ournal 1. Am Ceram Soc, s1 [9) 2315-26 (1998) Properties of Multilayered Interphases in SiC/SiC Chemical-Vapor-Infiltrated Composites with Weak"and"Strong"In Francis Rebillat, Jacques Lamon, and Roger Naslain Laboratoire des Composites Thermostructuraux(LCTS), UMR 5801, CNRS-SEP-UB1, 33600 Pessac, France Edgar Lara-Curzio, Mattison K Ferber, and Theodore M. Besmann Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6064 The interfacial properties of SiC/SiC composites with in- requires thinner carbon interphases, to reduce the amount of terphases that consist of (C-SiC) sequences deposited or oxidizing phase and to enhance the self-healing capability the fibers have been determined by single-fiber push-out the material via formation of SiO2. However, interphase thick- tests, The matrix has been reinforced with either as- ness must exceed a lower limit to allow deviation of the matrix received or treated Nicalon fibers. The measured interfa- cracks. An alternative concept of multilayered interphases was cial properties are correlated with the fiber-coating bond developed at LCTS, by intercalating SiC sublayers within strength and the number of interlayers. For the composites the carbon coating. Thus, a new family of SiC/SiC composites reinforced with as-received(weakly bonded) fibers, inter- with multilayered interphases has been produced. The thick facial characteristics are extracted from the nonlinear por ness of the carbon sublayers in the interphase is now as low as tion of the stress-displacement curve by fitting Hsueh's 0.05 um. It is expected that oxidation of the matrix and of the push-out model. The interfacial characteristics are con- SiC sublayers will ensure production of SiO, that will be suf- trolled by the carbon layer adjacent to the fiber. The re- ficient to seal the matrix cracks and the associated narrow gaps sistance to interface crack growth and fiber sliding in- between the successive SiC sublayers, thus preventing the creases as the number of(c-SiC) sequences increa complete oxidation of the carbon sublayers. Furthermore, much the osites reinforced with treated(strongly self-protection of the interphases should allow efficient load fibers, the push-out curves exhibit an uncommon transfer at high temperatures in aggressive environments curvature, which reflects different modes of interphase Good mechanical properties require that a balance in fiber/ cracking and a contribution of fiber roughness matrix interactions is found to maximize load transfer while retaining the ability of the fiber to debond and slide. -y Fiber/ . Introduction matrix interactions may be tailored by selecting an appropriate T IS now well acknowledged that the properties of fiber/ to matrix interfaces determine the mechanical behavior of forced with Nicalonas' (Nippon Carbon Co., Tokyo, Japan) brittle-matrix composites 1, 2 Damage tolerance results from the fibers that have been treated prior to deposition of the Pyc deviation of matrix cracks into the fiber/matrix interface. this henomenon can be controlled via deposition of a coating on bonded, whereas the fiber-carbon coating bond is weak the fibers. The most-efficient interphase materials exhibit an those composites reinforced with as-received fibers. Trans anisotropic microstructure and a low shear modulus (-30 GP mission electron microscopy(TEM) examination of the inter- Two materials are commonly used as interphases in SiC/SiC cial region in various families of SiC/SiC composites ha composites: pyrolytic carbon or pyrocarbon(PyC) and boron shown that the sublayers of SiOz and anisotropic carbon are no nitride(BN). Carbon has remained the most-efficient inter- present anymore at the surface of the fibers that have been hase.However,it is very sensitive to oxidation at tempera- treated before interphase deposition. o Furthermore, the surface oxygen, the carbon coating is consumed, which degrades the The primary objective of the reported research is to extract load-transfer capability. At temperatures >8000-1000oC, pas- interface and interphase properties pertinent to these families of ive oxidation of the SiC matrix occurs. The silica(SiO2) that forms can seal the matrix cracks, thus inhibiting further deg effects of multilayering the g single-fiber push-out tests. The SiC/SiC composites by usil interphase and strengthening the adation of the interphase fiber bonding on interphase cracking and fiber sliding have Improving the oxidation resistance of SiC/C/SiC composites been investigated at room temperature Il. Materials and Their Mechanical Properties R. Kerans--contributing editor The SiC/SiC composites were produced via chemical vapor infiltration(CVI) of preforms of either as-received or treated Nicalon fibers. The number of( C-SiC)sequences(n)depos- No. 191642. Received June 2, 1997; appro ited on the fibers was I(single carbon-layered interphase), 2, and 4 (Table 1). at Oak Ridge National Laboratory w For the first family of composites reinforced with as- ergy, Ofto En partment of Energy, Office of Fossil Energy, Advanced Technolo Program)under Contract No. DE-AC05-96OR22464 Lockheed Martin Energy Research, Inc. Member. American Ceramic Society Proprietary treatment by SEP 2315
Properties of Multilayered Interphases in SiC/SiC Chemical-Vapor-Infiltrated Composites with ‘‘Weak’’ and ‘‘Strong’’ Interfaces Francis Rebillat,* Jacques Lamon,* and Roger Naslain* Laboratoire des Composites Thermostructuraux (LCTS), UMR 5801, CNRS-SEP-UB1, 33600 Pessac, France Edgar Lara-Curzio,* Mattison K. Ferber,* and Theodore M. Besmann* Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831–6064 The interfacial properties of SiC/SiC composites with interphases that consist of (C–SiC) sequences deposited on the fibers have been determined by single-fiber push-out tests. The matrix has been reinforced with either asreceived or treated Nicalon fibers. The measured interfacial properties are correlated with the fiber–coating bond strength and the number of interlayers. For the composites reinforced with as-received (weakly bonded) fibers, interfacial characteristics are extracted from the nonlinear portion of the stress–displacement curve by fitting Hsueh’s push-out model. The interfacial characteristics are controlled by the carbon layer adjacent to the fiber. The resistance to interface crack growth and fiber sliding increases as the number of (C–SiC) sequences increases. For the composites reinforced with treated (strongly bonded) fibers, the push-out curves exhibit an uncommon upward curvature, which reflects different modes of interphase cracking and a contribution of fiber roughness. I. Introduction I T IS now well acknowledged that the properties of fiber/ matrix interfaces determine the mechanical behavior of brittle-matrix composites.1,2 Damage tolerance results from the deviation of matrix cracks into the fiber/matrix interface. This phenomenon can be controlled via deposition of a coating on the fibers. The most-efficient interphase materials exhibit an anisotropic microstructure and a low shear modulus (∼30 GPa). Two materials are commonly used as interphases in SiC/SiC composites: pyrolytic carbon or pyrocarbon (PyC) and boron nitride (BN). Carbon has remained the most-efficient interphase. However, it is very sensitive to oxidation at temperatures >500°C. Therefore, when the interphase is exposed to oxygen, the carbon coating is consumed, which degrades the load-transfer capability. At temperatures >800°–1000°C, passive oxidation of the SiC matrix occurs. The silica (SiO2) that forms can seal the matrix cracks, thus inhibiting further degradation of the interphase.3 Improving the oxidation resistance of SiC/C/SiC composites requires thinner carbon interphases, to reduce the amount of oxidizing phase and to enhance the self-healing capability of the material via formation of SiO2. However, interphase thickness must exceed a lower limit to allow deviation of the matrix cracks. An alternative concept of multilayered interphases was developed at LCTS,4–9 by intercalating SiC sublayers within the carbon coating. Thus, a new family of SiC/SiC composites with multilayered interphases has been produced.5 The thickness of the carbon sublayers in the interphase is now as low as 0.05 mm. It is expected that oxidation of the matrix and of the SiC sublayers will ensure production of SiO2 that will be sufficient to seal the matrix cracks and the associated narrow gaps between the successive SiC sublayers, thus preventing the complete oxidation of the carbon sublayers. Furthermore, much self-protection of the interphases should allow efficient load transfer at high temperatures in aggressive environments. Good mechanical properties require that a balance in fiber/ matrix interactions is found to maximize load transfer while retaining the ability of the fiber to debond and slide.5–9 Fiber/ matrix interactions may be tailored by selecting an appropriate combination of constituents and by modifying the fiber surface topography. For this purpose, the SiC/SiC composites are reinforced with Nicalony (Nippon Carbon Co., Tokyo, Japan) fibers that have been treated prior to deposition of the PyC interphase. The fiber and the carbon coating are strongly bonded, whereas the fiber–carbon coating bond is weak in those composites reinforced with as-received fibers.5–8 Transmission electron microscopy (TEM) examination of the interfacial region in various families of SiC/SiC composites has shown that the sublayers of SiO2 and anisotropic carbon are not present anymore at the surface of the fibers that have been treated before interphase deposition.10 Furthermore, the surface of these fibers seems to be rather smooth.8 The primary objective of the reported research is to extract interface and interphase properties pertinent to these families of SiC/SiC composites by using single-fiber push-out tests. The effects of multilayering the interphase and strengthening the fiber bonding on interphase cracking and fiber sliding have been investigated at room temperature. II. Materials and Their Mechanical Properties The SiC/SiC composites were produced via chemical vapor infiltration (CVI) of preforms of either as-received or treated† Nicalon fibers.5 The number of (C–SiC) sequences (n) deposited on the fibers was 1 (single carbon-layered interphase), 2, and 4 (Table I). For the first family of composites reinforced with asR. J. Kerans—contributing editor Manuscript No. 191642. Received June 2, 1997; approved December 4, 1997. Supported by LCTS (Pessac, France) and SEP through a grant given to author FR. Work performed at Oak Ridge National Laboratory was supported by the U.S. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Transportation Technologies as part of the HTML User Program (U.S. Department of Energy, Office of Fossil Energy, Advanced Research and Technology Development Materials Program) under Contract No. DE-AC05-96OR22464 with Lockheed Martin Energy Research, Inc. *Member, American Ceramic Society. † Proprietary treatment by SEP. J. Am. Ceram. Soc., 81 [9] 2315–26 (1998) Journal 2315
Journal of the American Ceramic Society-Rebillat et al. Vol 81. No 9 Table L. Description of the Investigated Materials Material the fabrics sequences, I Nature of the C-SiC sequence in the interphase NT 2244 0.1 0.1 B F K NT SiC 0.05 T F/C/ Si 0.05 0.0 0.0 .05 0.15 nformation regarding the investigated materials was taken from Droillard. The abbreviations"NT"and"T"denote nontreated and treated conditions, respectively. 8-F" represent fiber and matrix. respectively. The numbers directly below each phase represent the thickness of the phase(in units of um). received fibers(materials 1, A, and K), the tensile stress-strain um for the composites reinforced with as-received fibers) curves(Fig. 1)exhibit relatively high fracture strains and This feature is in opposition to the concept that significant fiber stresses(0.9% and 300 MPa, respectively ).5,6 The presence of pull-out is necessary to achieve high toughness. As a result, a typical plateaulike behavior during matrix cracking indicates natrix crack spacing at saturation is rather short(Table II)and that the fiber/interphase bond is weak and load transfer is the density of cracks is very high. The residual strains are very small and the hysteresis loops width on unloading-reloading is Similar failure strains and 50%-higher stresses are obtained very narrow, the widest hysteresis loop is approximately one for those composites reinforced with treated fibers(materials J, tenth that observed for composites reinforced with as-receivec and L). The nonlinear stress-strain behavior induced by matrix cracking is observed up to the ultimate strength of the omposite(Fig. 1). This feature demonstrates rather stron ber/matrix interactions. In those composites reinforced with lL. Mechanical Characterization of interfaces treated fibers the deviation of matrix cracks involves extensive and Interphases crack branching within the interphase, as opposed to a single crack in the fiber/carbon interface of composites reinforced () Experimental Conditions with as-received fibers(Fig. 2)5-79 In the former, crack The push-out tests were conducted using the Interfacial Test propagate first along a short distance within the first carbon System developed at Oak Ridge National Laboratory. The sublayer near the matrix, then they cross the sic sublayer and top of the fibers was loaded at a constant displacement rate of continue propagating and branching in the subsequent carbon 0. 1 mm/s, using a diamond indentor mounted on a load cell blayer. Crack arrest is observed as a result of energy release The samples consisted of 500-pum-thick wedges, prepared us- ia the creation of several failure surfaces and the frictional ing standard metallographic techniques sliding of crack surfaces. Push-back tests were also conducted on the The composites that contain treated fibers have high stra inforced with treated fibers(composites J and B) energies (Table Il the fracture surfaces exhibit short es were urned over once fibers fiber pull-out lengths -30 um, in comparison to 1 15-300 completely debonded by pushout and the load was 400 00 曰20 100 Fig. 1. Tensile stress-strain curves for 2D-SiC/SiC composites with multilayered interphases n(C-SiC)and reinforced with as-received(A, L, K) eated(B, J, L) Nicalon fibers(from Droillard5)
received fibers (materials I, A, and K), the tensile stress–strain curves (Fig. 1) exhibit relatively high fracture strains and stresses (0.9% and 300 MPa, respectively).5,6 The presence of a typical plateaulike behavior during matrix cracking indicates that the fiber/interphase bond is weak and load transfer is poor.5–11 Similar failure strains and 50%-higher stresses are obtained for those composites reinforced with treated fibers (materials J, B, and L). The nonlinear stress–strain behavior induced by matrix cracking is observed up to the ultimate strength of the composite (Fig. 1). This feature demonstrates rather strong fiber/matrix interactions. In those composites reinforced with treated fibers, the deviation of matrix cracks involves extensive crack branching within the interphase, as opposed to a single crack in the fiber/carbon interface of composites reinforced with as-received fibers (Fig. 2).5–7,9 In the former, cracks propagate first along a short distance within the first carbon sublayer near the matrix, then they cross the SiC sublayer and continue propagating and branching in the subsequent carbon sublayer. Crack arrest is observed as a result of energy release via the creation of several failure surfaces and the frictional sliding of crack surfaces. The composites that contain treated fibers have high strain energies (Table II), although the fracture surfaces exhibit short fiber pull-out lengths (∼20–30 mm, in comparison to 115–300 mm for the composites reinforced with as-received fibers).6 This feature is in opposition to the concept that significant fiber pull-out is necessary to achieve high toughness.1 As a result, matrix crack spacing at saturation is rather short (Table II) and the density of cracks is very high. The residual strains are very small and the hysteresis loops width on unloading–reloading is very narrow; the widest hysteresis loop is approximately one tenth that observed for composites reinforced with as-received fibers.5,6 III. Mechanical Characterization of Interfaces and Interphases (1) Experimental Conditions The push-out tests were conducted using the Interfacial Test System developed at Oak Ridge National Laboratory.12 The top of the fibers was loaded at a constant displacement rate of 0.1 mm/s, using a diamond indentor mounted on a load cell. The samples consisted of 500-mm-thick wedges, prepared using standard metallographic techniques.13 Push-back tests were also conducted on the composites reinforced with treated fibers (composites J and B). For this purpose, the samples were turned over once fibers had been completely debonded by pushout and the load was applied to Table I. Description of the Investigated Materials† Material Nature of the fabric‡ Number of (C–SiC) sequences, n Nature of the C–SiC sequence in the interphase§ I NT 1 F / C / M 0.5 J T 1 F/ C / M 0.5 A NT 2 F / C / SiC / C / M 0.1 0.3 0.1 B T 2 F / C / SiC / C / M 0.1 0.3 0.1 K NT 4 F / C / SiC / C / SiC / C / SiC / C / M 0.05 0.05 0.05 0.1 0.05 0.15 0.05 L T 4 F / C / SiC / C / SiC / C / SiC / C / M 0.05 0.05 0.05 0.1 0.05 0.15 0.05 † Information regarding the investigated materials was taken from Droillard.5 ‡The abbreviations ‘‘NT’’ and ‘‘T’’ denote nontreated and treated conditions, respectively. § ‘‘F’’ and ‘‘M’’ represent fiber and matrix. respectively. The numbers directly below each phase represent the thickness of the phase (in units of mm). Fig. 1. Tensile stress–strain curves for 2D-SiC/SiC composites with multilayered interphases n(C–SiC) and reinforced with as-received (A,I,K) or treated (B,J,L) Nicalon fibers (from Droillard5 ). 2316 Journal of the American Ceramic Society—Rebillat et al. Vol. 81, No. 9
September 1998 Multilayered Interphases in SiC/SiC CH Composites with"Weak"and" Strong"Interfaces microcrack the top of the protruding fiber. Now, only the frictional sliding phenomenon is characterized by the stress-displacement curve. (2) Push-Out Curves and Extraction of Interface Data for Composites Reinforced with As-Received Fibers 露[ interphase] Figure 3 shows a typical plot of stress versus fiber-end dis- placement obtained for SiC/SiC composites reinforced with as-received fibers. A similar curve is obtained for SiC/SiC c fber composites with a single carbon layer or multilayered inter- phases and for other ceramic-matrix composites(CMCs) with a weak fiber-matrix bond., 15-19 An interfacial crack is created when the applied load overcomes the debond resistance of the interphase and the axial residual stress in the fiber(com- pressive in SiC/SiC)(point(b). The curvature of the subse- microcrack on the mode of propagation of the crack(stable or unstable Crack propagation is dictated by the interfacial shear stress, and the interfacial shear strength at the crack tip(debonding parameter). When the debond extends over the major portion interphase fiber length and when the stress at the crack tip reaches a critical level, crack propagation becomes catastrophic(point (c). This condition is observed in the decrease in load from the maximum stress(omax)(the so-called push-out stress) to the from frictional sliding of the fiber in the matrix(region(dh- (e). Fiber sliding occurs only over a short distance, because of contact of the diamond indentor with the matrix(point(e), Fig.2. Schematic diagram of TEM micrographs showing the modes which leads to an increase in curvature beyond point(e) of deviations of matrix microcracks observed in the multilayered in- terphases of SiC/SiC composites5-7reinforced with(a)as-received fibers and(b) treated fibers Table IL. Main Mechanical Properties of the Investigated Materials stress (nontreated, n =1) 107183 15.2-21.3 J(treated, n= 1) 24-293 B l8-22 L(treated, n=4) 3500.7 174-26 Data was taken from Droillard and Lamon . The variable n represents the number of ( C-SiC)sequences. 2500 TTTTTTTTTTTTTTTTTT 2000 00 LeBen Displacement (um) Fig 3. Plot of single-fiber push-out stress versus fiber-end displacement for a composite with untreated Nicalon fibers(composite A).(The of the load-train compliance to the measured displ ment has been subtracted
the top of the protruding fiber. Now, only the frictional sliding phenomenon is characterized by the stress–displacement curve. (2) Push-Out Curves and Extraction of Interface Data for Composites Reinforced with As-Received Fibers Figure 3 shows a typical plot of stress versus fiber-end displacement obtained for SiC/SiC composites reinforced with as-received fibers. A similar curve is obtained for SiC/SiC composites with a single carbon layer or multilayered interphases14 and for other ceramic-matrix composites (CMCs) with a weak fiber–matrix bond.2,15–19 An interfacial crack is created when the applied load overcomes the debond resistance of the interphase and the axial residual stress in the fiber (compressive in SiC/SiC) (point (b)). The curvature of the subsequent nonlinear domain (region (b)–(c)) is dependent directly on the mode of propagation of the crack (stable or unstable). Crack propagation is dictated by the interfacial shear stress, which is proportional to the radial residual stress component and the interfacial shear strength at the crack tip (debonding parameter). When the debond extends over the major portion of fiber length and when the stress at the crack tip reaches a critical level, crack propagation becomes catastrophic (point (c)). This condition is observed in the decrease in load from the maximum stress (smax) (the so-called push-out stress) to the pseudo-plateau (splateau) (point (d)). The pseudo-plateau results from frictional sliding of the fiber in the matrix (region (d)– (e)). Fiber sliding occurs only over a short distance, because of contact of the diamond indentor with the matrix (point (e)), which leads to an increase in curvature beyond point (e). Fig. 2. Schematic diagram of TEM micrographs showing the modes of deviations of matrix microcracks observed in the multilayered interphases of SiC/SiC composites5–7 reinforced with (a) as-received fibers and (b) treated fibers. Fig. 3. Plot of single-fiber push-out stress versus fiber-end displacement for a composite with untreated Nicalon fibers (composite A). (The contribution of the load-train compliance to the measured displacement has been subtracted.) Table II. Main Mechanical Properties of the Investigated Materials† Material‡ Failure stress (MPa) Failure strain (%) Young’s modulus (GPa) Matrix crack spacing at saturation (mm) Interfacial shear stress (MPa) Maximum strain energy (kJ/m2 ) I (nontreated, n 4 1) 241 1.07 183 185 4 15.2–21.3 J (treated, n 4 1) 356 1.00 170 20 370 24–29.3 A (nontreated, n 4 2) 235 0.79 196 340 2 7.8–9 B (treated, n 4 2) 317 0.63 204 30 150 18–22 K (nontreated, n 4 4) 267 0.86 205 115 9 15.8–21 L (treated, n 4 4) 350 0.76 215 20 90 17.4–26 † Data was taken from Droillard and Lamon.6 ‡The variable n represents the number of (C–SiC) sequences. September 1998 Multilayered Interphases in SiC/SiC CVI Composites with ‘‘Weak’’ and ‘‘Strong’’ Interfaces 2317
2318 Journal of the American Ceramic SocieryRebillat et al. Vol 81. No 9 m (b 1 um Fig. 4. Micrographs showing the surface of a protruding fiber after push-out tests performed on SiC/SiC composite A with untreated fi. Fig. 6. SEM micrograph rs protruding after push-out tests on bers and a multilayered interphase(n(C-SiC)=2)(a) protruding composites with treated f multilayered interphases(a) SiC/ fiber and(b)sliding fiber surface) SiC composite B and(b) y00 f 4000 Displacement (um) Fig. 5. Plot of push-out stress versus fiber-end displacement for a composite with treated Nicalon fibers(composite B). (The contribution of the load-train compliance to the measured displacement has been subtracted
Fig. 4. Micrographs showing the surface of a protruding fiber after push-out tests performed on SiC/SiC composite A with untreated fibers and a multilayered interphase (n(C–SiC) 4 2) ((a) protruding fiber and (b) sliding fiber surface). Fig. 5. Plot of push-out stress versus fiber-end displacement for a composite with treated Nicalon fibers (composite B). (The contribution of the load-train compliance to the measured displacement has been subtracted.) Fig. 6. SEM micrographs of fibers protruding after push-out tests on composites with treated fibers and multilayered interphases ((a) SiC/ SiC composite B and (b) SiC/SiC composite L). 2318 Journal of the American Ceramic Society—Rebillat et al. Vol. 81, No. 9
September 1998 Multilayered Interphases in SiC/SiC CHI Composites with"Weak"and" Strong" Interfaces 3000 °531n 20 UCCUEKEf aeg( rcan 0 Fig. 7. Typical curve for a first push-back(composite B). The average frictional shear stress(T)may be estimated from the displacement of the top of the fiber is below the range of the platear accuracy of the equipment. The subsequent load decrease(re- lon(bHc))corresponds to debonding and sliding along a distance shorter than the embedded fiber length. At point( b),a ( critical stress level is reached at the crack tip and the resulting nonlinear portion(region(cHd) displays an upward curvature where t is the embedded length of the fiber and r is the fiber that is attributed to increasing compression along the debond radius. As previously mentioned, the curvature of the nonlinear that does not extend. From point(d)to point(e), the downward domain(region(bH(c) is dictated by the propagation of the curvature is indicative of controlled crack propagation. finally crack and sliding of the debonded fiber. Thus, fitting the in the load decrease from point(e) to the plateau(f) can be at- terfacial model of Hsueh b allows extraction of the interfacial tributed to the mechanisms that were previously described for shear stress. Fitting involves adjustment of three parameters composites reinforced with untreated fibers. Again, Hsueh's the equations have been detailed by Rebillat et al. ) the model b does not apply to push-out curves with an upward oefficient of friction(u), the residual clamping stress(o ), and curvature and a decreasing compliance, because this behavior the residual axial stress in the fiber(o,). 6 The interfacial sl is different from that described in the model. Therefore. the stress is then given by the following equation: frictional shear stress was estimated from the plateau using (2) Observation of protruding fibers using SEM revealed a where o.(the contribution of the Poisson expansion of the uther rough surface (Fig. 6). The debris from the surface have fiber)is derived from the applied load using the equation given by Hsueh. 16 (4) Push-Back Curves and Extraction of slidin The SEM image of a protruding fiber after a push-out test Properties for Composites Reinforced with Treated Fibers (Fig. 4(a)) indicates that the debonding occurs at the fiber/ The stress-displacement curves from push-back tests(Fig coating interface. The fiber surface has rather smooth features, 7)are similar to those from the standard push-out tests(Fig 3). s is usually observed for composites with weak interfaces. involving an initial linear region, a nonlinear region with a However, at higher magnification, a level of roughness is de- downward slope, a load decrease, and a plateau. A previous tected at the fiber surface(Fig. 4(b)) study on analogous two-dimensional(2D) SiC/SiC composites with a single carbon interphase 4 noted a similar push-back 3 Push-Out Curves and Extraction of interphase data behavior. The initial linear domain indicates the elastic defor- for Composites Reinforced with Treated Fibers In contrast to the"well-behaved"push-out curves shown in attributed to the progressive sliding of the fiber via the increas- Fig 3 for composites with untreated fibers, the push-out curves ing sliding distance as the stress overcomes the static frictional obtained for the composites with treated fibers are quite dif- resistance to sliding. The load decrease reflects movement of ferent and follow trends that differ from those assumed in the the fiber end prior to sliding over the length of the test push-out models. Figure 5 shows a typical plot of stress men, as indicated by the pseudo-plateau versus fiber-end displacement obtained for a composite rein- The push-back curve can be analyzed using Hsueh's push forced with treated fibers(composite B). This behavior has out model. 6 The reseating load decrease(point (d))in the been investigated on SiC/SiC composites that possess a single plateau shows that the fiber recovers its initial position before carbon interphase. The results can be applied to the compos- protruding from the other side. The wavelength of the reseating ites with multilayered interphases load decrease is directly related to the wavelength of roughness ss In the apparently linear portion orrig. 5(region (a(b), along the sliding surface. 21-25 rt cracks are created in the interphase. The presence of such Observations of the protruding fiber surface after a push short cracks in SiC/SiC composites that possess a single carbon back test using SEM show that the interphase is partially de- terphase was indicated by nanoindentation curves. The stroyed(Fig. 8). Fragments are present at the fiber end, and short cracks cannot be detected by microindentation because hese pieces seem to be made of carbon because crystal fractu
The average frictional shear stress (t) may be estimated from the plateau: tplateau = splateaur 2t (1) where t is the embedded length of the fiber and r is the fiber radius. As previously mentioned, the curvature of the nonlinear domain (region (b)–(c)) is dictated by the propagation of the crack and sliding of the debonded fiber. Thus, fitting the interfacial model of Hsueh16 allows extraction of the interfacial shear stress. Fitting involves adjustment of three parameters (the equations have been detailed by Rebillat et al.14): the coefficient of friction (m), the residual clamping stress (sc), and the residual axial stress in the fiber (sz).16 The interfacial shear stress is then given by the following equation: t 4 −m(sc + sp) (2) where sp (the contribution of the Poisson expansion of the fiber) is derived from the applied load using the equation given by Hsueh.16 The SEM image of a protruding fiber after a push-out test (Fig. 4(a)) indicates that the debonding occurs at the fiber/ coating interface. The fiber surface has rather smooth features, as is usually observed for composites with weak interfaces. However, at higher magnification, a level of roughness is detected at the fiber surface (Fig. 4(b)). (3) Push-Out Curves and Extraction of Interphase Data for Composites Reinforced with Treated Fibers In contrast to the ‘‘well-behaved’’ push-out curves shown in Fig. 3 for composites with untreated fibers, the push-out curves obtained for the composites with treated fibers are quite different and follow trends that differ from those assumed in the push-out models.17–20 Figure 5 shows a typical plot of stress versus fiber-end displacement obtained for a composite reinforced with treated fibers (composite B). This behavior has been investigated on SiC/SiC composites that possess a single carbon interphase.14 The results can be applied to the composites with multilayered interphases. In the apparently linear portion of Fig. 5 (region (a)–(b)), short cracks are created in the interphase. The presence of such short cracks in SiC/SiC composites that possess a single carbon interphase was indicated by nanoindentation curves.14 The short cracks cannot be detected by microindentation because the displacement of the top of the fiber is below the range of accuracy of the equipment. The subsequent load decrease (region (b)–(c)) corresponds to debonding and sliding along a distance shorter than the embedded fiber length. At point (b), a critical stress level is reached at the crack tip and the resulting nonlinear portion (region (c)–(d)) displays an upward curvature that is attributed to increasing compression along the debond that does not extend. From point (d) to point (e), the downward curvature is indicative of controlled crack propagation. Finally, the load decrease from point (e) to the plateau (f) can be attributed to the mechanisms that were previously described for composites reinforced with untreated fibers. Again, Hsueh’s model16 does not apply to push-out curves with an upward curvature and a decreasing compliance, because this behavior is different from that described in the model. Therefore, the frictional shear stress was estimated from the plateau using Eq. (1). Observation of protruding fibers using SEM revealed a rather rough surface (Fig. 6). The debris from the surface have the appearance of carbon flakes.5–7 (4) Push-Back Curves and Extraction of Sliding Properties for Composites Reinforced with Treated Fibers The stress–displacement curves from push-back tests (Fig. 7) are similar to those from the standard push-out tests (Fig. 3), involving an initial linear region, a nonlinear region with a downward slope, a load decrease, and a plateau. A previous study on analogous two-dimensional (2D) SiC/SiC composites with a single carbon interphase14 noted a similar push-back behavior. The initial linear domain indicates the elastic deformation of the fiber. The nonlinear domain (region (b)–(c)) is attributed to the progressive sliding of the fiber via the increasing sliding distance as the stress overcomes the static frictional resistance to sliding. The load decrease reflects movement of the fiber end prior to sliding over the length of the test specimen, as indicated by the pseudo-plateau. The push-back curve can be analyzed using Hsueh’s pushout model.16 The reseating load decrease (point (d)) in the plateau shows that the fiber recovers its initial position before protruding from the other side. The wavelength of the reseating load decrease is directly related to the wavelength of roughness along the sliding surface.21–25 Observations of the protruding fiber surface after a pushback test using SEM show that the interphase is partially destroyed (Fig. 8). Fragments are present at the fiber end, and these pieces seem to be made of carbon because crystal fracture Fig. 7. Typical curve for a first push-back (composite B). September 1998 Multilayered Interphases in SiC/SiC CVI Composites with ‘‘Weak’’ and ‘‘Strong’’ Interfaces 2319
Journal of the American Ceramic SocieryRebillat et al. Vol 81. No 9 M F 1 gm Fig 8. SEM micrographs on protruding fibers after a push-back test on composite b((a) protruding fiber and(b) sliding fiber surfaces). surfaces are not apparent. Deep grooves can also be observed significantly larger than Plateau (Table Ill). Moreover, T data parallel to the direction of sliding. These reflect fiber- for the multilayered interphases(composites A and K)are sim- ilar to those estimated for carbon interphases thinner than 0.5 IV. Results The highest axial and radial residual stresses were also ob- tained for the composites with multilayered interphases(Fig A 30% scatter was obtained for all the interfacial 11). However, these residual stresses, derived from Hsueh ments(Tables III-V). The influence of the number of ( C-SiC) model, are too high when compared to those generally de- ayers(n) can be observed in comparison of these characteris- rmined for SiC/C/SiC composites that possess a carbon tics with previous results terphase 28-30 The magnitude of the residual radial stres at the fiber surface estimated using a coaxial cylinder-based pendent on the interphase thickness; it de- ()Composites Reinforced with As-Received Fibers creases from 100 MPa to -200 MPa as the carbon thickness Higher resistances to fiber debonding and to frictional slid decreases from I um to 0 um. 9 Magnitudes of 40 MPa and g seem to be observed for composites that possess multilay 50 MPa have been obtained for typical thicknesses of 0.5 and ered interphases, as indicated by the plots of debonding stress 0.2 um, The overestimation of the clamping stress highlights a (Fig. 9)and frictional shear stress(Fig. 10)and by the fric limit of hsueh's model in that it does not consider the effect of tion coefficients(Table III). The frictional shear stress(t) de- surface roughness during sliding. This shortcoming was over- rived from the nonlinear domain of the push-out curves are come by introducing a radial compressive stress at the inter-
surfaces are not apparent. Deep grooves can also be observed parallel to the direction of sliding. These grooves reflect fibersurface roughness and confirm that fiber sliding was restricted. IV. Results A 30% scatter was obtained for all the interfacial measurements (Tables III–V). The influence of the number of (C–SiC) layers (n) can be observed in comparison of these characteristics with previous results on composites with a single carbon interphase.14 (1) Composites Reinforced with As-Received Fibers Higher resistances to fiber debonding and to frictional sliding seem to be observed for composites that possess multilayered interphases, as indicated by the plots of debonding stress (Fig. 9) and frictional shear stress (Fig. 10) and by the friction coefficients (Table III). The frictional shear stress (t) derived from the nonlinear domain of the push-out curves are significantly larger than tplateau (Table III). Moreover, t data for the multilayered interphases (composites A and K) are similar to those estimated for carbon interphases thinner than 0.5 mm.15,26,27 The highest axial and radial residual stresses were also obtained for the composites with multilayered interphases (Fig. 11). However, these residual stresses, derived from Hsueh’s model,16 are too high when compared to those generally determined for SiC/C/SiC composites that possess a carbon interphase.28–30 The magnitude of the residual radial stress at the fiber surface estimated using a coaxial cylinder-based model28–30 is dependent on the interphase thickness; it decreases from 100 MPa to −200 MPa, as the carbon thickness decreases from 1 mm to 0 mm.29 Magnitudes of 40 MPa and −50 MPa have been obtained for typical thicknesses of 0.5 and 0.2 mm. The overestimation of the clamping stress highlights a limit of Hsueh’s model in that it does not consider the effect of surface roughness during sliding. This shortcoming was overcome by introducing a radial compressive stress at the interFig. 8. SEM micrographs on protruding fibers after a push-back test on composite B ((a) protruding fiber and (b) sliding fiber surfaces). 2320 Journal of the American Ceramic Society—Rebillat et al. Vol. 81, No. 9
September 1998 Multilayered Interphases in SiC/SiC CHI Composites with"Weak"and" Strong" Interfaces 5000 苏 untreated 1000fe 5 n(C,SIC)layer Fig 9. Debond and maximum stresses in SiC/SiC composites, as a function of the number of(C-SiC) layers(n)in the inter o untreated fiber plateau+treated fibers n(C,SIC) layer Fig. 10. Interfacial shear stress in SiC/SiC composites, as a function of the number of( C-Sic) layers(n)in the interphase face. Thus, o is expressed as the sum of the true thermally bonding has not been yet modeled, only the following values induced residual stress and a compressive stress component could be determined: (i) the debonding stress(oa)(also called that accounts for a roughness contribution. The amplitude(A) the critical debonding stress for unstable debonding ),(ii)the of the roughness was derived from o using the following maximum stress, (iii) the displacement at the maximum load, and (iv) the frictional shear stress(Plateau)(Table IV) E(1 +vm)+Em(l-ve It was very difficult to push the fibers out of the matrix, AaArr(3) and satisfactory fiber push-out was possible only for embed- ded lengths of <190 um. At these lengths, the applied loads ete to unity for an infinite remained moderate so that the fibers were not damaged On the basis of the above estimates of radial (maximum load is -120 g for a Nicalon fiber) and the dia- stresses, a roughness amplitude A of -50 nm was mond indentor did not penetrate the matrix before the load Table Ill). The above-mentioned results apply to the multilay- decrease(diamond penetration is evidenced by a steep stiffness ered interphases. The residual stresses operating on the fiber are determined by the amount of carbon in the interphase. They The stresses required to debond(2500 MPa)or to push out are dependent on the total thickness of carbon layers (4000 MPa) the fibers are uncommonly high for such thin (2) Composites Reinforced with Treated Fibers samples(-150 um)(Fig. 9). Although fiber displacements are similar to those measured on composites with weakly bonded (A) Push-Out Tests: Because the push-out behavior ob- fibers, the embedded lengths are only half as long. The fric- served for the composites with strong fiber/coating interface tional shear stresses calculated from the plateau are -100 MPa
face. Thus, sc is expressed as the sum of the true thermally induced residual stress and a compressive stress component that accounts for a roughness contribution. The amplitude (A) of the roughness was derived from sc using the following expression:18 A = H− scF Ef~1 + nm! + Em~1 − nf! qEmEf G − DaDTJr (3) where q is a parameter equal to unity for an infinite matrix. On the basis of the above estimates of radial residual stresses, a roughness amplitude A of ∼50 nm was obtained (Table III). The above-mentioned results apply to the multilayered interphases. The residual stresses operating on the fiber are determined by the amount of carbon in the interphase. They are dependent on the total thickness of carbon layers. (2) Composites Reinforced with Treated Fibers (A) Push-Out Tests: Because the push-out behavior observed for the composites with strong fiber/coating interface bonding has not been yet modeled, only the following values could be determined: (i) the debonding stress (sd) (also called the critical debonding stress for unstable debonding14), (ii) the maximum stress, (iii) the displacement at the maximum load, and (iv) the frictional shear stress (tplateau) (Table IV). It was very difficult to push the fibers out of the matrix, and satisfactory fiber push-out was possible only for embedded lengths of 2500 MPa) or to push out (>4000 MPa) the fibers are uncommonly high for such thin samples (∼150 mm) (Fig. 9). Although fiber displacements are similar to those measured on composites with weakly bonded fibers, the embedded lengths are only half as long. The frictional shear stresses calculated from the plateau are ∼100 MPa Fig. 9. Debond and maximum stresses in SiC/SiC composites, as a function of the number of (C–SiC) layers (n) in the interphase. Fig. 10. Interfacial shear stress in SiC/SiC composites, as a function of the number of (C–SiC) layers (n) in the interphase. September 1998 Multilayered Interphases in SiC/SiC CVI Composites with ‘‘Weak’’ and ‘‘Strong’’ Interfaces 2321
Journal of the American Ceramic Society-Rebillat et al. Vol 81. No 9 8 事2 aE58ag edw sens 送到 EE
Table III. Interface Characteristics, As a Function of the Number of (C–SiC) Sequnces (n), for the SiC/SiC Composites Reinforced with Untreated Fibers† Sample Embedded length (mm) Debonding stress (MPa) Maximum stress (MPa) Stress drop with extraction (MPa) Imposed displacement (mm) Shear stress from plateau (MPa) Shear stress from fitting (MPa) Friction coefficient, m Axial stress (MPa) Clamping stress (MPa) Debond length (mm) Amplitude of roughness (nm) Ratio of debonding/drop stresses I (n 4 1) 390 510 1380 735 0.8 9.4 14 0.022 −345 −720 250 44 1.1 (30) (280) (735) (530) (0.6) (4.7) (8) (0.011) (300) (240) (80) (20) (0.5) A (n 4 2) 440 810 2930 915 1.7 19.3 31 0.031 −670 −1000 280 63 0.9 (10) (280) (725) (315) (0.6) (5) (7) (0.007) (340) (66) (80) (6) (0.3) K (n 4 4) 480 640 1850 500 0.6 12.53 28 0.037 −635 −890 190 50 1.6 (10) (160) (600) (250) (0.3) (3.5) (13) (0.016) (175) (250) (60) (20) (0.8) †The value in parentheses given beneath each measurement is the standard deviation. Fig. 11. Residual stresses in SiC/SiC composites, as a function of the number of (C–SiC) layers (n) in the interphase. 2322 Journal of the American Ceramic Society—Rebillat et al. Vol. 81, No. 9
September 1998 Multilayered Interphases in SiC/SiC CH Composites with"Weak"and" Strong" Interfaces 2323 Table IV. Interface Characteristics, As a Function of the Number of (C-SiC) Sequences(n), for the SiC/SiC Composites Reinforeed with Treated Fibers Embedded Stress drop Shear stress Ratio of cement with extraction Sample (MPa) MPa) (MPa) n (1400) 4370 133 (13 (0.59) (26) L(=4) 3070 4230 (930) (0.38) 6 he value in parentheses given beneath each measurement is the standard deviation, 'From Hsueh. 6 (Table IV), which is almost one order of magnitude higher than grains that correspond to the SiC sublayer (i.e, microcrystal- those determined for untreated fibers. However, they are in line grains on the order of the sublayer thickness ). Con- L from hysteresis loop widths measured using specimens tested showed the presence of thin sections of carbon that appeared under tension(Table II). ,6 The discrepancy between the push- torn and remained partially bonded to the surface. Thus, one out and the hysteresis-loop techniques that is observed for com- the sliding surfaces corresponds to a Sic sublayer and the other posite J has not been elucidated yet. However, it may be related corresponds to a carbon sublayer terface in those specimens tested under tension. Indeed, a par- (B) Push-Back Tests: The critical stress that corresponds to the presence of a much-more-tortuous debonding/sliding in- the limit of linearity during the pushback of strongly bonded ticularly intense crack branching has been detected in compos- fibers is higher than the debonding stress obtained during the te J tensile specimens. 5-7 push-out tests on composites with weak interfaces. This obser Contrary to previous results on composites reinforced with vation confirms that fiber sliding is significantly limited in as-received fibers, the composite that shows the strongest re- composites reinforced with initially strongly bonded fibers sistance to interfacial cracking is that with only a single carbon (Figs. 3 and 7, Tables Ill and V layer interphase, because it exhibits the highest debonding and The interfacial parameters extracted from the push-back maximum stress(Fig. 9). Nevertheless, the displacements at curves show the previously mentioned dependence on the num- maximum stress (Table IV) and the frictional shear stresses ber of n(C-SiC) sequences(n I and 2). The frictional re- ( Fig. 10)that were observed for the SiC/SiC composites with sistance to fiber sliding in a composite with a multilayered multilayered interphases are also indicative of a higher resis- interphase is unambiguously higher than that obtained for a tance to sliding. composite with a single carbon interphase, as indicated by the seM observations of the debond location indicated debonding and the maximum stress and the interfacial shea ences between single-carbon-layer and multiple-layer stress (Table v) he composites with a single carbon The following characteristics of roughness have been de- (0.5 um thick), the debond crack was always detected in the rived from the clamping stresses and the reseating load de- interior of the carbon layer, 4 whereas in composites with mul crease: amplitude of -60 nm and wavelength of -350 nm(note tilayered interphases, it is located at the first carbon/SiC sub- that the wavelength is -500 nm for a single carbon interphase) layer interface near the fiber or near that interface. The crack Figure 13 shows that the interfacial shear stresses are much deviations were similar in tensile specimens. -7 As previously smaller during pushback than those derived from push-out mentioned, TEM and SEM analyses- have shown cracks lo tests. A study on the wear of sliding surfaces during push-back cated within the interphase(single carbon layer)or in the car- tests performed on bon sublayer near the fiber in multilayered interphases, as sum- showed that wear is severe at the onset from reverse sliding. 31 marized by the schematic diagram in Fig. 2. Furthermore, changes in the sliding parameters for a glass- SEM observation of the loaded side of the fibers(Fig. 12(b)) matrix system have been attributed to wear of the interface showed that the sliding surface consists of submicrometer asperities that occur during pushout. 24 In composites with M 0.5 Fig. 12. SEM micrographs of sliding surfaces involved in SiC/SiC composites with treated fibers and multilayered interphases(a) fiber side, composite B, and(b)matrix side, composite L)
(Table IV), which is almost one order of magnitude higher than those determined for untreated fibers. However, they are in excellent agreement with those extracted for composites B and L from hysteresis loop widths measured using specimens tested under tension (Table II).5,6 The discrepancy between the pushout and the hysteresis-loop techniques that is observed for composite J has not been elucidated yet. However, it may be related to the presence of a much-more-tortuous debonding/sliding interface in those specimens tested under tension. Indeed, a particularly intense crack branching has been detected in composite J tensile specimens.5–7 Contrary to previous results on composites reinforced with as-received fibers, the composite that shows the strongest resistance to interfacial cracking is that with only a single carbon layer interphase, because it exhibits the highest debonding and maximum stress (Fig. 9). Nevertheless, the displacements at maximum stress (Table IV) and the frictional shear stresses (Fig. 10) that were observed for the SiC/SiC composites with multilayered interphases are also indicative of a higher resistance to sliding. SEM observations of the debond location indicated differences between single-carbon-layer and multiple-layer interfaces (Fig. 12). In the composites with a single carbon layer (0.5 mm thick), the debond crack was always detected in the interior of the carbon layer,14 whereas in composites with multilayered interphases, it is located at the first carbon/SiC sublayer interface near the fiber or near that interface. The crack deviations were similar in tensile specimens.5–7 As previously mentioned, TEM and SEM analyses5–7 have shown cracks located within the interphase (single carbon layer) or in the carbon sublayer near the fiber in multilayered interphases, as summarized by the schematic diagram in Fig. 2. SEM observation of the loaded side of the fibers (Fig. 12(b)) showed that the sliding surface consists of submicrometer grains that correspond to the SiC sublayer (i.e., microcrystalline grains on the order of the sublayer thickness.5 ). Conversely, SEM observation of the surface of protruding fibers showed the presence of thin sections of carbon that appeared torn and remained partially bonded to the surface. Thus, one of the sliding surfaces corresponds to a SiC sublayer and the other corresponds to a carbon sublayer. (B) Push-Back Tests: The critical stress that corresponds to the limit of linearity during the pushback of strongly bonded fibers is higher than the debonding stress obtained during the push-out tests on composites with weak interfaces. This observation confirms that fiber sliding is significantly limited in composites reinforced with initially strongly bonded fibers (Figs. 3 and 7, Tables III and V). The interfacial parameters extracted from the push-back curves show the previously mentioned dependence on the number of n(C–SiC) sequences (n 4 1 and 2). The frictional resistance to fiber sliding in a composite with a multilayered interphase is unambiguously higher than that obtained for a composite with a single carbon interphase, as indicated by the debonding and the maximum stress and the interfacial shear stress (Table V). The following characteristics of roughness have been derived from the clamping stresses and the reseating load decrease: amplitude of ∼60 nm and wavelength of ∼350 nm (note that the wavelength is ∼500 nm for a single carbon interphase). Figure 13 shows that the interfacial shear stresses are much smaller during pushback than those derived from push-out tests. A study on the wear of sliding surfaces during push-back tests performed on composites with weakly bonded fibers showed that wear is severe at the onset from reverse sliding.31 Furthermore, changes in the sliding parameters for a glassmatrix system have been attributed to wear of the interface asperities that occur during pushout.24 In composites with Fig. 12. SEM micrographs of sliding surfaces involved in SiC/SiC composites with treated fibers and multilayered interphases ((a) fiber side, composite B, and (b) matrix side, composite L). Table IV. Interface Characteristics, As a Function of the Number of (C–SiC) Sequences (n), for the SiC/SiC Composites Reinforced with Treated Fibers† Sample Embedded length (mm) Debonding stress (MPa) Maximum stress (MPa) Imposed displacement (mm) Stress drop with extraction (MPa) Shear stress from plateau (MPa) Ratio of debonding/drop stresses J ‡ (n 4 1) 140 3730 4950 1.01 890 100 5.2 (20) (400) (1400) (0.43) (390) (38) (2.6) B (n 4 2) 110 2650 4370 1.22 960 133 3.7 (40) (600) (1390) (0.59) (400) (51) (2.6) L (n 4 4) 160 3070 4230 1.13 1120 90 3.2 (80) (630) (930) (0.38) (420) (26) (1.6) † The value in parentheses given beneath each measurement is the standard deviation. ‡ From Hsueh.16 September 1998 Multilayered Interphases in SiC/SiC CVI Composites with ‘‘Weak’’ and ‘‘Strong’’ Interfaces 2323
Journal of the American Ceramic Society-Rebillat et al. Vol 81. No 9 28 药兰 EE32E兰89月8a 导88房8导月 edw sans a三 ≌月9巴
Table V. Evolution of the Interface Characteristics during Push-Back Tests Performed on SiC/SiC Composites Reinforced with Treated Fibers† Number of push-back tests Debonding stress (MPa) Maximum stress (MPa) Imposed displacement (mm) Stress drop with extraction (MPa) Shear stress from plateau (MPa) Shear stress from fitting (MPa) Friction coefficient, m Axial stress (MPa) Clamping stress (MPa) Debonded length (mm) Amplitude of roughness (nm) Wavelength of roughness (nm) Ratio of debonding/drop stresses Sample J, n 4 1, thickness of 150 mm (from Hsueh16) 0 (push-out) 3630 4345 0.8 935 80 4.2 (590) (1570) (0.3) (240) (30) (1.5) 1 970 2040 0.6 490 40 35 0.034 −850 −1030 114 60 490 2.5 (350) (650) (0.3) (220) (8) (20) (0.02) (390) (50) (35) (30) (110) (1.8) Sample B, n 4 2, thickness of 115 mm 0 (push-out) 2675 5000 2.1 1240 120 2.2 (350) (800) (0.4) (215) (20) (0.4) 1 1160 3200 0.3 1360 60 160 0.143 −780 −1150 44 62 370 0.9 (100) (300) (0.1) (130) (6) (50) (0.04) (450) (220) (5) (12) (60) (0.1) †The value in parentheses beneath each measurement is the standard deviation. Fig. 13. Evolution of the interfacial shear stress with n(C–SiC) in the interphase, from push-back tests on composites reinforced with treated fibers. 2324 Journal of the American Ceramic Society—Rebillat et al. Vol. 81, No. 9