ournal J.Am.Ceum.Soc.8104-1212004 Effect of a boron Nitride Interphase that Debonds between the Interphase and the matrix in SiC/SiC Composites Gregory N Morscher Ohio Aerospace Institue, Cleveland, Ohio 44142 Hee Mann yun*, Cleveland State University, Cleveland, Ohio 44115 James A. Dicarlo NASA Glenn Research Center. Cleveland, Ohio 44135 Linus Thomas-Ogbuji·t QSs Group, Inc, Cleveland, Ohio 44135 Typically, the debonding and sliding interface enabling fiber interfaces as well as oxidation of the fiber surface(Fig. 1(a). The pullout for SiC-fiber-reinforced SiC-matrix composites with liquid boria reaction product reacts with the Sic fiber to form a BN-based interphases occurs between the fiber and the inter borosilicate liquid that increases in SiO2 content with phase. Recently, composites have been fabricated where inter oxidation of the SiC. Also, B,, reacts with water vapor face debonding and sliding occur between the BN interphase atmosphere to form volatile B-containing hydrated species os result ind the matrix. This results in two major improvements in ing in an even higher Sio content in the oxidation product. These mechanical properties. First, significantly higher failure phenomena result in a solid oxidation product(glass) that strongly strains were attained due to the lower interfacial shear bonds fibers bridging the matrix crack to one another or to the strength with no loss in ultimate strength properties of the matrix itself and causes subsequent composite embrittlement( Fig composites. Second, significantly longer stress-rupture times at 1(a) higher stresses were observed in air at s15C. In addition, no One proposal to curtail this type of rapid oxidative process that loss in mechanical properties was observed for composites that leads to composite embrittlement would be for the debonding and did not possess a thin carbon layer between the fiber and th sliding interface to be some distance away from the reinforcing interphase when subjected to burner- rig exposure Two pri fibers. For SiC/SiC composites this has been attempted with mary factors were hypothesized for the occurrence of debond C/SiC multilayers as the"interphase"3-5 and more recently with ng and sliding between the bn interphase and the siC matrix: BN/SiC multilayers. In theory, debonding and sliding would a weaker interface at the BN/matrix interface than the fi- occur in some of the outer layers, prohibiting or complicating the ber/BN interface and a residual tensile/shear stress-state at the diffusion of oxidizing species to the inner fiber/interphase region BN/matrix interface of melt-infiltrated composites. Also, the that leads to composite embrittlement. Some benefit has been occurrence of outside debonding was believed to occur during demonstrated for stress-rupture of minicomposites with multilayer cooldown after molten silicon infiltration For SiC/SiC composites with Bn interphases, if the debonding and sliding layer was between the bn and the matrix, a similar benefit proposed for the multilayer approach could be I. Introduction Oxidation of the bn would occur from the"outside" f the bn OR woven SiC/Sic composites with BN interphases, the typical would react with the Sic matrix to eventually form a borosilicate interface where debonding and sliding occur is between the fI er and the bn interphase. We refer to this phenomenon glass that would act as a"sealant "slowing diffusion of oxidizing species to the Bn. In order for the fibers to be fused together or to interphase exacerbates the environmental durability problem of the mcat i:ioi dabo h s m en tr thiconsid er ble am wn of ae terphases at intermediate temper- considering the effects of sealing and the reduced surface area of atures(600 to 1000C)in the presence of oxidizing atmo- BN exposed to oxidizing species when compared with the typical direct access to the fibers themselves. This causes oxidation of the benefit expected from an outside debonded interphase in SiC/Sic Bn interphase preferentially at both the fiber/BN and BN/CVI SiC composites would be improved intermediate-temperature mechan ical properties, e. g, stress-rupture, in oxidizing environments Such behavior has been demonstrated and will be described and discussed in this work I. Experimental Procedure se NASA UF: 2002: approved April 23, 2003 0784 Received A SiC-fiber-reinforced melt-infiltrated SiC-matrix Ceram els that exhibited outside debonding were fabricated from 2D- cientist at NASA Glenn Research Center, Cleveland, OH woven, balanced, 5 harness satin, 0/90 fabric, by General Electric
J. Am. Cerum. Soc., 87 [I] 104-12 (2004) journal Effect of a Boron Nitride lnterphase That Debonds between the lnterphase and the Matrix in SiC/SiC Composites Gregory N. Morscher*.+ Ohio Aerospace Institue, Cleveland, Ohio 44142 Hee Mann Yun*.t Cleveland State University, Cleveland, Ohio 44 I 15 James A. DiCarlo* NASA Glenn Research Center, Cleveland, Ohio 44135 Linus Thomas-Ogbuji*7t QSS Group, Inc., Cleveland, Ohio 44135 Typically, the debonding and I ding interface enabling fiber pullout for Sic-fiber-reinforced Sic-matrix composites with BN-based interphases occurs between the fiber and the interphase. Recently, composites have been fabricated where interface debonding and sliding occur between the BN interphase and the matrix. This results in two major improvements in mechanical properties. First, significantly higher failure strains were attained due to the lower interfacial shear strength with no loss in ultimate strength properties of the composites. Second, significantly longer stress-rupture times at higher stresses were observed in air at 815°C. In addition, no loss in mechanical properties was observed for composites that did not possess a thin carbon layer between the fiber and the interphase when subjected to burner-rig exposure. Two primary factors were hypothesized for the Occurrence of debonding and sliding between the BN interphase and the Sic matrix: a weaker interface at the BNhatrix interface than the fiber/BN interface and a residual tensilekhear stress-state at the BN/matrix interface of melt-infiltrated composites. Also, the occurrence of outside debonding was believed to occur during composite fabrication, i.e., on cooldown after molten silicon infiltration. I. Introduction OR woven SiClSiC composites with BN interphases, the typical F interface where debonding and sliding occur is between the fiber and the BN interphase. We refer to this phenomenon as “inside debonding.” Unfortunately, the inside debonding of the interphase exacerbates the environmental durability problem of SiC/SiC composites with BN interphases at intermediate temperatures (600” to lO00”C) in the presence of oxidizing atmospheres.’.’ When matrix cracks are formed, the environment has direct access to the fibers themselves. This causes oxidation of the BN interphase preferentially at both the fiberlBN and BN/CVI Sic R. Naslain-ontributing editor Manuscript No. 186784. Received August 7, 2002: approved April 23,2003. This work was supported by the NASA UEET program. ‘Member, American Ceramic Society. ‘Senior Research Scientist at NASA Glenn Research Center, Cleveland, OH. interfaces as well as oxidation of the fiber surface (Fig. l(a)). The liquid boria reaction product reacts with the Sic fiber to form a borosilicate liquid that increases in SiO, content with further oxidation of the Sic. Also, B,O, reacts with water vapor in the atmosphere to form volatile B-containing hydrated species resulting in an even higher SiO, content in the oxidation product. These phenomena result in a solid oxidation product (glass) that strongly bonds fibers bridging the matrix crack to one another or to the matrix itself and causes subsequent composite embrittlement (Fig. One proposal to curtail this type of rapid oxidative process that leads to composite embrittlement would be for the debonding and sliding interface to be some distance away from the reinforcing fiber^.^ For SiC/SiC composites this has been attempted with C/SiC multilayers as the “interpha~e”~-~ and more recently with BN/SiC multilayers.6 In theory, debonding and sliding would occur in some of the outer layers, prohibiting or complicating the diffusion of oxidizing species to the inner fibedinterphase region that leads to composite embrittlement. Some benefit has been demonstrated for stress-rupture of minicomposites with multilayer C/SiC coating^.'.^ For SiC/SiC composites with BN interphases, if the debonding and sliding layer was between the BN and the matrix. a similar benefit proposed for the multilayer approach could be achieved. Oxidation of the BN would occur from the “outside” of the BN inwards toward the fiber. The resulting boria oxidation product would react with the Sic matrix to eventually form a borosilicate glass that would act as a “sealant” slowing diffusion of oxidizing species to the BN. In order for the fibers to be fused together or to the matrix, oxidation of the entire thickness of the BN would have to occur (Fig. l(b)). This may take a considerable amount of time considering the effects of sealing and the reduced surface area of BN exposed to oxidizing species when compared with the typical “inside” debonding case (Figs. I(a) and (b)). Therefore, the major benefit expected from an outside-debonded interphase in SiC/SiC composites would be improved intermediate-temperature mechanical properties, e.g., stress-rupture, in oxidizing environments. Such behavior has been demonstrated and will be described and discussed in this work. 1 (a)). II. Experimental Procedure Sic-fiber-reinforced melt-infiltrated SiC-matrix composite panels that exhibited outside debonding were fabricated from 2Dwoven, balanced, 5 harness satin, 0/90 fabric, by General Electric 104
January 2004 Effect of a BN interphase That Debonds between the Interphase and the Matrix in SiC/SiC Composite a: Inside Debonding Fiber Oxidation Matrix Crack SiO +B O, Oxidation SiO2+BO, b: Outside Debonding Fig. 1. Schematic representation of oxidation of the interphase for(a)debonding and sliding between the fiber and the Bn interphase, i. e, "inside debonding. "and(b)between the bn interphase and the matrix, i.e, " outside debonding Power Systems Composites(Newark, DE). The composite fabri- machine(Instron Model 8562, Instron, Ltd, Canton, MA). Modal cation process involves the following steps: chemical vapo acoustic emission (AE) was monitored during the room infiltration(CVi)of a stacked(152 mm X 229 mm)2D-woven temperature tests with two wide-band (50 kHz to 2.0 MHz)sensors fabric with bN, cvI SiC infiltration Sic ele slurry infiltra- laced outside the tapered region of the tensile bar The ae ion, and final liquid Si infiltration. The occurrence of outside waveforms were recorded and digitized using a fracture wave debonding was initially a processing aberration, but has since been detector(FWD, Digital Wave Corp, Englewood, CO). The AE under study to optimize and control its occurrence. Outside data were filtered using the location software provided by the debonding was observed for over 20 different SiC/SiC composite FWD manufacturer, after the tensile test, to separate out the aE panels fabricated with Sylramic( Dow Corming, Midland, MI) S(NI Intermediate-temperature stress-rupture tests were performed to as HNS in the following), and Sylramic-iBN (treated Sylrami on dogbone specimens using a different universal-testing machine fibers that possess an in situ Bn coating). Most of the pane (Instron Model 4502, Instron, Ltd, Canton, MA)in air at 815C as were fabricated with Sylramic in Ref. 3. Specimens were tabbed with graphite-epoxy composite in fiber volume fraction in the (i. e, total fiber volume fraction to 0.40) Table I lists some grips, and a very low load(100 N) was applied to account for of the variations in the physical characteristics of composite thermal expansion of the material during heating The specimens were exposed to elevated temperature using a resistance-heated Mechanical property evaluation included room- and furnace(MoSi2 elements). Although the furnace was 75 mm long intermediate-temperature tensile testing. Room-temperature ten- the hot zone region was only about 15 mm. When the furnace sile testing was performed on at least two dogbone specimens from reached the desired temperature, 815oC, the load was raised to the each panel. Dogbone specimens, 152 mm long, were cut so that the pture stress where it was held until failure Specimens from some panels were also subjected to an atmo wide. Both monotonic and load/unload/reload hysteresis tensile spheric pressure burmer-rig under zero-stress exposure at 815C tests were performed at room temperature using a universal-testing i.e., uncracked, and then tensile tested at room temperature to Table I. Physical and Mechanical Properties of Some of the SiC/SiC Composites Tested Estimated from epcm/No plies E(GPa) 0.13 224 SYL-outside 0.15 YL-inside SYL-inside 8.7/8 389 0. SYL-iBN outside 8.7/8 SYL-ibN outside 0.17 0.49 R 248 N inside 50/8 0.12 79 Tow ends per centimeter
January 2004 Effect of a BN Interphase That Debonds between the Interphase and the Matrix in SiC/SiC Composites 105 Fig. 1. Schematic representation of oxidation of the interphase for (a) debonding and sliding between the fiber and the BN interphase, i.e., “inside debonding,” and (b) between the BN interphase and the matrix, i.e., “outside debonding.” Power Systems Composites (Newark, DE). The composite fabrication process involves the following steps: chemical vapor infiltration (CVI) cf a stacked (-152 mm X 229 mm) 2D-woven fabric with BN, CVI Sic infiltration, Sic particle slurry infiltration, and final liquid Si infiltration.” The occurrence of outside debonding was initially a processing aberration, but has since been under study to optimize and control its occurrence. Outside debonding was observed for over 20 different SiCISiC composite panels fabricated with SylramicO (Dow Coming, Midland, MI) fibers, Hi-Nicalon type S (Nippon Carbon, Tokyo, Japan, referred to as HNS in the following), and Sylramic-iBN (treated SylramicO fibers that possess an in situ BN coating’’). Most of the panels were fabricated with Sylramic-iBN or SylramicO fibers and ranged in fiber volume fraction in the loading direction from 0.13 to 0.2 (i.e., total fiber volume fraction of 0.26 to 0.40). Table I lists some of the variations in the physical characteristics of composite panels. Mechanical property evaluation included room- and intermediate-temperature tensile testing. Room-temperature tensile testing was performed on at least two dogbone specimens from each panel. Dogbone specimens, 152 mm long, were cut so that the gauge section was 10 mm wide and the grip section was 12.5 mm wide. Both monotonic and loadlunloadlreload hysteresis tensile tests were performed at room temperature using a universal-testing machine (Instron Model 8562, Instron, Ltd, Canton, MA). Modal acoustic emission (AE) was monitored during the roomtemperature tests with two wide-band (50 kHz to 2.0 MHz) sensors placed outside the tapered region of the tensile bar.” The AE waveforms were recorded and digitized using a fracture wave detector (FWD, Digital Wave Corp., Englewood, CO). The AE data were filtered using the location software provided by the FWD manufacturer, after the tensile test, to separate out the AE that occurred outside the gauge section. Intermediate-temperature stress-rupture tests were performed on dogbone specimens using a different universal-testing machine (Instron Model 4502, Instron, Ltd., Canton, MA) in air at 815°C as in Ref. 3. Specimens were tabbed with graphite-epoxy composite tabs. The test specimens were gripped with water-cooled hydraulic grips, and a very low load (100 N) was applied to account for thermal expansion of the material during heating. The specimens were exposed to elevated temperature using a resistance-heated furnace (MoSi, elements). Although the furnace was 75 mm long, the hot zone region was only about 15 mm. When the furnace reached the desired temperature, 815°C. the load was raised to the rupture stress where it was held until failure. Specimens from some panels were also subjected to an atmospheric pressure burner-rig under zero-stress exposure at 8 15”C, i.e., uncracked, and then tensile tested at room temperature to Table I. Physical and Mechanical Properties of Some of the SiC/SiC Composites Tested T (MPa) Specimen-location of Estimated from Measured from debonding epcm+/No. plies f E (GPa) u,,~, (MPa) e,,, (%) ak curve push-in test HNS-outside 7.118 0.17 200 352 0.46 - HNS-inside 7.118 0.18 240 311 0.38 - SYL-outside 7.116 0.13 224 224 0.27 37 - SYL-outside 5.018 0.15 219 297 0.43 25 - SYL-mixed 7.918 0.19 246 353 0.33 45 26 SYL-inside 6.318 0.19 246 397 0.36 - - SYL-inside 8.718 0.2 265 389 0.3 65 64 SYL-inside 7.118 0.17 270 310 0.3 1 63 70 SYL-iBN outside 8.718 0.2 216 456 0.5 18 7 SYL-iBN outside 7.118 0.17 220 395 0.49 11 6 SYL-iBN mixed 7.918 0.19 228 >476 0.5 I 43 31 SYL-iBN inside 8.718 0.2 277 404 0.31 73 83 SYL-iBN inside 7.918 0.2 248 502 0.42 - - SYL-iBN inside 5.018 0.12 279 284 0.21 63 - ‘Tow ends per centimeter
Journal of the American Ceramic Society--Morscher et al etermine the retained strength. 2 The low-pressure burner rig(1.0 measure of the residual stress can be mated from th tm)uses a high-velocity(Mach 0.)flame and is designed to intersection of the average slopes of the hysteresis loops for simulate the combustion environments of turbine engines tresses higher than approximately half the peak stress of the fracture surfaces of the failed composites were examined with hysteresis loop(Fig. 2), 5-60 MPa for the inside-debonding a field emission scanning electron microscope (FESEMD), Hitachi composite and -35 MPa for the outside-debonding composite polished sections of untested panels to determine the interfacial removed) for the same architecture MI composites with"outside shear stress of the sliding interface. At least 20 different fibers and"inside"debonding. In general, although similar in ultimate were tested for each specimen. Finally, the interphase region of strength, two differences between outside- and inside-debonding small slivers of composite material were fractured in bending in moduli (Table I)and (2)a higher strain at a given applied stress nation. Depth profiles were then performed at regions wher mi situ under vacuum to prevent the fracture surface from conta including higher strains to failure(Table I and Fig. 3). However, one panel, which exhibited a mixture of inside and outside Bn layer adhered to the matrix and at other regions where the Bn debonding, was an exception and had a high elastic modulus (246 layer adhered to the fiber GPa) Figure 4 shows examples of composite fracture surfaces after II. Results room-temperature tensile failure. Some bundle pullout was ob- served for both types of composites; however, individual fiber (1) Room-Temperature Tensile Stress-Strain Behavior pullout was significantly longer for outside-debonding composites Typical unload-reload tensile hysteresis stress-strain curves Figs. 4(a) and(b) than for inside-debonding composites(Figs and AE activity are shown in Fig. 2 for MI SYL-iBN/SiC 4(c)and(d ). Note the adherence of the BN layer to the fibers for composite that displays inside and outside debonding. It was he outside -debonding composites(Fig. 4(b))compared with the observed that the first detectable aE that occurs in the gauge outside-debonding composites(Fig. 4(d). It would be ideal if section occurs at 110 20 MPa for both inside- and outside- debonding outside the bn interphase occurred for each fiber debonding composites. Also note that on unloading the material independently from one another(e. g, Fig. 1). However, because of tiffens, indicating that the matrix is in residual compression. a the close packing of fibers in woven bundles, debonding between 8.epcm 8 ply: f=0.2 E=280 GPa 0.2 Strain. a) 500 8.epcm; 8 ply: f0.2 E=216 GPa 350 00 0.1 02 0.5 Strain, % Fig. 2. Tensile load-unload-reload hysteresis curves for(a) inside-debonding and (b)outside-debonding SYL-iBN SiC/SiC composites. Also plotted is the normalized cumulative AE energy. Squares are stress-strain model for best-fit interfacial shear stress
106 Journal of the American Ceramic Society--Morscher et al. Vol. 87, No. 1 determine the retained strength.” The low-pressure burner rig (1.0 atm) uses a high-velocity (Mach 0.3) flame and is designed to simulate the combustion environments of turbine engines. Fracture surfaces of the failed composites were examined with a field emission scanning electron microscope (FESEM), Hitachi Model S-4700. A fiber push-in was performed on polished sections of untested panels to determine the interfacial shear stress of the sliding interface. At least 20 different fibers were tested for each specimen. Finally, the interphase region of some specimens was examined using Auger electron spectroscopy (AES) and transmission electron microscopy (TEM). For AES, small slivers of composite material were fractured in bending in situ under vacuum to prevent the fracture surface from contamination. Depth profiles were then performed at regions where the BN layer adhered to the matrix and at other regions where the BN layer adhered to the fiber. III. Results (I) Room-Temperature Tensile Stress-Strain Behavior Typical unload-reload tensile hysteresis stress-strain curves and AE activity are shown in Fig. 2 for MI SYL-iBN/SiC composite that displays inside and outside debonding. It was observed that the fist detectable AE that occurs in the gauge section occurs at 110 2 20 MPa for both inside- and outsidedebonding composites. Also note that on unloading the material stiffens, indicating that the matrix is in residual compression. A measure of the residual stress can be approximated from the intersection of the average slopes of the hysteresis loops for stresses higher than approximately half the peak stress of the hysteresis loop (Fig. 2),15 -60 MPa for the inside-debonding composite and -35 MPa for the outside-debonding composite. Figure 3 shows typical stress-strain curves (hysteresis loops removed) for the same architecture MI composites with “outside” and “inside” debonding. In general, although similar in ultimate strength, two differences between outside- and inside-debonding composites were evident for room-temperature stress-strain behavior: “outside-debonding” composites had (1) lower elastic moduli (Table I) and (2) a higher strain at a given applied stress including higher strains to failure (Table I and Fig. 3). However, one panel, which exhibited a mixture of inside and outside debonding, was an exception and had a high elastic modulus (246 GPa). Figure 4 shows examples of composite fracture surfaces after room-temperature tensile failure. Some bundle pullout was observed for both types of composites; however, individual fiber pullout was significantly longer for outside-debonding composites (Figs. 4(a) and (b)) than for inside-debonding composites (Figs. 4(c) and (d)). Note the adherence of the BN layer to the fibers for the outside-debonding composites (Fig. 4(b)) compared with the outside-debonding composites (Fig. 4(d)). It would be ideal if debonding outside the BN interphase occurred for each fiber independently from one another (e.g., Fig. 1). However, because of the close packing of fibers in woven bundles, debonding between 450 i I SY 400 4 8.7e~cm: L-iBN 8 ply: f = 0.2 Ir #Y 1.4 * 1.2 i1 15 0.4 0 0.4 ’/’ Strain, % (a) 1 ______ 500 I M0 O*I 0.2 0.3 0.4 0.5 Straln, % (b) 1.4 I 0.6 * C 1w W 0.8 a 1.2 f E 0.4 10.2 s -40 0.6 Fig. 2. Tensile load-unload-reload hysteresis curves for (a) inside-debonding and (b) outside-debonding SYL-iBN SiC/SiC composites. Also plotted is the normalized cumulative AE energy. Squares are stress-strain model for best-fit interfacial shear stress
January 2004 Effect of a BN Interphase That Debonds between the interphase and the Matrix in SiC/Sic Composites was estimated from the measured final crack density of failed composites multiplied by the normalized cumulative AE energy (Fig. 2), assuming the latter represented the stress-dependent Inside Debonair distribution of matrix cracks, which has been demonstrated for inside Debonding" similar systems'.8 Therefore, the only variable not known was T which was adjusted to best fit the predicted stress-strai 3008吵 the experimental stress-strain curve. For the case where SYL-IBN lengths overlap Ahn and Curtin"showed that if the ° Outside Debonding still equally spaced, the composite strain could then be modeled by o/(fE+ao/ -(o+o4ES(o)p] 00.10.20.304050607 E E品 <28 Strain stress-strain curves for ide-debonding Fig. 3. Room-temperature tensile stress-strain curves for 8.7 epcm ecimens are shown in Figs. 2(a) and SYL- iBN SiC/SiC composites and 7, I epcm HNS SiC/SiC composites (b), respectively. steresis loops removed). Note the HNS composites are displaced by The interfacial shear stress was also measured directly from the 0. 2% in strain for clarity. fiber push-in technique. Results of the two techniques are listed fo individual specimens in Table I for systems that displayed global outside debonding, mixed outside/inside debonding, and global the bn interphase and the matrix was often observed to occur inside debonding. Both techniques confirmed that the interfacial around groups of fibers that were linked to one another by the thin shear strength of global outside-debonding composites(-10 MPa Bn that was deposited on two closely spaced fibers. Usually, these was significantly less than that of inside-debonding composites fiber groups were made up of a few fibers that formed a row of (-70 MPa). Mixed outside/inside debonding had intermediate fibers as shown in Fig 4(b ). Debonding at the BN interphase/Sic values of interfacial shear strength. It is important to note that even matrix was observed for individual fibers that were well separated though the interfacial shear strength of outside debonding is lower from other fibers. For some composites, regions of outside than that of inside-debonding composites, there was no loss in debonding and inside debonding were observed in different re- ultimate strengths for outside-debonding composites and often u gions or bundles of the fracture surface, i.e., mixed debonding ultimate strength increased(e.g, compare the f= 0.2 composites In addition to a low elastic modulus, outside-debonding com- in Fig 3(a). posites often displayed a secondary modulus before significant matrix cracking. Figure 5 shows a family of stress-strain curves for a number of different outside-debonding composites with (2) Intermediate-Temperature Mechanical Behavior ifferent volume fractions, The initial elastic moduli were very Stress rupture at 815C was performed on SYL and SYL- iBN consistent(218 GPa)and all of the curves showed an inflection composites with inside and outside debonding(fig. 6).The x70 MPa that resulted in stress-rupture data for SYL SiC/SiC composites displaying inside MPa). This inflection was not associated with any AE activity; i.e debonding have been reported in refs 20 and 21. since the par it appears that this inflection was not due to matrix crack varied in fiber volume fraction, the rupture stress data are plotted ormation as the stress on the fibers i e. the load in a matrix crack that was Finally, the interfacial shear strength of several different inside- carried by the fibers. For comparison, the rupture stress corre- debonding and outside-debonding composites was determined sponding to a composite with f=0.2 in the loading direction is using two techniques. 4 First, the interfacial shear strength was shown on the right axis. Each set of data for the different types of estimated by modeling the stress-strain curve based on the composites had at least one panel with f =0.2. stress-dependent crack density(from AE) Composite strain was First, note that there is a difference in rupture behavior between determined in the same fashion as Pryce and Smith. Using the inside-debonding SYL-ibN fiber composites and SYL nomenclature of Curtin et al, composite strain can be modeled posites. Inside-debonding SYL-iBn composites outperform (i.e assuming equally spaced cracks fail after a longer time at a given stress) inside-debonding SYL composites because the fibers in SYL-iBn composites E=a/Ee+a6()pE0+σt) rally spread apart from one another with the formation of the 100 the fiber surface. 2I The where o is the applied stress. the residual (thermal)stress in more time with increasing separation distance. In addition, the the matrix(compression is negative), E is the elastic modulus, debonding interface for inside-debonding SYL-ibN actually oc- subscripts m, f, and c refer to matrix, fiber, and composite, curs between the in situ BN and the CVI-deposited BN. 0In other respectively, and p. is the matrix crack density. The fir rst part of the words, for inside-debonding SYL-iBN composites, the debonding equation corresponds to the elastic strain response of an uncracked and sliding interface was some distance(-100 nm)away from the composite and the second part of the equation corresponds to the fiber surface, which contained Sic in(displacement) of the fibers at and away from a or both fiber composite systems possessing an outside- through-thickness matrix crack dictated by the sliding lengt debonding interface, further improvements in intermediate mperature stress-rupture life were observed(Fig. 6). For SYL (2) composites with outside debonding compared with SYL inside bonding co ed by over 25 a=(1-∫)Em/E outside-debonding SYL-ibn composites in comparison to inside- debonding SYL-BN composites, the and r is the fiber radius, f is the fiber volume fraction in the magnitude in time improvement at resses and -200 MPa direction, and T is the interfacial shear strength. E and o improvement in fiber stress(40 determined from the stress-strain curves. Er is 380 GPa and at lower stresses near the run-out c It should be noted that was determined from the rule-of-mixtures. The stress-dependent p. these high-stress conditions for stress-rupture are significantly
January 2004 Effect of a BN Interphase That Debonds between the Interphase and the Matrix in SiC/SiC Composites 107 6oo 1 500 - 'Outside Debonding' "Inside Debonding" 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Strain, % Fig. 3. Room-temperature tensile stress-strain curves for 8.7 epcm SYL-iBN SiC/SiC composites and 7. I epcm HNS SiC/SiC composites (hysteresis loops removed). Note the HNS composites are displaced by 0.2% in strain for clarity. the BN interphase and the matrix was often observed to occur around groups of fibers that were linked to one another by the thin BN that was deposited on two closely spaced fibers. Usually, these fiber groups were made up of a few fibers that formed a row of fibers as shown in Fig. 4(b). Debonding at the BN interphaselSiC matrix was observed for individual fibers that were well separated from other fibers. For some composites, regions of outside debonding and inside debonding were observed in different regions or bundles of the fracture surface, i.e., mixed debonding. In addition to a low elastic modulus, outside-debonding composites often displayed a secondary modulus before significant matrix cracking. Figure 5 shows a family of stress-strain curves for a number of different outside-debonding composites with different volume fractions. The initial elastic moduli were very consistent (-218 GPa) and all of the curves showed an inflection at -70 MPa that resulted in a lower secondary modulus (-177 MPa). This inflection was not associated with any AE activity; i.e., it appears that this inflection was not due to matrix crack formation. Finally, the interfacial shear strength of several different insidedebonding and outside-debonding composites was determined using two technique^.'^ First, the interfacial shear strength was estimated by modeling the stress-strain curve based on the stress-dependent crack density (from AE). Composite strain was determined in the same fashion as Pryce and Smith.I6 Using the nomenclature of Curtin et aZ.,I7 composite strain can be modeled assuming equally spaced cracks: E zz U/E, + CX~(U)~,JE~(U + uth) (1) (for pi' > 26) where u is the applied stress, uIh is the residual (thermal) stress in the matrix (compression is negative), E is the elastic modulus, subscripts m, f, and c refer to matrix, fiber, and composite, respectively, and p, is the matrix crack density. The first part of the equation corresponds to the elastic strain response of an uncracked composite and the second part of the equation corresponds to the extra strain (displacement) of the fibers at and away from a through-thickness matrix crack dictated by the sliding length: 6 = cir(u + u,h)/2T (2) where a = (1 -f)E,,/fE, (3) and r is the fiber radius, f is the fiber volume fraction in the loading direction, and T is the interfacial shear strength. Ec and ulh were determined from the stress-strain curves. Ef is 380 GPa and Em was determined from the rule-of-mixtures. The stress-dependent pc was estimated from the measured final crack density of failed composites multiplied by the normalized cumulative AE energy (Fig. 2), assuming the latter represented the stress-dependent distribution of matrix cracks, which has been demonstrated for similar systems.''*'8 Therefore, the only variable not known was T, which was adjusted to best fit the predicted stress-strain curve to the experimental stress-strain curve. For the case where the sliding lengths overlap, Ahn and Curtin'' showed that if the cracks are still equally spaced, the composite strain could then be modeled by E = a/(fEJ + CKU~/E~ - CK(U + U~~)/[~E&(U)P,] (4) (for p;' < 26) Therefore, for higher applied stress conditions, if pC-' < 26 was predicted, Eq. (4) was used. Examples of best-fit stress-strain curves for inside-debonding (T - 73 MPa) and outside-debonding (T - 18 MPa) composite specimens are shown in Figs. 2(a) and (b), respectively. The interfacial shear stress was also measured directly from the fiber push-in technique. Results of the two techniques are listed for individual specimens in Table I for systems that displayed global outside debonding, mixed outsidelinside debonding, and global inside debonding. Both techniques confirmed that the interfacial shear strength of global outside-debonding composites (- 10 MPa) was significantly less than that of inside-debonding composites (-70 MPa). Mixed outsidehide debonding had intermediate values of interfacial shear strength. It is important to note that even though the interfacial shear strength of outside debonding is lower than that of inside-debonding composites, there was no loss in ultimate strengths for outside-debonding composites and often the ultimate strength increased (e.g., compare thef = 0.2 composites in Fig. 3(a)). (2) Intermediate-Temperature Mechanical Behavior Stress rupture at 815°C was performed on SYL and SYL-iBN composites with inside and outside debonding (Fig. 6). The stress-rupture data for SYL SiC/SiC composites displaying inside debonding have been reported in Refs. 20 and 21. Since the panels varied in fiber volume fraction, the rupture stress data are plotted as the stress on the fibers, i.e., the load in a matrix crack that was carried by the fibers. For comparison, the rupture stress corresponding to a composite withf = 0.2 in the loading direction is shown on the right axis. Each set of data for the different types of composites had at least one panel with f = 0.2. First, note that there is a difference in rupture behavior between inside-debonding SYL-iBN fiber composites and SYL fiber composites. Inside-debonding SYL-iBN composites outperform (i.e., fail after a longer time at a given stress) inside-debonding SYL composites because the fibers in SYL-iBN composites are naturally spread apart from one another with the formation of the - 100 nm BN layer on the fiber surface.*' The rupture life depends on the time it takes to bond nearest-neighbor fibers together, which takes more time with increasing separation distance. In addition, the debonding interface for inside-debonding SYL-iBN actually occurs between the in situ BN and the CVI-deposited BN." In other words, for inside-debonding SYL-iBN composites, the debonding and sliding interface was some distance (-100 nm) away from the fiber surface, which contained Sic. For both fiber composite systems possessing an outsidedebonding interface, further improvements in intermediatetemperature stress-rupture life were observed (Fig. 6). For SYL composites with outside debonding compared with SYL insidedebonding composites, stress-rupture improved by over 250 MPa in fiber stress (-50 MPa for an f = 0.2 composite). For outside-debonding SYL-iBN composites in comparison to insidedebonding SYL-BN composites, there was over an order of magnitude in time improvement at high stresses and -200 MPa improvement in fiber stress (-40 MPa for an f = 0.2 composite) at lower stresses near the run-out condition. It should be noted that these high-stress conditions for stress-rupture are significantly
Journal of the American Ceramic Society-Morscher et aL. 3w× 80SE(M 11/e2 11bm号2aw1 nox SE1an Fiber BN Fiber BN sav 113mmx20c4 0401110620v14 Fig. 4. FESEM images of fracture surfaces of SYL- iBn composites showing outside debonding(a, b)and inside debonding(c, d) higher than the stresses for matrix cracks to penetrate the load- contact, the thinner areas of BN earing fibers (determined from the onset of hyste op fusion occurred for rupture times activity, 175 MPa for f=0. 2 composites used in this study ) In several regions of significant fib other words, the SYL-iBN composites are significantly cracked at section of the fracture surface the stress-rupture conditions of this study, even for specimens that A few specimens(SYL and SYL-iBN)were precracked at room did not fail after long periods of time. temperature and compared with the rupture behavior of pristine Examination of the rupture specimen fracture surfaces con- from the same panel(Fig. 8). It was evident that firmed the survival of most of the bn around the fibers in the onding SYL composites with nominally good rupture matrix crack even though significant oxidation had occurred in the were significantly poorer in rupture behavior with matrix crack( Fig. 7). However, at regions of near fiber-to-fiber ng as has been observed in another study. On the other 600 f=0.18 f。=0.2 f。=017 200 Change in slope at-70 MPa not associated with occurrence of ae 0203040.50.6 Strain. Fig. 5. Room-temperature tensile stress-strain curves for a number of outside-debonding composites with different fiber volume fractions
108 Journal of the American Ceramic Society-Morscher et al. Vol. 87, No. 1 Fig. 4. FESEM images of fracture surfaces of SYL-iBN composites showing outside debonding (a,b) and inside debonding (c,d). higher than the stresses for matrix cracks to penetrate the loadbearing fibers (determined from the onset of hysteresis loop activity, - 175 MPa for f = 0.2 composites used in this study). In other words, the SYL-iBN composites are significantly cracked at the stress-rupture conditions of this study, even for specimens that did not fail after long periods of time. Examination of the rupture specimen fracture surfaces confirmed the survival of most of the BN around the fibers in the matrix crack even though significant oxidation had occurred in the matrix crack (Fig. 7). However, at regions of near fiber-to-fiber 600 500 6 300 b a 200 12 100 200 -I contact, the thinner areas of BN were oxidized and fiber-to-fiber fusion occurred for rupture times greater than 80 h. There were several regions of significant fiber pullout throughout the cross section of the fracture surface. A few specimens (SYL and SYL-iBN) were precracked at room temperature and compared with the rupture behavior of pristine specimens from the same panel (Fig. 8). It was evident that inside-debonding SYL composites with nominally good rupture properties were significantly poorer in rupture behavior with precracking as has been observed in another study." On the other 1& Change in slope at - 70 MPa not associated with occurrence of AE 0 0 0.1 0.2 0.3 0.4 0.5 0.6 Strain, % , = 0.18 Fig. 5. Room-temperature tensile stress-strain curves for a number of outside-debonding composites with different fiber volume fractions
January 2004 Effect of a BN Interphase That Debonds between the interphase and the Matrix in SiC/SiC Composites 1800 Outside Debonding +350 o SYL-IBN (SYL-IBN 1600 L1200 SYL v1000 ”;……… nside Debonding 600f squares a diamonds"SYL as-produced 100 1000 Time to fail, hr Fig. 6. Stress-rupture of inside- and outside- debonding composites with Sylramic (SYL)and Sylramic-iBN(SYL-iBN) fiber reinforcement in air at 815C The data are plotted as stress on the fibers, i. e, composite stress divided by f. The composite stress for an f =0. 2 is plotted on the right axi hand the outside-debonding specimen with sYL-ibn fibers that for (1)Hi-Nicalon due to a carbon rich layer that occurs for MI was precracked did not fail after 330 h compared with the pristine rM due to a carbon-rich layer on and com possible to conclude that precracked outside -debonding specimens processing. 2 and (3)sytramacer com posits s her facie after fiber on the fibers that was not burned off completely before BN first time a precracked SiC/SiC specimen outperformed a nonprec interphase deposition. SYL and SYL-iBN composites with a racked specimen from the same panel re time to sizing that has low char yield are unaffected by burner-rig failure of the nonprecracked specimen was greater than 10 h exposure, Figure 9 compares the burner-rig degradation(or lack Sylramic, SYL-iBN, and HNS outside-debonding composites thereof). Also shown is an example of outside-debonding SYL- were subjected to burner-rig exposure at 815oC for-100 h with no ibN before and after burner-rig exposure No significant strength applied stress, and then were tested at room temperature to degradation has been observed for SYL-iBN and SYL composites determine the retained strength properties. Whereas the ruptu with outside debonding and complete sizing removal. Burner-rig- tests evaluate the durability of composites when cracked. the exposed SYL-iBN and Sylramic composites with outside debond- zero-stress burner- rig experiment has proved to be an effective test ing were often observed to stiffen and fracture at slightly lower to evaluate the ability of an undamaged composite material to ultimate strain(Fig. 9, Sylramic composite not shown). However, withstand severe intermediate- temperature oxidation through the HNS outside- debonding composites were degraded after burner exposed (as-machined)edges of the composite specimen. It has rig exposure due to the presence of a carbon layer on the fiber been found that if carbon exists on the surface of any fiber type the surface HNS outside - debonding composites were also observed to SiC/SiC MI composites will be significantly degraded after stiffen slightly after burmer-rig exposure Stiffening does not occur burner-rig exposure. This type of degradation has been observed Fig. 7. SEM micrograph from the fracture surface of outside-debonding composite after stress-rupture after 100 h at 815.C
January 2004 Effect of a BN Interphase That Debonds between the Interphase and the Matrix in SiC/SiC Composites 109 1800 2 1600 @ 1400 9) LL 1200 E g c 0 8 1000 9) L 800 600 Outside Debonding I -0. *%. Inside SYL-IEN Debonding -I (SYL-\EN) 1'" 8 SYL circles = SYL-IBN v) 250 8 200 3 R u) + Y t lbO z v P) I squares (L diamonds = SYL as-produced I 10 100 1000 Time to fail, hr Fig. 6. The data are plotted as stress on the fibers, i.e., composite stress divided by5 The composite stress for anf = 0.2 is plotted on the right axis. Stress-rupture of inside- and outside-debonding composites with Sylramic@ (SYL) and Sylramic-iBN (SYL-iBN) fiber reinforcement in air at 815°C. hand, the outside-debonding specimen with SYL-iBN fibers that was precracked did not fail after 330 h compared with the pristine specimen which failed after -190 h. With so few data it is not possible to conclude that precracked outside-debonding specimens are superior to pristine specimens in rupture. However, this is the first time a precracked SiC/SiC specimen outperformed a nonprecracked specimen from the same panel at a stress where time to failure of the nonprecracked specimen was greater than 10 h. Sylramic, SYL-iBN, and HNS outside-debonding composites were subjected to burner-rig exposure at 815°C for - 100 h with no applied stress, and then were tested at room temperature to determine the retained strength properties. Whereas the rupture tests evaluate the durability of composites when cracked, the zero-stress burner-rig experiment has proved to be an effective test to evaluate the ability of an undamaged composite material to withstand severe intermediate-temperature oxidation through the exposed (as-machined) edges of the composite specimen. It has been found that if carbon exists on the surface of any fiber type, the SiC/SiC MI composites will be significantly degraded after burner-rig exposure. This type of degradation has been observed for (1) Hi-Nicalon due to a carbon-rich layer that occurs for MI composites after fiber and composite processing," (2) Hi-Nicalon STM due to a carbon-rich layer on the fiber surface after fiber processing,22 and (3) Sylramica composites when a sizing is used on the fibers that was not burned off completely before BN interphase depo~ition.~~ SYL and SYL-iBN composites with a sizing that has low char yield are unaffected by burner-rig exposure. Figure 9 compares the burner-rig degradation (or lack thereof). Also shown is an example of outside-debonding SYLiBN before and after burner-rig exposure. No significant strength degradation has been observed for SYL-iBN and SYL composites with outside debonding and complete sizing removal. Burner-rigexposed SYL-IBN and Sylramic composites with outside debonding were often observed to stiffen and fracture at slightly lower ultimate strain (Fig. 9, Sylramic composite not shown). However, HNS outside-debonding composites were degraded after burnerrig exposure due to the presence of a carbon layer on the fiber surface. HNS outside-debonding composites were also observed to stiffen slightly after burner-rig exposure. Stiffening does not occur for inside-debonding composites. Fig. 7. SEM micrograph from the fracture surface of outside-debonding composite after stress-rupture after -100 h at 815°C
Joumal of the American Ceramic Sociery-Morscher et al. Vol. 87. No. I (1) The increased strain to failure of outside-debonding composi can be attributed to the lower t of the bn-cvi sic interface over that of the fiber/BN interface. However, if T decreased and global load sharing exists, one would expect the ultimate strength properties to decrease. "4 The converse has been observed for strongly bonded interfaces compared with weakly bonded inter Ined Debonding faces in CVI SiC matrix composites where the higher T interface composites exhibit higher ultimate strengths. In this study, the lower T composites did not lose strength and in some cases were stronger than comparable high-T composites. Two explanations 0 can account for this. First, it is possible that high-T composites TIme, hr exhibit local load sharing and the lower-- composites exhibit global load sharing. If the high-T composites exhibit local loac 8. Stress-rupture of as-received and precracked SYL and SYL-iBn sharing, stress concentrations would develop for load-bearing IC composites at 815C in air. fibers surrounding individual or groups of broken fibers in a matrix crack. This would result in lower composite ultimate strengths than expected based on global load sharing.2 Second, global load sharing may occur for both low-and high- composites, and as Xi ome fracture surfaces were examined from burner-nig-exposed and Curtin" have theorized, fibers with an adhered coating would SYL and SYL-ibN composites. The composites exhibited long be effectively stronger than fibers without a coating because the allout lengths similar to as-produced specimens(Fig. 4(a)and a flaws on the fiber surface are constrained to some degree by the Sio2-containing layer was often observed on the surface of the Bn coating, even for low-modulus coatings such as C or BN. In fact in between the BN and the matrix throughout the cross section their model would predict approximately the same composite (Fig. 10). Evidently, oxidation occurred through the BN/CVI Sic strength for outside-debonding composites with a T= 10 MPa and interface region at the exposed cut edge into the interior of the inside-debonding composites with a T =70 MPa composite. HNS composites exhibited a flat fracture surface and strong bonding of fibers as has been reported for other systems The improved intermediate-temperature rupt with carbon layers that exist at the fiber surface. 3 bonding composites occurs in the manner put forward in Fi 1(b)(Fig. 7). Even after 100 h at 815C in a bridged matrix crack, a significant portion of the bn remained as a barier between the (3)Analysis of the BN-CVI SiC interface oxidation reaction product and the fiber surface for the majority of aes depth profiles were conducted on specimen surfaces that fiber circumference. However, thinner regions of BN separating were fractured in the aES chamber for several specimens exhib- nearest-neighbor fibers were oxidized and appear to have led to the iting inside and outside debonding. Depth profiles through BN time-dependent strength degradation. This would explain why the layers adhered to the CVi SiC matrix were performed(not shown). Sylramic composites with outside debonding are poorer in A mild enrichment of C appeared to exist at the BN-CVI Sic stress-rupture than the Sylramic- iBN composites( Fig. 7).The interface for both inside-and outside-debonding specimens. How former consists of many fibers nearly contacting another fiber with er, no difference in the amount of C enrichment at the bn-cVI little or no bn interphase in between, whereas the latter possesses SiC interface for inside-and outside-debonding composites could the in situ BN layers on the fiber surface that enable greater be discerned given the error in the aEs measurement (-10%). protection of the fibers as well as some degree of fiber separation Representative TEM micrographs of inside- and outside- Finally, for typical inside-debonding composites that fail at sig bonding specimens are shown in Fi of the nificantly shorter lives and lower stresses at the same temperature ame region are also shown. There does appear to be some c no bn was detectable in the oxidized nrichment at the BN-CVi SiC interface for the outside-debonding (see, e.g, Refs. 1 and 27). In other words the entire matrix and composites and little if any C enrichment for the inside-debonding interphase region of inside-debonding composites would be com mposites.However, this cannot be considered conclusive evi- pletely oxidized, with the fibers strongly bonded to the matrix dence of a C-layer (2)Why Outside Debonding Two potential mechanisms are considered for outside debond ng for these MI composite systems:(1)a weaker BN-CVI Sic 450 nterface than BN-fiber interface and(2)sufficient residual stress 400 at the interface to cause debonding of the weak interface probabl as-produced on cooling after infiltration of molten Si composites possess a lower T than inside-debonding composites interface is also lower than the debond energy of the fiber/BN nterface as well. Residual compression exists in the matrix( Fig 2), which presumably forces the fibers into residual tension. Thi 100 is due to free Si. The volume expansion of Si from the liquid to solid state is -9%. Therefore, expansion of the Si phase takes 50 place during cooling of the composite from its fabrication temper ature for MI(1400C depending on the additives to the Si). this 1020304050.6 places the Si in compression. Si also has a lower thermal expansion coefficient than SiC, -3 X 10rC compared with -4. x Strain. 10/C, respectively. Therefore, on further cooling, the Si placed in further compression. The crack closure effect( Fig. 2) side debonding. The HNS stress-strain curves are offset on the strain axis fiber/interphase bundles taken together are in residual compression for clarity necessitating residual tension in the fibers. Locally, the residual
110 Journal of the American Ceramic Society--Morscher et al. Vol. 87, No. 1 6 170 1 150 4 I 0 100 200 300 400 lime, hr Fig. 8. Stress-rupture of as-received and precracked SYL and SYL-iBN SiClSiC composites at 815°C in air. Some fracture surfaces were examined from burner-rig-exposed SYL and SYL-iBN composites. The composites exhibited long pullout lengths similar to as-produced specimens (Fig. 4(a)) and a SO,-containing layer was often observed on the surface of the BN in between the BN and the matrix throughout the cross section (Fig. 10). Evidently, oxidation occurred through the BN/CVI Sic interface region at the exposed cut edge into the interior of the composite. HNS composites exhibited a flat fracture surface and strong bonding of fibers as has been reported for other systems with carbon layers that exist at the fiber surface.23 (3) Analysis of the BN-CVZ Sic Interface AES depth profiles were conducted on specimen surfaces that were fractured in the AES chamber for several specimens exhibiting inside and outside debonding. Depth profiles through BN layers adhered to the CVI Sic matrix were performed (not shown). A mild enrichment of C appeared to exist at the BN-CVI Sic interface for both inside- and outside-debonding specimens. However, no difference in the amount of C enrichment at the BN-CVI Sic interface for inside- and outside-debonding composites could be discerned given the error in the AES measurement (-10%). Representative TEM micrographs of inside- and outsidedebonding specimens are shown in Fig. 11. Carbon maps of the same region are also shown. There does appear to be some C enrichment at the BN-CVI Sic interface for the outside-debonding composites and little if any C enrichment for the inside-debonding composites. However, this cannot be considered conclusive evidence of a C-layer. 500 1 450 400 2 350 B 300 # 250 150 100 50 0 4 200 I 0 0.1 0.2 0.3 0.4 0.5 0.6 Strain, % Fig. 9. Room-temperature tensile stress-strain curves of as-received and burner-rig-exposed SYL-iBN and HNS SiC/SiC specimens showing outside debonding. The HNS stress-strain curves are offset on the strain axis for clarity. IV. Discussion (1) Zmproved Mechanical Properties The increased strain to failure of outside-debondjng composites can be attributed to the lower 7 of the BN-CVI Sic interface over that of the fiber/BN interface. However, if 7 decreased and global load sharing exists, one would expect the ultimate strength properties to decrease.24 The converse has been observed for strongly bonded interfaces compared with weakly bonded interfaces in CVI Sic matrix composites where the higher 7 interface composites exhibit higher ultimate strengths5 In *his study, the lower T composites did not lose strength and in some cases were stronger than comparable high-? composites. Two explanations can account for this. First, it is possible that high-? composites exhibit local load sharing and the lower-? composites exhibit global load sharing. If the high-? composites exhibit local load sharing, stress concentrations would develop for load-bearing fibers surrounding individual or groups of broken fi6ers in a matrix crack. This would result in lower composite ultimate strengths than expected based on global load sharing.25 Second, global load sharing may occur for both low- and high-? composites, and as Xia and CurtinZ6 have theorized, fibers with an adhered coating would be effectively stronger than fibers without a coating because the flaws on the fiber surface are constrained to some degree by the coating, even for low-modulus coatings such as C or BN. In fact, their model would predict approximately the same composite strength for outside-debonding composites with a 7 = 10 MPa and inside-debonding composites with a T = 70 MPa. The improved intermediate-temperature rupture life of outsidedebonding composites occurs in the manner put forward in Fig. l(b) (Fig. 7). Even after 100 h at 815°C in a bridged matrix crack, a significant portion of the BN remained as a barrier between the oxidation reaction product and the fiber surface for the majority of fiber circumference. However, thinner regions of BN separating nearest-neighbor fibers were oxidized and appear to have led to the time-dependent strength degradation. This would explain why the Sylramic@ composites with outside debonding are poorer in stress-rupture than the Sylramic-iBN composites (Fig. 7). The former consists of many fibers nearly contacting another fiber with little or no BN interphase in between, whereas the latter possesses the in situ BN layers on the fiber surface that enable greater protection of the fibers as well as some degree of fiber separation. Finally, for typical inside-debonding composites that fail at significantly shorter lives and lower stresses at the same temperature, no BN was detectable in the oxidized portion of a fracture surface (see, e.g., Refs. 1 and 27). In other words, the entire matrix and interphase region of inside-debonding composites would be completely oxidized, with the fibers strongly bonded to the matrix. (2) Why Outside Debonding? Two potential mechanisms are considered for outside debonding for these MI composite systems: (1) a weaker BN-CVI Sic interface than BN-fiber interface and (2) sufficient residual stress at the interface to cause debonding of the weak interface probably on cooling after infiltration of molten Si. The outside-debonding composites possess a lower 7 than inside-debonding composites (Table I); presumably, the debond energy of the BN/CVI-Sic interface is also lower than the debond energy of the fiberlSN interface as well. Residual compression exists in the matrix (Fig. 2), which presumably forces the fibers into residual tension. This is due to free Si. The volume expansion of Si from the liquid to solid state is -9%. Therefore, expansion of the Si phase takes place during cooling of the composite from its fabrication temperature for MI (- 1400°C depending on the additives to the Si). This places the Si in compression. Si also has a lower thermal expansion coefficient than Sic, -3 X 10-6/oC compared with -4.5 X 10-6/oC, respectively. Therefore, on further cooling, the Si is placed in further compression. The crack closure effect (Fig. 2) reflects the global residual stress state of the entire matrix; i.e., the free Si, the particulate Sic, the CVI Sic, and the unbridged 90" fibedinterphase bundles taken together are in residual compression necessitating residual tension in the fibers. Locally, the residual
January 2004 Effect of a BN Interphase That Debonds between the interphase and the Matrix in SiC/SiC Composites 111 0.601201.802,40 0.601.201.80240 0.601.201.80240 101703009460kV11.3mmx10.0kSE(L)02/21/2001 500um Fig. 10. SEM micrograph and EDS spectra for an oxide layer on the outside of the BN, the BN interphase, and the SiC fiber surface of an outside debonding YL-iBN SiC/SiC composite fracture surface after 815oC burner- rig exposure and tensile testing at room temperatu stress states would be expected to be quite complex. Nevertheles existence of a"gap"between the BN and the CVI-SiC, i.e,an the interphase and interfaces between the fibers and matrix, should already debonded interface before testing be subjected to residual tensile and shear stress. This could creat the scenario where, if the strength of those interfaces were weak enough, the interface could debond during cooling of the compos V. Conclusion It in a stress state ahead of ial debonding of the BN-CVI SiC interface MI SiC/SiC composites with bn interp rather than the fiber-BN interface In this regard, since MI systems interface debonding and sliding at the BN-CVI SiC interface ill inherently have residual compression in the matrix, they may showed significantly higher strain capabilities and intermediate be an ideal composite system to enable outside debonding. temperature stress-rupture life over conventional composites that although outside-debonding type of behavior has been observed in exhibit interface debonding and sliding at the Bn-fiber interface SiC/BN/CVI SiC minicomposites with tailored BN interfaces Higher strain to failure was attributed to lower interfacial shear Regarding the observance of a weaker BN-matrix interface for stress at the BN-CVI Sic interface, Improved intermediate outside-debonding CMCs, the presence of carbon either as a thin temperature properties were attributed to the protection from the layer outside of the bn or in an enriched form appears to be the oxidizing environment due to the adherence of the bn layer to the most likely factor, even though the detection of carbon enrichment fiber surface, which is not the situation for the inside-debonding not compelling. One other possible explanation is that differ- composites. Thus the environment does not have direct access to nces in processing conditions led to more shrinkage of the the fibers, which prohibits or stalls the rapid strength-degrading w-temperature-deposited BN. BN shrinkage, the formation of oxidative process of strongly bonding fibers to I-neighbor gap between the bn and the CVI SiC, and outside debonding have fibers. In addition, no degradation in retained th was ob- een observed for fiber/BN/CVI SiC preforms that have been served after 100 h burner-rig exposure, which typically occurs heat-treated to higher temperatures. Nevertheless, oxidation the when carbon exists at the fiber/BN occurs between the bn and the CVI-SiC matrix during burner- rig The cause of outside debonding w to(1)a exposure clearly implies either the presence of a C layer, as was eaker BN/CVI-SiC interface than he case for the earlier-mentioned composite systems where a thin caused by the presence of C at the BN/CVI-SiC C layer existed between the fiber and the BN or the residual tensile/shear stress at the bn-cvi sic interface c
January 2004 Effect of a BN Interphase That Debonds between the Interphase and the Matrix in SiC/SiC Composites 111 Fig. 10. SEM micrograph and EDS spectra for an oxide layer on the outside of the BN, the BN interphase, and the Sic fiber surface of an outside-debonding SYL-iBN SiClSiC composite fracture surface after 8 15°C burner-rig exposure and tensile testing at room temperature. stress states would be expected to be quite complex. Nevertheless, the interphase and interfaces between the fibers and matrix, should be subjected to residual tensile and shear stress. This could create the scenario where, if the strength of those interfaces were weak enough, the interface could debond during cooling of the composite or result in a stress state ahead of an approaching crack that could lead to preferential debonding of the BN-CVI Sic interface rather than the fiber-BN interface. In this regard, since MI systems will inherently have residual compression in the matrix, they may be an ideal composite system to enable outside debonding, although outside-debonding type of behavior has been observed in SiC/BN/CVI Sic minicomposites with tailored BN interfaces2’ Regarding the observance of a weaker BN-matrix interface for outside-debonding CMCs, the presence of carbon either as a thin layer outside of the BN or in an enriched form appears to be the most likely factor, even though the detection of carbon enrichment is not compelling. One other possible explanation is that differences in processing conditions led to more shrinkage of the low-temperature-deposited BN. BN shrinkage, the formation of a gap between the BN and the CVI Sic, and outside debonding have been observed for fiber/BN/CVI Sic preforms that have been heat-treated to higher temperature^.^^ Nevertheless, oxidation that occurs between the BN and the CVI-SIC matrix during burner-rig exposure clearly implies either the presence of a C layer, as was the case for the earlier-mentioned composite systems where a thin C layer existed between the fiber and the BN,’”22-23 or the existence of a “gap” between the BN and the CVI-SIC, i.e., an already debonded interface before testing. V. Conclusion MI SiC/SiC composites with BN interphases that exhibited interface debonding and sliding at the BN-CVI Sic interface showed significantly higher strain capabilities and intermediatetemperature stress-rupture life over conventional composites that exhibit interface debonding and sliding at the BN-fiber interface. Higher strain to failure was attributed to lower interfacial shear stress at the BN-CVI Sic interface. Improved intermediatetemperature properties were attributed to the protection from the oxidizing environment due to the adherence of the BN layer to the fiber surface, which is not the situation for the inside-debonding composites. Thus the environment does not have direct access to the fibers, which prohibits or stalls the rapid strength-degrading oxidative process of strongly bonding fibers to nearest-neighbor fibers. In addition, no degradation in retained strength was observed after 100 h burner-rig exposure, which typically occurs when carbon exists at the fiber/BN interface. The cause of outside debonding was believed to be due to (1) a weaker BN/CVI-Sic interface than BN/fiber interface, perhaps caused by the presence of C at the BNICVI-Sic interface, and (2) residual tensilekhear stress at the BN-CVI Sic interface that was
Journal of the American Ceramic Society-Morscher et al Vol 87. No Sic BN 200mm 200mm Fig. 11. TEM micrographs(top)and carbon map(bottom, labeled"C )for SYL-iBN SiC/SiC composites showing(a)inside-debonding composite and(b) outside-debonding composite sufficient to cause interface debonding during cooldown after L.U.J.T. Ogbuji,“ ive Mode of Oxidation Degradation in a SiC-SiC The residual stress state in SiC fiber/MI caused by the infiltration of molten Si in processing. The volume expansion of the si IAG.N. Morscher and J. L. Eldridge, " Constituent Effects tress-Strain liquid to solid phase transformation coupled with a lower thermal expansion coefficient for Si compared with SiC results in residual on Fracture. Elsevier Science, Oxford, U. K in press ompression in the matrix of SiC fiber/MI matrix composites. For 1M. Steen and】.L.val this reason, outside debonding is expected to be easier to tailor ASTM Special Technical Publication, VoL. STP 1309. Edited by M. G al or MI composites compared with other Sic fiber. SiC matrix American Society for Testing and Materials, West Conshohocken, PA ons Composites Under Quasi-Static and Cyclic Loading, " Acta Metall. Mater. 41 141 1269-8l01993) References w. A. Curtin. B. K, Ahn, and N. Takeda, " Modeling Brittle and 46[0]3409-20(1998) G. N. Morscher, "" Modal Acoustic Emission Source Determination in JAm, Ceran,So,83|6l1441-4902000 Carbide Matrix Composites -N, S. Jacobson, G, N. Morscher. D. R. Bryant, and R. E. Tressler. Nondestructive evaluation, t Temperature Oxidation of Boron Nitride: Il, Boron Nitride Layers in Composites. Thompson and D. E. Chimet rican Institute of Physics. Melville, NY. 2000 JAm, Ceram.Soc,82|6l1473-82(99) Coatings for Ceramic Matrix B, K. Ahn and w. A Curtin, "Strain and Hysteresis by Stochastic Matrix Cracking Composites, "Ceram. Eng. Sci. Proc 13 [7-81 23-56(1992) 2G. N. Morscher and J. Hurst, "Stress-Rupture and Stress-R ature, "Ceram. Eng. Sci. Proc:, 22 [31 539-46(2001 2G. N Morscher and I. D. Cawley. "Intermediate Temperature Strength Degr OH,1995 Rebillat, Lamon, R, Naslain, E. Lara-Curzio m, K. Ferber, and T, M. Besmann and Y, L. Chen " Tensile Behavior of Itilayered Interphases in SiC/SiC Chemical-Vap erfaces, J. Am. Fibe ram. Eng, Sci. Prm,, in pi 2L. U J. T Ogbuji, D. R. Wheeler, and T.R. McCue,"Process-Induced Carbon Multilayered Interphases by VI, "Key Eng- Sub-Layer in SiC/BN dater,164-165,164-65 Eng.Sci.Proc2213]379-87(201 2W. A, Curtin, "Theory of Mechanical Properties of Ceramic-Matrix Compos yC-SiC) Interphases at High Tempe ites, "J. Am. Ceram. Soc., 74 [11] 2837-45(199 aler,164-165.249-52(1999) abdeljalil and w. A. Curtin, "Strength and Reliability of Fiber-Reinforced Bertrand, R. Pailler, and J. Lamon. " Influence of Strong Composites: Localized Load-sharing and Associated Size Effects. " int. J, Solids Interfaces on the Mechanical Behavior and Lifetime of Minicomposites, J. Am. Ceram Soc., 84(41787-94(200 26z. xia and W.A "HSREP lopment Program. Mater Scl. ro,22(3]371-78(2001 H. M. Yun and J. A. DiCarlo, "Comparison of the Tensile, Creep. Interphases for Woven Sic Strength Properties of Stoichiometric SiC Fibers, "Ceram. Eng. Sci. Proc., 20 [31 e7-Marure c dFI Sci oating Systems for High Strength in Ceram. Eng. Sci. Proc.,in C. Vincent, H nL, and J. Bouix, ""SiC/SiC aded Bn Interphases, " J. cumulation in a -38(2000 R. Bhatt, private communication
112 Journal of the American Ceramic Society-Morscher et al. Vol. 87, No. I Fig. 11. TEM micrographs (top) and carbon map (bottom, labeled “C“) for SYL-iBN SiClSiC composites showing (a) inside-debonding composite and (b) outside-debonding composite. sufficient to cause interface debonding during cooldown after composite processing. The residual stress state in Sic fiber/MI composites is primarily caused by the infiltration of molten Si in the final step of matrix processing. The volume expansion of the Si liquid to solid phase transformation coupled with a lower thermal expansion coefficient for Si compared with Sic results in residual compression in the matrix of Sic fiber/MI matrix composites. For this reason, outside debonding is expected to be easier to tailor for MI composites compared with other Sic fiber, Sic matrix combinations. References ‘G. N. Morscher, J. Hurst, and D. Brewer, “Intermediate-Temperature Stress Rupture of a Woven Hi-Nicalon. BN-lnterphase. Sic-Matrix Composite in Air,” J. 9. Ceram. Soc.. 83 [6l 1441-49 (2000). -N. S. Jacobson. G. N. Morscher. D. R. Bryant, and R. E. Tressler. “HighTemperature Oxidation of Boron Nitride: 11, Boron Nitride Layers in Composites.” J. Am. Ceram. Soc., 82 161 1473-82 (1999). ’H. W. Carpenter and J. W. Bohlen, “Fiber Coatings for Ceramic Matrix Composites.” Ceram. Eng. Sci. Proc., 13 [7-81 23-56 (1992). ‘R. Naslain. ‘The Concept of Layered Interphases in SiCISiC”: pp. 23-39 in Ceramic Transactions, Vol. 58. High-Temperalure Ceramic-Matrix Composites /I: Manufacturing and Materials Development. Edited by A. G. Evans and R. Naslain. American Ceramic Society, Westerville, OH, 1995. ’F. Rebillat. J. Lamon, R. Naslain, E. Lara-Curzio, M. K. Ferber. and T. M. Besmann, “Properties of Multilayered Interphases in SiUSiC Chemical-Vapor-Infiltrated Composites with ‘Weak’ and ‘Strong’ Interfaces.” J. Am Ceram. SOC., 81 191 2315-26 (1998). ‘S. Betrand. 0. Boisron. R. Pailler, J. Lamon. and R. Naslain, “(PyCISiC),, and (BN/SiC),, Nanoscale-Multilayered Interphases by Pressure-Pulsed CVI,” Key Eng. Mufer., 164-165. 164-65 (1999). ’S. Pasquier. J. Lamon, and R. Naslain, “Static Fatigue of 2D SiCISiC Composites with Multilayered (PyC-Sic),, Interphases at High Temperatures in Air.” Key Eng. Mafer., 164-165, 249-52 (1999). “S. Bertrand, R. Pailler, and J. Lamon. “Influence of Strong FiheKoating Interfaces on the Mechanical Behavior and Lifetime of Hi-Nicalonl(PyC/SiC)~SiC Minicomposites.” J. Am. Ceram. Soc., 84 I41 787-94 (2001). ’D. Brewer, “HSREPM Combustor Materials Development Program.” Mater. Sci. Eng. A. A261, 284-91 (1999). “’H. M. Yun and J. A. DiCarlo, ‘Comparison of the Tensile, Creep, and Rupture Strength Properties of Stoichiometric Sic Fibers.” Ceram. Eng. Sci. Proc.. u) [3] 259-72 (1999). “G. N. Morscher. “Modal Acoustic Emission of Damage Accumulation in a Woven SiCISiC Composite,” Comp. Sci. Tech., 59. 687-97 (1999). I2L. U. 1. T. Ogbuji, “A Pervasive Mode of Oxidation Degradation in a Sic-Sic Composite.” J. Am. Ceram., 81 [I I] 2777 (1998). ”J. 1. Eldridge. “Desktop Fiber Push-Out Apparatus,’’ NASA Technical Mrmarandum 105341, December 1991. I4G. N. Morscher and J. 1. Eldridge, “Constituent Effects on the Stress-Strain Behavior of Woven Melt-Infiltrated Sic Composites.” ICF 10, International Congress on Fracture. Elsevier Science, Oxford, U.K., in press. I’M. Steen and J. L. Valles, “Unloading-Reloading Sequences and the Analysis of Mechanical Test Results for Continuous Fiber Ceramic Composites”: pp. 49-65 in ASTMSpecial Technical Publication, Vol. STP 1309. Edited by M. G. Jenkins el ol. American Society for Testing and Materials, West Conshohocken, PA, 1997. ”A. W. Pryce and P. A. Smith, “Matrix Cracking in Unidirectional Ceramic Matrix Composites Under Quasi-Static and Cyclic Loading,” Acta Metall. Muter.. 41 141 1269-81 (1993). ”W. A. Curtin, B. K. Ahn, and N. Takeda. “Modeling Brittle and Tough Stress-Strain Behavior in Unidirectional Ceramic Matrix Composites,” Acta Mnter.. 46 [lo] 3409-20 (1998). ‘*G. N. Morscher, “Modal Acoustic Emission Source Determination in Silicon Carbide Matrix Composites”; pp. 383-90 in Review of Progress in Qeantitarive Nondestructive Evaluation. Conference Proceedings 509. Vol. 19A. Edited by D. 0. Thompson and D. E. Chimenti. American Institute of Physics, Melville, NY, 2oOO. I9B. K. Ahn and W. A. Curtin, “Strain and Hysteresis by Stochastic Matrix Cracking in Ceramic Matrix Composites,’’ J. Mech. Phys. Solid.7, 45 121 177-209 (1997). ’“G. N. Morscher and J. Hunt, “Stress-Rupture and Stress-Relaxation of SiC/SiC Composites at Intermediate Temperature.” Ceram. Eng. Sci. Pror., 22 [3] 539-46 (2001 ). ”G. N. Morscher and I. D. Cawley, “Intermediate Temperature Strength Degradation in SiC/SiC Composites.” J. Eur. Ceram. Soc.. in press. 22H. M. Yun, J. A. DiCarlo, L. T. Ogbuji, and Y. L. Chen, “Tensile Behavior of As-Fabricated and Burner-Rig Exposed SiClSiC Composites with Hi-Nicalon Type-S Fibers,” Ceram. Eng. Sci. Proc., in press. ”L. U. J. T. Ogbuji, D. R. Wheeler, and T. R. McCue, “Process-Induced Carbon Sub-Layer in SiCIBNISiC Composites: Characterization and Consequences.” Ceram. Gig. Sci. Proc., 22 131 379-87 (2001). “W. A. Curtin, ”Theory of Mechanical Properties of Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 74 [Ill 2837-45 (1991). 25M. Ibnabdeljalil and W. A. Curtin, “Strength and Reliability of Fiber-Reinforced Composites: Localized Load-Sharing and Associated Size Effects,” lnt. J. Solids Sfruct.. 34 [21] 2649-88 (1997). ’62. Xia and W. A. Curtin, “Design of FiberKoating Systems for High Strcngth in Ceramic Matrix Composites,” Ceram. Eng. Sci. Proc., 22 [3] 371-78 (2001). ”G, N. Morscher. H. Y. Yun. and F. I. Hunvitz. “High Temperature Si-Doped BN Interphases for Woven SiC/SiC Composites,” Ceram. Eng. Sci. Proc., in press. ”S. Jacques, A. Lopez-Mmre, C. Vincent, H. Vincent, and J. Bouix. “SiCISiC Minicomposites with Structure-Graded BN Interphases.” J. Eur. Ceram. Soc. 20. 1929-38 (2000). 29R. Bhatt, private communication. 0