718 M. Leparoux et al Table 1. Minimum number of cycles and duration to cause failure of 2 D-SiC/SiC composites with different interphases. They were tested in tension-tension, in air at 600 C, with various applied stresses bN thickness PyC MPa) 01-02m μm 02 4000(7h30min)68400(9h30min)108000-172800(1524h) 9000(h15min) 19800(2h45min) 0-200 2000(18min) 9500(8min) Table 2. Time to cause failure of 2D-SiC/ BN/SiC composite microscopy (TEM)and clectron energy loss spec- submitted to static tensile stress. The interphase thickness was troscopy(EELS). The analyses were radially per cates no rupture formed, from the fiber to the matrix through the BN interphase and its interfaces both with the bn thickne (MPa) treated Nicalon fiber and the Sic matrix ccording to a procedure which has been described elsewhere. 5 It is interesting to recall that the 145 15h I h interfaces were initially carbon-rich, before any exposure to the oxidative environment. 5Because 3.2 Evolution of the fiber/matrix interfacial zones our interest was principally focused on the chem in an oxidizing environment istry of the interfacial zone, no attempt was made to characterize the matrix and the fiber 3.2.1 SEM observations 3.2.2.1 The bn interphase. The thickness of These observations have been performed on a the Bn interphase is homogeneous(about 0.5 um) sample made with a bn thickness of 0.5 um which and no voids are observed(Fig. 9). The eels had failed after a tensile test of about 70 h at 600 C. spectra of this region show boron and nitrogen In that case another test procedure was employed edges(Fig. 10) typical of these elements involved that permits to evaluate quickly the performances of in hexagonal boron nitride. u In addition, they a new material: the applied stress was increased by evidence significant amounts of carbon and oxy increments of 10 MPa. After each step the following gen in the bn phase. Silicon was not dctcctcd, increment was applied when the strain of the com- however it could be below the detection limit of posite nearly reached a stable state. Finally, failure of EELS (<I at%). The concentration of B, C, the composite occurred at 170 MPa. and O was determined by a classical quantitative SEM micrographs of polished cross-sections analysis technique. The average atomic com show that the bn coating is uniformly deposited position was 40% N, 48%B, 7%C and 5%O inside the preform [Fig 3(aH(c). The fiber/matrix In comparison with the average composition interfacial zone either appears smooth or presents measured on the same composite tested at room some crystalline or amorphous masses [Fig 4(a) temperature (41. 5% N, 39%B, 10% C and and(b)]. Microcracks are observed at the fiber/ Bn 9.5%O), 5 we note a decrease of the o, C and N interface but also in the interphase itself and at the contents BN/matrix interface Fig. 5(a) and(b). They 3. 2.2.2 The fiber/bn interface. The TEM propagate randomly in these three regions, on the micrographs show many large decohesion at this contrary to the results obtained at room tempera- interface as clearly shown in Fig. 9. Otherwise, ture where these cracks mainly propagate at the when the fiber and the interphase are still bonded, fiber/BN interface. 12 Many links that present aa bright zone of about 10 nm thick borders the vitreous appearance are seen between Bn and the fiber. This sublayer is mainly composed of silicon fiber(Fig. 6)and between BN and the matrix. These and oxygen with traces of boron and carbon. The particular glassy phases are also observed in cracks near-edge fine structure of the Si-L2,3 edge(double that propagate into the matrix between two neigh- peak at 109 and 116 ev)(Fig. 11)and the dis boring fibers(Fig. 7). The fracture surface shows symmetrical plasmon peak at 22. 5 eV in the range gencrally a short pull-out lcngth [Fig 8(a)and(b)]. of low cncrgy loss(Fig. ll, inset) are consistent Moreover, it appears that the Bn coating adheres with the presence of silica at this interface. The both to the fiber and the matrix [Figs 5(b)and 8(c)]. quantitative analyses give an avcrage Si/ o atomic ratio of 0. 53. It can be recalled here th 3. 2. 2 Microstructure and chemistry of the fiber/ treated Nicalon@ NLM 202 fiber present matrix interfacial zone carbon-rich surface which was maintained at the Further microstructure investigations have been fiber/BN interface, at the end of the composite made on the same sample by transmission electron manufacturing. I5718 M. Leparoux et al. Table 1. Minimum number of cycles and duration to cause failure of 2D-Sic/Sic composites with different interphases. They were tested in tension-tension, in air at 6OO”C, with various applied stresses Stress (MPa) Frequency (Hz) 0.7 pm BN thickness 0.4 pm 0.2 pm PYC 0.14.2 pm Cl20 G150 &200 2 54 000 (7 h 30 min) 68 400 (9 h 30 min) 108 000-172 800 (15-24 h) 2 9000 (lh 15 min) 19 800 (2 h 45 min) 20 22 000 (18min) 9500 (8 min) Table 2. Time to cause failure of 2D-SiC/BN/SiC composites submitted to static tensile stress. The interphase thickness was 0.7pm. R indicates a rupture of the materials and NR indi￾cates no rupture Stress BN thickness (MPa) 0.7 pm 120 > 1OOh 145 15h 185 lh NR R R 3.2 Evolution of the fiber/matrix interfacial zones in an oxidizing environment 3.2.1 SEA4 observations These observations have been performed on a sample made with a BN thickness of 0.5pm which had failed after a tensile test of about 70 h at 600°C. In that case another test procedure was employed that permits to evaluate quickly the performances of a new material: the applied stress was increased by increments of 10 MPa. After each step the following increment was applied when the strain of the com￾posite nearly reached a stable state. Finally, failure of the composite occurred at 170 MPa. SEM micrographs of polished cross-sections show that the BN coating is uniformly deposited inside the preform [Fig. 3(a)-(c)]. The fiber/matrix interfacial zone either appears smooth or presents some crystalline or amorphous masses [Fig. 4(a) and (b)]. Microcracks are observed at the fiber/BN interface but also in the interphase itself and at the BN/matrix interface [Fig. 5(a) and (b)]. They propagate randomly in these three regions, on the contrary to the results obtained at room tempera￾ture where these cracks mainly propagate at the fiber/BN interface. l2 Many links that present a vitreous appearance are seen between BN and the fiber (Fig. 6) and between BN and the matrix. These particular glassy phases are also observed in cracks that propagate into the matrix between two neigh￾boring fibers (Fig. 7). The fracture surface shows generally a short pull-out length [Fig. 8(a) and (b)]. Moreover, it appears that the BN coating adheres both to the fiber and the matrix [Figs 5(b) and 8(c)]. 3.2.2 Microstructure and chemistry of the$ber/ matrix interfacial zone Further microstructure investigations have been made on the same sample by transmission electron microscopy (TEM) and electron energy loss spec￾troscopy (EELS). The analyses were radially per￾formed, from the fiber to the matrix through the BN interphase and its interfaces both with the treated Nicalon@ fiber and the SIC matrix, according to a procedure which has been described elsewhere.15 It is interesting to recall that the interfaces were initially carbon-rich, before any exposure to the oxidative environment.15 Because our interest was principally focused on the chem￾istry of the interfacial zone, no attempt was made to characterize the matrix and the fiber. 3.2.2.1 The BN interphase. The thickness of the BN interphase is homogeneous (about 0.5 pm) and no voids are observed (Fig. 9). The EELS spectra of this region show boron and nitrogen edges (Fig. 10) typical of these elements involved in hexagonal boron nitride.20 In addition, they evidence significant amounts of carbon and oxy￾gen in the BN phase. Silicon was not detected, however it could be below the detection limit of EELS (< 1 at%). The concentration of B, C, N and 0 was determined by a classical quantitative analysis technique. 21 The average atomic com￾position was 40% N, 48% B, 7% C and 5% 0. In comparison with the average composition measured on the same composite tested at room temperature (41.5% N, 39% B, 10% C and 9.5% 0),15 we note a decrease of the 0, C and N contents. 3.2.2.2 The fiber/BN interface. The TEM micrographs show many large decohesions at this interface as clearly shown in Fig. 9. Otherwise, when the fiber and the interphase are still bonded, a bright zone of about 10nm thick borders the fiber. This sublayer is mainly composed of silicon and oxygen with traces of boron and carbon. The near-edge fine structure of the Si-L2,3 edge (double peak at 109 and 116 eV) (Fig. 11) and the dis￾symmetrical plasmon peak at 22.5 eV in the range of low energy loss (Fig. 11, inset) are consistent with the presence of silica at this interface. The quantitative analyses give an average SijO atomic ratio of 0.53. It can be recalled here that the treated Nicalon@ NLM 202 fiber presented a carbon-rich surface which was maintained at the fiber/BN interface, at the end of the composite manufacturing.15