《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_j.1151-2916.1999.tb02207.x[1]

ournal JAm.Ceam.Soc,82l1l3087-96(1999 Notched fracture behavior of an Oxide/Oxide Ceramic-Matrix Composite Victoria A. Kramb, .t Reji John, and Larry P Zawada aterials and Manufacturing Directorate, Air Force Research Laboratory (AFRL/MLLN), Wright-Patterson Air Force Base, Ohio 45433-7817 The fracture behavior of an oxide/oxide ceramic-matrix vicinity of a notch. John et al, 3 Cady et al, and Mall et al. 5 showed that NicalonTM-fiber-reinforced glass matrix compos- vestigated at 23 and 950 C using a single edge notched ites exhibit multiple matrix cracking around notches and holes specimen geometry with clamped ends. Crack growth and resulting in notch-insensitive behavior at room temperature damage progression were monitored during the tests using Fiber-dominated CMCs, including some C/C composites, show optical microscopy, ultrasonic C-scans, and crack mouth a nearly linear unnotched stress-strain response in the [0/901 opening displacement. The net section strength of orientation. However, increasing nonlinearity with increasing Nextel610/AS was less than the unnotched ultimate tensile notch length was observed for edge notched specimens ,9For strength. The failure mode was nonbrittle with consider- these composite systems, damage ahead of the notch was found able nonlinear deformation prior to and after the peak load to extend as a narrow zone normal to the notch plane. The at 23 and 950%C. The effect of temperature on the notched damage zone effectively redistributed the stresses away from strength was significant. Net section failure stress de he notch tip thus increasing fracture toughness at room tem- creased 50% when temperature was increased from 23to rature. Similar to other fiber-dominated systems, the oxide/ 950C. Observations of damage progression indicated that oxide CMC Nextel610/AS exhibits nearly linear unnotched he reduction in notch strength with temperature was as- stress-strain behavior. 2 However, the notched behavior of sociated with self-similar crack growth at 950%C. Ultrasonic Nextel610/AS has not been examined C-scans were found to be an effective method of monitoring For CMCs to be used effectively in aerospace applications damage progression. Ultrasonic attenuation ahead of the with sites of stress concentration, the decrease in fracture notch tip was correlated with surface matrix cracks and toughness at elevated temperatures must be minimized. For exposed fiber lengths on the fracture surface. most of the CMC systems studied thus far, the enhanced frac ture toughness relies on the existence of a weak fiber/matrix L. Introduction nterphase which allows fiber debonding and sliding during fracture. At elevated temperature, oxidation of the fiber/matrix C ERAMIC-MATRIX COMPOSITES(CMCs) consisting of an oxide nterphase resulted in embrittlement of these composites as the matrix and oxide fibers with no engineered fiber-matrix matrix became tightly bonded to the fiber 3,6 Consequently, the interphase are currently under consideration for high fracture toughness at elevated temperature was significantly temperature aerospace applications due to their inherent resis lower than that at room temperature. 3,Oxidation occurring at tance to oxidation. The CMc produced by General Electric must therefore be prevented to maintain fracture toughness in nsisting of alumina fibers and an these CMC systems. a different approach to enhanced CMC 87% alumina-13%silica matrix(Nextel610/AS)exhibits ex- fracture toughness utilizes a very weak, porous matrix bonded ellen room-temperature fatigue and tensile strength(170 and 205 MPa, respectively). 1, 2 When tested at 1000C, fatigue and to the fiber with no engineered interphase. Oxide-based CMCs nsile strength decreased by only 15% 1,2 Current component are inherently oxidation resistant at elevated temperature and designs. however. include stress concentration sites such as are expected to show damage tolerance when produced using a strong interface and weak matrix. 1 2, 10 The limited experimen- bolted attachment points and cooling holes. Local stresses in tal data available in the open literature on the notched fracture posite, thus acting as crack and damage initiation sites. Com. behavior of an oxide/oxide CMC are restricted to room tem- these regions often exceed the proportional limit of the co ponent design Nextel61O/AS composites will therefore perature. This paper discusses the results of a study of the require knowledge of the notched fracture behavior and dam- notched fracture behavior and damage progression of an oxide. age progression at room temperature and expected service composite(Nextel610/AS)at roor temperatures Fracture studies of other CMC systems-9 have shown that damage mechanisms such as multiple matrix cracking, fiber IL. Material bridging, and fiber pullout act to redistribute stresses in the The Nextel610/AS CMC used in this investigation was pro- duced by geae under the trade name Gen IV. The Nextel610 fibers, produced by the 3M Company, consisted of fine polycrystalline a-alumina. These fibers J.J. Petrovic--contributing editor were woven into an eight harness satin weave(8HSW)cloth The resulting cloth consisted of fibers primarily in the war direction on one side of the ply and fill fibers on the oth lies were warp aligned di g plles n s G0贴R阳时 tract io. a4 szu-9s. matched with fill fibers, warp fibers matched with warp fibers el warpage. The resulting panel contained primarily fill University of Dayton Research Institute, Dayton, Ohio 45469-0128 fibers on the exposed outer plies. The composite panel used in 3087

Notched Fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite Victoria A. Kramb,*,† Reji John,* and Larry P. Zawada* Materials and Manufacturing Directorate, Air Force Research Laboratory (AFRL/MLLN), Wright-Patterson Air Force Base, Ohio 45433-7817 The fracture behavior of an oxide/oxide ceramic-matrix composite, alumina/alumina–silica (Nextel610/AS), was in￾vestigated at 23° and 950°C using a single edge notched specimen geometry with clamped ends. Crack growth and damage progression were monitored during the tests using optical microscopy, ultrasonic C-scans, and crack mouth opening displacement. The net section strength of Nextel610/AS was less than the unnotched ultimate tensile strength. The failure mode was nonbrittle with consider￾able nonlinear deformation prior to and after the peak load at 23° and 950°C. The effect of temperature on the notched strength was significant. Net section failure stress de￾creased 50% when temperature was increased from 23° to 950°C. Observations of damage progression indicated that the reduction in notch strength with temperature was as￾sociated with self-similar crack growth at 950°C. Ultrasonic C-scans were found to be an effective method of monitoring damage progression. Ultrasonic attenuation ahead of the notch tip was correlated with surface matrix cracks and exposed fiber lengths on the fracture surface. I. Introduction CERAMIC-MATRIX COMPOSITES (CMCs) consisting of an oxide matrix and oxide fibers with no engineered fiber-matrix interphase are currently under consideration for high￾temperature aerospace applications due to their inherent resis￾tance to oxidation. The CMC produced by General Electric Aircraft Engines (GEAE) consisting of alumina fibers and an 87% alumina–13% silica matrix (Nextel610/AS) exhibits ex￾cellent room-temperature fatigue and tensile strength (170 and 205 MPa, respectively).1,2 When tested at 1000°C, fatigue and tensile strength decreased by only 15%.1,2 Current component designs, however, include stress concentration sites such as bolted attachment points and cooling holes. Local stresses in these regions often exceed the proportional limit of the com￾posite, thus acting as crack and damage initiation sites. Com￾ponent design using Nextel610/AS composites will therefore require knowledge of the notched fracture behavior and dam￾age progression at room temperature and expected service temperatures. Fracture studies of other CMC systems3–9 have shown that damage mechanisms such as multiple matrix cracking, fiber bridging, and fiber pullout act to redistribute stresses in the vicinity of a notch. John et al., 3 Cady et al., 4 and Mall et al.5 showed that Nicalon™-fiber-reinforced glass matrix compos￾ites exhibit multiple matrix cracking around notches and holes resulting in notch-insensitive behavior at room temperature. Fiber-dominated CMCs, including some C/C composites, show a nearly linear unnotched stress–strain response in the [0/90] orientation. However, increasing nonlinearity with increasing notch length was observed for edge notched specimens.8,9 For these composite systems, damage ahead of the notch was found to extend as a narrow zone normal to the notch plane. The damage zone effectively redistributed the stresses away from the notch tip, thus increasing fracture toughness at room tem￾perature. Similar to other fiber-dominated systems, the oxide/ oxide CMC Nextel610/AS exhibits nearly linear unnotched stress–strain behavior.1,2 However, the notched behavior of Nextel610/AS has not been examined. For CMCs to be used effectively in aerospace applications with sites of stress concentration, the decrease in fracture toughness at elevated temperatures must be minimized. For most of the CMC systems studied thus far, the enhanced frac￾ture toughness relies on the existence of a weak fiber/matrix interphase which allows fiber debonding and sliding during fracture. At elevated temperature, oxidation of the fiber/matrix interphase resulted in embrittlement of these composites as the matrix became tightly bonded to the fiber.3,6 Consequently, the fracture toughness at elevated temperature was significantly lower than that at room temperature.3,7 Oxidation occurring at the fiber/matrix interface and degradation of the fiber itself must therefore be prevented to maintain fracture toughness in these CMC systems. A different approach to enhanced CMC fracture toughness utilizes a very weak, porous matrix bonded to the fiber with no engineered interphase. Oxide-based CMCs are inherently oxidation resistant at elevated temperature and are expected to show damage tolerance when produced using a strong interface and weak matrix.1,2,10 The limited experimen￾tal data available in the open literature on the notched fracture behavior of an oxide/oxide CMC are restricted to room tem￾perature.11 This paper discusses the results of a study of the notched fracture behavior and damage progression of an oxide/ oxide composite (Nextel610/AS) at room temperature and 950°C. II. Material The Nextel610/AS CMC used in this investigation was pro￾duced by GEAE under the trade name Gen IV. The Nextel610 fibers, produced by the 3M Company,10 consisted of fine￾grained (<0.5 mm) polycrystalline a-alumina. These fibers were woven into an eight harness satin weave (8HSW) cloth. The resulting cloth consisted of fibers primarily in the warp direction on one side of the ply and fill fibers on the other. The plies were warp aligned during lay-up, but alternating plies were rotated about the warp direction so that fill fibers were matched with fill fibers, warp fibers matched with warp fibers. This lay-up provided better nesting of the piles and minimized panel warpage. The resulting panel contained primarily fill fibers on the exposed outer plies. The composite panel used in J. J. Petrovic—contributing editor Manuscript No. 190148. Received May 29, 1998; approved June 1, 1999. Support for V. A. Kramb was provided by the Dayton Area Graduate Studies Institute (DAGSI) and by the AFOSR/AASERT Program (Contract No. F49620-95- 1-0500). Support for L. P. Zawada was provided by Dr. W. S. Coblenz at DARPA under Contract No. in-house and Order No. A565. *Member, American Ceramic Society. † University of Dayton Research Institute, Dayton, Ohio 45469-0128. J. Am. Ceram. Soc., 82 [11] 3087–96 (1999) Journal 3087

3088 Journal of the American Ceramic Sociery-Kramb et al Vol. 82. No. 11 this study contained 12 plies. Sections of the Nextel610 cloth δ↑ were prepregged with alumina powder and a silica-forming lymer before stacking. No coating was applied to the fi before prepregging. The laminate was then warm molded to produce the green state ceramic tile. Sintering the green tile in air at 1000oC removed the organic binders and produced a y porous alumina-silica matrix. The resulting composite con tained sintered matrix which was bonded to the fibers with no CMOD red hase. The matrix con- tained approximately 87 wt% alumina and 13 wt% silica. Im age analysis of polished cross sections( Fig. 1)showed that the fiber volume fraction was 33%. Extensive microcracking present throughout the matrix was a result of the shrinkage which occurred during the pyrolysis processing. These micre racks were distributed throughout the interior matrix as well as on the specimen surface δ↓ Fig. 2. Schematic of single edge notched specimen with clamped IlL. Experimental Procedure ends, MSE(). W= 19.0 and 25.4 mm, B= 2.9 mm, and H/W=4 Test Procedure Unnotched tension tests were conducted using an automated Decimen notch. Slotted center of the quartz servo-controlled, hydraulic, horizontal test system. 3, 4 The lamps allowed for visual inspecti rack growth from the tests were conducted under load-line displacement control at a notch tip during testing ual inspection of crack the test equip rate(8)of 0.05 mm/s Specimen endpoint displacement was have been described elsewhere 13, 14 increased monotonically to specimen failure. Tension tests arious methods of thermocouple attachment were investi- were conducted at 23 and 1000C. Further discussion of the gated. The method which produced the most repeatable tem- unnotched tension test procedures and results are described perature measurements with minimal damage to the specimen use in tester Notched fracture tests were conducted using an automated S-type, beaded thermocouples were first wire tied to the speci- servo-controlled, hydraulic, horizontal test system. 3, 14The men with the thermocouple bead in contact with the specimen single edge notched surface. A small drop of CeramabondM(Ceramabond 503 MSE(T)'S shown in Fig. 2, was used for all of the tests. The Aremco Products, Inc, Ossining, NY) ceramic adhesive was constrained end conditions. The overall dimensions of the bead surface. The adhesive was cured for 30 min at 300%C specimens were as follows: width(W)=190 or 25.4 mm resulting in a hard, adhesive, insulating cover for the thermo- thickness(B)= 2.9 mm, and height-to-width ratio(H/H couple bead while assuring contact with the specimen surface 4.0. The notches were cut using a diamond saw to produce an during testing. The adhesive was white in color, matching the initial notch height of 0. 4 mm, and length(ao)equal to 0. 2W for color of the Nextel610/AS specimen, which eliminated tem- all specimens. Crack mouth opening displacement(CMOD was measured continuously using a high-resolution knife edge perature errors due to differential radiant heat absorption. The extensometer at room temperature and a high-temperature ex- surface after testing of matrix crack extension were made on both sides of the specimen lengtap of the temperature distribution across the tensometer with alumina rods at 950.. Optical measurements th, and thickness was obtained using a specimen during testing notched Nextel6l0/AS specimen with W= 25.4 mm and a (2) Elevated-Temperature Testing 0. 2W. The temperature variation was <+2.9% across the gage length, and <+1.0% through the thickness. during testing, six Heating of the test specimen was achieved with closed-loop controlled, four-zone quartz lamps. The heated section of the surfaces. Four thermocouples controlled the quartz lamp output specimen was approximately 3 in. in length centered on the nd three monitored temperature near the notch plane (3) Fracture Test Parameters Fracture tests were conducted under load-line displacement control at a rate of 0.001 and 0.01 mm/s. During the tests conducted at 0.001 mm/s, load-line displacement was periodi- cally held constant for crack length measurements. During this hold time, the specimen was also unloaded and reloaded to determine residual CMOD and changes in specimen compli ance. The fracture test conducted at 0.01 mm/s w monotonically(without unloading loops or hold specimen failure. Applied load, CMOD, and load-line ment(8) were recorded continuously as a function during all tests. Whenever possible, specimens were unloaded and removed from the test machine after attaining the peak load for ultrasonic and optical evaluation. Mode I stress intensity factor (K,), and elastic modulus(E), from specimen compli- ance(=CMOD/P), were calculated using isotropic expres- sions given by Eq(1). Assuming crack length(a)equal to 200um he saw-cut notch length, elastic modulus was calculated from the initial loading compliance below a far-field applied stress Fig.1.Nextel610/AS composite polished section optical (oa)=15 MPa. Applied stress was calculated as a P/(BW) where P is the applied load, and B and w are shown in Fig. 2

this study contained 12 plies. Sections of the Nextel610 cloth were prepregged with alumina powder and a silica-forming polymer before stacking. No coating was applied to the fibers before prepregging. The laminate was then warm molded to produce the green state ceramic tile. Sintering the green tile in air at 1000°C removed the organic binders and produced a porous alumina–silica matrix. The resulting composite con￾tained sintered matrix which was bonded to the fibers with no naturally occurring or engineered interphase. The matrix con￾tained approximately 87 wt% alumina and 13 wt% silica. Im￾age analysis of polished cross sections (Fig. 1) showed that the fiber volume fraction was 33%. Extensive microcracking present throughout the matrix was a result of the shrinkage which occurred during the pyrolysis processing. These micro￾cracks were distributed throughout the interior matrix as well as on the specimen surface. III. Experimental Procedure (1) Test Procedure Unnotched tension tests were conducted using an automated, servo-controlled, hydraulic, horizontal test system.13,14 The tests were conducted under load-line displacement control at a rate (d . ) of 0.05 mm/s. Specimen endpoint displacement was increased monotonically to specimen failure. Tension tests were conducted at 23° and 1000°C. Further discussion of the unnotched tension test procedures and results are described elsewhere.2 Notched fracture tests were conducted using an automated, servo-controlled, hydraulic, horizontal test system.13,14 The single edge notched specimen geometry with clamped ends, MSE(T)15 shown in Fig. 2, was used for all of the tests. The specimen ends were rigidly clamped, resulting in rotationally constrained end conditions. The overall dimensions of the specimens were as follows: width (W) 4 19.0 or 25.4 mm, thickness (B) 4 2.9 mm, and height-to-width ratio (H/W) 4 4.0. The notches were cut using a diamond saw to produce an initial notch height of 0.4 mm, and length (a0) equal to 0.2W for all specimens. Crack mouth opening displacement (CMOD) was measured continuously using a high-resolution knife edge extensometer at room temperature and a high-temperature ex￾tensometer with alumina rods at 950°C. Optical measurements of matrix crack extension were made on both sides of the specimen during testing. (2) Elevated-Temperature Testing Heating of the test specimen was achieved with closed-loop controlled, four-zone quartz lamps. The heated section of the specimen was approximately 3 in. in length centered on the specimen notch. Slotted windows in the center of the quartz lamps allowed for visual inspection of crack growth from the notch tip during testing. Further details of the test equipment have been described elsewhere.13,14 Various methods of thermocouple attachment were investi￾gated. The method which produced the most repeatable tem￾perature measurements with minimal damage to the specimen surface was chosen for use in testing. Platinum–10% rhodium, S-type, beaded thermocouples were first wire tied to the speci￾men with the thermocouple bead in contact with the specimen surface. A small drop of Ceramabond™ (Ceramabond 503, Aremco Products, Inc., Ossining, NY) ceramic adhesive was then placed over the thermocouple bead and spread to cover the bead surface. The adhesive was cured for 30 min at 300°C, resulting in a hard, adhesive, insulating cover for the thermo￾couple bead while assuring contact with the specimen surface during testing. The adhesive was white in color, matching the color of the Nextel610/AS specimen, which eliminated tem￾perature errors due to differential radiant heat absorption. The Ceramabond™ was also easily removed from the specimen surface after testing. A thermal map of the temperature distribution across the specimen length, width, and thickness was obtained using a notched Nextel610/AS specimen with W 4 25.4 mm and a0 4 0.2W. The temperature variation was #±2.9% across the gage length, and #±1.0% through the thickness. During testing, six thermocouples were attached to the top and one to the bottom surfaces. Four thermocouples controlled the quartz lamp output and three monitored temperature near the notch plane. (3) Fracture Test Parameters Fracture tests were conducted under load-line displacement control at a rate of 0.001 and 0.01 mm/s. During the tests conducted at 0.001 mm/s, load-line displacement was periodi￾cally held constant for crack length measurements. During this hold time, the specimen was also unloaded and reloaded to determine residual CMOD and changes in specimen compli￾ance. The fracture test conducted at 0.01 mm/s was loaded monotonically (without unloading loops or hold times) to specimen failure. Applied load, CMOD, and load-line displace￾ment (d) were recorded continuously as a function of time during all tests. Whenever possible, specimens were unloaded and removed from the test machine after attaining the peak load for ultrasonic and optical evaluation. Mode I stress intensity factor (KI ), and elastic modulus (E), from specimen compli￾ance (4CMOD/P), were calculated using isotropic expres￾sions15 given by Eq. (1). Assuming crack length (a) equal to the saw-cut notch length, elastic modulus was calculated from the initial loading compliance below a far-field applied stress (sa) 4 15 MPa. Applied stress was calculated as sa 4 P/(BW) where P is the applied load, and B and W are shown in Fig. 2. Fig. 1. Nextel610/AS composite polished section optical micrograph. Fig. 2. Schematic of single edge notched specimen with clamped ends, MSE(T). W 4 19.0 and 25.4 mm, B 4 2.9 mm, and H/W 4 4. 3088 Journal of the American Ceramic Society—Kramb et al. Vol. 82, No. 11

November 1999 Notched fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite K=vAf(a/w) 至 where F(alW)and G(a/w)are given in Ref. 15 (4) Damage Characterization Ultrasonic C-scans of the test specimens were recorded be- ore testing and during interrupted tests. The C-scans were 950c obtained using a through transmission, reflector plate scanning technique described elsewhere. 16, 17 Each ultrasonic C-scan was librated, and referenced to a calibrated ultrasonic attenuation 0060,080.109.120.14 cale. During the C-scan, the ultrasonic signal passing through the specimen was acquired at regularly spaced x-y locations CMOD(mm) digitized and color coded. A 12.7 mm diameter, 10 MHz, 76 mm focused transducer produced by KB Aerotech was used to tg 4: d versus or tkm mperature, loading ra nloading loops emit and receive the ultrasonic energy. The C-scan technique line displacement controlled fracture tests, W esponse for load required immersion of the test specimen in water during the scan. After each C-scan, excess moisture was removed during I h bakeout at 70 C. Zawada and Lee2,1& showed that water 004 exposure did not result in a change in the mechanical behavior of nextel6l0/as Optical inspection of the notch tip region was performed on both sides of the specimen during testing. Because of the ex tensive preexisting matrix cracking on the specimen surface, the notch tip region appeared cracked before testing. Crack growth from the notch tip was identified as a clearly observable 950c opening of a preexisting surface crack(identified as the dom- inant crack), or the appearance of a new crack at the notch ti 0.01 23°c was performed using optical and scanning electron microscopy 000包…2--…“ (SEM). Specimens were sputter coated with gold/palladium 000.020040.0e0.080.10 12 before SEM imaging Backscatter electron imaging was used to minimize charging effects and highlight microcracks. Destruc tive evaluation of interrupted test specimens was performed by Fig. 5. Effect of temperature on residual CMOD after unld sectioning and polishing using a diamond impregnated lapping om a maximum load with corresponding CMODmax CMOD film. Polishing the specimens with light pressure removed sur the peak load is also indicated on the plc face layers slowly enough to prevent damage to fibers within the underlying layers compliance ranged from 69 to 71 GPa for all fracture tests IV. Results and Discussion Table I). No dependence of elastic modulus on temperature was observed The load-cMoD room-temperature () Notched fracture Behavior tests conducted on 19.0 and 25. 4 mm wide specimens is The experimental results of the notched fracture test in Fig. 3. Unloading the imen from applied loads own in Figs. 3-5 and summarized in Table 1. Each experi- the initial linear region resulted in no residual CMOD mental condition listed in the table represents a single test hysteresis. Beyond the linear region, loading behavior was in- result. For each fracture test, a low-load(<1.0 kN) check-out creasingly nonlinear and associated with increasing residual cycle was used to determine the initial loading compliance CMOD(CMODres)and measurable hysteresis in the unloading Elastic modulus calculated using Eq. (1)and the initial loading The onset of nonlinear loading behavior was determined by fitting a linear curve to the initial loaddisplacement data. Mea sured load deviation from the linear fit exceeding 5%was =25.4mm defined as the onset of nonlinear loading behavior. Fracture tests conducted at room temperature exhibited linear loading behavior up to an applied stress intensity factor (Ka)=6 MPam, corresponding to an applied load, P=2. 8 kN for the 25.4 mm wide specimen. Unloading the specimen from Ka w=190 mm ak= 16.8 MPavm levels above 6 MPa. m/ resulted in measurable CMOD and hysteresis in the load-CMOD loops. Residual CMOD and un loading loop hysteresis increased with applied load up to the peak load. After reaching the peak load, CMOD continued to increase under increasing load-line displacement, while the ap- plied load remained nearly constant( Fig. 3) The stress intensity factor(Kpeak) based on the peak load and initial notch length was calculated to be 16.8 and 15.5 MPam for the 19.0 and 25.4 mm wide specimens, respec- tively. The 19.0 mm wide specimen failed abruptly after reach- Load versus crack mouth opening displacement for load-lin ing a CMOD cement controlled fracture tests at room temperature, w= 19.0 reached the peak load when CMOD =0.08 mm and was removed from the test frame when CMod = 0.10 mm for

KI = sa=paF~a/W ! (1a) E = 2saa CMODG~a/W ! (1b) where F(a/W) and G(a/W) are given in Ref. 15. (4) Damage Characterization Ultrasonic C-scans of the test specimens were recorded be￾fore testing and during interrupted tests. The C-scans were obtained using a through transmission, reflector plate scanning technique described elsewhere.16,17 Each ultrasonic C-scan was calibrated, and referenced to a calibrated ultrasonic attenuation scale. During the C-scan, the ultrasonic signal passing through the specimen was acquired at regularly spaced x–y locations, digitized, and color coded. A 12.7 mm diameter, 10 MHz, 76 mm focused transducer produced by KB Aerotech was used to emit and receive the ultrasonic energy. The C-scan technique required immersion of the test specimen in water during the scan. After each C-scan, excess moisture was removed during a 1 h bakeout at 70°C. Zawada and Lee2,18 showed that water exposure did not result in a change in the mechanical behavior of Nextel610/AS. Optical inspection of the notch tip region was performed on both sides of the specimen during testing. Because of the ex￾tensive preexisting matrix cracking on the specimen surface, the notch tip region appeared cracked before testing. Crack growth from the notch tip was identified as a clearly observable opening of a preexisting surface crack (identified as the dom￾inant crack), or the appearance of a new crack at the notch tip. Post-test inspection of the crack growth and fracture surfaces was performed using optical and scanning electron microscopy (SEM). Specimens were sputter coated with gold/palladium before SEM imaging. Backscatter electron imaging was used to minimize charging effects and highlight microcracks. Destruc￾tive evaluation of interrupted test specimens was performed by sectioning and polishing using a diamond impregnated lapping film. Polishing the specimens with light pressure removed sur￾face layers slowly enough to prevent damage to fibers within the underlying layers. IV. Results and Discussion (1) Notched Fracture Behavior The experimental results of the notched fracture tests are shown in Figs. 3–5 and summarized in Table I. Each experi￾mental condition listed in the table represents a single test result. For each fracture test, a low-load (<1.0 kN) check-out cycle was used to determine the initial loading compliance. Elastic modulus calculated using Eq. (1) and the initial loading compliance ranged from 69 to 71 GPa for all fracture tests (Table I). No dependence of elastic modulus on temperature was observed. The load–CMOD response for room-temperature fracture tests conducted on 19.0 and 25.4 mm wide specimens is shown in Fig. 3. Unloading the specimen from applied loads within the initial linear region resulted in no residual CMOD or loop hysteresis. Beyond the linear region, loading behavior was in￾creasingly nonlinear and associated with increasing residual CMOD (CMODres) and measurable hysteresis in the unloading loops. The onset of nonlinear loading behavior was determined by fitting a linear curve to the initial load/displacement data. Mea￾sured load deviation from the linear fit exceeding 5% was defined as the onset of nonlinear loading behavior. Fracture tests conducted at room temperature exhibited linear loading behavior up to an applied stress intensity factor (Ka) 4 6 MPa?m1/2, corresponding to an applied load, P 4 2.8 kN for the 25.4 mm wide specimen. Unloading the specimen from Ka levels above 6 MPa?m1/2 resulted in measurable CMODres and hysteresis in the load–CMOD loops. Residual CMOD and un￾loading loop hysteresis increased with applied load up to the peak load. After reaching the peak load, CMOD continued to increase under increasing load-line displacement, while the ap￾plied load remained nearly constant (Fig. 3). The stress intensity factor (Kpeak) based on the peak load and initial notch length was calculated to be 16.8 and 15.5 MPa?m1/2 for the 19.0 and 25.4 mm wide specimens, respec￾tively. The 19.0 mm wide specimen failed abruptly after reach￾ing a CMOD 4 0.10 mm. The 25.4 mm wide specimen reached the peak load when CMOD 4 0.08 mm and was removed from the test frame when CMOD 4 0.10 mm for Fig. 3. Load versus crack mouth opening displacement for load-line displacement controlled fracture tests at room temperature, W 4 19.0 and 25.4 mm. Fig. 4. Effect of temperature, loading rate, and unloading loops on the load versus crack mouth opening displacement response for load￾line displacement controlled fracture tests, W 4 25.4 mm. Fig. 5. Effect of temperature on residual CMOD after unloading from a maximum load with corresponding CMODmax. CMODmax at the peak load is also indicated on the plot. November 1999 Notched Fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3089

Journal of the American Ceramic Socieny-Kramb et al. Vol. 82. No. 11 Table I. Notched and Unnotched Test Results Unnotched emperature net secton stress MPa Unnotched 4 13.6 Notched l51 16.8 25.4 5.4 62 7.9 Reported values are an average from three tests conducted at 23C and two tests at 1000oC. ultrasonic and optical evaluations. Although the current data on the notched behavior must be examined under sustained are limited, the fracture results indicate that within the range of load conditions specimen widths examined, the effect of width on Kpeak at (3) Damage Accumulation The load versus CMOD responses for fracture tests con- (A) Ultrasonic Evaluation: Because of the extensive pre- ducted at 23 and 950C on 25.4 mm wide specimens are existing matrix cracks present throughout the Nextel610/AS shown in Fig. 4. Unloading loops for the 23C test were elimi composite, characterization of damage accumulation based on nated from Fig. 4 for clarity. Loading behavior at 950C was surface matrix crack growth alone was determined to be inap- linear to Ka =4 MPa m2(P= 1.7 kN), which is =33% lower propriate. A nondestructive ultrasonic C-scan technique 16, 17 than that at 23C. Similar to the 23C fracture test, unloading was therefore adapted for use with the Nextel610/AS compos- the specimen from applied loads within the initial linear region ite system. 9 Some of the Nextel610/AS specimens were at 950C resulted in no CMODres or loop hysteresis. Increase in canned prior to testing. Because of the extensive preexisting applied load beyond the linear region resulted in sudden de- matrix cracks and porosity, C-scans of the untested composite crease in specimen compliance nowed regions of varying ultrasonic attenuation. These re- Nonlinear loading behavior can be compared for the two gions typically varied from 0% to 50% attenuation, with iso- fracture tests by examining the residual CMODres. Figure 5 lated regions of high porosity which attenuated the signal up to shows the CMODres after unloading the specimen from a maxi 75%. C-scans of specimens which had been loaded to suffi- mum load with corresponding maximum CMOD(CMODmax) ciently high stress showed levels of ultrasonic attenuation near or the room-temperature specimen, CMODres increased he notch tip region which exceeded 75%. Specimens whicl slowly as CMOD the Cmod at the at he eeak load this eesp isn cons stent with g the sndwna ned regges fan ere the notch nic therefore buted matri racking with attenuation of the ultrasonic signal exceeding 75% was ide inimal 0 fiber breakage. In contrast, the 950%C specimen tified as that which indicated damage accumulation during test- showed a sudden rapid increase in CMODres prior to the pea For clarity in this publication, the original color scale used load with values =5 times higher than that at room temperature r the C-scans was mapped into the gray scale image shown in The sudden rapid increase in CMODres indicates a change in Fig8. With reference to the gray scale, white indicates that the damage mode, consistent with observations of dominant crack ultrasonic signal through the specimen was attenuated less than growth discussed later in Section Ill. The increase in CMOD 75%, gray shades indicate attenuation of 75-100%. Further and dominant crack growth at 950 C are consistent with the discussion of the ultrasonic test technique and detailed color onset of 0 fiber breakage. However, additional interrupted C-scan images can be found in Ref. 19 tests are required to determine if 0 fiber breakage occurred The room-temperature fracture tests were conducted prior to the peak load at 950C. periodic unloading, but without interruption until clearly be- As a result of the enhanced nonlinear deformation prior to yond the peak load. Optical inspection of the notch tip region the peak at 950 C, a decrease in peak load =50%, compared to during testing did not reveal the formation of either new matrix that at 23C, was observed. The decrease in peak load corre cracks or a dominant matrix crack. The lack of dominant ma- sponded to a decrease in Kpeak from 16 MPa m at 23 C to 7.9 trix crack extension prior to the peak load correlated with the MPa'mn at 950 C. The higher CMODres at 950C, and de- small CMODres after unloading(Fig. 5). Although CMODres crease in notch strength, indicated a change in damage mecha- increased more rapidly after the peak load, clear dominant nism from23°C surface crack growth was not observed. However, nonlinear (2) Effect of loading Rate on 950C Fracture Behavior loading behavior prior to and after the peak load indicated that a progressive damage mechanism was operat Therefore The effect of loading rate and unloading loops on the frac- subsurface damage was examined using ultrasonic C-scans ture behavior at 950C was examined by conducting an addi- The ultrasonic C-scan of the specimen after the peak load tional fast-rate monotonic fracture test. The load versus CMOd showed a region of enhanced ultrasonic attenuation approxi- responses for the slow and fast rate tests are shown in Fig. 4. mately 10 mm in length and 10 mm in height( Fig. 6(a)). The vith unloading loops was 8=0.001 mm/s, resulting in a total gression from the notch tip at room temperature was distributed elapsed time to peak load(tp) of 668 s. The test without un away from the notch plane, consistent with the absence of loading loops was conducted at a rate of 8= 0.01 mm/s, primary crack growth from the notch. sulting in t, 15 s. As shown in Fig. 4, the overall load During the 950oC fracture test with periodic unloading versus CMob response was nearly the same for both tests, tical inspection of the notch tip region first identified the indicating that loading rate and unloading loops had no effect mation of a dominant matrix crack at an applied load of 3.0 kN on the peak stress intensity factor or overall fracture behavior. Unloading the specimen from this applied load resulted in the However, a complete assessment of time at temperature effects significant CMODres(Fig. 5). Increasing load-line displace-

ultrasonic and optical evaluations. Although the current data are limited, the fracture results indicate that within the range of specimen widths examined, the effect of width on Kpeak at room temperature is minimal. The load versus CMOD responses for fracture tests con￾ducted at 23° and 950°C on 25.4 mm wide specimens are shown in Fig. 4. Unloading loops for the 23°C test were elimi￾nated from Fig. 4 for clarity. Loading behavior at 950°C was linear to Ka 4 4 MPa?m1/2 (P 4 1.7 kN), which is ≈33% lower than that at 23°C. Similar to the 23°C fracture test, unloading the specimen from applied loads within the initial linear region at 950°C resulted in no CMODres or loop hysteresis. Increase in applied load beyond the linear region resulted in sudden de￾crease in specimen compliance. Nonlinear loading behavior can be compared for the two fracture tests by examining the residual CMODres. Figure 5 shows the CMODres after unloading the specimen from a maxi￾mum load with corresponding maximum CMOD (CMODmax). For the room-temperature specimen, CMODres increased slowly as CMODmax increased. Prior to the peak load, CMODres remained below 0.01 mm, corresponding to 13% of the CMODmax at the peak load. This result is consistent with CMODres due primarily to distributed matrix cracking with minimal 0° fiber breakage. In contrast, the 950°C specimen showed a sudden rapid increase in CMODres prior to the peak load with values ≈5 times higher than that at room temperature. The sudden rapid increase in CMODres indicates a change in damage mode, consistent with observations of dominant crack growth discussed later in Section III. The increase in CMODres and dominant crack growth at 950°C are consistent with the onset of 0° fiber breakage. However, additional interrupted tests are required to determine if 0° fiber breakage occurred prior to the peak load at 950°C. As a result of the enhanced nonlinear deformation prior to the peak at 950°C, a decrease in peak load ≈50%, compared to that at 23°C, was observed. The decrease in peak load corre￾sponded to a decrease in Kpeak from 16 MPa?m1/2 at 23°C to 7.9 MPa?m1/2 at 950°C. The higher CMODres at 950°C, and de￾crease in notch strength, indicated a change in damage mecha￾nism from 23°C. (2) Effect of Loading Rate on 950°C Fracture Behavior The effect of loading rate and unloading loops on the frac￾ture behavior at 950°C was examined by conducting an addi￾tional fast-rate monotonic fracture test. The load versus CMOD responses for the slow and fast rate tests are shown in Fig. 4. The load-line displacement rate (d . ) used during the fracture test with unloading loops was d . 4 0.001 mm/s, resulting in a total elapsed time to peak load (tp) of 668 s. The test without un￾loading loops was conducted at a rate of d . 4 0.01 mm/s, resulting in tp 4 15 s. As shown in Fig. 4, the overall load versus CMOD response was nearly the same for both tests, indicating that loading rate and unloading loops had no effect on the peak stress intensity factor or overall fracture behavior. However, a complete assessment of time at temperature effects on the notched behavior must be examined under sustained load conditions. (3) Damage Accumulation (A) Ultrasonic Evaluation: Because of the extensive pre￾existing matrix cracks present throughout the Nextel610/AS composite, characterization of damage accumulation based on surface matrix crack growth alone was determined to be inap￾propriate. A nondestructive ultrasonic C-scan technique16,17 was therefore adapted for use with the Nextel610/AS compos￾ite system.19 Some of the Nextel610/AS specimens were scanned prior to testing. Because of the extensive preexisting matrix cracks and porosity, C-scans of the untested composite showed regions of varying ultrasonic attenuation. These re￾gions typically varied from 0% to 50% attenuation, with iso￾lated regions of high porosity which attenuated the signal up to 75%. C-scans of specimens which had been loaded to suffi￾ciently high stress showed levels of ultrasonic attenuation near the notch tip region which exceeded 75%. Specimens which were tested and showed increased attenuation in the notch tip region showed no change in the level of ultrasonic attenuation of the undamaged regions far from the notch tip. Therefore, attenuation of the ultrasonic signal exceeding 75% was iden￾tified as that which indicated damage accumulation during test￾ing. For clarity in this publication, the original color scale used for the C-scans was mapped into the gray scale image shown in Fig. 8. With reference to the gray scale, white indicates that the ultrasonic signal through the specimen was attenuated less than 75%; gray shades indicate attenuation of 75–100%. Further discussion of the ultrasonic test technique and detailed color C-scan images can be found in Ref. 19. The room-temperature fracture tests were conducted with periodic unloading, but without interruption until clearly be￾yond the peak load. Optical inspection of the notch tip region during testing did not reveal the formation of either new matrix cracks or a dominant matrix crack. The lack of dominant ma￾trix crack extension prior to the peak load correlated with the small CMODres after unloading (Fig. 5). Although CMODres increased more rapidly after the peak load, clear dominant surface crack growth was not observed. However, nonlinear loading behavior prior to and after the peak load indicated that a progressive damage mechanism was operative. Therefore, subsurface damage was examined using ultrasonic C-scans. The ultrasonic C-scan of the specimen after the peak load showed a region of enhanced ultrasonic attenuation approxi￾mately 10 mm in length and 10 mm in height (Fig. 6(a)). The extent of ultrasonic attenuation indicated that the damage pro￾gression from the notch tip at room temperature was distributed away from the notch plane, consistent with the absence of primary crack growth from the notch. During the 950°C fracture test with periodic unloading, op￾tical inspection of the notch tip region first identified the for￾mation of a dominant matrix crack at an applied load of 3.0 kN. Unloading the specimen from this applied load resulted in the significant CMODres (Fig. 5). Increasing load-line displace￾Table I. Notched and Unnotched Test Results Temperature (°C) Width (mm) d . (mm/s) Elastic modulus (GPa) Unnotched peak stress (MPa) Notched peak net section stress (MPa) Peak stress intensity factor (MPa√m) Time to peak stress (s) Unnotched 23† 10.1 0.05 73 205 9.4 1000† 13.6 0.05 77 173 8.9 Notched 23 19.0 0.001 69 151 16.8 736 23 25.4 0.001 71 124 15.5 665 950 25.4 0.001 71 62 7.9 668 950 25.4 0.01 70 62 7.9 15 † Reported values are an average from three tests conducted at 23°C and two tests at 1000°C. 3090 Journal of the American Ceramic Society—Kramb et al. Vol. 82, No. 11

November 1999 Notched fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3091 ment resulted in extension of the dominant crack as it grew and 0 fiber tows extend from the failure plane as fiber bundles inked with other preexisting surface matrix cracks. When Farther from the failure plane, the bundles separated into indi- test was stopped, the continuous matrix crack extension fre vidual fibers with uncorrelated fiber failure locations the notch tip was equal to 9 mm(Fig. 6(c). Figure 6(b)shows Matrix cracking throughout the 90 fiber tows resulted the corresponding ultrasonic C-scan at the same point of un- long unbroken lengths of exposed 90 fibers and a peeling type oading. The region of ultrasonic attenuation was 5 mm in of failure mode. This damage mode led to disintegration of the height and extended 1l mm beyond the notch, slightly longer 90 tows(Fig. 8(b) and 7(a)). Matrix degradation within 90 an the surface matrix crack extension the dominant crack tows was also observed above and below the primary failure extension at 950C was associated with a more narrow damage plane. The distributed damage resulting from degradation of zone compared to that observed at room temperature the matrix within 90 tows resulted in the enhanced ultrasonic Fracture surface profiles of the 23 and 950C specimens are attenuation away from the notch plane in the room-temperature shown in Fig. 7. Note that the specimen in Fig. 7(a)corre specimen. 19,20 ds to that in Fig. 6(a), and the specimen in Fig. 7(b) was The observed damage indicated that crack growth from de nically loaded to failure as shown in Fig 4. The 23C graded 90 tows and preexisting matrix cracks was dissipated are specimen exhibited fiber bundles which ex in the crack plane by the 0 fiber tows as a whole. Near the mm in either direction from the notch plane( Fig. 7(a)). In ary crack plane broken matrix remained attached to the ontrast, fibers extended a maximum of 1.0 mm from the crack fiber tows as shown in Fig 8(c). Away from the crack plane lane of the fracture surface on the 950C specimen( Fig. 7(b). matrix crack growth parallel to individual fibers further pro- These relative fiber lengths are consistent with the C-scan at- moted energy dissipation. This mechanism of stress redistri tenuation regions which indicated a damage zone which was bution is in contrast to the debonding/sliding mechanism op- more concentrated in the notch plane at 950C. For both tem- erating in CMc with an engineered interphase. The ma- res,the height of the C-scan damage zone is approxi trix remaining within the tow >l mm away from the primary mately 2 times larger than the maximum length of exposed failure plane was effectively disintegrated as fibers broke dur fibers. This result can be attributed to matrix damage which ing the final fast fracture. As a result of the matrix disintegra extends beyond the point where fiber breakage occurs. Hence, tion, no matrix sockets were observed on the fracture surfaces fibers do not necessarily break at the end of the damage zone. as are observed in CMCs which exhibit a fiber/matrix sliding and exposed fiber lengths on the fracture surface underestimate mechanism. The absence of sockets is similar to the observa- the extent of damage above and below the notch plane tions of damage in an all-oxide CMC by Levi et al As a (B) Observed Damage Modes: Micrographs of the room- result of matrix cracking and disintegration along 0 fibers, temperature specimen fracture surface are shown in Fig 8. A exposed fibers far from the primary crack plane are relatively rOss-sectional view of the crack growth from the machined smooth and nearly devoid of matrix( Fig 8(d). These damage notch tip showed the primary failure plane advanced from the mechanisms associated with extensive matrix cracking resul notch along 90 tows and matrix-rich regions(Fig. 8(a). The in redistribution of stresses at the notch tip, thus decreasing friable nature of the brittle matrix resulted in multiple distrib- notch sensitivity and increasing fracture toughness at room uted matrix cracking within the 90 tows. This damage mode allowed the 90 fibers to be pulled away from the fracture Observations of surface and subsurface damage occurring at surface at the notch tip and thus cannot be seen in Fig 8(a). The 950C were obtained from fracture surfaces and polished sec- 10 mm 10 mm notch notch notch dominant matrix 2 mm crack extension (c) Fig. 6. (a) Ultrasonic C-scan of entire gage section, room- temperature specimen(b)Ultrasonic C-scan of entire 950C specimen gage section. The aw-cut notch region is indicated by the solid white bar in both C-scans.(c)Optical micrograph of 950C fracture test specimen unloaded after peak

ment resulted in extension of the dominant crack as it grew and linked with other preexisting surface matrix cracks. When the test was stopped, the continuous matrix crack extension from the notch tip was equal to 9 mm (Fig. 6(c)). Figure 6(b) shows the corresponding ultrasonic C-scan at the same point of un￾loading. The region of ultrasonic attenuation was 5 mm in height and extended 11 mm beyond the notch, slightly longer than the surface matrix crack extension. The dominant crack extension at 950°C was associated with a more narrow damage zone compared to that observed at room temperature. Fracture surface profiles of the 23° and 950°C specimens are shown in Fig. 7. Note that the specimen in Fig. 7(a) corre￾sponds to that in Fig. 6(a), and the specimen in Fig. 7(b) was monotonically loaded to failure as shown in Fig. 4. The 23°C fracture specimen exhibited fiber bundles which extend 2–3 mm in either direction from the notch plane (Fig. 7(a)). In contrast, fibers extended a maximum of 1.0 mm from the crack plane of the fracture surface on the 950°C specimen (Fig. 7(b)). These relative fiber lengths are consistent with the C-scan at￾tenuation regions which indicated a damage zone which was more concentrated in the notch plane at 950°C. For both tem￾peratures, the height of the C-scan damage zone is approxi￾mately 2 times larger than the maximum length of exposed fibers. This result can be attributed to matrix damage which extends beyond the point where fiber breakage occurs. Hence, fibers do not necessarily break at the end of the damage zone, and exposed fiber lengths on the fracture surface underestimate the extent of damage above and below the notch plane. (B) Observed Damage Modes: Micrographs of the room￾temperature specimen fracture surface are shown in Fig. 8. A cross-sectional view of the crack growth from the machined notch tip showed the primary failure plane advanced from the notch along 90° tows and matrix-rich regions (Fig. 8(a)). The friable nature of the brittle matrix resulted in multiple distrib￾uted matrix cracking within the 90° tows. This damage mode allowed the 90° fibers to be pulled away from the fracture surface at the notch tip and thus cannot be seen in Fig. 8(a). The 0° fiber tows extend from the failure plane as fiber bundles. Farther from the failure plane, the bundles separated into indi￾vidual fibers with uncorrelated fiber failure locations. Matrix cracking throughout the 90° fiber tows resulted in long unbroken lengths of exposed 90° fibers and a peeling type of failure mode. This damage mode led to disintegration of the 90° tows (Fig. 8(b) and 7(a)). Matrix degradation within 90° tows was also observed above and below the primary failure plane. The distributed damage resulting from degradation of the matrix within 90° tows resulted in the enhanced ultrasonic attenuation away from the notch plane in the room-temperature specimen.19,20 The observed damage indicated that crack growth from de￾graded 90° tows and preexisting matrix cracks was dissipated in the crack plane by the 0° fiber tows as a whole. Near the primary crack plane, broken matrix remained attached to the fiber tows as shown in Fig. 8(c). Away from the crack plane, matrix crack growth parallel to individual fibers further pro￾moted energy dissipation. This mechanism of stress redistri￾bution is in contrast to the debonding/sliding mechanism op￾erating in CMC with an engineered interphase. The ma￾trix remaining within the tow >1 mm away from the primary failure plane was effectively disintegrated as fibers broke dur￾ing the final fast fracture. As a result of the matrix disintegra￾tion, no matrix sockets were observed on the fracture surfaces as are observed in CMCs which exhibit a fiber/matrix sliding mechanism. The absence of sockets is similar to the observa￾tions of damage in an all-oxide CMC by Levi et al.11 As a result of matrix cracking and disintegration along 0° fibers, exposed fibers far from the primary crack plane are relatively smooth and nearly devoid of matrix (Fig. 8(d)). These damage mechanisms associated with extensive matrix cracking result in redistribution of stresses at the notch tip, thus decreasing notch sensitivity and increasing fracture toughness at room temperature. Observations of surface and subsurface damage occurring at 950°C were obtained from fracture surfaces and polished sec￾Fig. 6. (a) Ultrasonic C-scan of entire gage section, room-temperature specimen. (b) Ultrasonic C-scan of entire 950°C specimen gage section. The saw-cut notch region is indicated by the solid white bar in both C-scans. (c) Optical micrograph of 950°C fracture test specimen unloaded after peak load. November 1999 Notched Fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3091

3092 Journal of the American Ceramic Socieny-Kramb et al. Vol. 82. No. 11 1 mm notch tip 1 mm tch t Fig. 7. Optical fracture surface profiles for specimens tested at(a)23 and( b)950C monotonically loaded to failure tions of two specimens. The fracture surface images were observed along the entire length of the surface crack. a com- ined from the specimen loaded monotonically to failure. The candle behavior at 950 and 23 C. At 23 C, multiple matrix son of Figs. 8(c)and 10 shows the differences in fiber polished sections were obtained from the specimen shown ig. 6(c)which was unloaded after reaching the peak load cracking within fiber tows allowed individual fibers to fail igure 9(a)shows a cross-sectional view of the machined notch independently. This damage mode resulted in fiber failure far ip in the unloaded specimen. The micrograph shows that the from the primary crack plane and long exposed individual 0 fibers on the fracture surface. In contrast, at 950C, energy rich regions and 90 tows. The residual crack opening displace dissipation through matrix cracking along individual 0 ment(CODres)within the 90 tows is clearly seen in the mi- crograph. The average 90 tow CODes =40 um, which is as bundles near the crack plane as the crack propagated comparable to the Cmod after final unloading 42 theo°tow Fig. 5). A cross-sectional view of the notch tip region of the The observed differences in matrix cracking behavior at 23 failed specimen is shown in Fig 9(b). As observed on the and 950C may be related to changes in the alumina-silica unloaded specimen, crack growth within the 90 tows and matrix with temperature. Studies of crack growth in alumina trix-rich regions was restricted to a single plane. However, both ceramics containing glassy grain boundary phases showed that micrographs in Fig 9 show that the primary crack was initiall crack healing and blunting occurred at temperatures near deflected in the notch plane by matrix cracking along 0 tows. 1000 C.22, 23In Nextel610/AS, stress redistribution occurred, at This initial crack deflection allowed 0 fiber failure to occur room temperature, through distributed matrix cracking. This away from the crack plane. Unlike the room-temperature speci- damage mechanism relied on the friable nature of the weak men, however, distributed matrix cracking within both the 0 matrix. At 950C, crack healing and blunting would inhibit and 90 tows was greatly reduced. Within the 0o tows, the lack distributed matrix cracking. The lack of distributed matrix of distributed matrix cracking along 0 fibers resulted in failure cracking around the notch would result in increased notch sen- ocations significantly closer to the crack plane than at 23C. sitivity at 950C. Further studies are currently under way to hin the 90 tows, individual fibers remained bonded to the investigate these mechanisms matrix and were broken in the crack plane as the crack ad The assessment of damage progression in Nextel61O/AS dis- vanced(Figs. 9(b)and 7(b). Thus, individual 90 fibers were cussed above indicated there was a change in damage mode as not pulled away from the crack plane as was observed on the temperature was increased from 23. to 950C. The lack of room-temperature fracture surface. Similarly, polished sections distributed damage at 950C resulted in a higher stress con- of the unloaded specimen showed that matrix degradation centration at the notch tip than at room temperature, and the within 90 fiber tows was limited. Thus. the lack of distributed propagation of a single dominant crack. This change in damage matrix cracking between fibers within the 0 and 90 tows at mode, coupled with the decrease in fiber strength at 950oC 950 C resulted in a change in damage mechanism from that (=15%),21 resulted in significant decrease(=50%)in apparent observed at23°C. fracture toughness further details of the o° tow behavior at950°C The applicability of linear elastic fracture mechanics tained by closer examination of crack propagation from the (LEFM) to characterize the notched behavior has not been notch tip. Figure 10 shows the notch tip region of the specimen established for the present CMC system. The large post-peak in Fig. 6(c) after polishing the specimen surface to remove the damage zone size indicated in the C s(Fig. 6) implies top matrix layer and some of the 90 fibers At 950C, matrix LEFM may not be applicable. Further interrupted testing is king between fibers within the 0 tows was minimal, re- required to determine damage progression prior to the peak Iting in 0 fibers breaking as bundles. Broken 0o fibers were load. The damage zone size at the peak load can then be used

tions of two specimens. The fracture surface images were ob￾tained from the specimen loaded monotonically to failure. The polished sections were obtained from the specimen shown in Fig. 6(c) which was unloaded after reaching the peak load. Figure 9(a) shows a cross-sectional view of the machined notch tip in the unloaded specimen. The micrograph shows that the primary failure plane grew from the notch within the matrix￾rich regions and 90° tows. The residual crack opening displace￾ment (CODres) within the 90° tows is clearly seen in the mi￾crograph. The average 90° tow CODres ≈ 40 mm, which is comparable to the CMODres after final unloading 4 42 mm (Fig. 5). A cross-sectional view of the notch tip region of the failed specimen is shown in Fig. 9(b). As observed on the unloaded specimen, crack growth within the 90° tows and ma￾trix-rich regions was restricted to a single plane. However, both micrographs in Fig. 9 show that the primary crack was initially deflected in the notch plane by matrix cracking along 0° tows. This initial crack deflection allowed 0° fiber failure to occur away from the crack plane. Unlike the room-temperature speci￾men, however, distributed matrix cracking within both the 0° and 90° tows was greatly reduced. Within the 0° tows, the lack of distributed matrix cracking along 0° fibers resulted in failure locations significantly closer to the crack plane than at 23°C. Within the 90° tows, individual fibers remained bonded to the matrix and were broken in the crack plane as the crack ad￾vanced (Figs. 9(b) and 7(b)). Thus, individual 90° fibers were not pulled away from the crack plane as was observed on the room-temperature fracture surface. Similarly, polished sections of the unloaded specimen showed that matrix degradation within 90° fiber tows was limited. Thus, the lack of distributed matrix cracking between fibers within the 0 and 90° tows at 950°C resulted in a change in damage mechanism from that observed at 23°C. Further details of the 0° tow behavior at 950°C were ob￾tained by closer examination of crack propagation from the notch tip. Figure 10 shows the notch tip region of the specimen in Fig. 6(c) after polishing the specimen surface to remove the top matrix layer and some of the 90° fibers. At 950°C, matrix cracking between fibers within the 0° tows was minimal, re￾sulting in 0° fibers breaking as bundles. Broken 0° fibers were observed along the entire length of the surface crack. A com￾parison of Figs. 8(c) and 10 shows the differences in 0° fiber bundle behavior at 950° and 23°C. At 23°C, multiple matrix cracking within fiber tows allowed individual fibers to fail independently. This damage mode resulted in fiber failure far from the primary crack plane and long exposed individual 0° fibers on the fracture surface. In contrast, at 950°C, energy dissipation through matrix cracking along individual 0° fibers within the tow did not occur. As a result, 0° fibers were broken as bundles near the crack plane as the crack propagated through the 0° tow. The observed differences in matrix cracking behavior at 23° and 950°C may be related to changes in the alumina–silica matrix with temperature. Studies of crack growth in alumina ceramics containing glassy grain boundary phases showed that crack healing and blunting occurred at temperatures near 1000°C.22,23 In Nextel610/AS, stress redistribution occurred, at room temperature, through distributed matrix cracking. This damage mechanism relied on the friable nature of the weak matrix. At 950°C, crack healing and blunting would inhibit distributed matrix cracking. The lack of distributed matrix cracking around the notch would result in increased notch sen￾sitivity at 950°C. Further studies are currently under way to investigate these mechanisms. The assessment of damage progression in Nextel610/AS dis￾cussed above indicated there was a change in damage mode as temperature was increased from 23° to 950°C. The lack of distributed damage at 950°C resulted in a higher stress con￾centration at the notch tip than at room temperature, and the propagation of a single dominant crack. This change in damage mode, coupled with the decrease in fiber strength at 950°C (≈15%),21 resulted in significant decrease (≈50%) in apparent fracture toughness. The applicability of linear elastic fracture mechanics (LEFM) to characterize the notched behavior has not been established for the present CMC system. The large post-peak damage zone size indicated in the C-scans (Fig. 6) implies LEFM may not be applicable. Further interrupted testing is required to determine damage progression prior to the peak load. The damage zone size at the peak load can then be used Fig. 7. Optical fracture surface profiles for specimens tested at (a) 23° and (b) 950°C monotonically loaded to failure. 3092 Journal of the American Ceramic Society—Kramb et al. Vol. 82, No. 11

November 1999 Notched fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 0.5m 0°tow matrix 90°tow thickness 0.10mm GE F 0.05mm 0° fibers matrix 90° fibers Fig.8. Scanning electron micrographs of room-temperature fracture surface:(a) cross-sectional view of machined notch tip, (b)top view of fracture surface (15 tilt from normal),(c)0 fibers near matrix crack plane, (d)0 fibers=l mm from primary crack plar Ko wit we applicability of LEFM. In addition, variation in lower than the UTS at 1000C of 176 MPa. The notched trength decreased =50% as temperature was increased from determine the applicability of K peak to predict failure. However 23 to 950C. In contrast, the unnotched UTS decreased by ndependent of LEFM, notch sensitivity of the Nextel61O/As only 15% as temperature was increased from 23 to 1000C system can be studied in terms of the net section strength(on) Hence, the relative decrease in notch strength with increasing as discussed in the next section temperature was significantly greater than the corresponding decrease in unnotched tensile strength. The time to peak stress (4 Comparison of Notched and Unnotched Behavior rp =9 s in both tensile tests, was close to tp =15 s in the notched Table I summarizes the results for the notched fracture tests fracture test without unloading loops. A similar time to peak and unnotched tensile tests 1, 2 The elastic moduli of 73 and 77 stress in the tension and fracture tests implies that the observed GPa at 23 and 1000 C, respectively, were close to the avesoved due to time at temperature decrease in fracture toughness between 23 and 950%C was not value of 70 GPa for the notched specimens. The unnotche tensile stress-strain response is shown in Fig. 11.At temperature the tensile behavior was nearly linear to the ulti mate failure stress(UTS)(Fig. 11). In contrast, the notched E he room-temperature unnotched tensile specimen fracture surfaces showed that failure was associated with interply de mination and 0o fiber failure far from the fracture plane(Fig specimens exhibited nonlinear loading behavior prior to the 12(a). The absence of a stress concentration in the unnotched ak stress, and nonbrittle failure after reaching the peak load specimen distributed the applied stress across the entire width (Fig. 4). The unnotched tensile response at 1000 C was slightly and gage length, allowing the composite plies to fail on differ nonlinear before abrupt failure at the ultimate stress(Fig. 11). ent planes. Examination of the 0o fiber tows on the fracture This is in sharp contrast to the notched fracture behavior at surface showed that close to the ply failure plane matrix re 950C which resulted in extensive nonlinear loading prior to mained attached to the fibers(Fig. 12(c)). This is similar to the and after reaching the peak load observed failure features near the crack plane of the notched The net section failure strength(on)of the notched speci- specimen(Fig. 8(c). Far from the ply failure plane, relatively nens was significantly less than the UTS (Table I). At room smooth fiber surfaces nearly devoid of matrix were observ mperature, the average on was 138 MPa, 34% lower than the similar to the notched specimen( Figs. 12(c)and 8(d). The notched UTS of 208 MPa. At 950 C, 0, was 62 MPa, 65% smooth appearance of fibers in both the notched and unnotched

to assess the applicability of LEFM. In addition, variation in Kpeak with notch length and specimen size must be examined to determine the applicability of Kpeak to predict failure. However, independent of LEFM, notch sensitivity of the Nextel610/AS system can be studied in terms of the net section strength (sn) as discussed in the next section. (4) Comparison of Notched and Unnotched Behavior Table I summarizes the results for the notched fracture tests and unnotched tensile tests.1,2 The elastic moduli of 73 and 77 GPa at 23° and 1000°C, respectively, were close to the average value of 70 GPa for the notched specimens. The unnotched tensile stress–strain response is shown in Fig. 11. At room temperature the tensile behavior was nearly linear to the ulti￾mate failure stress (UTS) (Fig. 11). In contrast, the notched specimens exhibited nonlinear loading behavior prior to the peak stress, and nonbrittle failure after reaching the peak load (Fig. 4). The unnotched tensile response at 1000°C was slightly nonlinear before abrupt failure at the ultimate stress (Fig. 11). This is in sharp contrast to the notched fracture behavior at 950°C which resulted in extensive nonlinear loading prior to and after reaching the peak load. The net section failure strength (sn) of the notched speci￾mens was significantly less than the UTS (Table I). At room temperature, the average sn was 138 MPa, 34% lower than the unnotched UTS of 208 MPa. At 950°C, sn was 62 MPa, 65% lower than the UTS at 1000°C of 176 MPa. The notched strength decreased ≈50% as temperature was increased from 23° to 950°C. In contrast, the unnotched UTS decreased by only 15% as temperature was increased from 23° to 1000°C. Hence, the relative decrease in notch strength with increasing temperature was significantly greater than the corresponding decrease in unnotched tensile strength. The time to peak stress, tp ≈ 9 s in both tensile tests, was close to tp ≈ 15 s in the notched fracture test without unloading loops. A similar time to peak stress in the tension and fracture tests implies that the observed decrease in fracture toughness between 23° and 950°C was not due to time at temperature. The room-temperature unnotched tensile specimen fracture surfaces showed that failure was associated with interply de￾lamination and 0° fiber failure far from the fracture plane (Fig. 12(a)). The absence of a stress concentration in the unnotched specimen distributed the applied stress across the entire width and gage length, allowing the composite plies to fail on differ￾ent planes. Examination of the 0° fiber tows on the fracture surface showed that close to the ply failure plane matrix re￾mained attached to the fibers (Fig. 12(c)). This is similar to the observed failure features near the crack plane of the notched specimen (Fig. 8(c)). Far from the ply failure plane, relatively smooth fiber surfaces nearly devoid of matrix were observed, similar to the notched specimen (Figs. 12(c) and 8(d)). The smooth appearance of fibers in both the notched and unnotched Fig. 8. Scanning electron micrographs of room-temperature fracture surface: (a) cross-sectional view of machined notch tip, (b) top view of fracture surface (15° tilt from normal), (c) 0° fibers near matrix crack plane, (d) 0° fibers ≈1 mm from primary crack plane. November 1999 Notched Fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3093

Journal of the American Ceramic Sociery-Kramb et al Vol. 82, No. 11 0.5mm 9o° tow cOD o°tow thickness 0.5mm 0°tow 90°tow thickness Fig 9. Cross-sectional view of machined notch tip region in 950oC specimens:(a) interrupted after the peak load, (b) monotonically loaded to failure 0.20mm notch tip 0"c 23" 100 1 tow width Fig. 11. Tensile stress/strain response for unnotched Nextel6lO/AS 1g specimen shown in Fig. 6(c), after t23°and1000 removin some90° fibers. Failed 0° fibers the dominant crack Unnotched tensile failure at 1000C resulted in a more o- specimens is a direct result of matrix cracking and disintegra- calized fracture zone with 0 fiber breaks closer to the crack tion during fracture, not due to a debonding/sliding mechanism lane than at 23C(Fig. 12(b). Similar to the notched speci- as occurs in CMCs with an engineered interphase. Multiple men at 950C, matrix cracks within 90 tows were restricted to matrix cracking within the 90 fiber tows in the unnotched a single plane. Matrix remained bonded to the fibers, resulting specimen resulted in long exposed 90 fiber lengths, similar to in 90 fiber failure within the crack plane(Figs. 9(b)and the notched specimen(Figs. 12(a)and 8(b). Although over- 12(d ). Although the failure locations of exposed 0 tows were all features of the fracture surfaces are similar in the notched close to the failure plane, fracture was not planar. This result and unnotched specimens, exposed 0o fiber lengths were indicated that propagation of preexisting matrix cracks was onsiderably longer (2-3 times)in the unnotched tension initially deflected by the 0o tows. In contrast to the observed ature fracture behavior. close examination of the

specimens is a direct result of matrix cracking and disintegra￾tion during fracture, not due to a debonding/sliding mechanism as occurs in CMCs with an engineered interphase. Multiple matrix cracking within the 90° fiber tows in the unnotched specimen resulted in long exposed 90° fiber lengths, similar to the notched specimen (Figs. 12(a) and 8(b)). Although over￾all features of the fracture surfaces are similar in the notched and unnotched specimens, exposed 0° fiber lengths were considerably longer (2–3 times) in the unnotched tension specimen. Unnotched tensile failure at 1000°C resulted in a more lo￾calized fracture zone with 0° fiber breaks closer to the crack plane than at 23°C (Fig. 12(b)). Similar to the notched speci￾men at 950°C, matrix cracks within 90° tows were restricted to a single plane. Matrix remained bonded to the fibers, resulting in 90° fiber failure within the crack plane (Figs. 9(b) and 12(d)). Although the failure locations of exposed 0° tows were close to the failure plane, fracture was not planar. This result indicated that propagation of preexisting matrix cracks was initially deflected by the 0° tows. In contrast to the observed room-temperature fracture behavior, close examination of the Fig. 9. Cross-sectional view of machined notch tip region in 950°C specimens: (a) interrupted after the peak load, (b) monotonically loaded to failure. Fig. 10. Notch tip region of 950°C specimen shown in Fig. 6(c), after removing the top matrix layer and some 90° fibers. Failed 0° fibers were identified along the length of the dominant crack. Fig. 11. Tensile stress/strain response for unnotched Nextel610/AS at 23° and 1000°C. 3094 Journal of the American Ceramic Society—Kramb et al. Vol. 82, No. 11

November 1999 Notched fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3095 W=13.6mm W=10.1mm (a) 0.10mm matrix coated fibers near crack plane 020m Fig. 12. Nextel610/AS unnotched tension fracture surfaces: (a)23. and(b)1000oC. Details of matrix and 0 fiber behavior;(c)23 and(d) 1000°C. 0 fiber tows showed no evidence of the extensive matrix was characterized by nonbrittle failure with considerable re- cracking within the tows. As a result of the change in matrix tained load-bearing capacity. Load-line displacement rate and cracking behavior at 1000%C, most fibers failed as bundles rather than individual fibers. These observations showed that behavior at 950C Exposed fiber lengths on the fracture sur- the elevated temperature fracture surfaces for notched and un- face were reduced at elevated temperature with fiber bundle notched specimens exhibited similar features. Hence, the av failure significantly closer to the primary crack plane than at Although the fracture surfaces of the notched and unnotched away from the matrix crack plane with distributed matrix pecimens exhibited similar features, the overall tensile behav or was dramatically different for the two test geometries. Ap- tween the notched fracture behavior and the unnotched tensile parently the lack of matrix cracking within fiber tows at tem- response of Nextel610/AS showed that the effect of tempera- eratures near 1000%C increased notch sensitivity without ture is significantly different. greatly affecting unnotched tensile strength Ultrasonic C-scans were shown to be an effective method of neasuring the extent of damage accumulation. Post-peak room-temperature test specimens showed no evidence of dam- Summary and conclusion age on the specimen surface, however, the attenuated zone indicated by the C-scan provided a measure of the extent and Fracture tests conducted at 23 and 950C showed that the distribution of damage. Dominant matrix crack growth was net section strength of Nextel610/AS composite was less tha observed on the specimen surface during testing at 950@C. The the unnotched strength. The net section strength of the notched surface crack length correlated well with the length of the specimens was =65% and 35% of UTS at 23 and 950C. C-scan damage zone. C-scans indicated that, at 950C, the respectively. The net sections strength decreased 50% with an height of the damage zone was approximately half that at 23.C ncrease in temperature from 23 to 950C. Fracture behavior The reduction in damage zone height correlated with the was nonlinear prior to the peak load. The post-peak behavior duction in exposed 0 fiber lengths on fracture surfaces

0° fiber tows showed no evidence of the extensive matrix cracking within the tows. As a result of the change in matrix cracking behavior at 1000°C, most fibers failed as bundles rather than individual fibers. These observations showed that the elevated temperature fracture surfaces for notched and un￾notched specimens exhibited similar features. Hence, the av￾erage length of exposed 0° fibers was also similar, ≈0.5 mm. Although the fracture surfaces of the notched and unnotched specimens exhibited similar features, the overall tensile behav￾ior was dramatically different for the two test geometries. Ap￾parently the lack of matrix cracking within fiber tows at tem￾peratures near 1000°C increased notch sensitivity without greatly affecting unnotched tensile strength. V. Summary and Conclusion Fracture tests conducted at 23° and 950°C showed that the net section strength of Nextel610/AS composite was less than the unnotched strength. The net section strength of the notched specimens was ≈65% and 35% of UTS at 23° and 950°C, respectively. The net sections strength decreased 50% with an increase in temperature from 23° to 950°C. Fracture behavior was nonlinear prior to the peak load. The post-peak behavior was characterized by nonbrittle failure with considerable re￾tained load-bearing capacity. Load-line displacement rate and unloading loops were found to have no effect on the fracture behavior at 950°C. Exposed fiber lengths on the fracture sur￾face were reduced at elevated temperature with fiber bundle failure significantly closer to the primary crack plane than at room temperature. At room temperature individual fibers failed away from the matrix crack plane with distributed matrix cracking within the 0° and 90° fiber tows. A comparison be￾tween the notched fracture behavior and the unnotched tensile response of Nextel610/AS showed that the effect of tempera￾ture is significantly different. Ultrasonic C-scans were shown to be an effective method of measuring the extent of damage accumulation. Post-peak room-temperature test specimens showed no evidence of dam￾age on the specimen surface; however, the attenuated zone indicated by the C-scan provided a measure of the extent and distribution of damage. Dominant matrix crack growth was observed on the specimen surface during testing at 950°C. The surface crack length correlated well with the length of the C-scan damage zone. C-scans indicated that, at 950°C, the height of the damage zone was approximately half that at 23°C. The reduction in damage zone height correlated with the re￾duction in exposed 0° fiber lengths on fracture surfaces. Fig. 12. Nextel610/AS unnotched tension fracture surfaces: (a) 23° and (b) 1000°C. Details of matrix and 0° fiber behavior: (c) 23° and (d) 1000°C. November 1999 Notched Fracture Behavior of an Oxide/Oxide Ceramic-Matrix Composite 3095

3096 Journal of the American Ceramic Socieny-Kramb et al. Vol. 82. No. 11 Observations of damage on fracture surfaces and polished Nicalon/CAS Continuous Fibre-Reinforced Glass-Ceramic Matrix Composite, tions revealed a change in matrix crack behavior between Composites,24口2]150-56(1993) 3°and950°C.At23°C, extensive matrix cracking within fiber F E. Heredia, S M. Spearing, T.J. Mackin, M.Y. He, A G. Evans, Mosher and P. Bron tows allowed for stress redistribution around the notch. Evi Am. Ceran.Soc,772817-27(1994) lence of this damage mechanism was indicated by exposed K.R. Turner, J.S. Speck, and A G, Evans, ""Mechanisms of Deformation 0o fiber lengths, matrix cracking between fibers, and the ex and Failure in Carbon-Matrix Composites Subject to Tensile and Shear Load- tent of ultrasonic C-scan attenuation. At 950%C. a change in ing,J. Am. Cera. Soc., 78[7)1841-48(1995) ow. C. Tu, F. F. Lange, and A. G. Evans, "Concept for a Damage-Tolerant ailure mode to self-similar crack growth occurred. Examina Ceramic Composite with'Strong'Interfaces, "J. Am. Ceram Soc., 79[ 417- ions of exposed 0 fibers revealed at 950C matrix cracking fiber lengths and the extent of ultrasonic C-scan attenuation mpany Product Data Sheet, 3M Ceramic Fiber Products, 3M Center, erified that the damage zone was confined to the notch plane t 950C. Crack growth beyond the peak load was character- IG. A. Hartman and D. J. Buchanan, "Methodologies for Thermal and Me- ized by fiber breakage. Although load-bearing capacity de hanical Testing of TMC Materials, "AGARD Report 796, Characterizatio creased slightly after the peak load, clearly fiber breakage in StuD, Bordeaux pals Pane bOrdeaux, France, Septeting of the AGARD er27-28,1993) the crack wake did not immediately result in catastrophic fail- ure at950°C 4G. A. Hartman and S. M. Russ, Techniques for Mechanical and T Testing of Ti3Al/SCS-6 Metal Matrix Composites"; pp. 43-53 in ASTM Spe- Acknowledgment: This research was conducted at the Materials and Analsis, and Failure Modes. Edited S Johnson. American Society for Manufacturing Directorate, Air Force Research Laboratory (AFRL/MLLN), Wright-Patterson Air Force Base, OH 45433-7817. Solutions for a Single edge Cracked t eeight to Width Ratio on K and CMOD Eng. fract.Mech,602147-56(1997) A. Stubbs and G. S. Clem Screening Metal Matrix Composites References Using Ultrasonic Reflector Plate an L P. Zawada and S.S. Lee, "The Effect of Hold Times on the Fatig Behavior of an Oxide/Oxide Ceramic Matrix Con vior and Damage Tolerance of TMCs, NASP Techni Memorandum 1199. NASP Joint Program Office. Wright-Patterson Air Force ds and Behavior of Continous- Fiber Ceramic Matrix Composite Base. OH. 1995. M. G. Jenkins, S. T. Gonczy, E. Lara-Curzio, N. E. Ashbaugh Stubbs and G. S. Clemons. "Guidelines for Standardizing the gain of S S. Lee, L P. Zawada, R.S. Hay, and J. Staehler, High Tem coind Characteriation of Titanium Matrix Composites, Vol. 7, Mechanice Mechanical Behavior and Characterization of an Oxide/Oxide Composite, "un- R. John and N. E. Ashbaugh, " Fatigue Crack Growth in Ceramics and Ce amic Matrix Composites, Pp 28-50 in ASTM Speci nical Publication Turbine Engine Divergent Flaps and Seals, "Ceram. Eng. Sci. Proc., 16 [41 ion, Fracture and Nondestructive Evaluation of 337-39(1995 Adranced Materials. Edited by M. R. Mitchell and O Buck. American Society 9V. Kramb, R. John and D. Stubbs, to be submitted for ication A. Kramb and R. John, "Room Temperature Fracture Behavior of C. M. Cady, T J Mackin, and A G. Evans, "Silicon Carbide/Calcium Alu- minosil Notch-Insensitive Ceramic-Matrix Composite, "J. Am. Ceram. 2D. M. Wilson, S.L. Lieder, and D. C. Lueneburg, "Microstructure Soc,78u]77-82(1995 High Temperature Properties of Nextel 720 Fibers, "Ceram. Eng. Sci. Proc., J J. Pernot, " Tensile Fracture Behaviour of ced Ceramic Composite with Hole, " Composite, 25 3J Y. Donaldson, A. Venkateswaran, and D. P H. Hasselman, ""Observa- 273-42(1994) tions on the Crack-Enhanced Creep- Fracture of a Poly G. Evans, F w. Zok, R. M. ne. and ZZ. Du. "Models of a Glassy Grain-Boundary Phase, "J. Mater. Sci., 27 [16 4501-10(199 25S. R. Choi and V. Tikare. "Crack Healing of Alumina with a Residual Composites, J. Am. Ceram. Soc., 79 Glassy Phase: Strength, Fracture Toughness and Fatigue, " Mater. Sci. E 7R. F. Allen. C. J. Beevers, and n, "Fracture and Fatigue of a Al7l,77-83(1993)

Observations of damage on fracture surfaces and polished sections revealed a change in matrix crack behavior between 23° and 950°C. At 23°C, extensive matrix cracking within fiber tows allowed for stress redistribution around the notch. Evi￾dence of this damage mechanism was indicated by exposed 0° fiber lengths, matrix cracking between fibers, and the ex￾tent of ultrasonic C-scan attenuation. At 950°C, a change in failure mode to self-similar crack growth occurred. Examina￾tions of exposed 0° fibers revealed at 950°C matrix cracking between fibers within the tows was greatly reduced, resulting in a higher stress concentration in the notch tip area. Exposed 0° fiber lengths and the extent of ultrasonic C-scan attenuation verified that the damage zone was confined to the notch plane at 950°C. Crack growth beyond the peak load was character￾ized by fiber breakage. Although load-bearing capacity de￾creased slightly after the peak load, clearly fiber breakage in the crack wake did not immediately result in catastrophic fail￾ure at 950°C. Acknowledgment: This research was conducted at the Materials and Manufacturing Directorate, Air Force Research Laboratory (AFRL/MLLN), Wright-Patterson Air Force Base, OH 45433-7817. References 1 L. P. Zawada and S. S. Lee, “The Effect of Hold Times on the Fatigue Behavior of an Oxide/Oxide Ceramic Matrix Composite”; pp. 69–101 in ASTM Special Technical Publication, Vol. 1309, Thermal and Mechanical Test Meth￾ods and Behavior of Continuous-Fiber Ceramic Matrix Composites. Edited by M. G. Jenkins, S. T. Gonczy, E. Lara-Curzio, N. E. Ashbaugh, and L. P. Zawada. American Society for Testing and Materials, Philadelphia, PA, 1996. 2 S. S. Lee, L. P. Zawada, R. S. Hay, and J. Staehler, “High Temperature Mechanical Behavior and Characterization of an Oxide/Oxide Composite,” un￾published work. 3 R. John and N. E. Ashbaugh, “Fatigue Crack Growth in Ceramics and Ce￾ramic Matrix Composites”; pp. 28–50 in ASTM Special Technical Publication, Vol. 1157, Cyclic Deformation, Fracture and Nondestructive Evaluation of Advanced Materials. Edited by M. R. Mitchell and O. Buck. American Society for Testing and Materials, Philadelphia, PA, 1992. 4 C. M. Cady, T. J. Mackin, and A. G. Evans, “Silicon Carbide/Calcium Alu￾minosilicate: A Notch-Insensitive Ceramic-Matrix Composite,” J. Am. Ceram. Soc., 78 [1] 77–82 (1995). 5 S. Mall, D. E. Bullock, and J. J. Pernot, “Tensile Fracture Behaviour of Fibre-Reinforced Ceramic-Matrix Composite with Hole,” Composite, 25 [3] 273–42 (1994). 6 A. G. Evans, F. W. Zok, R. M. McMeeking, and Z. Z. Du, “Models of High-Temperature, Environmentally Assisted Embrittlement in Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 79 [9] 2345–52 (1996). 7 R. F. Allen, C. J. Beevers, and P. Bowen, “Fracture and Fatigue of a Nicalon/CAS Continuous Fibre-Reinforced Glass-Ceramic Matrix Composite,” Composites, 24 [2] 150–56 (1993). 8 F. E. Heredia, S. M. Spearing, T. J. Mackin, M. Y. He, A. G. Evans, P. Mosher, and P. Brondsted, “Notch Effects in Carbon Matrix Composites,” J. Am. Ceram. Soc., 77 [11] 2817–27 (1994). 9 K. R. Turner, J. S. Speck, and A. G. Evans, “Mechanisms of Deformation and Failure in Carbon-Matrix Composites Subject to Tensile and Shear Load￾ing,” J. Am. Ceram. Soc., 78 [7] 1841–48 (1995). 10W. C. Tu, F. F. Lange, and A. G. Evans, “Concept for a Damage-Tolerant Ceramic Composite with ‘Strong’ Interfaces,” J. Am. Ceram. Soc., 79 [2] 417– 24 (1996). 11C. G. Levi, J. Y. Yang, B. J. Dalgleish, F. W. Zok, and A. G. Evans, “Pro￾cessing and Performance of an All-Oxide Composite,” J. Am. Ceram. Soc., 81 [8] 2077–86 (1998). 123M Company Product Data Sheet, 3M Ceramic Fiber Products, 3M Center, St. Paul, MN 55144-1000. 13G. A. Hartman and D. J. Buchanan, “Methodologies for Thermal and Me￾chanical Testing of TMC Materials,” AGARD Report 796, Characterization of Fibre Reinforced Titanium Matrix Composites, 77th Meeting of the AGARD Structures and Materials Panel (Bordeaux, France, September 27–28, 1993). AGARD, Bordeaux, France, 1994. 14G. A. Hartman and S. M. Russ, “Techniques for Mechanical and Thermal Testing of Ti3Al/SCS-6 Metal Matrix Composites”; pp. 43–53 in ASTM Spe￾cial Technical Publication, Vol. 1032, Metal Matrix Composites: Testing, Analysis, and Failure Modes. Edited by W. S. Johnson. American Society for Testing and Materials, Philadelphia, PA, 1989. 15R. John and B. Rigling, “Effect of Height to Width Ratio on K and CMOD Solutions for a Single Edge Cracked Geometry with Clamped Specimen Ends,” Eng. Fract. Mech., 60 [2] 147–56 (1997). 16D. A. Stubbs and G. S. Clemons, “Screening Metal Matrix Composites Using Ultrasonic Reflector Plate and X-ray Radiography Nondestructive Evalu￾ation Techniques”; in Characterization of Titanium Matrix Composites, Vol. 7, Mechanical Behavior and Damage Tolerance of TMCs, NASP Technical Memorandum 1199. NASP Joint Program Office, Wright-Patterson Air Force Base, OH, 1995. 17D. A. Stubbs and G. S. Clemons, “Guidelines for Standardizing the Gain of Ultrasonic Inspection Systems used to Acquire Ultrasonic Reflector Plate C￾scans”; in Characterization of Titanium Matrix Composites, Vol. 7, Mechanical Behavior and Damage Tolerance of TMCs, NASP Technical Memorandum 1199. NASP Joint Program Office, Wright-Patterson Air Force Base, OH, 1995. 18L. P. Zawada and S. S. Lee, “Evaluation of Four CMCs for Aerospace Turbine Engine Divergent Flaps and Seals,” Ceram. Eng. Sci. Proc., 16 [4] 337–39 (1995). 19V. Kramb, R. John and D. Stubbs, to be submitted for publication. 20V. A. Kramb and R. John, “Room Temperature Fracture Behavior of an Oxide/Oxide CMC,” unpublished work. 21D. M. Wilson, S. L. Lieder, and D. C. Lueneburg, “Microstructure and High Temperature Properties of Nextel 720 Fibers,” Ceram. Eng. Sci. Proc., 16 [5] 1005–14 (1995). 22K. Y. Donaldson, A. Venkateswaran, and D. P. H. Hasselman, “Observa￾tions on the Crack-Enhanced Creep-Fracture of a Polycrystalline Alumina with a Glassy Grain-Boundary Phase,” J. Mater. Sci., 27 [16] 4501–10 (1991). 23S. R. Choi and V. Tikare, “Crack Healing of Alumina with a Residual Glassy Phase: Strength, Fracture Toughness and Fatigue,” Mater. Sci. Eng., A171, 77–83 (1993). h 3096 Journal of the American Ceramic Society—Kramb et al. Vol. 82, No. 11