JAm. Ceram Soc,879l1726-173302004) urna Tensile and Stress-Rupture behavior of SiC/SiC Minicomposite Containing Chemically Vapor Deposited Zirconia Interphase Hao Li, Gregory N. Morscher, Jinil Lee, and Woo Young Lee Department of Chemical, Biochemical and Materials Engineering, Stevens Institute of Technology, Hoboken, New Jersey 07030 Ohio Aerospace Institute and NASA Glenn Research Center, MS 106-5, Cleveland, Ohio 44 135 The tensile and stress-rupture behavior of sic/Sic minicom failure load. These results suggested that a single CVd-zro osite containing a chemically vapor deposited(CvD)Zro2 layer, if properly prepared, might be more useful than the mult interphase was evaluated. Fractographic analyses showed that layer coating because of (1)no fiber strength degradation caused in situ fiber strength and minicomposite failure loads were by the first SiO, layer,(2) processing simplicity, and (3)elimi- strongly dependent on the phase contents and microstructure nation of Sio, -ZrO, reaction as a source of thermochemical of the Zro2 interphase. When the zro2 interphase structure instability at the interface region ossessed a weakly bonded interface within the dense ZrO, The objectives of this study were(i )to understand the effects of nterphase coating layer, the interphase sufficiently protected the phase contents and microstructure of the zro, coating on the he fiber surface from processing degradation and promot tensile behavior of SiC/SiC minicomposite samples (i.e, SiC/ matrix crack deflection around the fibers. With this weakl ZrO,/SiC)at room temperature,(ii) to investigate the stress- onded interphase, the stress-rupture properties of sic/Sic rupture properties of a selected number of SiC/ZrO,/SiC minicom- minicomposite at 950 and 1200C appeared to be controlle posite samples at elevated temperatures, and (iii) to formulate the by fiber rupture properties, and compared favorably to those basic criteria for further optimization of the ZrO2 interphase previously measured for state-of-the-art BN fiber coatings concept for environmentally durable SiC/SiC composites . Introduction IL. Experimental Procedure and Analysis A WEAK"interface or interphase layer is required between the (Coating and Minicomposite Preparation reinforcing fiber and the matrix in SiC/Sic ceramic matrix Three batches of CVD-ZrO, coating were prepared in a hot- common interphases, pyrocarbon( PyC)and boron nitride( BN), samples from these batches were used for this study Hi-Nicalon 4 composites for strong and tough mechanical behavior. The most all CVD reactor. as described elsewhere. ,A selected number of oxidize at relatively low temperatures(>-500oC). The oxidation fiber tows(Nippon Carbon Co, Ltd, Japan)were wrapped around problem is considered to be one of the major problems remaining a fused Sio, holder to fabricate about nine fiber tow samples(25 to be solved before SiC/Sic composites can be reliably introduced cm long) for each CVD experiment. For Batch I experiments, into next-generation aircraft engines, power generation turbines, deposition time(5, 10, 20, 30, 60, and 120 min) was varied while space vehicles, and other industrial applications. cOnsiderable other processing conditions were kept constant. The detailed effort has been exerted to improve the interphase oxidation experiment conditions of the sample deposited for 120 min resistance: however. most of this research has been aimed at W1-2h) are shown in Table I. The same CVD system and base line edifying the PyC or BN interpha procedure were used for Batch II and Batch Ill experiments exc Lee et al.6 reported that SiC/SiC minicomposite samples con- that the pressure and the precursor flow rates were different, as taining a Sio2/ZrO/SiO, oxide coating prepared by summarized in Table I. For each deposition experiment, the leak vapor deposition(CVD)exhibited graceful tensile failure b rate of air into the CVd system was measured using a pressure and extensive crack deflection within the coating. The m auge and isolation valve located between the cvd chamber and the vacuum pump. This is described in detail elsewhere, espe- CVD-ZrO, coating revealed that tetragonal-to-monoclinic phase ment is reported in Table I. Some Batch I samples and all Batch Il transformation is most likely the key mechanism responsible for the weak interface behavior of the SiO, /ZrO,/SiO, fiber coating.7 tows(BN)obtained from NASA Glenn Research Center(Cleve When longitudinal cracks formed within the ZrO2 layer as a land, OH), were infiltrated with a Sic matrix by chemical vapor result of the phase transformation, the Sic fiber tows did not infiltration(CVI) at BF Goodrich(Brecksville, OH) exhibit strength degradation regardless of coating thickness in the range of o I um to several micrometers. In contrast, a single Sio, (2) Room-Temperature Tensile Tests and fractographic layer(as thin as 100 nm) significantly decreased the fiber tow Analyses The coated fiber tows and minicomposite samples were tested at room temperature to measure their tensile failure loads. All of the uin--contributing editor fiber tow tests and most of minicomposite tests were conducted with the same method previously used for fiber strength measure- ments.Essentially, 25 mm of tow or minicomposite ends were mounted in epoxy attached to cardboard tabs. The minicomposite Manuscript No. 10030. Received March 9, 2003; approved May In an,Force ends encased in epoxy were gripped with mechanical clamps and tested in tension. The gauge length was the length of tow or minicomposite not in epoxy, i.e., the length of tow or minicom- tOhio Aerospace Institute and NASA Glenn Research Center posite between the two epoxy-encased ends. The
Tensile and Stress-Rupture Behavior of SiC/SiC Minicomposite Containing Chemically Vapor Deposited Zirconia Interphase Hao Li,* Gregory N. Morscher,* ,† Jinil Lee, and Woo Young Lee* Department of Chemical, Biochemical and Materials Engineering, Stevens Institute of Technology, Hoboken, New Jersey 07030 Ohio Aerospace Institute and NASA Glenn Research Center, MS 106-5, Cleveland, Ohio 44135 The tensile and stress-rupture behavior of SiC/SiC minicomposite containing a chemically vapor deposited (CVD) ZrO2 interphase was evaluated. Fractographic analyses showed that in situ fiber strength and minicomposite failure loads were strongly dependent on the phase contents and microstructure of the ZrO2 interphase. When the ZrO2 interphase structure possessed a weakly bonded interface within the dense ZrO2 interphase coating layer, the interphase sufficiently protected the fiber surface from processing degradation and promoted matrix crack deflection around the fibers. With this weakly bonded interphase, the stress-rupture properties of SiC/SiC minicomposite at 950° and 1200°C appeared to be controlled by fiber rupture properties, and compared favorably to those previously measured for state-of-the-art BN fiber coatings. I. Introduction A“WEAK” interface or interphase layer is required between the reinforcing fiber and the matrix in SiC/SiC ceramic matrix composites for strong and tough mechanical behavior. The most common interphases, pyrocarbon (PyC) and boron nitride (BN), oxidize at relatively low temperatures (500°C). The oxidation problem is considered to be one of the major problems remaining to be solved before SiC/SiC composites can be reliably introduced into next-generation aircraft engines, power generation turbines, space vehicles, and other industrial applications.1,2 Considerable effort has been exerted to improve the interphase oxidation resistance; however, most of this research3–5 has been aimed at modifying the PyC or BN interphase. Lee et al.6 reported that SiC/SiC minicomposite samples containing a SiO2/ZrO2/SiO2 oxide coating prepared by chemical vapor deposition (CVD) exhibited graceful tensile failure behavior and extensive crack deflection within the coating. The minicomposite possessed a reasonable degree of load-carrying capability and retained much of the composite characteristics after short-term oxidation. Our recent study of morphologic evolution of the CVD-ZrO2 coating revealed that tetragonal-to-monoclinic phase transformation is most likely the key mechanism responsible for the weak interface behavior of the SiO2/ZrO2/SiO2 fiber coating.7 When longitudinal cracks formed within the ZrO2 layer as a result of the phase transformation, the SiC fiber tows did not exhibit strength degradation regardless of coating thickness in the range of 0.1 m to several micrometers. In contrast, a single SiO2 layer (as thin as 100 nm) significantly decreased the fiber tow failure load.8 These results suggested that a single CVD-ZrO2 layer, if properly prepared, might be more useful than the multilayer coating because of (1) no fiber strength degradation caused by the first SiO2 layer,8 (2) processing simplicity, and (3) elimination of SiO2–ZrO2 reaction as a source of thermochemical instability at the interface region. The objectives of this study were (i) to understand the effects of the phase contents and microstructure of the ZrO2 coating on the tensile behavior of SiC/SiC minicomposite samples (i.e., SiC/ ZrO2/SiC) at room temperature, (ii) to investigate the stressrupture properties of a selected number of SiC/ZrO2/SiC minicomposite samples at elevated temperatures, and (iii) to formulate the basic criteria for further optimization of the ZrO2 interphase concept for environmentally durable SiC/SiC composites. II. Experimental Procedure and Analysis (1) Coating and Minicomposite Preparation Three batches of CVD-ZrO2 coating were prepared in a hotwall CVD reactor, as described elsewhere.7,9 A selected number of samples from these batches were used for this study. Hi-NicalonTM fiber tows (Nippon Carbon Co., Ltd., Japan) were wrapped around a fused SiO2 holder to fabricate about nine fiber tow samples (25 cm long) for each CVD experiment. For Batch I experiments, deposition time (5, 10, 20, 30, 60, and 120 min) was varied while other processing conditions were kept constant. The detailed experiment conditions of the sample deposited for 120 min (WI-2h) are shown in Table I. The same CVD system and base line procedure were used for Batch II and Batch III experiments except that the pressure and the precursor flow rates were different, as summarized in Table I. For each deposition experiment, the leak rate of air into the CVD system was measured using a pressure gauge and isolation valve located between the CVD chamber and the vacuum pump. This is described in detail elsewhere,10 especially the important role of minimizing air leaks on promoting the nucleation of tetragonal ZrO2. The air leak rate for each experiment is reported in Table I. Some Batch I samples and all Batch II and Batch III samples, along with BN-coated Hi-Nicalon fiber tows (BN) obtained from NASA Glenn Research Center (Cleveland, OH), were infiltrated with a SiC matrix by chemical vapor infiltration (CVI) at BF Goodrich (Brecksville, OH). (2) Room-Temperature Tensile Tests and Fractographic Analyses The coated fiber tows and minicomposite samples were tested at room temperature to measure their tensile failure loads. All of the fiber tow tests and most of minicomposite tests were conducted with the same method previously used for fiber strength measurements.7 Essentially, 25 mm of tow or minicomposite ends were mounted in epoxy attached to cardboard tabs. The minicomposite ends encased in epoxy were gripped with mechanical clamps and tested in tension. The gauge length was the length of tow or minicomposite not in epoxy, i.e., the length of tow or minicomposite between the two epoxy-encased ends. The minicomposite R. Naslain—contributing editor Manuscript No. 10030. Received March 9, 2003; approved May 10, 2004. Supported by NSF (DMR-GOALI 9971623) with cofunding from the Air Force Research Laboratory (AFRL). *Member, American Ceramic Society. † Ohio Aerospace Institute and NASA Glenn Research Center. J. Am. Ceram. Soc., 87 [9] 1726–1733 (2004) 1726 journal
eptember 2004 Tensile and Stress-Rupture Behavior of siC/SiC Minicomposite 1727 Table 1. Processing Conditions and Tensile Test Result Air leak Tensile Temperature failure loads Correspondi ( Pa/min) failure loads(N) WI-2h 12012040/1.5 120 0.4 118(27)MW-2h(a):288 Mw-2hb):118 12012040/1.5 MSII-1: brittle MSI-2:10 WII-2h12012040/1.51.3 0.4 MWIll-2h: 133 WII-3h12012040/1.51.3180 1050 0.4 MWI1-3h:132↑ 11014)MBN:114(1)1 tGauge length default: 2.5 cm. Numbers in parentheses are standard deviations. Gauge length 5 cm. ' Failed during sample handling. f Gauge lengt sample containing WI-2h(MWl-2h)was tested with a The stress-rupture tests were performed with a simple dead- gauge length of 5 cm; however, it failed at mount. This eight load stress-rupture rig at 950% or 1200.C. The distance pecimen was remounted and retested with gth of 2.5 from the top to the bottom of the outside walls of the furnace was cm(in Table 1). 35 mm. The hot zone was -14 mm. The load was applied at roo Typical fracture morphology of the SiC fiber consists of mirror temperature The temperature was increased with a rate of 100C/ (smooth region), mist (region with misty appearance between min to the desired test temperature, and the time was monitored mirror and hackle), and hackle(region of multiple fracture plane until failure unless the failure did not occur after significantly long The minicomposite in situ fiber strength(S) was estimated based time, e.g., over 100 h, in which case the experiment was stopped on the empirical relationship between the fiber strength and mirror radius for the sic fiber (4 Microstructure Characterization A A field-emission scanning electron microscope(FEG-SEM, (1) LEO DSM 982, LEO Electron Microscopy, Inc, Thornwood, NY) was used to study coating morphology and minicomposite fracture here A is the fracture mirror constant and r is the fracture mirror surface. Transmission electron microscope (TEM), electron dif- fraction, and Raman microprobe were also used for determination Morscher's work. Around 50 fibers were measured to calculate or the middle location(12.5 cm)of the fiber tow actur.Unless adius. The value of a varies over the range of 20-4. 0 MPa/m of the phase contents and morphology of Batch I samples for the SiC fiber. 1 1-17 A value of 2.5 MPa/m was used following noted otherwise, SEM micrographs were from the the in situ fiber strength. Minicomposite failure loads were also estimated with a fractographic analyses based on a global load sharing(GLS)model. Ia (I Room-Temperature Tensile behavior () Tensile Stress-Rupture Tests The tensile failure load was used to the tensile The room-temperature tensile tests were used as a screening behavior of a single tow of Zro2-coated rocess for producing a small number of minicomposite samples without the SiC matrix)and a single tow mir site with the or stress-rupture tests at elevated temperatures. Only Batch Il matrix (i.e, SiC/ZrO, /SiC). It was reported- composite minicon posite samples were tested for their tensile stress-rupture strength is independent of gauge length, when a sample is les. The tensile stress-rupture testing procedure was iden- sufficiently long, in the context of the GLs assumption. Therefore, tical to that used in a previous study: a minicomposite san for most samples, we expect that their tensile failure loads would was precracked at room te to a constant-load stress-rupt e s a t a re in tension and then subjected not be significantly affected by the gauge length. Table I shows est at an elevated temperature art of the room-temperature tensile test data obtained for the The reason for precracking is that it is essential that transverse samples before and after the Cvi-siC matrix infiltration. More matrix cracks are present during stress-rupture testing to determine detailed room-temperature tensile test results of the fiber tows the extent of strength reduction that occurs as a result of fiber/ coated from Batch I and Batch Il experiments could be found interphase oxidation. The volume fraction of fibers in the elsewhere. As expected, MBN(SIC/BN/SiC)exhibited a fairly minicom sites was relatively low (-0. 15). In fact, the stress high average failure load, which was similar to that of the required to form matrix cracks in the unidirectional minicomposite BN- coated fiber tows(SIC/BN). a slightly higher failure load was usually exceeds the stress required to fail the matrix-cracked eported for similar SiC/BN/SiC samples(PBN-HN and 3MBN minicomposite under stress-rupture conditions if the CVI matrix is HN),142N. In contrast, SiC/ZrO, and SiC/ZrO,/SiC samples already cracked. In other words, if no precrack existed, no exhibited a range of failure loads. Batch I(Wl-2h) and Batch Ill stress-rupture degradation would occur because the fiber/inter (WIll-2h and-3h) fiber tow failure loads were comparable to hase region would not be exposed to the environment and the BN-coated fiber tows, whereas Batch Il (Sin) coated fiber tows ep-resistant CVI-SiC matrix would carry most of the applied were very weak. Minicomposite failure loads followed the same trend. However, the strength of Batch I minicomposites varied A room-temperature tensile test was performed to precrack the along the length of the minicomposite. For example, the Batch I CVI-SiC matrix using a universal testing machine(Model 4502 pecimen mwl-2h initially failed at the edge of the epoxy, 2.5 Instron, Canton, MA). Acoustic emission(AE) monitoring has cm from the center of the minicomposite length, at a load of 28 N been shown to be a good method for Loring matrIx crack The same specimen was remounted with a 2.5 cm gauge length and formation during mposite testing2223 and was used in the ailed at 118N in the gauge region. The strength difference for the same manner in this study. An AE transducer (pico model was most likely due to coating nonunifor hysical Acoustics Corp., Princeton, N) was attached to the epox mity along the length of the fiber tow region of the minicomposite mount and ae was monitored Locan An intensive TEM study was previously conducted for the 20, Physical Acoustics Corp., Princeton, NJ) during the tensile Batch I ZrO, coating. The study revealed that the Batch I Zro test. Usually, a load of at least 110 N was required to ensure a coating, before and after CVI-SiC process, usually consisted of significant crack density(at least 0.5 crack/mm) four distinct regions: an inner laver retained on the fiber surface. a
sample containing WI-2h (MWI-2h) was initially tested with a gauge length of 5 cm; however, it failed at the epoxy mount. This specimen was remounted and retested with a gauge length of 2.5 cm (in Table I). Typical fracture morphology of the SiC fiber consists of mirror (smooth region), mist (region with misty appearance between mirror and hackle), and hackle (region of multiple fracture plane). The minicomposite in situ fiber strength (S) was estimated based on the empirical relationship between the fiber strength and mirror radius for the SiC fiber: S A rm (1) where A is the fracture mirror constant and rm is the fracture mirror radius. The value of A varies over the range of 2.0–4.0 MPa/m1/2 for the SiC fiber.11–17 A value of 2.5 MPa/m1/2 was used following Morscher’s work.14 Around 50 fibers were measured to calculate the in situ fiber strength. Minicomposite failure loads were also estimated with a fractographic analyses based on a global loadsharing (GLS) model.18–21 (3) Tensile Stress-Rupture Tests The room-temperature tensile tests were used as a screening process for producing a small number of minicomposite samples for stress-rupture tests at elevated temperatures. Only Batch III minicomposite samples were tested for their tensile stress-rupture properties. The tensile stress-rupture testing procedure was identical to that used in a previous study:14 a minicomposite sample was precracked at room temperature in tension and then subjected to a constant-load stress-rupture test at an elevated temperature. The reason for precracking is that it is essential that transverse matrix cracks are present during stress-rupture testing to determine the extent of strength reduction that occurs as a result of fiber/ interphase oxidation.14 The volume fraction of fibers in the minicomposites was relatively low (0.15). In fact, the stress required to form matrix cracks in the unidirectional minicomposite usually exceeds the stress required to fail the matrix-cracked minicomposite under stress-rupture conditions if the CVI matrix is already cracked.14 In other words, if no precrack existed, no stress-rupture degradation would occur because the fiber/interphase region would not be exposed to the environment and the creep-resistant CVI-SiC matrix would carry most of the applied load. A room-temperature tensile test was performed to precrack the CVI-SiC matrix using a universal testing machine (Model 4502, Instron, Canton, MA). Acoustic emission (AE) monitoring has been shown to be a good method for monitoring matrix crack formation during minicomposite testing22,23 and was used in the same manner in this study. An AE transducer (pico model, Physical Acoustics Corp., Princeton, NJ) was attached to the epoxy region of the minicomposite mount and AE was monitored (Locan 320, Physical Acoustics Corp., Princeton, NJ) during the tensile test. Usually, a load of at least 110 N was required to ensure a significant crack density (at least 0.5 crack/mm). The stress-rupture tests were performed with a simple deadweight load stress-rupture rig14 at 950° or 1200°C. The distance from the top to the bottom of the outside walls of the furnace was 35 mm. The hot zone was 14 mm. The load was applied at room temperature. The temperature was increased with a rate of 100°C/ min to the desired test temperature, and the time was monitored until failure unless the failure did not occur after significantly long time, e.g., over 100 h, in which case the experiment was stopped. (4) Microstructure Characterization A field-emission scanning electron microscope (FEG-SEM, LEO DSM 982, LEO Electron Microscopy, Inc., Thornwood, NY) was used to study coating morphology and minicomposite fracture surface. Transmission electron microscope (TEM), electron diffraction, and Raman microprobe were also used for determination of the phase contents and morphology of Batch I samples.7 Unless noted otherwise, SEM micrographs were from the fracture surface or the middle location (12.5 cm) of the fiber tows. III. Results (1) Room-Temperature Tensile Behavior The tensile failure load was used to compare the tensile behavior of a single tow of ZrO2-coated fibers (i.e., SiC/ZrO2 without the SiC matrix) and a single tow minicomposite with the matrix (i.e., SiC/ZrO2/SiC). It was reported24 that minicomposite strength is independent of gauge length, when a sample is sufficiently long, in the context of the GLS assumption. Therefore, for most samples, we expect that their tensile failure loads would not be significantly affected by the gauge length. Table I shows part of the room-temperature tensile test data obtained for the samples before and after the CVI-SiC matrix infiltration. More detailed room-temperature tensile test results of the fiber tows coated from Batch I and Batch II experiments could be found elsewhere.9 As expected, MBN (SiC/BN/SiC) exhibited a fairly high average failure load, which was similar to that of the BN-coated fiber tows (SiC/BN). A slightly higher failure load was reported for similar SiC/BN/SiC samples (PBN-HN and 3MBNHN), 142 N.14 In contrast, SiC/ZrO2 and SiC/ZrO2/SiC samples exhibited a range of failure loads. Batch I (WI-2h) and Batch III (WIII-2h and -3h) fiber tow failure loads were comparable to BN-coated fiber tows, whereas Batch II (SII) coated fiber tows were very weak. Minicomposite failure loads followed the same trend. However, the strength of Batch I minicomposites varied along the length of the minicomposite. For example, the Batch I specimen MWI-2h initially failed at the edge of the epoxy, 2.5 cm from the center of the minicomposite length, at a load of 28 N. The same specimen was remounted with a 2.5 cm gauge length and failed at 118 N in the gauge region. The strength difference for the Batch I minicomposite was most likely due to coating nonuniformity along the length of the fiber tow. An intensive TEM study7 was previously conducted for the Batch I ZrO2 coating. The study revealed that the Batch I ZrO2 coating, before and after CVI-SiC process, usually consisted of four distinct regions: an inner layer retained on the fiber surface, a Table I. Processing Conditions and Tensile Test Results Sample Flow rate (cm3 /min) Pressure (kPa) Time (min) Temperature (°C) Air leak rate (Pa/min) Tensile failure loads (N) Corresponding minicomposite and tensile failure loads† CO (N) 2 H2 Ar/ZrCl4 WI-2h 120 120 40/1.5 4 120 1050 0.4 118(27)‡ MWI-2h(a): 28§ MWI-2h(b): 118 SII-1 120 120 40/1.5 4 35 1050 2 27 MSII-1: brittle¶ SII-2 200 100 80/3 1.3 30 1050 2 25 MSII-2: 10 WIII-2h 120 120 40/1.5 1.3 120 1050 0.4 126 MWIII-2h: 133†† WIII-3h 120 120 40/1.5 1.3 180 1050 0.4 138 MWIII-3h: 132†† BN - - - - - - - 110(14)‡ MBN: 114(11)‡ † Gauge length default: 2.5 cm. ‡ Numbers in parentheses are standard deviations. § Gauge length 5 cm. ¶ Failed during sample handling. ††Gauge length 10 cm. September 2004 Tensile and Stress-Rupture Behavior of SiC/SiC Minicomposite 1727
Journal of the American Ceramic Sociery-Li et al. Vol. 87. No 9 (b) 5 um 500nm Fig. 1. SEM images of CVD-ZrO,: (a)surface of WI-2h, 'and(b)cross section of SIl-1. 1 debonded region, a dense and columnar layer in the middle part of monoclinic ZrO, phase, and no debonding occurred the coating, and an outer layer which appeared to be somewhat coating when cut for SEM observation; i. e, it was strongl gorous. As shown in Fig. I(a)(W1-2h), part of the coating was ( Figs. I(b). Therefore, the five SiC/ZrO2 samples in Tab aminated most likely when the sample was prepared for the be categorized, based on their coating microstructure, SEM observation. The coating from Batch Ill experiments had a bonded"coating(Wl-2h, WIll-2h, and WIll-3h) and"strongly microstructure very similar to that from Batch I experiments, bonded"coating(SlI-I and SIl-2) except that the outer porous layer for Batch Ill samples was much Figures 2(a) and(b)show the fracture surface images of the thinner. A coating with this type of microstructure will be referred minicomposite containing the weakly bonded coating (MWI to as a"weakly bonded"coating in this paper. a different coating 2h(b)at the failure location after the second test. The coating microstructure was observed for Batch Il samples compared with morphology in the minicomposite appears the same as the as- Batch I and Batch Ill samples. This coating only consisted of coated fiber tow(Fig. 2(a)). Fiber pullout was observed for the i-Nicalon mIrror CVI-SiC O um (c) (d) CVI-SiC Hi-Nicalon 500nm Fig. 2. Fracture surface of minicomposite: (a, b)MWl-2h(b),(c)MSIl-1, and(d) MSIl-2
debonded region, a dense and columnar layer in the middle part of the coating, and an outer layer which appeared to be somewhat porous. As shown in Fig. 1(a) (WI-2h), part of the coating was delaminated most likely when the sample was prepared for the SEM observation. The coating from Batch III experiments had a microstructure very similar to that from Batch I experiments, except that the outer porous layer for Batch III samples was much thinner. A coating with this type of microstructure will be referred to as a “weakly bonded” coating in this paper. A different coating microstructure was observed for Batch II samples compared with Batch I and Batch III samples. This coating only consisted of a monoclinic ZrO2 phase,25 and no debonding occurred within the coating when cut for SEM observation; i.e., it was strongly bonded (Figs. 1(b)). Therefore, the five SiC/ZrO2 samples in Table I could be categorized, based on their coating microstructure, as “weakly bonded” coating (WI-2h, WIII-2h, and WIII-3h) and “strongly bonded” coating (SII-1 and SII-2). Figures 2(a) and (b) show the fracture surface images of the minicomposite containing the weakly bonded coating (MWI- 2h(b)) at the failure location after the second test. The coating morphology in the minicomposite appears the same as the ascoated fiber tow (Fig. 2(a)). Fiber pullout was observed for the Fig. 1. SEM images of CVD-ZrO2: (a) surface of WI-2h,7 and (b) cross section of SII-1.21 Fig. 2. Fracture surface of minicomposite: (a,b) MWI-2h(b), (c) MSII-1, and (d) MSII-2. 1728 Journal of the American Ceramic Society—Li et al. Vol. 87, No. 9
eptember 2004 Tensile and Stress-Rupture Behavior of siC/SiC Minicomposite 1729 Table Il. Properties of Minicomposite Tested and Estimated Number of Tensile failure load (N diameter(um) fibers per to s(GPam*↑ a(gRamm Estimated 2.0/2.8 PBN-HN2 13.3 3.1/3.3 3.2/3.2 142(9)° MWI-2h(b) 2.5/6.3 2.5/6.5 MWl-2h(a) 490 n situ fiber strength. Corrected ength. ' Number in parentheses are standard deviations. 'Among 50 fibers observed, only 20 fibers with discernable mirr incOmposite terms of in situ fiber strength and tensile the MW1-2h(b) had a higher Weibull nd pullout occurred within the arrow distribution of fiber strengt onded region formed during the CVD process(Fig. 2(a). In a few he estimation of the tensile failure load for the minicomposite cases, when the outer porous layer was too thick for the CVI-SiC samples based on fractographic analyses and the Gls model was matrix to penetrate, sliding occurred between the outer porous fairly consistent with measured values. The interfacial shear stress layer and the middle dense layer(Fig. 2(b). Whether fiber pullout estimated for MWl-2h(b)was about 140 MPa.25 ccurred or not, most fibers had an easily identified small mirror indicative of high fiber strength. The fracture surfaces of Batch Ill minicomposites(Mwlll-2h and MwIll-3h) were essentially the (2) Tensile Stress-Rupture Tests same as that in Fig. 2(a). Pullout was observed for the majority of Since none of the Batch I or II posite samples could be the fibers and fiber pullout and crack deflection occurred within uccessfully precracked, only the Batch Ill minicomposite samples the zro, interphase close to the fiber. For Batch Ill minicom were precracked and tested for stress rupture. Typical precracking ites, no fiber pullout occurred between the infiltrated porous Zro, load versus time curves(constant crosshead displacement rate)for layer and the middle dense ZrO2 layer, as in the case of Fig. 2(b both MWill-2h and MwIll-3h samples are shown in Fig 3(a). The ite samples might be attributed to the small thickness of the outer load curves are shown in Fig 3(b). An AE value greater than 100 l Such a difference between MW1-2h(b)and Batch Ill minicompos retracking load was - 1ll+5N. The cumulative Ae energy vers porous layer in WIll-2h and wIll-3h. It was also found that most was desired for precracking. The two MWill-3h samples showed bers in Mwlll-2h and Mwill-3h had a relatively small fracture nearity in the load-time curves and a higher mirror, which indicated that retained fiber strength was high cumulative AE energy compared with MWlll-2h. For MWlll-3h Figures 2(c)and(d) show the typical fracture surface and the samples, there was significant deviation from linearity between 80 interphase microstructure of the Batch Il e samples and 90 N in the load-time curves. It is evident in Fig. 3 that the For both MSIl-1 and MSIl-2. no crack deflection and no fiber deviation points are associated with increased AE activity, which is allout were observed at the fiber/matrix interface region, as the attributed to matrix cracking. 4 Table Ill summarizes the precrack ZrO, coating appeared to be strongly bonded to both Hi-Nicalon ing information of the samples used in this paper fiber and CvI-SiC matrix or the zrOz coating was not continuous The room-temperature failure loads of MWIll-2h and In some areas MWinl-3h based on a single test were 133 and 132 N respectively a quantitative fractographic analysis was conducted for PBN- as shown in Table I, which corresponded to 2004 and 1989 MPa fiber strength to the m nico posite failure load s the analysis tow ibe radius g 6. h e m for all of the samples). The tensile result is shown in Table Il. There was no significant difference stress-rupture data of MWIll-2h and MWlll-3h samples, between the Mw1-2h(b) and the other two types of SiC/BN/SiC BN interphase Hi-Nicalon reinforced SiC minicomposite = 300000 2500 200000 E 100000 MWIlI-2h-1 MWIll-25-1 w MWllI-3h-1 MWIll-3h-1 50000 MWllk3h-2 MWIlI-3h-2 20406080100120 020406080100120140 Time(sec) Load( N) (a) ig. 3. Typical (a)load versus time curves and(b) cumulative AE energy versus load curves for precracking samples
majority of the fibers. The average pullout length is about 50 m based on measurements on about 100 fibers. In most cases, fiber sliding and pullout occurred within the ZrO2 coating at the weakly bonded region formed during the CVD process (Fig. 2(a)). In a few cases, when the outer porous layer was too thick for the CVI-SiC matrix to penetrate, sliding occurred between the outer porous layer and the middle dense layer (Fig. 2(b)). Whether fiber pullout occurred or not, most fibers had an easily identified small mirror, indicative of high fiber strength. The fracture surfaces of Batch III minicomposites (MWIII-2h and MWIII-3h) were essentially the same as that in Fig. 2(a). Pullout was observed for the majority of the fibers and fiber pullout and crack deflection occurred within the ZrO2 interphase close to the fiber. For Batch III minicomposites, no fiber pullout occurred between the infiltrated porous ZrO2 layer and the middle dense ZrO2 layer, as in the case of Fig. 2(b). Such a difference between MWI-2h(b) and Batch III minicomposite samples might be attributed to the small thickness of the outer porous layer in WIII-2h and WIII-3h. It was also found that most fibers in MWIII-2h and MWIII-3h had a relatively small fracture mirror, which indicated that retained fiber strength was high. Figures 2(c) and (d) show the typical fracture surface and the interphase microstructure of the Batch II minicomposite samples. For both MSII-1 and MSII-2, no crack deflection and no fiber pullout were observed at the fiber/matrix interface region, as the ZrO2 coating appeared to be strongly bonded to both Hi-Nicalon fiber and CVI-SiC matrix or the ZrO2 coating was not continuous in some areas. A quantitative fractographic analysis was conducted for PBNHN, MBN, MWI-2h(a), and MWI-2h(b) to correlate the in situ fiber strength to the minicomposite failure load.25 The analysis result is shown in Table II. There was no significant difference between the MWI-2h(b) and the other two types of SiC/BN/SiC minicomposite samples in terms of in situ fiber strength and tensile failure load, except that the MWI-2h(b) had a higher Weibull modulus, which indicated narrow distribution of fiber strength. The estimation of the tensile failure load for the minicomposite samples based on fractographic analyses and the GLS model was fairly consistent with measured values. The interfacial shear stress estimated for MWI-2h(b) was about 140 MPa.25 (2) Tensile Stress-Rupture Tests Since none of the Batch I or II minicomposite samples could be successfully precracked, only the Batch III minicomposite samples were precracked and tested for stress rupture. Typical precracking load versus time curves (constant crosshead displacement rate) for both MWIII-2h and MWIII-3h samples are shown in Fig. 3(a). The precracking load was 111 5 N. The cumulative AE energy versus load curves are shown in Fig. 3(b). An AE value greater than 100 000 was desired for precracking. The two MWIII-3h samples showed more deviation from linearity in the load–time curves and a higher cumulative AE energy compared with MWIII-2h. For MWIII-3h samples, there was significant deviation from linearity between 80 and 90 N in the load–time curves. It is evident in Fig. 3 that the deviation points are associated with increased AE activity, which is attributed to matrix cracking.22,23 Table III summarizes the precracking information of the samples used in this paper. The room-temperature failure loads of MWIII-2h and MWIII-3h based on a single test were 133 and 132 N respectively, as shown in Table I, which corresponded to 2004 and 1989 MPa on fibers assuming that the fibers were fully loaded (500 fibers per tow, fiber radius 6.5 m for all of the samples). The tensile stress-rupture data of MWIII-2h and MWIII-3h samples, as well as BN interphase Hi-Nicalon reinforced SiC minicomposite samples Fig. 3. Typical (a) load versus time curves and (b) cumulative AE energy versus load curves for precracking samples. Table II. Properties of Minicomposite Tested and Estimated25 Sample Average fiber diameter (m) Number of fibers per tow S*(GPa)/m*† (GPa)/m‡ Tensile failure load (N) Estimated Measured MBN 13.3 490 2.0/2.8 2.1/2.7 98 114(11)§ PBN-HN12 13.3 490 3.1/3.3 3.2/3.2 140 142(9) § MWI-2h(b) 14.4 490 2.5/6.3 2.5/6.5 148 118 MWI-2h(a) 14.4 490 - - 42¶ 28 † In situ fiber strength. ‡ Corrected in situ fiber strength. § Number in parentheses are standard deviations. ¶ Among 50 fibers observed, only 20 fibers with discernable mirror were used for calculation. September 2004 Tensile and Stress-Rupture Behavior of SiC/SiC Minicomposite 1729
1730 Journal of the American Ceramic Sociery-Li et al. Vol. 87. No 9 2000 1800 ◆BN1←HN ◆BN1HN 1800 I BN2-HN 1600 n BN2HN o MWlll-2h o MWIl1-2h 1600 MWlll-3h 00 1400 1200 1000 950° 1200°C 1000 Time (hour) Time(hour (b) Fig 4. Stress-rupture data of ZrO, and BN, 14 interphase Hi-Nicalon reinforced Sic matrix minicomposite samples in air at(a)950 and(b)1200.C (BN1-HN and BN2-HN), are shown in Figs. 4(a)and(b) as the for this sample at the fracture surface. Obviously, the observed pplied stress(assuming fibers were fully loaded) versus the time fracture surface was a matrix crack and experienced long time to rupture. This corresponded to the stress on the fibers in a oxidation exposure. Fiber creep occurred and the fiber was transverse matrix crack, which was always the location of stress- oxidized. An annular SiO, layer was usually observed around the rupture failure. At 950%C, MWIll-2h exhibited longer time to fiber at the fracture surface(Fig. 6(a). ZrO, particles were rupture at the same stress than BN1-HN; however, MWlll-3h embedded within the annular Sio, layer(Fig. 6(b). EDS analysis exhibited a similar stress-rupture property compared with BNI showed that the Sio2 layer and ZrO2 particles were relatively pure HN. At 1200oC, MWlll-2h exhibited a better stress-rupture prop- in composition. However, limited in-plane and through-thickness erty than both Bnl-hn and bn2-HN. Note that bn1-hn and spatial resolutions made it difficult to determine whether there was BN2-HN are two different batches of SiC/BN/SiC samples. The reaction at the zro2-SiO2 interface to form Zrsioa room-temperature strength of BN2-HN (2750 MPa) was greater than that of BNI-HN (2100 MPa). The stronger batch was found to have proportionately higher stress-rupture properties than the weaker batch IV. Discussion The number of the cracks formed within the hot zone for sever (n Current Understanding of the Observed weak Interface amples was also counted by SEM(not shown here). No cracks Mechanism could be detected for Mwll-2h-I within the hot zone other than Our previous study showed that the microstructure of the ZrO, e failure cracks, while several cracks were clearly observed with coating was strongly dictated by the air leak rate with other SEM close to the fracture surface of MwIll-3h-l and MWIl-3h-2 experimental parameters kept constant. At a low air leak rate(e.g Because all three samples were not polished or etched before the 0.4 Pa/min), a weakly bonded coating developed on the Hi- SEM observation, small cracks might not be observed with SEM Our SEM observation probably indicated that more cracks oc- Nicalon fiber surface for Batch I and Ill experiments. As we have curred during precracking for MWIll-3h samples. eviously postulated and confirmed through an in-depth TEM The fracture surface of the SiC/ZrO,/Sic after the 950c study, the weak bonding most likely occurred as a result of () tress-rupture test is shown in Fig. 5. For the sample with shorter continuous formation of tetragonal Zro, nuclei on the deposition time to stress rupture(1.2 h), as shown in Figs. 5(a) and(b), fiber pullout was observed but with shorter pullout lengths compared the monoclinic ZrO, on reaching a critical grain size, and (iii) ith the sample failed at room temperature. For the sample with development of significant compressive hoop stress because of the the longer time to stress rupture(13 h), as shown in Fig. 5(c),a volume dilation associated with the transformation In this mech- limited number of fibers were observed to pull out. At this anism, the stabilization of the tetragonal phase, when its grain size hange was observed for the Zro, is small, was attributed to the lower surface free energy of the interphase(Fig. 5(d)) tetragonal phase compared with that of the monoclinic phase. This Figure 6 shows the fracture surface of the sic/zro /Sic after effect has been well documented and is generally referred to as the 57 h at 1200C. Significant microstructural change was observed size effect." For Batch Il experiments, an increase in the air leak rate from 0.4 to 2.0 Pa/min during the CVD-zrO2 process mitigated the nucleation of the tetragonal ZrO2 phase. Without th Table Ill. Precracking Information of Zro, Interphase nucleation of the tetragonal ZrO, and martensitic phase transfo Minicomposite Samples mation, we observed that the zrO, layer remained strongly bonded to the Sic fiber surface. This evidence supported our postulation load (N) AE energy for the above weak interface mechanism In a more recent study o we determined that the effect of the air MWIIL-2h-I 116 113 145×10 leak rate on the Zio, microstructure was attributed to the associ- MWIll-2h- MWIlI-21-3 13×105 ated oxygen partial pressure increase in the CVD chamber. The MWIll-3h-1 oxygen partial pressure was systematically controlled in the range MWIl-3h- 2.50×105 of 0.004 to 1.6 Pa by controlling the flow rate of O, diluted in Ar Experimental results showed that high oxygen pressures, similar to
(BN1-HN14 and BN2-HN5 ), are shown in Figs. 4(a) and (b) as the applied stress (assuming fibers were fully loaded) versus the time to rupture. This corresponded to the stress on the fibers in a transverse matrix crack, which was always the location of stressrupture failure. At 950°C, MWIII-2h exhibited longer time to rupture at the same stress than BN1-HN; however, MWIII-3h exhibited a similar stress-rupture property compared with BN1- HN. At 1200°C, MWIII-2h exhibited a better stress-rupture property than both BN1-HN and BN2-HN. Note that BN1-HN and BN2-HN are two different batches of SiC/BN/SiC samples. The room-temperature strength of BN2-HN (2750 MPa) was greater than that of BN1-HN (2100 MPa). The stronger batch was found to have proportionately higher stress-rupture properties than the weaker batch.5 The number of the cracks formed within the hot zone for several samples was also counted by SEM (not shown here). No cracks could be detected for MWIII-2h-1 within the hot zone other than the failure cracks, while several cracks were clearly observed with SEM close to the fracture surface of MWIII-3h-1 and MWIII-3h-2. Because all three samples were not polished or etched before the SEM observation, small cracks might not be observed with SEM. Our SEM observation probably indicated that more cracks occurred during precracking for MWIII-3h samples. The fracture surface of the SiC/ZrO2/SiC after the 950°C stress-rupture test is shown in Fig. 5. For the sample with shorter time to stress rupture (1.2 h), as shown in Figs. 5(a) and (b), fiber pullout was observed but with shorter pullout lengths compared with the sample failed at room temperature. For the sample with the longer time to stress rupture (13 h), as shown in Fig. 5(c), a limited number of fibers were observed to pull out. At this temperature no microstructural change was observed for the ZrO2 interphase (Fig. 5(d)). Figure 6 shows the fracture surface of the SiC/ZrO2/SiC after 57 h at 1200°C. Significant microstructural change was observed for this sample at the fracture surface. Obviously, the observed fracture surface was a matrix crack and experienced long time oxidation exposure. Fiber creep occurred and the fiber was oxidized. An annular SiO2 layer was usually observed around the fiber at the fracture surface (Fig. 6(a)). ZrO2 particles were embedded within the annular SiO2 layer (Fig. 6(b)). EDS analysis showed that the SiO2 layer and ZrO2 particles were relatively pure in composition. However, limited in-plane and through-thickness spatial resolutions made it difficult to determine whether there was reaction at the ZrO2–SiO2 interface to form ZrSiO4. IV. Discussion (1) Current Understanding of the Observed Weak Interface Mechanism Our previous study9 showed that the microstructure of the ZrO2 coating was strongly dictated by the air leak rate with other experimental parameters kept constant. At a low air leak rate (e.g., 0.4 Pa/min), a weakly bonded coating developed on the HiNicalon fiber surface for Batch I and III experiments. As we have previously postulated and confirmed through an in-depth TEM study,7 the weak bonding most likely occurred as a result of (i) continuous formation of tetragonal ZrO2 nuclei on the deposition surface, (ii) martensitic transformation of the tetragonal ZrO2 to the monoclinic ZrO2 on reaching a critical grain size, and (iii) development of significant compressive hoop stress because of the volume dilation associated with the transformation. In this mechanism, the stabilization of the tetragonal phase, when its grain size is small, was attributed to the lower surface free energy of the tetragonal phase compared with that of the monoclinic phase. This effect has been well documented and is generally referred to as the size effect.26 For Batch II experiments, an increase in the air leak rate from 0.4 to 2.0 Pa/min during the CVD-ZrO2 process mitigated the nucleation of the tetragonal ZrO2 phase.9 Without the nucleation of the tetragonal ZrO2 and martensitic phase transformation, we observed that the ZrO2 layer remained strongly bonded to the SiC fiber surface. This evidence supported our postulation for the above weak interface mechanism. In a more recent study,10 we determined that the effect of the air leak rate on the ZiO2 microstructure was attributed to the associated oxygen partial pressure increase in the CVD chamber. The oxygen partial pressure was systematically controlled in the range of 0.004 to 1.6 Pa by controlling the flow rate of O2 diluted in Ar. Experimental results showed that high oxygen pressures, similar to Fig. 4. Stress-rupture data of ZrO2 and BN5,14 interphase Hi-Nicalon reinforced SiC matrix minicomposite samples in air at (a) 950° and (b) 1200°C. Table III. Precracking Information of ZrO2 Interphase Minicomposite Samples Sample Gauge length (mm) Precracking load (N) Cumulative AE energy MWIII-2h-1 116 113 1.45 105 MWIII-2h-2 - - - MWIII-2h-3 121 121 1.3 105 MWIII-3h-1 111 111 2.50 105 MWIII-3h-2 117 111 2.50 105 1730 Journal of the American Ceramic Society—Li et al. Vol. 87, No. 9