May 2006 Creep and After-Creep Stress-Strain Behavior 1655 Table lll. offset stress at 1315C and matrix Cracking observations Creep stress 1315C initial loading 1315C after creep Creep strain Panel Experim (MPa) (MPa) Observed matrix cracking due to cre 11315°C 1315°C 150/167 1315°C 82/>103 147/185 0.09 A21315°C 103/18 1315°C 6/188 1315°C 88/103 59/84 0.3 Some bridged surface cracks propagate 4 pI 0.23 Cracks only in surface 90 minicomposites 1315°C 0.44 Some bridged cracks containing local unbridged 95/>103 08/143 0.07 1315°C 138 90/107 36 Some bridged surface cracks propagate mately 60 MPa(Fig. 4a). This compressive stress was Cracks formed during after-creep fast-fracture were generally cantly higher than the 20 MPa residual stress obtained much finer, and could be observed only after plasma etching the as-polished sections. The difficulty in observing these cracks is matrix relaxes during creep so that upon stress removal, it was again because of the residual compressive stress in the matrix placed in even greater compression which had to be overcome in hich results in very little crack opening. The polished micro- order to cause matrix crack formation and growth. Micro- structures described here are as-polished sections which were not ructural observations of the specimen after fracture, as shown plasma etched in order to show the effects of creep on micro- below, are consistent with this notion, as only minor surface ructural change. micro-cracks were observed in this specimen. No matrix cracking attributed to creep could be discerned for The tensile spe n that had been crept for 138 MPa and imens from all three panels that were subject to creep at 103 100 h from panel B(high Si content, low CVI SiC content)was MPa. The specimen from panel Al that was subjected to tensile also tensile hysteresis tested at room temperature (Fig. 5(a)) reep at 138 MPa for 100 h and subsequently tested at room This specimen had the highest creep strain without failing during temperature did have relatively wide open, oxide- filled surface 100 h creep. The differences between the tensile stress-strain cracks emanating from the 90 tows at the surface of the com- havior of the specimen before creep and after creep was not posite; however, the matrix cracks did not appear to penetrate the same as for the panel A specimen (Table IV). For the panel b into the adjacent 0 fiber tow(Fig. 6). For the specimen from specimen no real difference was observed in offset-stress or AE panel A2 with a creep strain of 0.3% after 138 MPa creep which activity(Fig. 5(b)and only a mild improvement was observed as immediately fast fractured at 1315.C and the specimen from n residual compressive stress. Also the elastic modulus at room panel B subjected to 138 MPa tensile creep and tested at room temperature for this crept specimen was particularly low in temperature, similar surface 90 matrix cracks were observed to comparison with its as-produced modulus at room temperature ropagate several plies into the composite and were bridged by the specimen showed the specimen had bridged matrix cracks 0 fibers (Table IIf). The specimen from panel A2 that was crept at 172 MPa was the only specimen that showed fiber failure as- that had to occur during creep(next section). As matrix cracks lready existed before testing at room temperature after the cracks. However, with the exception of the failure crack, the creep test, it is not surprising that the elastic modulus was lower unbridged portion of the matrix cracks were localized and were and that no real improvement in matrix cracking stress was ob- all associated with the same 0 tow, three plies from one side of the specimen(Fig. 7). Six unbridged cracks were observed for the section polished and were spaced 2.3, 3.2, 2.6, 1.6, 1.2, and 1. 6 mm apart from one another moving away from the fracture 3) Matrix Cracking surface. Some of the unbridged matrix cracks did extend The matrix cracking behavior of several specimens is described er-bridged cracks to neighboring plies and/or extend to th failed specimens. Only the 172 MPa crept specimen had failed in appeared to extend a total of four plies, i. e, half the thickness pture. All of the other specimens were failed during the afte However, some of the cracks did not appear to propagate more creep tensile testing, which always occurred at a higher load thanthan one ply away from the unbridged bundle. The rest of the the creep stress. These latter specimens would be expected te matrix cracks were in between one and four plies in length. The have larger amounts of matrix cracks because of the fast-frac- polished surface in Fig. 7 corresponds to the region of the gage ture condition. It was generally easy to distinguish matrix cracks section that was 2 mm from one edge of the specimen. To de- formed during creep from those that were formed during the termine if any unbridged cracks existed in the interior of the after-creep fast-fracture test Matrix cracks formed during creep specimen, an additional 4 mm was sliced from the mounted were easy to observe in polished sections because they possessed polished section (i. e, 6 mm from the same edge or 4 mm from gnificant crack openings were generally accompanied by edge). None were observed oxide formation within the crack and along the pores, especially that significant fiber-bridged matrix cracking oc- those cracks that clearly extended to the composite surface. curred in specimens which had exhibited significant creep strainsmately 60 MPa (Fig. 4a). This compressive stress was signifi- cantly higher than the 20 MPa residual stress obtained for the as-produced specimen.7 These data support the notion that the matrix relaxes during creep so that upon stress removal, it was placed in even greater compression which had to be overcome in order to cause matrix crack formation and growth.11–13 Micro￾structural observations of the specimen after fracture, as shown below, are consistent with this notion, as only minor surface micro-cracks were observed in this specimen. The tensile specimen that had been crept for 138 MPa and 100 h from panel B (high Si content, low CVI SiC content) was also tensile hysteresis tested at room temperature (Fig. 5(a)). This specimen had the highest creep strain without failing during 100 h creep. The differences between the tensile stress–strain behavior of the specimen before creep and after creep was not the same as for the panel A specimen (Table IV). For the panel B specimen no real difference was observed in offset-stress or AE activity (Fig. 5(b)) and only a mild improvement was observed in residual compressive stress. Also the elastic modulus at room temperature for this crept specimen was particularly low in comparison with its as-produced modulus at room temperature and at temperature before creep (see Table II). Microscopy of the specimen showed the specimen had bridged matrix cracks that had to occur during creep (next section). As matrix cracks already existed before testing at room temperature after the creep test, it is not surprising that the elastic modulus was lower and that no real improvement in matrix cracking stress was ob￾served in this specimen. (3) Matrix Cracking The matrix cracking behavior of several specimens is described in Table III as determined from polished longitudinal sections of failed specimens. Only the 172 MPa crept specimen had failed in rupture. All of the other specimens were failed during the after￾creep tensile testing, which always occurred at a higher load than the creep stress. These latter specimens would be expected to have larger amounts of matrix cracks because of the fast-frac￾ture condition. It was generally easy to distinguish matrix cracks formed during creep from those that were formed during the after-creep fast-fracture test. Matrix cracks formed during creep were easy to observe in polished sections because they possessed significant crack openings and were generally accompanied by oxide formation within the crack and along the pores, especially those cracks that clearly extended to the composite surface. Cracks formed during after-creep fast-fracture were generally much finer, and could be observed only after plasma etching the as-polished sections.6 The difficulty in observing these cracks is again because of the residual compressive stress in the matrix which results in very little crack opening. The polished micro￾structures described here are as-polished sections which were not plasma etched in order to show the effects of creep on micro￾structural change. No matrix cracking attributed to creep could be discerned for specimens from all three panels that were subject to creep at 103 MPa. The specimen from panel A1 that was subjected to tensile creep at 138 MPa for 100 h and subsequently tested at room temperature did have relatively wide open, oxide-filled surface cracks emanating from the 901 tows at the surface of the com￾posite; however, the matrix cracks did not appear to penetrate into the adjacent 01 fiber tow (Fig. 6). For the specimen from panel A2 with a creep strain of 0.3% after 138 MPa creep which was immediately fast fractured at 13151C and the specimen from panel B subjected to 138 MPa tensile creep and tested at room temperature, similar surface 901 matrix cracks were observed to propagate several plies into the composite and were bridged by 01 fibers (Table III). The specimen from panel A2 that was crept at 172 MPa was the only specimen that showed fiber failure as￾sociated with matrix cracks, i.e., unbridged portions of matrix cracks. However, with the exception of the failure crack, the unbridged portion of the matrix cracks were localized and were all associated with the same 01 tow, three plies from one side of the specimen (Fig. 7). Six unbridged cracks were observed for the section polished and were spaced 2.3, 3.2, 2.6, 1.6, 1.2, and 1.6 mm apart from one another moving away from the fracture surface. Some of the unbridged matrix cracks did extend as fib￾er-bridged cracks to neighboring plies and/or extend to the sur￾face. The bridged portion of some of these matrix cracks appeared to extend a total of four plies, i.e., half the thickness. However, some of the cracks did not appear to propagate more than one ply away from the unbridged bundle. The rest of the matrix cracks were in between one and four plies in length. The polished surface in Fig. 7 corresponds to the region of the gage section that was 2 mm from one edge of the specimen. To de￾termine if any unbridged cracks existed in the interior of the specimen, an additional 4 mm was sliced from the mounted polished section (i.e., 6 mm from the same edge or 4 mm from the opposite edge). None were observed. It appears that significant fiber-bridged matrix cracking oc￾curred in specimens which had exhibited significant creep strains Table III. Offset Stress at 13151C and Matrix Cracking Observations Panel Experiment Creep stress (MPa) 0.002%/0.005% offset stress Creep strain (%) Observed matrix cracking due to creep 13151C initial loading (MPa) 13151C after creep (MPa) A1 13151C — 70 — — — 13151C creep 103 NA 150/167 0.18 — 13151C creep 103 82/4103 147/185 0.09 None A2 13151C — 103/118 — — — 13151C creep 103 67/97 166/188 0.05 None 13151C creep 138 88/103 159/184 0.3 Some bridged surface cracks propagate up to 4 plies 13151C creep 138 79/98 — 0.23 Cracks only in surface 901 minicomposites 13151C creep 172 85/103 — 0.44 Some bridged cracks containing local unbridged regions B 13151C creep 103 95/4103 108/143 0.07 None 13151C creep 138 90/107 — 0.36 Some bridged surface cracks propagate up to 4 plies Each row corresponds to an individual tensile specimen. May 2006 Creep and After-Creep Stress–Strain Behavior 1655