N P Bansal Journal of the European Ceramic Society 29(2009)525-535 Table 2 =194 Effects of annealing on fiber push-in test resul directional Triction=10 MP Hi-Nicalon/BN/SIC/BSAS composite; 12 plies, Vi=0.32(#Hi-NIC-BSAS-6- 24-97) Annealing conditions Friction (MPa) Debond As-fab 0.31士0.14 100h, 0.53±0.47a 8.33±1.72a Measured for fibers in the interior of the annealed CMC sample Fiber displacement, 1 um/div Poisson expansion of the fibers leads to an overestimation of Load versus fiber displacement curve recorded during fiber T friction values, the relative changes in friction with load cycling as-fabricated celsian matrix composite reinforced with BNS could be followed using Eq (2). In addition, a debond initiation stress, od, could be calculated from the debond initiation load Fa (load at which fiber end begins to move during first loading larly in the unannealed sample. Surface of the debonded fibers cycle)by the relation appears to be smooth indicating no fiber degradation or chemi- cal reaction with the matrix during high-temperature annealing od atll00°C The results of fiber push-in data are summarized in Table 2. 3.8. Fiber-matrix interface The scatters in the values of debond stress (od) and frictional sliding stress(tf) are due to the accuracy of measurements In as-fabricated fiber-reinforced composites, the fibers may for a particular fiber as well from the variations in experience a several hundred MPa clamping force upon cool- ues from fiber-to-fiber. The variation in frictional sliding stress ing from the CMC processing temperature. Thermal mismatch (tr) during successive loading-unloading cycles was small and stresses result from smaller thermal expansion of reinforcing within the standard deviation. Values of debond stress, od, and fibers as compared to the matrix. A compliant layer is necessary frictional sliding stres to reduce the stresses. The importance of a compliant interface 0.31+0.14 GPa and 10.4+3. 1 MPa, respectively, compared yer for a strong and tough CMC has been emphasized by with 0.53+0.47 GPa and 8.33+1.72 MPa for the fibers in the various workers. -Both graphitic and pyrolytic carbon and interior of the 1200C annealed sample. These results indicate hexagonal or turbostratic BN have exceptionally low moduli. that only the outer ply of the 1200 C annealed CMC specimens Ceramic composites that demonstrate good damage tolerance has been degraded from oxidation whereas the bulk interior part generally contain carbon or BN layer between the fiber and remains unaffected matrix with some exceptions such as porous oxide matrix CMCs For strong and particularly tough CMCs, the fiber-matrix 3.9. Degradation mechanism at 1200oC interface must be sufficiently weak to allow debonding at the interface, yet strong enough for effective load transfer from the a possible mechanism explaining the various steps involved matrix to the fiber. Fiber-matrix debonding and frictional slid the degradation of Hi-Nicalon/BN/SiC/BSAS CMC on ing stresses at the fiber-matrix interface were evaluated from annealing in air at 1200Cispresented in Fig 12. During anneal fiber push-in tests. About 12-15 fibers/coupon were individu- ing at 1200 C, the surface matrix layer delaminates from the ally pushed in for the air annealed and as-fabricated composites. composite ply underneath(SEM micrograph of Fig. 5), probably A typical cyclic push-in curve at room temperature for the as- due to the large CTE mismatch between the Hi-Nicalon fibers cycle, is given in Fig. 11. The data were analyzed by first sub- gen into the composite. This causes the oxidation of SiC ibig eceived composite, along with reloading part of the second and the celsian matrix. This facilitates the ingression of ox tracting the appropriate load-train compliance correction from particularly those which have lost the duplex CVD coating(see the measured displacements. An estimate of frictional slidi Fig. 2), resulting in the formation of Sioz stress, friction, was obtained using the constant Friction model of Marshall and Oliver25 which includes effects of residual SiC()+O2(g)- SiO2(s)+Co(g)+ CO2(g) stresses, but does not consider fiber roughness or Poisson expan- sion. Value of friction was determined by fitting the compliance tion of surface matrix layer in FRC corrected data from each reloading curve to the relationship Oxidation of sic fibers O2→Si02+00+cO2 2r3 Reaction of SiOz and Celsian: Et friction Low m.p. ternary phase where u is the fiber end displacement, uo is the residual fiber end On cooling: formation of Celsian crystals in glass matrix displacement after the previous unloading, Fis the applied load, Fig. 12. Proposed mechanism for degradation of Hi-Nicalon/BN/SiC/BSAS rf is the fiberradius, and Ef is the fibermodulus. While neglecting composites during annealing at 1200C in ai532 N.P. Bansal / Journal of the European Ceramic Society 29 (2009) 525–535 Fig. 11. Load versus fiber displacement curve recorded during fiber push-in test of as-fabricated celsian matrix composite reinforced with BN/SiC-coated Hi-Nicalon fibers. larly in the unannealed sample. Surface of the debonded fibers appears to be smooth indicating no fiber degradation or chemi￾cal reaction with the matrix during high-temperature annealing at 1100 ◦C. 3.8. Fiber–matrix interface In as-fabricated fiber-reinforced composites, the fibers may experience a several hundred MPa clamping force upon cool￾ing from the CMC processing temperature. Thermal mismatch stresses result from smaller thermal expansion of reinforcing fibers as compared to the matrix. A compliant layer is necessary to reduce the stresses. The importance of a compliant interface layer for a strong and tough CMC has been emphasized by various workers.26–29 Both graphitic and pyrolytic carbon and hexagonal or turbostratic BN have exceptionally low moduli. Ceramic composites that demonstrate good damage tolerance generally contain carbon or BN layer between the fiber and matrix with some exceptions such as porous oxide matrix CMCs. For strong and particularly tough CMCs, the fiber–matrix interface must be sufficiently weak to allow debonding at the interface, yet strong enough for effective load transfer from the matrix to the fiber. Fiber–matrix debonding and frictional slid￾ing stresses at the fiber–matrix interface were evaluated from fiber push-in tests. About 12–15 fibers/coupon were individu￾ally pushed in for the air annealed and as-fabricated composites. A typical cyclic push-in curve at room temperature for the as￾received composite, along with reloading part of the second cycle, is given in Fig. 11. The data were analyzed by first sub￾tracting the appropriate load-train compliance correction from the measured displacements. An estimate of frictional sliding stress, τfriction, was obtained using the constant τfriction model of Marshall and Oliver25 which includes effects of residual stresses, but does not consider fiber roughness or Poisson expan￾sion. Value of τfriction was determined by fitting the compliance corrected data from each reloading curve to the relationship u = u0 +  F2 8π2 r3 f Ef τfriction  (2) where u is the fiber end displacement, u0 is the residual fiber end displacement after the previous unloading, F is the applied load, rf is the fiber radius, andEf is the fiber modulus. While neglecting Table 2 Effects of annealing on fiber push-in test results for unidirectional Hi-Nicalon/BN/SiC/BSAS composite; 12 plies, Vf = 0.32 (#Hi-NIC-BSAS-6- 24-97) Annealing conditions σd (GPa) τfriction (MPa) As-fabricated 0.31 ± 0.14 10.4 ± 3.1 100 h, air, 1200 ◦C 0.53 ± 0.47a 8.33 ± 1.72a a Measured for fibers in the interior of the annealed CMC sample. Poisson expansion of the fibers leads to an overestimation of τfriction values, the relative changes in τfriction with load cycling could be followed using Eq. (2). In addition, a debond initiation stress, σd, could be calculated from the debond initiation load, Fd (load at which fiber end begins to move during first loading cycle) by the relation σd = Fd πr2 f (3) The results of fiber push-in data are summarized in Table 2. The scatters in the values of debond stress (σd) and frictional sliding stress (τf) are due to the accuracy of measurements for a particular fiber as well as from the variations in val￾ues from fiber-to-fiber. The variation in frictional sliding stress (τf) during successive loading–unloading cycles was small and within the standard deviation. Values of debond stress, σd, and frictional sliding stress, τf, for the as-fabricated CMC were 0.31 ± 0.14 GPa and 10.4 ± 3.1 MPa, respectively, compared with 0.53 ± 0.47 GPa and 8.33 ± 1.72 MPa for the fibers in the interior of the 1200 ◦C annealed sample. These results indicate that only the outer ply of the 1200 ◦C annealed CMC specimens has been degraded from oxidation whereas the bulk interior part remains unaffected. 3.9. Degradation mechanism at 1200 ◦C A possible mechanism explaining the various steps involved in the degradation of Hi-Nicalon/BN/SiC/BSAS CMC on annealing in air at 1200 ◦C is presented in Fig. 12. During anneal￾ing at 1200 ◦C, the surface matrix layer delaminates from the composite ply underneath (SEM micrograph of Fig. 5), probably due to the large CTE mismatch between the Hi-Nicalon fibers and the celsian matrix. This facilitates the ingression of oxy￾gen into the composite. This causes the oxidation of SiC fibers, particularly those which have lost the duplex CVD coating (see Fig. 2), resulting in the formation of SiO2: SiC (s) + O2 (g) → SiO2 (s) + CO (g) + CO2 (g) (4) Fig. 12. Proposed mechanism for degradation of Hi-Nicalon/BN/SiC/BSAS composites during annealing at 1200 ◦C in air