AH.Cea.Sox86161981-90(2003) urna Characterization and High-Temperature Mechanical Behavior of an Larry P. Zawada, Randall S. Hay, Shin S. Lee, and James Staehler Materials an Air Force Research Laboratory, Wright-Patterson Air For Ohio 45433 An oxide/oxide ceramic fiber-matrix composite(CMC) has been extensively characterized for high-temperature aerospac porous, cracked matrices for high-temperature aerospace applica- structural applications. This CMC is called GEN-IV, and it and an aluminosilicate matrix is called GEN-IV. 4, 5, 2I and it is has a porous and cracked aluminosilicate matrix reinforced by referenced in the following text as N6lO/AS. A review of the 3M Nextel 610 M alumina fibers woven in a balanced eight mechanisms and mechanical properties of porous-matrix CMCs harness weave(SHSW). This CMC has been specifically has been given elsewhere. 22 designed without an interphase between the fiber and matrix The following characteristics of N610/AS porous-matrix com- and it relies on the porous matrix for flaw tolerance. Stress- posites are evaluated: (i) long-time phase and microstructural strain response is nearly linear to failure and without a stability: (ii) high-temperature, short-time stress-strain well-defined proportional limit in tension and compression (iii)fatigue; and (iv) creep and creep rupture. Compari In-plane shear and interlaminar strength increases with in other CMCs are made and possible explanations for the creasing temperature. The 1000 C fatigue limit in air at 105 behavior are discussed cycles is 160 MPa, and the residual tensile strength of run-out pecimens is not affected by the fatigue loading. The creep- rupture resistance above 1000'C is relatively poor, but it can lL. Materials and Experiments be improved with a more-creep-resistant fiber. (1) Composite Fabrication 3M Nextel 610 fibers are used to manufacture tile of NolO/AS The Nextel 610 fiber is 99 wt% polycrystalline a-Al,O,, with a L. Introduction density of 3.88 g/cm, an average grain size of 0. I um, and an average filament diameter of 12 um. Eight harness satin weave H fracture toughness and damage tolerance is engineered (8HSW) cloth of Nextel 610 was prepregged with a mixture of fine o most fiber-reinforced ceramic-matrix composites (CMCs)by tailoring properties of the fiber-matrix interface. The A Os powder and a SiO,forming polymer. Twelve individ- fiber-matrix interface must deflect matrix cracks and allow fiber ual prepregged cloths were stacked on top of each other as a aminate. The laminate was warm molded in an autoclave to pullout afterward. Mechanical properties of CMCs break down if the coating is not stable in the application environment. 3.s ered in air at -1000C. This pr roduce a dense green-state ceramic tile. The tile was pressureless Carbon- and BN- are the usual fiber-matrix interphases in SiC-fiber CMCs. Unfortunately, carbon coatings begin to oxidize converted the polymer to porous So cess removed organics and at-450C, and the gap left by oxidation may fill with the Sio oxidation product of SiC and form a strong fiber-matrix bond that (2) Experiments seriously degrades CMC mechanical properties. .m Amorphous (A) Microstructure Characterization: The composi and imperfectly crystalline BN are moisture sensitive and easily oxidize. o, 2 Similar to SiO2, B, O, forms a strong fiber-matrix sity (seven specimens) was measured using the Archi pycnometer. bond that degrades CMC properties. Oxidation is a serious adsorbs moisture from the air. the specimens were carefully obstacle to long-term use of CMCs with carbon or BN fiber- outgassed before measurement. The total pore surface area was matrix interfaces at intermediate and high temperatures measured using the Brunauer-Emmett-Teller (BET) method An approach to flaw-tolerant CMCs that are also oxidation (eight different measurements on seven specimens). Fiber volume resistant is oxide/oxide CMCs with fibers that are " strongly fractions were measured in three specimens polished at 45 to the bonded to a matrix deliberately made weak by incorporation of fiber axes. A microscope( Model Metallovert, Leitz) mounted with high porosity and microcracks. Instead of crack deflection and a video camera, video monitor, and computer running image diffuse microcracking in the s matrix. Early modeling of mage analysis of the fiber volume fractions pus)was used for hese materials suggests that the fiber bundles must be heterog Fracture surfaces of failed specimens were characterized using enously distributed in the matrix, have higher coefficient of thermal expansion (CTE) than the matrix, and have a Mode I As-fabricated specimens and specimens heat-treated for 3000 h at fracture energy twice the Mode I fracture energy of the ma- 982 C(1800 F) were characterized using optical microscopy trix.9.2 General Electric has developed oxide/oxide CMCs with SEM(Model 360FE, Leica), and TEM operating at 200 kV(Model 2000 FX, JEOL Tokyo, Japan ). SEM specimens and TEM thin sections were prepared using a method described elsewhere. TEM specimens were mounted on copper grids and ion milled at F. Zok--contnbuting editor 7kV, with the last 15 min at 4 kv. Most specimens were carbon coated. Some of the TEM specimens were observed without carbon coating so that the fine structure of the matrix porosity could be better resolved. For these specimens, care was taken to uscript No, 188004. Received January 17, 2002; approved January 28, 2003 align and stigmate the microscope for the particular condenser lens ber, American Ceramic societ
Journal of the American Ceramic Sociery-Zawada et al (B) Mechanical Test Apparatus: A horizontal servohydraulic rous SiO A micrograph taken 45 to the warp and fill fibers is machine with rigid hydraulic clamping grips and quartz lamp heating shown in Fig. I. Parallel arrays of cracks in the matrix were was used for the tension, in-plane shear, creep re and fa perpendicular to the cloth layers. The crack spacing was wider in tests. Test control, data acquisition, and interactive data analysis wa the matrix-rich regions and smaller within the fiber tows, Presum- done using the MATE on an IBM-compatible n ably, these cracks formed by matrix shrinkage during sintering that Ten, uter (PC) linked to the test frame by an analog-to-digital board. was constrained by the cloth layers, similar to that observed for perature was measured using five S-type thermocouples bonded constrained sintering of films" and around inclusions 27-29These to each specimen with an alumina-based ceramic adhesive. A detailed cracks formed under relatively low stresses that could not exceed description of the test equipment was given elsewhere. The inter- twice the sintering stress. 0 The fiber-matrix interfaces and the laminar strength and in-plane shear tests were done using a standard fibers themselves were not cracked vertical servohydraulic test machine with a box furmace that used iter elements g/cm' using the Archimedes method and 3.65 g/cm'using the Monotonic tests consisted of testing two to three test specimens pycnometer, with a standard deviation of =0.06. An average bulk per condition. Fatigue and creep rupture testing consisted of testing density of 2.90 g/cm' was calculated from the immersion meas- only one test specimen at each stress level investigated. It was urements. The average fiber volume fraction was -30.7%+ recognized by the authors that this was an extremely limited set of 2.29%0, The SiO, and AlO, volume fractions of the matrix were data on which to make scientific observations. However, the not measured, but the expected proportions following processing objective was to explore the boundaries of mechanical behavior fo oxide/oxide CMCs and to determine if they warranted a more Therefore, the interconnected porosity in the entire composite was rigorous investigation for use in aerospace applications. 24%, or 35%o in the matrix alone. Al,O, occupied 5I vol% of the (C) Monotonic Loading: All tension and compression tests matrix and SiO, 14%. Because this matrix porosity was concen were performed using stroke control with a 0.05 mm/s displace trated in the SiO,, initial SiO, porosity was 71%.However, ment rate. Tension tests were performed using dogbone test sintering shrinkage cracks accounted for some of this porosity; the specimens, whereas the compression tests used straight-sided test remainder was finely distributed in the SiO, that cemented the pecimens. Tension tests were also done at 1000 and 100%C in Al, O, grains together. The high specific surface area measured by air with an -1. 5 cm hot zone BET nitrogen adsorption of 25 to 35 m'g(average of 31.25+ In high-temperature tests, each specimen was ramped to the test 3.98 m2/g) was consistent with a large amount of fine intercon- emperature in 15 min and then equilibrated for20 min: the nected porosity. In contrast to N610/AS, dense glass-ceramic stress then was ramped up until the specimen failed. For in-plane composites of Nicalon/MAS were found to have a specific surface shear measurement, tension tests were performed using test spec- area of <0.3 m"/s imens with =45 fiber orientations, Residual room-temperature The mixture of a-AlzO, and amorphous SiO that forms the tensile strength was measured on all specimens that reached run-out during fatigue and creep testing Below 1200 C, the diphasic Al,Ox-SiO, mixture is expected to be (D) Interlaminar Strength: A compressive load was applied kinetically stable with respect to mullite formation. 1-2 No evi- to a notched specimen of uniform width using ASTM standard test dence of mullite formation in the matrix has been found usin practice D3846 ("Standard Test Method for In-Plane Shear TEM or XRD Strength of Reinforced Plastics, Designation No, D3846. ASTM Book of Standards, Vol. 08.02, ASTM International, West Con 2) Microstructure and Residual Stress shohocken, PA). The specimens failed in shear between two centrally located notches machined halfway through the specimen No interphase was present between the matrix and fiber thickness at a standard distance apart on opposing faces. Tests same porous SiO, that bonded matrix Al O, grains together also were done in stroke control with a rate of I mm/min at 23,538, 610 fiber grain size was 0.11+0.03 um. Fine porosity was distributed unifor ormly through the fiber, without preference for (E) Cyclic Tension: Cyclic tension (fatigue) tests were intragranular or intergranular location. The average pore diameter performed on dogbone test specimens at room temperature and was 9.8+4.2 nm. Fiber porosity did not coarsen after 3000 h at 000C. The tests were conducted in load control with a load ratio 982.C(1800F of 0.05(R=omin/omax). Room-temperature tests were cycled at I Hz for the first 100) cycles and then at 5 Hz for an additional 900000 cycles, or until failure. The 1000"C fatigue tests were done at a frequency of I Hz and were allowed to run for 100 00C cycles. The 1000"C cycle count value was chosen to roughly duplicate the number of loadings expected in aerospace applica- tions at that temperature. Fatigue run-out limits were defined to be the stress level at or slightly above the highest run-out stress for the J-N relationships. Changes in hysteresis energy density and elastic modulus versus fatigue cycle were calculated for each fatigue test to assess the extent of damage that occurred to the compo (F) Creep Rupture: Creep rupture tests on dogbone test specimens were conducted under load control at 75. 100. 125, and 135 MPa at 1000 C. The 1 100 C tests were performed at 50 and 75 MPa. The run-out condition was defined as 100 h for both test temperatures. The specimens were ramped at -10 MPa/s to the test stress. Data were recorded during loading up to the test load 美 and from the time the specimen reached the test stress so that total strain and creep strain could be calculated. IlL. Results and Discussion N610/AS: A: (l) General Microstructure Fig. 1. Optical micrograph of Nextel 610/AS showing a cross section Fiber distribution in the matrix was not uniform. The matrix aken 45 to the warp direction of the fibers. Micrograph shows matrix-rich a homogenous mixture of a-Al,O, grains cemented together by regions between plies, matrix cracks, and fiber distribution
June 2003 Characterization and High-Temperature Mechanical Behavior of an Oxide/Oxide Composite No Heat-Treatmen Fig. 2. High-resolution TEM micrograph showing the complete absence of an interphase between the fiber and matrix in Nextel 6I0/AS (b In as-fabricated specimens, the matrix Al2O3 grain 19+0.08 um in diameter, with an average aspect ratio of 1.54 The SiO, between the Al, O, grains was amorphous with a large volume fraction of mostly interconnected porosity(Fig. 3). The SiO, pore volume fraction was so high that it was unclear whether the radii relevant for coarsening analysis were those of the pores or those of the SiO,, The pore radii had a lognormal distribution with an inverse log average of 0.58+0.18 log(nm)(Fig. 4(a)). In pecimens heat-treated for 3000 h at 982C(1800%). the Al,o natrIx ns averaged 0. 17+0.07 um in diameter ratio of 1. 43. Statistically, there was no change in AlO, grain size or shape with heat treatment at 982C. However, the average pore radii in the SiO, increased by a large amount(Fig 4(b). The SiO pore radii again had a lognormal distribution, but with an inverse log average of 1.04+0.31 log(nm). In some cases the Sio formed a dense coating around Al,O, grains(Fig 4(b)) Fig. 4.(a) TEM micrograph of as-processed Nextel 610/AS matrix showing the very fine porosity in the matrix SiO.(b) TEM micrograph of Nextel 610/AS showing coarsening of the porosity in the matrix SiO, after heat treatment at 982 C for 3000 h The pore-coarsening rate can be analyzed to obtain an imate SiO, viscosity. The pores are interconnected and tl radii distributions are lognormal rather than large-sized normal distributions typical for steady-state coarsening. This uggests that a"cylinder model for the porosity". is appropri ate. The tightly packed Al,O, grains that surround the SiO, and the fibers that surround the matrix constrain the porous SiO, and do not allow densification. Therefore, coarsening occurs instead of sintering. A pore-coarsening calculation using the as-fabricated and 3000 h/982"C coarsening data has been done by Scherer. 5 The calcu- lation assumes a SiO, surface energy of 0.28 J/m2. A SiO, viscosity of 1. 1 x 10 Pas is calculated, which is about the viscosity expected for SiO, with 0.. 12% hydroxyl groups. The"sintering stress" found from the same calculation is16 MPa. Because, in this case, sintering is constrained, the sintering stress is a real tensile stress in the porous SiO, supported by a small compressive stress at the Al,O3-Al,O, grain contacts in the matrix. The sintering stress can be present only at high temperatures when mass transport processes redistribute SiO, over the 0 I um spaces between Al,O, grains. A rough estimation of the matrix stress state now can be made. 50 The high CTE difference(8 x 10-C)and modulus difference between Al, O, and porous SiO, should cause-200 MPa residual hydrostat 3. High plane stress on this matrix residual stress. Residual stresses vanish
984 Journal of the American Ceramic Society--zawada et al Vol. 86. No. 6 as the composite is brought to the processing temperature (1000 C), where, with time, a negligible tensile sintering stress in SiO, may develop Load Rate 0.05 mm/s 3) Monotonic Loading Tests (A) Tension: The results of the monotonic tension tests are E15 shown in Figs. 5(aH(c). The stress-strain response is nearly linear-elastic to failure for all specimens and all temperatures investigated. Such linear behavior suggests that there is little opL 100 MPa additional matrix cracking during loading and that fiber-matrix debonding is insignificant. In tension, fiber fracture appears to be the dominant damage mode and is typical fiber-dominated com- posite behavior. The average room-temperature ultimate tensile strength and strain to failure are 205 MPa and 0.3%6, respectively E。=70GPa (Fig. 5(a)). The average room-temperature elastic modulus is70 GPa. Such a value is significantly lower than the value of 200 GPa measured for traditional SiC-fiber CMCs of the same fiber volume fraction. High-temperature tensile strength decreases only 15%o Strain(%) from room temperature, and the modulus is relatively unchanged. The short-term tensile behavior of this composite does not chang significantly for temperatures up to 1100oC The fracture surfaces were examined using low-magnification T=1000°c Load Rate =0.05 mm/s optical microscopy. The room-temperature specimens had irregu 200 lar fracture paths and " fiber bundle pullout"with matrix material =173MF remaining on the fiber surfaces. These were common fracture features in all the room-temperature tension tests(Fig. 6(a)), but re significantly less common in the high-temperature tests b)). Fiber bundle pullout was not caused by the fracture ms that caused single-fiber pullout in CMCs with weak 100 atrix interfaces, The apparent fiber bundle pullout was C. =85 MPa simply due to weak matrix material falling apart during failure fracture surfaces did not mate, Fiber failure was intergranular, as bserved in tension tests of other CMCs with the same fiber E=80 GPa One very important observation was that the room-temperature fracture surfaces were substantially more jagged. The crack path deviated substantially across the width as well as along the length of the specimen, and the fracture surface extended -10 mm along Strain (% the length of the gauge section. The fracture paths stepped across the specimen in the thickness direction, and some of the bundle lengths were several millimeters in length. In contrast, the elevated-temperature fracture surfaces were significantly more T=1100°c planar, with only very short bundle lengths evident on the fracture Load Rate 0.05 surfaces. In high-temperature tests, the fracture surface was con- 200 =171 MPa fined to only 3-4 mm along the specimen gauge length. The distinct change in fracture path morphology suggested a funda mental change in damage mechanisms with temperature. It was temperature, as discussed earlier. 100 The stress versus strain behavior of this N6lO/AS composite is very similar to composites made by other manufacturers. A Nextel 610 composite with a matrix of 80%o mullite and 20% Al,O opL=56 MPa (c) demonstrates tensile behavior similar to what has been measured E 75 GPa in this investigation Tensile strengths were found to be - 200 MPa, and the stress versus strain traces were essentially linear to failure. The room-temperature tensile properties of several oxide/ 02 0.3 oxide CMCs were measured by one of the authors in a separate Strain(%) investigation. Each of the oxide/oxide systems investigated had similar stress versus strain traces that were essentially linear to Fig. 5. (a) Tensile stress versus strain behavior for three test specimens of failure, Differences in modulus and ultimate strength were primar- 8HSW Nextel 61(/AS tension tested at 23C.(b) Tensile stress versus ily attributed to the volume fraction of fibers used and type of strain behavior for two test specimens of 8HSW Nextel 6I0/AS tension fiber. A review of the mechanical properties of porous-matri ceramic composites was given by Zok and Levi. One important OAS tested at 1 100°C statement in the review was that, because of the inherent nature of a porous low-energy matrix, much of the stress-strain behavior was controlled by the fiber and that ultimate strength was a Fig. 7 as shear stress versus shear strain. The stress-strain respons function of how much the fibers were damaged became nonlinear at-50% of ultimate stress. An average shear U (B) In-Plane Shear and Interlaminar Strength: Stress- strength of 27 GPa was observed for three tests. On the specimen in room-temperature in-plane shear testing on faces, there was a dominant shear band at a 45 angle across the specimen + fiber orientations was very different from gauge section that was-10 mm wide(7-10 fiber tows). At the specimen 0/90 fiber orientations. The results are plotted in edge of the failed specimens, the fiber tows along the shear band
June 2003 Characterization and High-Temperature Mechanical Behavior af an Oxide/Oxide Composite T=23° Load Rate 0.05 mm/s E苏95 uss= MPa 17 MPa G Shear Strain (%) Fig. 7. Shear stress versus shear strain at room temperature for 8HSw Nextel 6I0/AS tested using a +45 fiber orientation in the tensile 10 mm The in-plane strength increased slightly with ture(Fig. while the interlaminar shear strength increased significantly as the test temperature increased(Fig. 9). The low in-plane and interlaminar shear strength should have made this composite less sensitive to failure from center or edge notches. Increased shear strengths with increased temperature again suggested that the viscosity of the matrix SiO, and residual stress state may have affected the observed behavior (4) Cyclic Tension Results of the room-temperature and 1000"C fatigue tests are shown in Figs. 10(a)and(b), respectively. The room-temperature mit is -170 MPa. which averd room-temperature tensile strength. The 1000C fatigue limit is 150 MPa which is-85% of the average 1000C tensile strength. The room-temperature fatigue performance is similar to other CMCs many of which exhibit fatigue limits within 5%-20% of the 10m average tensile strength. However, the fatigue behavior at elevated temperature for this CMC is unlike that observed for any other CMC. CMCs with an interphase consisting of carbon or BN ypically exhibit run-out at stress levels of only 75-120 MPa and these run-out stress levels are always closely associated with the proportional limit and development of matrix crack ing. Once the matrix is cracked, there is rapid ingress of oxygen into the composite, resulting in oxidation of the Fig. 6. (a) Fracture surface of Nextel 6lO/AS tension tes Fracture surface shows extensive length, and macrofeatures associat the fiber tow Fracture surface of Nextel 6I0/: tested at1O00° 5 surface is very flat and is confined to a very narrow zone along the length ±45 Tension Test of the specimen. were pulled -I mm into the composite, while the fiber tows 90o to the shear band remained in place. There was extensive matrix damage, The fragmentation of the matrix allowed the tows to rotate as they withdrew. This may have promoted nonlinear stress-strain behavior beyond the onset of strain localization. The shear strengths were very low compared with the 0%/90%orienta- g10 tion and several times lower than those of Nicalon/SiC CMCs (100 MPa) measured using the losipescu test fixture, Very similar behavior was observed for a N610/mullite+ Al,O, CMC tested using the +45 fiber orientation coupon loaded in tension. 4 Maximum stress levels were measured to be-63 MPa, whereas Temperature(°c) this investigation measured an average failure stress of 54 MPa Photographs of a fractured +45 tensile specimen were nearly Fig. 8. Interlaminar shear strength versus temperature for SHSW Nextel identical to the fracture features produced in this investigation 6lOVAS. Data were generated using a double-notched compression specimen