《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_SiC-iBN-SiC

Composites Science and Technology 68(2008)3305-3313 Contents lists available at ScienceDirect Composites Science and Technology ELSEVIER journalhomepagewww.elsevier.com/locate/compscitech Tensile creep and fatigue of Sylramic-iBN melt-infiltrated Sic matrix composites Retained properties, damage development, and failure mechanisms Gregory N Morscher Greg Ojard, Robert miller Yasser Gowayed, Unni Santhosh Jalees Ahmad Reji john Materials and Manufacturing Directorate, Air Force Research Laboratory, AFRL/RXLMN, Wright-Patterson AFB, OH, USA SA Glenn Research Center, 21000 Brookpark Road, MS 106-5, Cleveland, OH 44135, US pRatt and Whitney, East Hartford, CT, USA d Auburn University, Auburn, AL, USA Research Applications Inc, San Diego, CA, USA ARTICLE INFO ABSTRACT An understanding of the elevated temperature tensile creep, fatigue, rupture, and retained properties of eceived in revised form 8 August 2008 ceramic matrix composites( CMC)envisioned for use in gas turbine engine applications is essential for omponent design and life-prediction. In order to quantify the effect of stress, time, temperature, and oxi- Accepted 21 August 2008 Available online 11 September 2008 dation for a state-of-the-art composite system, a wide variety of tensile creep, dwell fatigue, and cyclic fatigue experiments were performed in air at 1204C for the SiC/SiC CMC system consisting of Sylram- ic-iBN SiC fibers, BN fiber interphase coating, and slurry-cast melt-infiltrated( Mi)SiC-based matrix. Tests were either taken to failure or interrupted. Interrupted tests were then mechanically tested at room tem- perature to determine the residual properties. The retained properties of most of the composites sub- jected to tensile creep or fatigue were usually within 20% of the as-produced strength and 10% of the s-produced elastic modulus. It was observed that during creep, residual stresses in the composite are Tered to some extent which results in an increased compressive stress in the matrix upon cooling nd a subsequent increased stress required to form matrix cracks Microscopy of polished sections and the fracture surfaces of specimens which failed during stressed-oxidation or after the room-temperature etained property test was performed on some of the specimens in order to quantify the nature and extent of damage accumulation that occurred during the test. It was discovered that the distribution f stress-dependent matrix cracking at 1204C was similar to the as-produced composites at room tem- perature; however, matrix crack growth occurred over time and typically did not appear to propagate through-the-thickness except at the final failure crack Failure of the composites was due to either oxi- dation- induced unbridged crack growth, which dominated the higher stress regime(>179 MPa)or con- trolled by degradation of the fibers, probably caused by intrinsic creep-induced flaw growth of the fibers or internal attack of the fibers via Si diffusion through the Cvi SiC and or microcracks at the lower stress regime(≤165MPa) e 2008 Elsevier Ltd. All rights reserved. understand the mechanisms that lead to degradation in stress-strain behavior and ultimate creep or fatigue rupture Silicon carbide fiber-reinforced silicon-carbide matrix compos- The effect of damage development in some 2D-woven 0/90 SiC- ites are currently being evaluated for aircraft engine hot-section reinforced non-oxide reinforced composites has been determined components [1-3. Elevated temperature creep and fatigue condi- for CVI SiC matrix systems with lower modulus Sic-based fiber- tions under oxidative environments are a primary concern for these types(Nicalon"and Hi-Nicalon")for creep and fatigue conditions types of applications. Therefore, it is essential that the effect of ele- between 1100 and 1400C in air and argon environments [4-6]. In vated temperature creep and fatigue conditions be well understood those composite systems, with increasing stress and time, micro in order to predict useful lives for these types of composites. As part cracks are formed in the 90 minicomposite bundles with increas- of this effort, it is critical to understand the development of damage ing stress and or time, the cracks grow and extend through the cvi in these materials over stress, time, and environment in order to Sic into the load-bearing 0 fiber minicomposite bundles produc ing through-the-thickness matrix cracks. Eventually a"master Corresponding author. Tel : +1 216 433 5512. crack"[6 will form that becomes the site of ultimate failure. For mail address: gmorscheresbcglobal net (G N Morscher these composite systems, fairly high strains to failure(0.5% to 3538/s-see front matter o 2008 Elsevier Ltd. All rights rese

Tensile creep and fatigue of Sylramic-iBN melt-infiltrated SiC matrix composites: Retained properties, damage development, and failure mechanisms Gregory N. Morscher b,*, Greg Ojard c , Robert Miller c , Yasser Gowayed d , Unni Santhosh e , Jalees Ahmad e , Reji John a a Materials and Manufacturing Directorate, Air Force Research Laboratory, AFRL/RXLMN, Wright-Patterson AFB, OH, USA bOhio Aerospace Institute, NASA Glenn Research Center, 21000 Brookpark Road, MS 106-5, Cleveland, OH 44135, USA c Pratt and Whitney, East Hartford, CT, USA d Auburn University, Auburn, AL, USA e Research Applications Inc., San Diego, CA, USA article info Article history: Received 28 March 2008 Received in revised form 8 August 2008 Accepted 21 August 2008 Available online 11 September 2008 Keywords: A. Ceramic matrix composites B. Creep B. Fatigue B. Matrix cracking D. Acoustic emission abstract An understanding of the elevated temperature tensile creep, fatigue, rupture, and retained properties of ceramic matrix composites (CMC) envisioned for use in gas turbine engine applications is essential for component design and life-prediction. In order to quantify the effect of stress, time, temperature, and oxi￾dation for a state-of-the-art composite system, a wide variety of tensile creep, dwell fatigue, and cyclic fatigue experiments were performed in air at 1204 C for the SiC/SiC CMC system consisting of Sylram￾ic-iBN SiC fibers, BN fiber interphase coating, and slurry-cast melt-infiltrated (MI) SiC-based matrix. Tests were either taken to failure or interrupted. Interrupted tests were then mechanically tested at room tem￾perature to determine the residual properties. The retained properties of most of the composites sub￾jected to tensile creep or fatigue were usually within 20% of the as-produced strength and 10% of the as-produced elastic modulus. It was observed that during creep, residual stresses in the composite are altered to some extent which results in an increased compressive stress in the matrix upon cooling and a subsequent increased stress required to form matrix cracks. Microscopy of polished sections and the fracture surfaces of specimens which failed during stressed-oxidation or after the room-temperature retained property test was performed on some of the specimens in order to quantify the nature and extent of damage accumulation that occurred during the test. It was discovered that the distribution of stress-dependent matrix cracking at 1204 C was similar to the as-produced composites at room tem￾perature; however, matrix crack growth occurred over time and typically did not appear to propagate through-the-thickness except at the final failure crack. Failure of the composites was due to either oxi￾dation-induced unbridged crack growth, which dominated the higher stress regime (P179 MPa) or con￾trolled by degradation of the fibers, probably caused by intrinsic creep-induced flaw growth of the fibers or internal attack of the fibers via Si diffusion through the CVI SiC and/or microcracks at the lower stress regime (6165 MPa). 2008 Elsevier Ltd. All rights reserved. 1. Introduction Silicon carbide fiber-reinforced silicon-carbide matrix compos￾ites are currently being evaluated for aircraft engine hot-section components [1–3]. Elevated temperature creep and fatigue condi￾tions under oxidative environments are a primary concern for these types of applications. Therefore, it is essential that the effect of ele￾vated temperature creep and fatigue conditions be well understood in order to predict useful lives for these types of composites. As part of this effort, it is critical to understand the development of damage in these materials over stress, time, and environment in order to understand the mechanisms that lead to degradation in stress-strain behavior and ultimate creep or fatigue rupture. The effect of damage development in some 2D-woven 0/90 SiC￾reinforced non-oxide reinforced composites has been determined for CVI SiC matrix systems with lower modulus SiC-based fiber￾types (NicalonTM and Hi-NicalonTM) for creep and fatigue conditions between 1100 and 1400 C in air and argon environments [4–6]. In those composite systems, with increasing stress and time, micro cracks are formed in the 90 minicomposite bundles. With increas￾ing stress and/or time, the cracks grow and extend through the CVI SiC into the load-bearing 0 fiber minicomposite bundles produc￾ing through-the-thickness matrix cracks. Eventually a ‘‘master crack” [6] will form that becomes the site of ultimate failure. For these composite systems, fairly high strains to failure (0.5% to 0266-3538/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2008.08.028 * Corresponding author. Tel.: +1 216 433 5512. E-mail address: gmorscher@sbcglobal.net (G.N. Morscher). Composites Science and Technology 68 (2008) 3305–3313 Contents lists available at ScienceDirect Composites Science and Technology journal homepage: www.elsevier.com/locate/compscitech

G.N. Morscher et al/ Composites Science and Technology 68 (2008)3305-3313 greater than 1%)are achieved. However, times to failure in air are Some of the elevated temperature tests were performed in usually less than 100 h for stresses that are in excess of the matrix other studies[9-10. All the tests were performed at 1204C in cracking stress (e.g, x100 MPa for the Hi-Nicalon m CVI SiC com- lab air at Southern research Institute, Birmingham AL or Cincinnati posite tested in Ref. 4]at 1300C Test Labs, Cincinnati, OH. Tensile creep tests were performed under For SiC/SiC composites reinforced with high modulus poly constant load using either universal testing or lever-arm machines talline SiC fibers(Hi-Nicalon S)and a slurry-cast melt-infiltrated The cyclic fatigue tests were all performed in hydraulic testing ma- (MI) SiC matrix composite system a somewhat different damage chines; while dwell fatigue testing was done in modified lever-arm development was observed at 1315C[7. For the stress ranges tensile machines to control loading rates. Several fatigue loading tested(up to 172 MPa), only minor microcracking in the 90 mini- rates were used. Dwell fatigue(DF)consisted of a 2 h hold followed composites was observed. In some cases, these cracks were by a 1 min unload-reload back to a two hour hold High cycle fati- observed to extend to the surface(138 MPa)where 90 minicom- gue(HCF)was performed at 1 or 30 Hz. All of the fatigue tests were osites were adjacent to the surface and into some 0 minicompos- performed for an R ratio of 0.05. Displacement for all tests was tes at higher stresses(172 MPa)resulting in local fiber failure but monitored using contact extensometers on the edge of the speci- not significant through-the-thickness matrix crack formation. men with a gage length of 25 mm. Creep strain was determined 00 h followed by determination of retained stress-strain behavior, perature along the gage section he furnace assured uniform tem hich was usually very high since minimal damage occurred in the For some of the specimens which were not taken to failure, a opposites. room temperature unload-reload hysteresis tensile test was per It is the goal of this study to further understand the develop- formed to failure to determine residual properties. aE sensors were ment of damage in a similar polycrystalline Sic fiber-reinforced placed above and below the extensometers(50-60 mm apart ) and slurry-cast MI SiC matrix system, the Sylramic-iBN MI SiC sys- AE was monitored during the room temperature test using a Digi em developed at NASA Glenn Research Center and referred to tal Wave Fracture Wave detector (Englewood Co)as described as N24A[ 8]. This material has undergone some of the most elsewhere [7, 11. After the test, the events were sorted as to loca exhaustive testing to date of any high-performance SiC/Sic com- tion along the specimen length and only those events which posite system under an US Air Force sponsored program, includ- occurred in the 25 mm gage section were used for AE analysis. tensile creep, 2 hour dwell fatigue, 1 Hz cyclic fatigue, and Analysis of the crept/fatigued specimens consisted of both opti 30 Hz cyclic fatigue at 1204C in air for stresses ranging from cal and scanning electron microscopy. Fig. 1 shows a typical failed 110 to 220 MPa for times up to 2000h [ 9-10. Specimens from specimen. One part of the fracture surface was used for observation this wide range of tests were acquired for this study in order of the fracture surface in a field emission scanning electron micro- to determine the damage accumulation for the wide range of scope(FESEM-Hitachi 4700, Tokyo Japan). The other half of the stress-time conditions which will be described here. Some of specimen was cut(between 10 and 15 mm long) and polished the specimens had been interrupted at predetermined times. along the edge (approximately 1 mm from the exposed edge For most of those specimens, a room temperature unload-reload and or face of the specimen in order to observe and quantify the test to failure was performed with acoustic emission(AE)mon- number of matrix cracks along the length Matrix cracks caused itoring in the same way as Ref. [7 in order to determine the by time-dependent deformation had significant crack-openings residual properties of the composite and were easy to distinguish from matrix cracks formed during the room temperature retained strength tests, which were often not discernable as-polished due to crack closure from the high 2. Experimental compressive stresses in the matrix For the edge-polished speci- mens, matrix cracks were counted along the length for both The composite system evaluated is described in more detail elsewhere [8 It consists of eight plies of 2D woven five-harness satin Sylramic-iBN fabric, a CVI BN fiber interphase coating, a CVI Sic coating of initial matrix to rigidize the preform and protect the fibers, slurry-infiltrated SiC particulates, and molten Si infil- trated to fill in the remaining open porosity. The initial five-harness fabric consisted of 7.9 tow ends per cm of Sylramic SiC fiber bal anced in the warp and weft 0/90 orthogonal directions. The fiber ows consisted of 800 fibers approximately 10 um in diameter. The fibers were originally produced by Dow Corning, Midland, Michigan, but are now produced by ATK Col Ceramics in San Diego CA The precursor Sylramic fabric plies were subjected to a NASA- proprietary treatment in order to improve fiber creep-resistance and also produce a thin(150 nm) in situ grown BN layer on the sEM|丙 urface of each fiber prior to composite fabrication. The composites were fabricated as 153 x 230 mm panels by ply lay-up at GE Cera- nic Composite Products, LLC in Newark, DE All plies were aligned /90 in plane but were randomly stacked through-the-thickness with a degree of fiber nesting. The fiber volume fraction of compos- ites was approximately 36-38% as measured by weight. The tensile specimens were machined from the as-fabricated panels with a length of 155 mm, a grip width of 12 mm, and a dogboned section in the middle with a length of 40 mm and width of 8.2 mm. One of the orthogonal fiber directions was aligned within +/-3 of the specimen length or tensile direction. Typical specimen thickness was 2 mm. Fig. 1. Typical failed specimen after creep or fatigue

greater than 1%) are achieved. However, times to failure in air are usually less than 100 h for stresses that are in excess of the matrix cracking stress (e.g., 100 MPa for the Hi-NicalonTM CVI SiC com￾posite tested in Ref. [4] at 1300 C). For SiC/SiC composites reinforced with high modulus polycrys￾talline SiC fibers (Hi-Nicalon S) and a slurry-cast melt-infiltrated (MI) SiC matrix composite system a somewhat different damage development was observed at 1315 C [7]. For the stress ranges tested (up to 172 MPa), only minor microcracking in the 90 mini￾composites was observed. In some cases, these cracks were observed to extend to the surface (138 MPa) where 90 minicom￾posites were adjacent to the surface and into some 0 minicompos￾ites at higher stresses (172 MPa) resulting in local fiber failure but not significant through-the-thickness matrix crack formation. However, for this study, creep times were usually limited to 100 h followed by determination of retained stress-strain behavior, which was usually very high since minimal damage occurred in the composites. It is the goal of this study to further understand the develop￾ment of damage in a similar polycrystalline SiC fiber-reinforced slurry-cast MI SiC matrix system, the Sylramic-iBN MI SiC sys￾tem developed at NASA Glenn Research Center and referred to as N24A [8]. This material has undergone some of the most exhaustive testing to date of any high-performance SiC/SiC com￾posite system under an US Air Force sponsored program, includ￾ing tensile creep, 2 hour dwell fatigue, 1 Hz cyclic fatigue, and 30 Hz cyclic fatigue at 1204 C in air for stresses ranging from 110 to 220 MPa for times up to 2000 h [9–10]. Specimens from this wide range of tests were acquired for this study in order to determine the damage accumulation for the wide range of stress-time conditions which will be described here. Some of the specimens had been interrupted at predetermined times. For most of those specimens, a room temperature unload-reload test to failure was performed with acoustic emission (AE) mon￾itoring in the same way as Ref. [7] in order to determine the residual properties of the composite. 2. Experimental The composite system evaluated is described in more detail elsewhere [8]. It consists of eight plies of 2D woven five-harness satin Sylramic-iBN fabric, a CVI BN fiber interphase coating, a CVI SiC coating of initial matrix to rigidize the preform and protect the fibers, slurry-infiltrated SiC particulates, and molten Si infil￾trated to fill in the remaining open porosity. The initial five-harness fabric consisted of 7.9 tow ends per cm of SylramicTM SiC fiber bal￾anced in the warp and weft 0/90 orthogonal directions. The fiber tows consisted of 800 fibers approximately 10 lm in diameter. The fibers were originally produced by Dow Corning, Midland, Michigan, but are now produced by ATK COI Ceramics in San Diego, CA. The precursor Sylramic fabric plies were subjected to a NASA￾proprietary treatment in order to improve fiber creep-resistance and also produce a thin (150 nm) in situ grown BN layer on the surface of each fiber prior to composite fabrication. The composites were fabricated as 153 230 mm panels by ply lay-up at GE Cera￾mic Composite Products, LLC in Newark, DE. All plies were aligned 0/90 in plane, but were randomly stacked through-the-thickness with a degree of fiber nesting. The fiber volume fraction of compos￾ites was approximately 36–38% as measured by weight. The tensile specimens were machined from the as-fabricated panels with a length of 155 mm, a grip width of 12 mm, and a dogboned section in the middle with a length of 40 mm and a width of 8.2 mm. One of the orthogonal fiber directions was aligned within +/3 of the specimen length or tensile direction. Typical specimen thickness was 2 mm. Some of the elevated temperature tests were performed in other studies [9–10]. All the tests were performed at 1204 C in lab air at Southern Research Institute, Birmingham AL or Cincinnati Test Labs, Cincinnati, OH. Tensile creep tests were performed under constant load using either universal testing or lever-arm machines. The cyclic fatigue tests were all performed in hydraulic testing ma￾chines; while dwell fatigue testing was done in modified lever-arm tensile machines to control loading rates. Several fatigue loading rates were used. Dwell fatigue (DF) consisted of a 2 h hold followed by a 1 min unload-reload back to a two hour hold. High cycle fati￾gue (HCF) was performed at 1 or 30 Hz. All of the fatigue tests were performed for an R ratio of 0.05. Displacement for all tests was monitored using contact extensometers on the edge of the speci￾men with a gage length of 25 mm. Creep strain was determined after reaching full applied stress. The furnace assured uniform tem￾perature along the gage section. For some of the specimens which were not taken to failure, a room temperature unload-reload hysteresis tensile test was per￾formed to failure to determine residual properties. AE sensors were placed above and below the extensometers (50–60 mm apart) and AE was monitored during the room temperature test using a Digi￾tal Wave Fracture Wave detector (Englewood CO) as described elsewhere [7,11]. After the test, the events were sorted as to loca￾tion along the specimen length and only those events which occurred in the 25 mm gage section were used for AE analysis. Analysis of the crept/fatigued specimens consisted of both opti￾cal and scanning electron microscopy. Fig. 1 shows a typical failed specimen. One part of the fracture surface was used for observation of the fracture surface in a field emission scanning electron micro￾scope (FESEM – Hitachi 4700, Tokyo Japan). The other half of the specimen was cut (between 10 and 15 mm long) and polished along the edge (approximately 1 mm from the exposed edge) and/or face of the specimen in order to observe and quantify the number of matrix cracks along the length. Matrix cracks caused by time-dependent deformation had significant crack-openings and were easy to distinguish from matrix cracks formed during the room temperature retained strength tests, which were often not discernable as-polished due to crack closure from the high compressive stresses in the matrix. For the edge-polished speci￾mens, matrix cracks were counted along the length for both Fig. 1. Typical failed specimen after creep or fatigue. 3306 G.N. Morscher et al. / Composites Science and Technology 68 (2008) 3305–3313

G N Morscher et al /Composites Science and Technology 68(2008 )3305-3313 surfaces (creep-formed cracks usually emanated from or to a sur face)and the average value was taken as the matrix crack density 165 MPa. 100h DF 192 MPa. 250h DF E=230 GPa 3. Results All of the creep and fatigue data are shown in Fig. 2 in terms of As-Produced maximum applied stress during the test versus exposure time, that E=259 GPa is, time for either test interruption or specimen rupture. Note that the open symbols indicate that the specimen did not fail and thus as esidual testing at room temperature. Df was only carried out to 250 h and there were no failures for this condi tion over the entire stress range. Creep-rupture occurred at the high stresses(220 MPa)for shorter times(<100 h)and for longer 50 times(up to 2000 h) at the lower stresses(110 and 165 MPa). Fa- tigue failure occurred for all of the 30 Hz HCf tests over a wide 020.30.40.50.60.7 stress range, 179-220 MPa, which indicates this was a more severe Strain. % condition for failure compared to the creep and dwell fatigue con- ditions. The 30 Hz HCF specimen tested for the peak stress condi b 220 MPa. 168h DF tion of 165 MPa exceeded the run-out condition of 10 cycles (42,000,000 cycles survived) and was tested at room temperature for determination of residual properties. For comparison, 0.33 Hz 165 MPa 100h DF fatigue data for the same type of composite system from Kalluri et al. [12] is plotted on Fig. 2 showing similar behavion 3. 1. Residual properties of specimens that did not fail 8 More than twenty five different specimens were tested at room 20.41 110 MPa, 100h DF temperature after being subject to tensile creep or fatigue, as well s two as-produced specimens for comparison. The stress-strain AS-Produced behavior and aE behavior is shown in Fig 3 for some representa- tive specimens. The 220 MPa DF specimens was the only specimen that survived the 220 MPa applied stress condition for any signifi 0 500 cant time. The degradation in ultimate strength after the 220 MPa. AE onset stresses Stress. MPa For the specimens that had experienced dwell fatigue or creep, Fig 3. Room temperature ously dwell fatigue specimens. Note that the stress-strain curve there was an increase in the stress above which non-linearity oc- in(a) are offset for clarity and the specimens were tested to failure. The modulus curs, i.e. the proportional limit stress. There was also an increase values refer to the initial linear(10-55 MPa) portion of the stress-strain curve in the stress at which significant high energy AE occurs ("AE Onset"in Fig. 3b). AE onset stress has been shown to be a good parameter for the onset of significant matrix cracking in this com- osite system[11]. Fig. 4 shows the increase in AE matrix cracking stress(Ae onset stress) with time-dependent strain for some of 280 the specimens. This can be explained by the increase in residual mpressive stress in the matri amount of residual compressive stress in the matrix can be deter- mined from the intersection of the top part of the hysteresis loop 220 6200 Creep specimens 5160 193 MPa (28ks 0.05 0.1 0.15 Fig 4. Room temperature AE onset stress versus time dependent strain for DF and creep(where noted) tests. Data points encircled refer to creep data as noted. x:10 +0.33 Hz(tailed)-Ref 12 with the original loading curve(Fig. 3a)[13]. For as-produced specimens, the residual compressive stress is about 50 MPa 0.1 1000 was observed in Ref [11. However, after tensile creep, the matrix residual compressive stress increases to over 100 MPa. Similar Fig. 2. Creep and fatigue data versus exposure time at 1204 C for the different behavior was reported in Ref. [7] which was attributed to matrix levated temperature tests. Open symbols indicate test specimens that did not fail. relaxation during creep. Upon unloading the fibers put the matrix

surfaces (creep-formed cracks usually emanated from or to a sur￾face) and the average value was taken as the matrix crack density. 3. Results All of the creep and fatigue data are shown in Fig. 2 in terms of maximum applied stress during the test versus exposure time, that is, time for either test interruption or specimen rupture. Note that the open symbols indicate that the specimen did not fail and thus was removed for residual testing at room temperature. DF was only carried out to 250 h and there were no failures for this condi￾tion over the entire stress range. Creep-rupture occurred at the high stresses (220 MPa) for shorter times (6100 h) and for longer times (up to 2000 h) at the lower stresses (110 and 165 MPa). Fa￾tigue failure occurred for all of the 30 Hz HCF tests over a wide stress range, 179–220 MPa, which indicates this was a more severe condition for failure compared to the creep and dwell fatigue con￾ditions. The 30 Hz HCF specimen tested for the peak stress condi￾tion of 165 MPa exceeded the run-out condition of 107 cycles (42,000,000 cycles survived) and was tested at room temperature for determination of residual properties. For comparison, 0.33 Hz fatigue data for the same type of composite system from Kalluri et al. [12] is plotted on Fig. 2 showing similar behavior. 3.1. Residual properties of specimens that did not fail More than twenty five different specimens were tested at room temperature after being subject to tensile creep or fatigue, as well as two as-produced specimens for comparison. The stress-strain behavior and AE behavior is shown in Fig. 3 for some representa￾tive specimens. The 220 MPa DF specimens was the only specimen that survived the 220 MPa applied stress condition for any signifi- cant time. The degradation in ultimate strength after the 220 MPa, 168 h, DF condition is evident. For the specimens that had experienced dwell fatigue or creep, there was an increase in the stress above which non-linearity oc￾curs, i.e., the proportional limit stress. There was also an increase in the stress at which significant high energy AE occurs (‘‘AE Onset” in Fig. 3b). AE onset stress has been shown to be a good parameter for the onset of significant matrix cracking in this com￾posite system [11]. Fig. 4 shows the increase in AE matrix cracking stress (AE onset stress) with time-dependent strain for some of the specimens. This can be explained by the increase in residual compressive stress in the matrix with creep or fatigue. The amount of residual compressive stress in the matrix can be deter￾mined from the intersection of the top part of the hysteresis loop with the original loading curve (Fig. 3a) [13]. For as-produced specimens, the residual compressive stress is about 50 MPa as was observed in Ref. [11]. However, after tensile creep, the matrix residual compressive stress increases to over 100 MPa. Similar behavior was reported in Ref. [7] which was attributed to matrix relaxation during creep. Upon unloading, the fibers put the matrix 0 50 100 150 200 250 300 0.1 1 10 100 1000 10000 Exposure Time (hrs) Max Applied Stress (MPa) Dwell Fatigue 1204ºC (did not fail) Creep 1204ºC (did not fail) Creep 1204ºC Failure 30 Hz (failed) 30 Hz (did not fail) 1 Hz (failed) 0.33 Hz (failed) - Ref 12 Fig. 2. Creep and fatigue data versus exposure time at 1204 C for the different elevated temperature tests. Open symbols indicate test specimens that did not fail. 0 50 100 150 200 250 300 350 400 450 500 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Strain, % Stress, MPa residual compressive stress As-Produced E = 259 GPa 110 MPa, 100h DF E = 250 GPa 165 MPa, 100h DF E = 230 GPa 192 MPa, 250h DF E = 230 GPa 220 MPa, 168h DF E = 221 GPa 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 100 200 300 400 500 Stress, MPa Norm Cum AE As-Produced 110 MPa, 100h DF 165 MPa, 100h DF 192 MPa, 250h DF 220 MPa, 168h DF AE onset stresses a b Fig. 3. Room temperature stress strain behavior (a) and AE behavior (b) for an as￾produced and previously dwell fatigue specimens. Note that the stress-strain curves in (a) are offset for clarity and the specimens were tested to failure. The modulus values refer to the initial linear (10–55 MPa) portion of the stress–strain curve. 100 120 140 160 180 200 220 240 260 280 300 0 Time-dependent DF or Creep Strain, % AE Onset Stress, MPa 193 MPa (28ksi) 165 MPa (24ksi) 110 MPa (16ksi) As-Produced creep specimens 0.05 0.1 0.15 0.2 Fig. 4. Room temperature AE onset stress versus time dependent strain for DF and creep (where noted) tests. Data points encircled refer to creep data as noted. G.N. Morscher et al. / Composites Science and Technology 68 (2008) 3305–3313 3307

G.N. Morscher et al/ Composites Science and Technology 68 (2008)3305-3313 in greater compression which must be overcome in order to pro- uce and propagate matrix cracks. It should be noted that oxida tion within surface cracks could also place the composite in esidual compression. However, multiple creep experiments have err performed on this same composite system at 1204C in in- 台 environment which showed the same magnitude of residual stress(150 MPa) for tensile creep at 220 MPa after hundreds of hours[14. DF, HCF or Creep Stress: Fig 5 shows the effect of residual compressive stress on AE set stress. There is nearly a direct relationship between the two. It 200x220MPa(32ks) 193 MPa(28ksi should be noted that for the higher stress conditions, 192 and in 2 100 4110 MPa(16ksi) 日165MPa(24ksi particular 220 MPa, significant matrix cracking had occurred the specimens tested at room temperature as will be discussed be- ●As- Produced low. Therefore the ae onset stresses for 192 and 220 MPa mask the fact that there already were some large matrix cracks formed dur ing the elevated temperature stressed-oxidation condition How- Time of test, hr ever, for specimens exposed to 165 MPa and lower stresses at elevated temperature, matrix cracking was minor(see below) Fig. 6. Ultimate tensile strength(UTS)at and the increase in aE onset stress corresponds to the stresses re- fter creep or fatigue specimens. Straight line indicates average of 193 MPa-tested pecimens indicating a loss in UTS at stresses above the onset stress for matrix quired to form and propagate initial matrix cracks of significant cracking. size, i.e., the matrix cracking stress increased with low stress creep he ultimate tensile strength properties are shown in Fig. 6 for he different specimens. Note that for applied creep/fatigue stres- and polished (usually about 1 to 1.5 mm from the edge surtace). ses of 165 MPa or lower, there is very little if any degradation in A few specimens were aligned along the face of the specimen. ultimate strength, even for specimens subjected to nearly 250h. For specimens polished along the length edge, no cracks were There appears to be a 10% drop in strength of specimens tested observed for the specimens subjected to 110 MPa For 165 MPa ap- at 193 MPa up to 250 h (straight line in Fig. 6)and about a 20% plied stresses and above, surface 90 minicomposite micro cracks drop for specimens tested at 220 MPa for 250 h. Specimens that (Fig. 7a)and inner back-to-back 90 minicomposite cracks were radation in room temperature ultimate strength. The room tem- For 193 MPa applied stresses and above, cracks which penetrated perature strength of the specimen that survived 1239 h at up to two plies in from the face were common( Fig 8).With 110 MPa was degraded about 20% compared to as-produced mate increasing time and stress, unbridged cracks were observed along rial. Whereas the room temperature strength of the two specimens (Fig. 7b) However, even for the specimens subjected to the highest the length for 165 MP that survived greater than 2000 h of creep at 110 and 165 graded almost 50% of the as-produced composite stre stresses(220 MPa), these surface cracks containing unbridged por- should also be noted that the retained room-temperature tions typically extended only two plies and sometimes three into odulus was only reduced slightly, as much as 10% after the cree the thickness of the specimen. The presence of unbridged regions or fatigue exposure [10] of cracks filled with a glass indicates that the fibers had failed at some significant period of time prior to ultimate failure of the 3. 2. Optical microscopy along the length composite In order to quantify matrix cracking, the crack densi age secte o f the tntey the extent of damage development in the可p,ap贴 acks were recorded tion).Most specimens were aligned so that the edge was ground cracks increase. However, for most cracks, even for specimens tested at 220 MPa, very few appear to propagate through-the- thickness of the specimen. When polished along the face of the specimen so that cracks can be observed along the 8.2 mm width, atrix cracks rarely appear to traverse the width, even for the 220 MPa stress-condition experiments, although they always ema- △110MPa(16ks 250-165MPa(24ksi) nate either from or to an edge(corner) ·192MPa(28ks) At room temperature, matrix cracks do appear to go through the 230x220MPa cross-section of the composite at the higher stre ≥200MP [11 In Ref [11. it was shown that the normalized cumulative AE energy was nearly directly proportional to measured matrix crack density when plotted versus stress. The matrix crack density versus stress measured in this study for elevated temperature tested specimens is compared to the room temperature data in Fig. 10. Note that at the lower stresses, low energy AE events pre 170 dominate and are caused by tunnel cracks which propagate An unbridged crack is defined as a crack which had originally propagate Residual Compressive Stress, MPa temperature/stress, some of those fibers in the crack wake, typically near the Fig. 5. Matrix cracking stress as determined from AE versus the residual compres- or edge of the specimen fail leaving a formerly bridged portion of the matrix crack sive stress in the matrix. Most of the data is for DF tests except where noted

in greater compression which must be overcome in order to pro￾duce and propagate matrix cracks. It should be noted that oxida￾tion within surface cracks could also place the composite in residual compression. However, multiple creep experiments have been performed on this same composite system at 1204 C in in￾ert environment which showed the same magnitude of residual stress (150 MPa) for tensile creep at 220 MPa after hundreds of hours [14]. Fig. 5 shows the effect of residual compressive stress on AE on￾set stress. There is nearly a direct relationship between the two. It should be noted that for the higher stress conditions, 192 and in particular 220 MPa, significant matrix cracking had occurred in the specimens tested at room temperature as will be discussed be￾low. Therefore, the AE onset stresses for 192 and 220 MPa mask the fact that there already were some large matrix cracks formed dur￾ing the elevated temperature stressed-oxidation condition. How￾ever, for specimens exposed to 165 MPa and lower stresses at elevated temperature, matrix cracking was minor (see below) and the increase in AE onset stress corresponds to the stresses re￾quired to form and propagate initial matrix cracks of significant size, i.e., the matrix cracking stress increased with low stress creep. The ultimate tensile strength properties are shown in Fig. 6 for the different specimens. Note that for applied creep/fatigue stres￾ses of 165 MPa or lower, there is very little if any degradation in ultimate strength, even for specimens subjected to nearly 250 h. There appears to be a 10% drop in strength of specimens tested at 193 MPa up to 250 h (straight line in Fig. 6) and about a 20% drop for specimens tested at 220 MPa for 250 h. Specimens that survived greater than 1000 h at lower stresses showed greater deg￾radation in room temperature ultimate strength. The room tem￾perature strength of the specimen that survived 1239 h at 110 MPa was degraded about 20% compared to as-produced mate￾rial. Whereas the room temperature strength of the two specimens that survived greater than 2000 h of creep at 110 and 165 MPa de￾graded almost 50% of the as-produced composite strength. It should also be noted that the retained room-temperature elastic modulus was only reduced slightly, as much as 10% after the creep or fatigue exposure [10]. 3.2. Optical microscopy along the length In order to quantify the extent of damage development in the gage section of the specimens subjected to creep and fatigue, over a dozen specimens were polished along the length (loading direc￾tion). Most specimens were aligned so that the edge was ground and polished (usually about 1 to 1.5 mm from the edge surface). A few specimens were aligned along the face of the specimen. For specimens polished along the length edge, no cracks were observed for the specimens subjected to 110 MPa. For 165 MPa ap￾plied stresses and above, surface 90 minicomposite micro cracks (Fig. 7a) and inner back-to-back 90 minicomposite cracks were evident for stress-temperature conditions up to 250 h (Fig. 7b). For 193 MPa applied stresses and above, cracks which penetrated up to two plies in from the face were common (Fig. 8). With increasing time and stress, unbridged cracks1 were observed along the length for 165 MPa and greater applied stress conditions (Fig. 7b). However, even for the specimens subjected to the highest stresses (220 MPa), these surface cracks containing unbridged por￾tions typically extended only two plies and sometimes three into the thickness of the specimen. The presence of unbridged regions of cracks filled with a glass indicates that the fibers had failed at some significant period of time prior to ultimate failure of the composite. In order to quantify matrix cracking, the crack density, number of cracks per mm, and the depth of cracks were recorded and are plotted in Fig. 9. It is evident that with increasing stress and time that the number of cracks increases and the depth of the matrix cracks increase. However, for most cracks, even for specimens tested at 220 MPa, very few appear to propagate through-the￾thickness of the specimen. When polished along the face of the specimen so that cracks can be observed along the 8.2 mm width, matrix cracks rarely appear to traverse the width, even for the 220 MPa stress-condition experiments, although they always ema￾nate either from or to an edge (corner). At room temperature, matrix cracks do appear to go through the cross-section of the composite at the higher stresses (P200 MPa) [11]. In Ref. [11], it was shown that the normalized cumulative AE energy was nearly directly proportional to measured matrix crack density when plotted versus stress. The matrix crack density versus stress measured in this study for elevated temperature tested specimens is compared to the room temperature data in Fig. 10. Note that at the lower stresses, low energy AE events pre￾dominate and are caused by tunnel cracks which propagate 150 170 190 210 230 250 270 0 Residual Compressive Stress, MPa AE Onset Stress, MPa As-produced 110 MPa (16ksi) 165 MPa (24ksi) 192 MPa (28ksi) 220 MPa (32ksi) 30 Hz Creep 50 100 150 200 Fig. 5. Matrix cracking stress as determined from AE versus the residual compres￾sive stress in the matrix. Most of the data is for DF tests except where noted. 0 100 200 300 400 500 600 1 10 100 1000 10000 Time of test, hr RT Residual Strength, MPa 220 MPa (32ksi) 193 MPa (28ksi) 165 MPa (24ksi) 110 MPa (16ksi) As-Produced DF, HCF, or Creep Stress: Fig. 6. Ultimate tensile strength (UTS) at room temperature for as-produced and after creep or fatigue specimens. Straight line indicates average of 193 MPa-tested specimens indicating a loss in UTS at stresses above the onset stress for matrix cracking. 1 An unbridged crack is defined as a crack which had originally propagated some distance so that load-bearing fibers bridge the crack. However, after some time/ temperature/stress, some of those fibers in the crack wake, typically near the surface or edge of the specimen fail leaving a formerly bridged portion of the matrix crack unbridged. 3308 G.N. Morscher et al. / Composites Science and Technology 68 (2008) 3305–3313

G N Morscher et al /Composites Science and Technology 68(2008 )3305-3313 a 0.1mm 0.1mr Stress direction Fig. 7. Typical creep-formed cracks: (a) surface 90 minicomposite(110 MPa, 2036 h, did not fail in rupture)and (b)inner back-to-back 90 minicomposite cracks which extended to the surface through a 0minicomposite(165 MPa; 1508 h creep rupture). mm several unbridged cracks emanating from both surfaces 0.2mm g 8. Matrix cracks that extend at least two plies from the surface and in some cases have fractured fibers in the matrix crack wake. This specimen had undergone 220 MPa 30 Hz fatigue and lasted approximately 1.2 h at 1204 C. ■220MPa(32ks)HcF through the 90 minicomposites The formation and propagation of bridged-matrix cracks at room temperature relates to the high en- ▲193MPa(28ks)HcF 08■ ergy ae events. If the low energy event data are removed from the 193 MPa(28 ksi) DF ae and only the high energy event data are used, there is a very °165MPa(24ksi) Creep good correlation between the room temperature and elevated tem- perature stress-dependent matrix crack density with the exception that the elevated temperature matrix crack density was not through-the-cross-section. The elevated temperature matrix crack 193 MPa(28 ksi) Dwell Fatigue densities tend to fall below the room temperature distribution, i.e less cracking at high temperature compared to room temperature. mostly surface cracks, a few tw brid em vated temperature with time(Fig. 9). the crack density tends to in- 165 MPa(24 ksi) Creep crease towards the room temperature derived matrix crack density value with time. Therefore, the room temperature stress-depen- 1000 0000 dent matrix crack is a reasonable, at least conservative, representa- tion for modeling matrix crack density at elevated temperature as Fig 9. Crack density along the length of the specimen for different stress, time, and long as the depth of matrix cracks at elevated temperatures at a given stress is taken into consideration

through the 90 minicomposites. The formation and propagation of bridged-matrix cracks at room temperature relates to the high en￾ergy AE events. If the low energy event data are removed from the AE and only the high energy event data are used, there is a very good correlation between the room temperature and elevated tem￾perature stress-dependent matrix crack density with the exception that the elevated temperature matrix crack density was not through-the-cross-section. The elevated temperature matrix crack densities tend to fall below the room temperature distribution, i.e., less cracking at high temperature compared to room temperature. Even though some increase in crack density was observed at ele￾vated temperature with time (Fig. 9), the crack density tends to in￾crease towards the room temperature derived matrix crack density value with time. Therefore, the room temperature stress-depen￾dent matrix crack is a reasonable, at least conservative, representa￾tion for modeling matrix crack density at elevated temperature as long as the depth of matrix cracks at elevated temperatures at a given stress is taken into consideration. Fig. 7. Typical creep-formed cracks: (a) surface 90 minicomposite (110 MPa, 2036 h, did not fail in rupture) and (b) inner back-to-back 90 minicomposite cracks which extended to the surface through a 0 minicomposite (165 MPa; 1508 h creep rupture). Fig. 8. Matrix cracks that extend at least two plies from the surface and in some cases have fractured fibers in the matrix crack wake. This specimen had undergone 220 MPa 30 Hz fatigue and lasted approximately 1.2 h at 1204 C. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1 10 100 1000 10000 Time, hr Crack density, mm-1 220 MPa (32 ksi) HCF 220 MPa (32 ksi) DF 193 MPa (28 ksi) HCF 193 MPa (28 ksi) DF 165 MPa (24 ksi) Creep most cracks go through two plies or more, ~ 1/3 of the cracks are unbridged most cracks go through one to two plies, a few unbridged mostly surface cracks, a few through one ply 165 MPa (24 ksi) Creep 193 MPa (28 ksi) Dwell Fatigue two plies, some unbridged Fig. 9. Crack density along the length of the specimen for different stress, time, and loading conditions. G.N. Morscher et al. / Composites Science and Technology 68 (2008) 3305–3313 3309