G.N. Morscher et al/ Composites Science and Technology 68 (2008)3305-3313 a10 1 HZ HCF RT Crack m:= 6 0.3 Tunel 2 Tunnel/microcrack 1 HZ HCF 30 HZ HCF 0. 400 Stress, MPa Fig. 10. Matrix crack density versus stress for cracks that propagate at least 1 ply. (b)is a the data in(a)magnified at the lower stress region. 3.3. FESEM fracture surface examination 110 MPa specimen which failed at 1953 h. Three specimens that did not fail at temperature and were subsequently tested at In order to assess the nature of failure for the composites, the room temperature showed no evidence of oxidation at the frac actual fracture surfaces were examined using a FESEM. Note that ture surface: an 165 MPa 30 Hz HCF specimen that survived when the specimen failed the furnace was immediately shut off 42,000,000 cycles, an 110 MPa creep specimen that survived so that the time the interior of the fracture surface was exposed 2036 h(Fig. 13), and an 110 MPa creep specimen that survived to significant temperatures and prolonged oxidation was only a 1269 h. few minutes at most. Thirteen specimens were examined from the HCF (220, 192, 179, and 165 MPa), DF (220 MPa)and creep 4. Discussion (110 and 165 MPa)specimens(Fig. 2)that failed at temperature oom temperature after the specimens had been subjected For this composite system, two mechanistic regimes appear to describe time-dependent strength-degradation: (1) oxidation-as- For specimens subjected to stresses of 179 MPa and above, the sisted unbridged matrix crack growth and(2)fiber degradatior predominant feature on the fracture surface was a large not associated with oxidation. These two mechanisms may be syn- where an unbridged crack existed prior to failure. In other ergistic at intermediate stresses; however, the former dominates gions of the fracture surface were oxidized including fib the higher stress -shorter time conditions whereas the latte ture surfaces, the bn region, and the matrix surface as observed mechanism controls the lower stress -long time conditions. from the polished sections. The rest of the fracture surface was not oxidized, i.e the Sic matrix surface, fiber fracture surface, 4.1. Oxidation-induced unbridged crack growth and Bn interphase were all not oxidized and fiber pullout was pre- plied stress conditions (greater than 165 MPa). nost often observed to propagate from the exposed edge of the non-through-the-thickness fiber-bridged matrix cracks that inter- ross-section some depth into the width of the specimen sect the surface of the composite are exposed to the oxidizing envi- (Fig. 11). There were a few cases where the unbridged crack prop- ronment. Oxygen ingress into the crack occurs, BN and Sic react to agated from the exposed face of the cross-section(Fig. 12)some form gaseous species and solid borosilicate reaction products that depth into the thickness of the specimen. In all cases, the matrix fuse fibers together (similar to intermediate temperatures-see acks appeared to emanate from at least one corner of the speci- men cross-section 1 for a number of possible reasons: intrinsic fiber degradation Again, the fact that fiber fracture surfaces were oxidized indi- stressed-oxidation degradation of fibers, and or local stress-con- cates that these fibers failed before the ultimate failure event of centrations created from local-load sharing conditions as a result the composite. the oxide layer that covers the fibers does appear of strongly bonded fibers The result is transverse unbridged micro- to be thicker near the edge and is thinner away from the edge indi- cracks of significant depth. One, several, or many of these cracks ating that the fibers closer to the edge probably failed earlier [15]. exist along the length of the specimen depending on the stress- the opposite edge indicates that the matrix crack that led to ulti- the specimen is essentially the same as observed for room temper mate failure was not through-the-cross-section ature stress-strain, except that they are not through-the-thickness The fracture surfaces of specimens subjected to stresses at and become unbridged with time at 1204C. These unbridged ma- 110 and 165 MPa either had a very small region of oxidation trix cracks result in redistribution of load to the intact region of the on the fracture surface or no real evidence of oxidation-induced composite cross-section and local stress-concentrations near the embrittlement on the fracture surface. For two specimens that crack tip. Ultimately one of these cracks becomes the source of ruptured during creep, a triangular-shaped oxidized region of rupture as time continues. the fracture surface emanated from one corner of the fracture surface about two plies deep at the edge (the deepest part) 42. Fiber strength degradation not due to oxidation width of th3.3. FESEM fracture surface examination In order to assess the nature of failure for the composites, the actual fracture surfaces were examined using a FESEM. Note that when the specimen failed the furnace was immediately shut off so that the time the interior of the fracture surface was exposed to significant temperatures and prolonged oxidation was only a few minutes at most. Thirteen specimens were examined from the HCF (220, 192, 179, and 165 MPa), DF (220 MPa) and creep (110 and 165 MPa) specimens (Fig. 2) that failed at temperature or at room temperature after the specimens had been subjected to a creep or fatigue. For specimens subjected to stresses of 179 MPa and above, the predominant feature on the fracture surface was a large region where an unbridged crack existed prior to failure. In other words, regions of the fracture surface were oxidized including fiber frac￾ture surfaces, the BN region, and the matrix surface as observed from the polished sections. The rest of the fracture surface was not oxidized, i.e., the SiC matrix surface, fiber fracture surface, and BN interphase were all not oxidized and fiber pullout was pre￾valent. At these stresses, the unbridged crack that led to failure was most often observed to propagate from the exposed edge of the cross-section some depth into the width of the specimen (Fig. 11). There were a few cases where the unbridged crack prop￾agated from the exposed face of the cross-section (Fig. 12) some depth into the thickness of the specimen. In all cases, the matrix cracks appeared to emanate from at least one corner of the speci￾men cross-section. Again, the fact that fiber fracture surfaces were oxidized indi￾cates that these fibers failed before the ultimate failure event of the composite. The oxide layer that covers the fibers does appear to be thicker near the edge and is thinner away from the edge indi￾cating that the fibers closer to the edge probably failed earlier [15]. The fact that no oxidation is observed on the fracture surface near the opposite edge indicates that the matrix crack that led to ulti￾mate failure was not through-the-cross-section. The fracture surfaces of specimens subjected to stresses at 110 and 165 MPa either had a very small region of oxidation on the fracture surface or no real evidence of oxidation-induced embrittlement on the fracture surface. For two specimens that ruptured during creep, a triangular-shaped oxidized region of the fracture surface emanated from one corner of the fracture surface about two plies deep at the edge (the deepest part) and about 5 mm long along the width of the cross-section for the 165 MPa creep specimen that failed at 478 h and about 2 mm long along the width of the cross-section for the 110 MPa specimen which failed at 1953 h. Three specimens that did not fail at temperature and were subsequently tested at room temperature showed no evidence of oxidation at the frac￾ture surface: an 165 MPa 30 Hz HCF specimen that survived 42,000,000 cycles, an 110 MPa creep specimen that survived 2036 h (Fig. 13), and an 110 MPa creep specimen that survived 1269 h. 4. Discussion For this composite system, two mechanistic regimes appear to describe time-dependent strength-degradation: (1) oxidation-as￾sisted unbridged matrix crack growth and (2) fiber degradation not associated with oxidation. These two mechanisms may be syn￾ergistic at intermediate stresses; however, the former dominates the higher stress – shorter time conditions whereas the latter mechanism controls the lower stress – long time conditions. 4.1. Oxidation-induced unbridged crack growth For higher applied stress conditions (greater than 165 MPa), non-through-the-thickness fiber-bridged matrix cracks that inter￾sect the surface of the composite are exposed to the oxidizing envi￾ronment. Oxygen ingress into the crack occurs, BN and SiC react to form gaseous species and solid borosilicate reaction products that fuse fibers together (similar to intermediate temperatures – see Ref. [15,16]). After some time fibers in the oxidized matrix cracks fail for a number of possible reasons: intrinsic fiber degradation, stressed-oxidation degradation of fibers, and/or local stress-con￾centrations created from local-load sharing conditions as a result of strongly bonded fibers. The result is transverse unbridged micro￾cracks of significant depth. One, several, or many of these cracks exist along the length of the specimen depending on the stress￾state (Fig. 9). The number of matrix cracks along the length of the specimen is essentially the same as observed for room temper￾ature stress-strain, except that they are not through-the-thickness and become unbridged with time at 1204 C. These unbridged ma￾trix cracks result in redistribution of load to the intact region of the composite cross-section and local stress-concentrations near the crack tip. Ultimately one of these cracks becomes the source of rupture as time continues. 4.2. Fiber strength degradation not due to oxidation At lower applied stresses, degradation in composite ultimate strength was due to a fiber-degradation mechanism not caused 0 1 2 3 4 5 6 7 8 9 10 0 100 200 300 400 500 Stress, MPa Crack Density, mm-1 RT Crack Density Based on All AE Events RT Crack Density Based on only High AE Energy Events Tunnel/microcrack formation 30 HZ HCF DF 1 HZ HCF Creep 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 Stress, MPa Crack Density, mm-1 RT Crack Density Based on All AE Events RT Crack Density Based on only High AE Energy Events Tunnel￾microcrack formation 30 HZ HCF DF 1 HZ HCF Creep 100 200 300 a b Fig. 10. Matrix crack density versus stress for cracks that propagate at least 1 ply. (b) is a the data in (a) magnified at the lower stress region. 3310 G.N. Morscher et al. / Composites Science and Technology 68 (2008) 3305–3313