly fractured SiC whisker surrounded by an even layer of amorphous phase, no cavitation or glassy ligaments are observed in the amorphous phase which appears to "close off" the cavities. The authors therefore feel that oxidation of the whiskers may equally well have occurred after cavitation as much as being the facilitator of cavitation as suggested in other works. Quantitative analysis was undertaken to investigate differences in cavitation between the samples. The results for the area Fig. 4. Grain boundary sliding of alumina grains observed in SEM. Fig. 5. TEM images showing whisker debonding (a) and pullout (b) in crept material. close to the fracture surfaces are given in Table 2. All the specimens showed a similar pore/cavity size distribution with a mean cavity size of approx. 0.5/z m. In all cases 95% of the cavities were under 1.5 /~m in length which agrees well with TEM observations. Even with the limited number of test points a linear-stress relationship at temperature is observed between both the area density of cavities and the cavity nucleation rate (total number of cavities/time to fracture) (Fig. 7). The pre-heat treated specimens show the same dependence but at lower values indicating some positive effect of the heat treatment. These results do imply that cavitation is the main deformation mechanism occurring during creep and that it was operative under all test conditions. However the results also imply that creep rupture does not occur when a similar level of cavitation has been attained but rather that the level of tolerable cavitation is highly stress/temperature dependent due to an interplay of additional microstructural factors as will be discussed later. As is shown in Fig. 8, more severe creep damage in the form of cracks ( ~> 10/tm) was observed: (i) through the AI20 3 in the rims around the clusters originating at the edge of the specimen and; (ii) in the interior of the clusters associated predominantly with the SiC whiskers. In both cases the cracking is mainly perpendicular to the tensile direction and is intergranular. The
C. O Meara er al./ Materials Science and Engineering A209(1996)251-259 frequency of the cracks decreases with distance from severity of the cracks varied with temperature and stress. The cracks in the Sic clusters(Fig. 9)were found in all specimens but were most frequent in speci mens tested at 1200oC. These areas were also observed g四 the fracture surface. The distribution, type, size and s in TEM as interconnecting cavitated networks of SiC grains where cohesion is maintained by an amorphous 9 oating on the surface of the SiC whiskers. In general o1200°c A1300°c these cracks remained isolated within the clusters, i.e 13o0°c, preheated61 did not join up via the rims with neighbouring cracks. 1300°, preheated Thus these cracks were generally 20-50 um in length 0.001 TTTTTT As is shown in Fig. 8, the cracks in the Al,O3 rich Stress(MPa) Fig. 7. Cavity nucleation rate vs. stress edge of specimens and were found to extend from 20 um up to half way through the specimen diameter and were responsible for creep rupture. As is seen in Fig. 10 the cracks also penetrated through these regions intergran deflected along the interface or remained bridged by the whiskers. Cracking through the Al2O, was most severe in specimens tested at 1300C. In the pre-heat treated specimens cracking was predominantly of this type with much fewer Sic cluster cracks developing. In specimens tested at 1200C a considerable amount of cracking or crack branching parallel to the tensile direction is ob served. Examination of etched specimens indicated this cracking was frequently intragranular and was most severe in the specimen tested at 67 MPa. The pre- heated specimens and the specimen tested at the highest stress showed the most severe crack damage The oxide scale thicknesses and extent of oxidatio of the cracks were examined in SEM and although in general the scale thickness varied with length of expo sure the scales are thicker at 1300C as compared to 1200C. However there was very little oxidation in the cracks from the surface in the pre-heat treated speci mens again indicating some effect of pre-heat treatment on the microstructure The results of this work indicate that in tension the secondary creep is dominated by a damage accumula tion process namely cavity and crack growth. This is consistent with a calculated stress exponent of approxi mately three for all test conditions and that of previous investigations on ceramic materials tested in tension showing and flexure where stress exponents of three have been 2 nm)on the surface of Sic whiskers in cavitated areas: (a) bright found to arise from creep cavitation [25]. However the field and; (b) dark field (35 MPa, 1300oC creep resistance is inferior to both monolithic alumina
256 C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251 259 frequency of the cracks decreases with distance from the fracture surface. The distribution, type, size and severity of the cracks varied with temperature and stress. The cracks in the SiC clusters (Fig. 9) were found in all specimens but were most frequent in specimens tested at 1200 °C. These areas were also observed in TEM as interconnecting cavitated networks of SiC grains where cohesion is maintained by an amorphous coating on the surface of the SiC whiskers. In general these cracks remained isolated within the clusters, i.e. did not join up via the rims with neighbouring cracks. Thus these cracks were generally 20-50 pm in length. As is shown in Fig. 8, the cracks in the A1203 rich 10~ w~ 0 0.1- 0.01- "4 0.001 - 10 • 0 O 0 0 1200°C A 1300°C • . 1300oe, preheated 61h & • 1300°C, preheated 72h 20 30 40 50 60 70 Stress (MPa) Fig. 7. Cavity nucleation rate vs. stress. regions originated predominantly at the edge of the specimens and were found to extend from 20 pm up to half way through the specimen diameter and were responsible for creep rupture. As is seen in Fig. 10 the cracks also penetrated through these regions intergranularly. At whiskers lying in their path they were either deflected along the interface or remained bridged by the whiskers. Cracking through the A1203 was most severe in specimens tested at 1300 °C. In the pre-heat treated specimens cracking was predominantly of this type with much fewer SiC cluster cracks developing. In specimens tested at 1200 °C a considerable amount of cracking or crack branching parallel to the tensile direction is observed. Examination of etched specimens indicated this cracking was frequently intragranular and was most severe in the specimen tested at 67 MPa. The preheated specimens and the specimen tested at the highest stress showed the most severe crack damage. The oxide scale thicknesses and extent of oxidation of the cracks were examined in SEM and although in general the scale thickness varied with length of exposure the scales are thicker at 1300 °C as compared to 1200 °C. However there was very little oxidation in the cracks from the surface in the pre-heat treated specimens again indicating some effect of pre-heat treatment on the microstructure. ~i iii?ili! Fig. 6. TEM images showing the presence of an amorphous coating ( ~ 2 nm) on the surface of SiC whiskers in cavitated areas: (a) bright field and; (b) dark field (35 MPa, 1300 °C). 4. Discussion The results of this work indicate that in tension the secondary creep is dominated by a damage accumulation process namely cavity and crack growth. This is consistent with a calculated stress exponent of approximately three for all test conditions and that of previous investigations on ceramic materials tested in tension and flexure where stress exponents of three have been found to arise from creep cavitation [25]. However the creep resistance is inferior to both monolithic alumina
C. O Meara et al. Materials Science and Engineering 4209(1996)251-259 and to similar composites tested previously in bending or compression. In addition creep controlled by damage ccumulation processes is occurring in this composite at stresses far below the threshold stress observed for this mechanism in compression and bending(60-170 MPa Ithough the mode of testing, tensile as opposed to bending or compression, is expected to have some effect on the creep behaviour in particular to promote the damage accumulation process and reduce creep failure strains it is thought that this microstructural examina tion has provided microstructural explanations for this poor creep behaviour The inhomogeneity of the microstructure means that Fig. 9. Crack development in the whisker clusters is associated the material is literally a composite within a composite predominantly with the SiC whiskers (11 MPa, 1100C. SEM) Sic whisker content 30 volume which is well in excess of the optimum for creep inhibition of alumina. These clusters are surrounded by an interconnecting alumina rich"skeleton"which makes up approximately 40% of the microstructure and has a sic content of less than 15 volume. Although this is a low fibre content other works have shown that additions of even 5 wt % sic whiskers causes improvement in creep resistance [7]. In addition the alumina grain size is very small (largely ub-micron) which pre-empts a lower creep resistance and is also bi-modal in distribution with the larger grains in the rims and very small grains in the whisker clusters. Improved creep resistance in this composite is not necessarily to be expected with this micostructural scenario and the results obtained need to be examined n this context Microstructural observations and quantitative analy sis indicated that the extent of cavitation at failure has definite stress dependence at temperature, the extent of crack growth and the volume of the two different types of crack was also stress dependent at temperature failure was different for the different specimens. The ne Fig. 8. SEM backscattered electron image showing bserved in crept specimens (a) through the Al,o the clusters originating at the edge of the specimen( 19 and: ()in the interior of the clusters associated predominantly with Fig. 10. SEM image showing an intergranular crack in a rim follow the Sic whiskers, (19 MPa, 1200C) ing creep testing at 19 MPa and 1300C
C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251 259 257 and to similar composites tested previously in bending or compression. In addition creep controlled by damage accumulation processes is occurring in this composite at stresses far below the threshold stress observed for this mechanism in compression and bending (60-170 MPa). Although the mode of testing, tensile as opposed to bending or compression, is expected to have some effect on the creep behaviour in particular to promote the damage accumulation process and reduce creep failure strains it is thought that this microstructural examination has provided microstructural explanations for this poor creep behaviour. The inhomogeneity of the microstructure means that the material is literally a composite within a composite Fig. 9. Crack development in the whisker clusters is associated predominantly with the SiC whiskers. (11 MPa, l l00 °C, SEM). and consists to approximately 60% of clusters with a SiC whisker content 30 volume which is well in excess of the optimum for creep inhibition of alumina. These clusters are surrounded by an interconnecting alumina rich "skeleton" which makes up approximately 40% of the microstructure and has a SiC content of less than 15 volume. Although this is a low fibre content other works have shown that additions of even 5 wt.% SiC whiskers causes improvement in creep resistance [7]. In addition the alumina grain size is very small (largely sub-micron) which pre-empts a lower creep resistance and is also bi-modal in distribution with the larger grains in the rims and very small grains in the whisker clusters. Improved creep resistance in this composite is not necessarily to be expected with this micostructural scenario and the results obtained need to be examined in this context. Microstructural observations and quantitative analysis indicated that the extent of cavitation at failure has a definite stress dependence at temperature, the extent of crack growth and the volume of the two different types of crack was also stress dependent at temperature so that the level and type of damage accumulation at failure was different for the different specimens. The net Fig. 8. SEM backscattered electron image showing severe cracking observed in crept specimens (a) through the A1203 in the rims around the clusters originating at the edge of the specimen (19 MPa, 1300 °C) and; (b) in the interior of the clusters associated predominantly with the SiC whiskers. (19 MPa, 1200 °C). Fig. 10. SEM image showing an intergranular crack in a rim following creep testing at 19 MPa and 1300 °C
258 C. O Meara et al. Materials Science and Engineering A209 (1996)251-259 damage accumulation will be a summation of damage 100 um) surrounded/separated by al2O3 rich rims(10 effects in the two different elements of the microstruc- um). Stress exponents of approximately three ure. The clusters have a high SiC content and can be spond well with the microstructural analysis expected to cavitate easily as observed. The skeleton of indicated that the secondary creep rate is dominated by Al2O3 with low whisker should be more creep resistant a damage accumulation process namely cavitation and and will act as the reinforcing phase encapsulating the crack growth in both the Sic clusters and the Al2O SiC damage and preventing it from linking up. Cavita- rims. Final fracture seems to occur through the alumina tion and crack formation in the clusters would then be rich regions. The poorer creep resistance of this com te and Sic cluster cracks are observed to be more frequent in attributed primarily to the inhomogeneity of the as-re- specimens with a high secondary creep rate. Cavitation ceived material also occurs in the Al,O3 rich rims phase but not as ly as in the Sic clusters. Final fracture seems to originate from the surface in the alumina regions and then to propagate through these regions intergranular either when the fracture stress is reached for alumina of References this particular grain size(specimen 4)or when a suffi cient number of contiguous cavities have been built up [] P F. Becher and G.C. Wei, Toughening behaviour in SiC such that the crack is subject to a stress intensity whisker- reinforced alumina, J, Am, Ceram, Soc,, 67C (1984)267. [2]GC. Wei and P F. Becher, Development of SiC-whisker-rein- K>Kh [26(specimens 1-3, 5-9). This interpretation forced ceramics, Am. Ceram Soc. Bull., 64(1985)298 is consistent with and can explain the differences in [3] J. Homeny and W.L. Vaughn, Silicon carbide whisker/alumina crept microstructures, times to failure and secondary atrix composites effect of whisker surface treatment on frac- creep rates of the tested samples re toughness, J. Am. Ceram. Soc., 73(1990)394. The effect of oxidation on the creep behaviour was 4 T. Hansson, R. Warren and J. Wasen, Fracture toughness ifficult to evaluate as all the samples had been sub anisotropy and toughening mechanisms of a hot-pressed alumina reinforced with silicon carbide whiskers, J. Am. Ceram. Soc.76 jected to high temperature air exposure. Although other (1993)841 microstructural factors are responsible for creep failure [5]A H Chokshi and J.R. Porter, Creep deformation of an alumina ith reference to other work it is highly likely that matrix composite reinforced with silicon carbide whiskers, J. oxidation during creep exposure does facilitate cavita- Am. Ceram. Soc., 68C(1985)144 tion in this work some oxidation of the sic whiskers [6]J. R. Porter, FF. Lange and A H. Chokshi, Processing and creep rformance of SiC-whisker-reinforced Al,O3, Am. Ceram. Soc in the matrix was observed in TEM although no signifi- Bu.,66(1987)343. cant increase could be detected in the volume of inter- [7] A.R. de arellano-Lopez, F L. Cumbrera, A. Domingues-Ro granular amorphous phase in non-cavitated regions of drigues, K C. Goretta and J. L. Routbort, Compressive creep of the microstructure as compared to as- received material SiC-whisker-reinforced Al,O,, J. Am. Ceram. Soc., 73(1990) The TEM evidence suggests that oxidation of the [8]H -T Lin and P F Becher, High-temperature creep deformation whiskers had occurred after cavitation and may bond of alumina-SiC-whisker composites, J. Am. Ceram. Soc. the whiskers or the whiskers and the matrix grains gether. In addition the specimens which were pre-heat [9]P, Lipetzky, S.R. Nutt, D.A. Koester and R F. Davis, Atmo- treated had a lower density of cavities, lower cavity spheric effects on compressive creep of Sic-whisker-reinforced nucleation rates, fewer Sic cluster cavities and lower (10) A R. de Arellano-Lopez, A.D. Domingues-Rodrigues, KC secondary creep rates than specimens subjected to the Goretta and J.L. Routbort Plastic deformation in SiC-whisker same conditions without pre-heat treatment. In general reinforced alumina, J. Am. Ceram. Soc., 76(1993)1425 the specimens with the lowest secondary creep rates 11]A H. Swan, M. V Swain and G L. Dunlop, Compressive creep of were also those which were the most severely oxidised SiC-whisker-reinforced alumina, J. Eur. Ceram. Soc., 10( 1992) At this stage it is therefore difficult to accurately iden- [12)J. L. Routbort, K.C. Goretta, A. D. Domingues-Rodrigues and tify the role of oxidation, whether positive or negative, A R. de Arellano- Lopez, Creep of whisker-reinforced ceramics. on the creep deformation process(es) J. Hard [13] T. Hansson, C. OMeara, K. Rundgren, P. Svensson and Warren, Tensile creep of alumina and Sic whisker reinforced 5. Conclusions alumina, Proc. Plastic Deformation of Ceramics, Engineering Foundation Conf, Snowbird, Utah, August 7-12, 1994 [14] W.R. Cannon and T.G. Langdon, Review: Creep of ceramics- The tensile creep behaviour of a Sic(25%)reinforced Part 1. Mechanical characteristics, J. Mater. Sci, 18(1983) alumina composite was investigated in air in the ranges 5]A H. Chokshi and J. R. Porter, High temperature mechanic 1100-1300 C and 11-67 MPa. The as-fabricated properties of single phase alumina, J. Mater. Sci., 21(1986)705 microstructure was found to be extremely inhomoge [16] A.G. Robertson, D.S. Wilkinson and C H. Caceres, Creep and reep fracture in hot-pressed alumina, J. Am. Ceram. Soc., 74 neous consisting of spherical whisker-rich clusters(20 (1991)915
258 C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251-259 damage accumulation will be a summation of damage effects in the two different elements of the microstructure. The clusters have a high SiC content and can be expected to cavitate easily as observed. The skeleton of A1203 with low whisker should be more creep resistant and will act as the reinforcing phase encapsulating the SiC damage and preventing it from linking up. Cavitation and crack formation in the clusters would then be a major contributor to the secondary creep rate and SiC cluster cracks are observed to be more frequent in specimens with a high secondary creep rate. Cavitation also occurs in the A1203 rich rims phase but not as easily as in the SiC clusters. Final fracture seems to originate from the surface in the alumina regions and then to propagate through these regions intergranularly either when the fracture stress is reached for alumina of this particular grain size (specimen 4) or when a sufficient number of contiguous cavities have been built up such that the crack is subject to a stress intensity K~> Kth [26] (specimens 1-3, 5-9). This interpretation is consistent with and can explain the differences in crept microstructures, times to failure and secondary creep rates of the tested samples. The effect of oxidation on the creep behaviour was difficult to evaluate as all the samples had been subjected to high temperature air exposure. Although other microstructural factors are responsible for creep failure, with reference to other work it is highly likely that oxidation during creep exposure does facilitate cavitation. In this work some oxidation of the SiC whiskers in the matrix was observed in TEM although no significant increase could be detected in the volume of intergranular amorphous phase in non-cavitated regions of the microstructure as compared to as-received material. The TEM evidence suggests that oxidation of the whiskers had occurred after cavitation and may bond the whiskers or the whiskers and the matrix grains together. In addition the specimens which were pre-heat treated had a lower density of cavities, lower cavity nucleation rates, fewer SiC cluster cavities and lower secondary creep rates than specimens subjected to the same conditions without pre-heat treatment. In general the specimens with the lowest secondary creep rates were also those which were the most severely oxidised. At this stage it is therefore difficult to accurately identify the role of oxidation, whether positive or negative, on the creep deformation process(es). 5. Conclusions The tensile creep behaviour of a SiC (25%) reinforced alumina composite was investigated in air in the ranges 1100-1300 °C and 11-67 MPa. The as-fabricated microstructure was found to be extremely inhomogeneous consisting of spherical whisker-rich clusters (20- 100/tm) surrounded/separated by A1203 rich rims (10 /~m). Stress exponents of approximately three correspond well with the microstructural analysis which indicated that the secondary creep rate is dominated by a damage accumulation process namely cavitation and crack growth in both the SiC clusters and the AlzO 3 rims. Final fracture seems to occur through the alumina rich regions. The poorer creep resistance of this composite compared with that of similar composites is attributed primarily to the inhomogeneity of the as-received material. References [1] P.F. Becher and G.C. Wei, Toughening behaviour in SiCwhisker-reinforced alumina, J. Am. Ceram. Soc., 67C (1984) 267. [2] G.C. Wei and P.F. Becher, Development of SiC-whisker-reinforced ceramics, Am. Ceram. Soc. Bull., 64 (1985) 298. [3] J. Homeny and W.L. Vaughn, Silicon carbide whisker/alumina matrix composites: effect of whisker surface treatment on fracture toughness, J. Am. Ceram. Soc., 73 (1990) 394. [4] T. Hansson, R. Warren and J. Was6n, Fracture toughness anisotropy and toughening mechanisms of a hot-pressed alumina reinforced with silicon carbide whiskers, J. Am. Ceram. Soc., 76 (1993) 841. [5] A.H. Chokshi and J.R. Porter, Creep deformation of an alumina matrix composite reinforced with silicon carbide whiskers, J. Am. Ceram. Soc., 68C (1985) 144. [6] J.R. Porter, F.F. Lange and A.H. Chokshi, Processing and creep performance of SiC-whisker-reinforced A1203, Am. Ceram. Soc. Bull., 66 (1987) 343. [7] A.R. de Arellano-L6pez, F.L. Cumbrera, A. Domingues-Rodrigues, K.C. Goretta and J.L. Routbort, Compressive creep of SiC-whisker-reinforced A1203, J. Am. Ceram. Soc., 73 (1990) 1297. [8] H.-T. Lin and P.F. Becher, High-temperature creep deformation of alumina-SiC-whisker composites, J. Am. Ceram. Soc., 74 (1991) 1886. [9] P. Lipetzky, S.R. Nutt, D.A. Koester and R.F. Davis, Atmospheric effects on compressive creep of SiC-whisker-reinforced alumina, J. Am. Ceram. Soc., 74 (1991) 1240. [10] A.R. de Arellano-L6pez, A.D. Domingues-Rodrigues, K.C. Goretta and J.L. Routbort, Plastic deformation in SiC-whiskerreinforced alumina, J. Am. Ceram. Soc., 76 (1993) 1425. [11] A.H. Swan, M.V. Swain and G.L. Dunlop, Compressive creep of SiC-whisker-reinforced alumina, J. Eur. Ceram. Soc., I0 (1992) 317. [12] J.L. Routbort, K.C. Goretta, A.D. Domingues-Rodrigues and A.R. de AreUano-L6pez, Creep of whisker-reinforced ceramics, J. Hard Mater., I (1990) 221. [13] T. Hansson, C. O'Meara, K. Rundgren, P. Svensson and R. Warren, Tensile creep of alumina and SiC whisker reinforced alumina, Proc. Plastic Deformation of Ceramics, Engineering Foundation Conf., Snowbird, Utah, August 7-12, 1994. [14] W.R. Cannon and T.G. Langdon, Review: Creep of ceramics-- Part 1. Mechanical characteristics, J. Mater. Sci., 18 (1983) 1. [15] A.H. Chokshi and J.R. Porter, High temperature mechanical properties of single phase alumina, J. Mater. Sci., 21 (1986) 705. [16] A.G. Robertson, D.S. Wilkinson and C.H. C~tceres, Creep and creep fracture in hot-pressed alumina, J. Am. Ceram. Soc., 74 (1991) 915
C. O Meara et al. Materials Science and Engineering A209(1996)251-259 [7A. Xu and AA. Solomon, Diffusional creep and cavitational alumina(Al,O3): IL, Basal slip and nonaccomo- strains in high purity alumina under tension and subsequent dated oundary sliding, J. Aim. Ceram. Soc., 63(1980) hydrostatic compression, J. Am. Ceram. Soc., 75( 1992)985 [18]R C. Folweiler, Creep behaviour of pore-free polycrystalline [23] J.R. Porter, w. Blumenthal and A G. Evans, Creep fracture in aluminum oxide. J. App/. Phys., 32(1961)773. ceramic polycrystals. Creep cavitation effects in polycrystalline [19]AH. Chokshi and J.R. Porter, Analysis of concurrent grain alumina, Acta Metall, 29(1981)1899 rowth during creep of polycrystalline alumina, J. Am. Ceram [24]RA. Page, J, Lankford and S. Spooner. Small-angie neutron Soc,69(1986)C37 [20R.A. Page and K.S. Chan. Improved creep resistance in a lumina, J. Mater. Sci., 19(1984)3360 lass-bonded alumina through concurrent devitrification, J. [25] D F. Carroll and R E. Tressler, Accumulation of creep damage Mater, Sci. Left, 8(1989)938 in a siliconized silicon carbide, J. Am., Ceram. Soc.. 71(1988) [21]R. M. Cannon, w.H. Rhodes and A H. Heuer, Plastic deforma- 472. tion of fine-grained alumina (Al,O3): 1. Interface-controlle [26] S M. Johnson, B.J. Dalgleish and A.G. Evans, High-temperature diffusional creep, J. Am. Ceram. Soc., 63(1980)46 failure of polycrystalline alumina: Ill. Failure times. J,Am [22]A H. Heuer, N.J. Tighe and R.M. Cannon, Plastic deformation cerm,Soc…67(1984)759
C. O'Meara et al. / Materials Science and Engineering A209 (1996) 251 259 259 [17] A. Xu and A.A. Solomon, Diffusional creep and cavitational strains in high purity alumina under tension and subsequent hydrostatic compression, J. Am. Ceram. Soc., 75 (1992) 985. [18] R.C. Folweiler, Creep behaviour of pore-free polycrystalline aluminum oxide, J. Appl. Phys., 32 (1961) 773. [19] A.H. Chokshi and J.R. Porter, Analysis of concurrent grain growth during creep of polycrystalline alumina, J. Am. Ceram. Soe., 69 (1986) C37. [20] R.A. Page and K.S. Chan, Improved creep resistance in a glass-bonded alumina through concurrent devitrification, J. Mater. Sei. Lett., 8 (1989) 938. [21] R.M. Cannon, W.H. Rhodes and A.H. Heuer, Plastic deformation of fine-grained alumina (A1203): I, Interface-controlled diffusional creep, J. Am. Ceram. Soc., 63 (1980) 46. [22] A.H. Heuer, N.J. Tighe and R.M. Cannon, Plastic deformation of fine-grained alumina (AIzO3): II, Basal slip and nonaccomodated grain-boundary sliding, J. Am. C2,ram. Sot., 63 (1980) 53. [23] J.R. Porter, W. Blumenthal and A.G. Evans, Creep fracture in ceramic polycrystals-l. Creep cavitation effects in polycrystalline alumina, Acta Metall., 29 (1981) 1899. [24] R.A. Page, J. Lankford and S. Spooner, Small-angle neutron scattering study of creep cavity nucleation and growth in sintered alumina, J. Mater. Sci., 19 (1984) 3360. [25] D.F. Carroll and R.E. Tressler, Accumulation of creep damage in a siliconized silicon carbide, J. Am. Ceram. Soc.. 71 (1988) 472. [26] S.M. Johnson, B.J. Dalgleish and A.G. Evans, High-temperature failure of polycrystalline alumina: Ill, Failure times. J. Am. Ceram. Soe., 67 (1984) 759