M. Leparoux et al Si L2,3 the thinnest Bn interphase could be explained. For the same reasons also when considering the very thin carbon sublayers (when compared to the B-K pyrocarbon interphases)and some possible protec- tion by oxide formers, SiC/BN SiC composites are more oxidation resistant than Sic/C/SiC materials O-K under static loading(Fig. 2 In contrast, the composites with a pyrocarbon interphase exhibit higher resistance to dynamic fatigue in tcnsion-tcnsion, at 600C in air. In that case, the mechanical behavior depends on the slid ing and wear resistance, besides the stability of each constituent in the oxidizing environment and Fig. 12. EELS spectrum of the BN/SiC-matrix interfac the ability for oxygen to diffuse into the composite. Even if it seems that the materials with a bn It is indeed particularly important to examine interphase are less altered by oxidation, it has been accurately the interfaces between the bn inter- shown'2 that the Sic/C/Sic composites present a phase and both the fibers and the matrix. Crack better dynamic fatigue resistance at room tem deflection from mode i to mode il occurs either at per perature. This has been related to the much higher the fiber/BN or BN / matrix interfaces where carbon interfacial shear resistance, twice to five times is the main initial constituent, as shown pre- greater than those of materials with a bn inter viously. 5 The deflection mechanism at the later phase. With regard to the influence of the BN interface was not (or rarely) observed at room thickness, thinner interphases lead to lower sliding temperature owing probably to a weaker fiber/BN resistance and less developed microcracking inside link and different stresses field. When the bn the 0 tows at room temperature. Then, it could be interphase is still bonded with the fibers and the supposed that the fiber/matrix bonding changes matrix, the initial carbon sub-layer has been con- slower during mechanical tests at 600C than those sumed with the formation of silica at the Bn/fiber of thicker interphases. The result is a better reten intcrface. At the BN SiC interface, the initial free tion of the initial mechanical properties.These carbon has been oxidized or transformed in silica hypotheses are based principally on characteristics nd silicon carbide. The csc evaluated at room temperature, espccially the slid con carbide formation clearly leads to strong ing resistance and the crack spacing. A more thor interfacial bondings. When compared to room ough study should consider the coupled influence temperature properties, a worse mechanical beha- of the temperature with the interphase thickness, vior can therefore result the influence of the stress field in the fiber/matrix Obviously, the evolution of the interfacial zones region, and the environmental effects on the inter at a given temperature depends on time but also on facial properties the ability for oxygen to reach the oxidizible layers At this temperature, a sealing of the sic/mati cracks is considered to be negligible, especially far 5 Conclusion from the bn interphase and when a dynamic loading is applied. Thus, the failures of the com- The large potentialities of ceramic composites es appear all the later since the applied stress require the preservation of their thermomech- w as the total crack opening is directly corre- anical properties for long duration, in oxidative to strain. On the other hand, the total environments, under high temperature fat amount of crack opening depends on both cracks conditions. This implies that the fibers, the matrix width and cracks density. A higher density of and the interfacial properties such as load transfer, microcracking might increase the oxidation rate of wear resistance and crack deflection could the interfacial sublayers if we suppose that the maintained gascous diffusion within the radial cracks is not Boron nitride interphases appeared to be more rate-limiting(a reasonable hypothesis at such mod- efficient than pyrocarbon ones under static fatigue erate temperature of 600C). In that casc,morc at 600C in air. Although boron nitride, deposited interfacial zones are damaged when the cracks by ICVI from BCl3-NHy-H2 mixtures, seems to be density increases. Because the crack density decrea- partially oxidized cven at this moderate tempera ses when the BN interphase thickness decreases(at ture, its resistance towards corrosion is still higher least in the range 0.207 um) the better behavior than carbon. As a result of this oxidation, a Si- under static loading of the composites that include b-o glassy phase has been found in the cracks722 M. Leparoux et al. 7 Si L2,3 J II “‘I’ 1 j n I I 8 b r I1 I I I I I I 1 100 200 300 400 500 600 Energy Loss (eV) Fig. 12. EELS spectrum of the BN/SiC-matrix interface. It is indeed particularly important to examine accurately the interfaces between the BN inter￾phase and both the fibers and the matrix. Crack deflection from mode I to mode II occurs either at the fiber/BN or BN/matrix interfaces where carbon is the main initial constituent, as shown pre￾viously.15 The deflection mechanism at the later interface was not (or rarely) observed at room temperature owing probably to a weaker fiber/BN link and different stresses field. When the BN interphase is still bonded with the fibers and the matrix, the initial carbon sub-layer has been con￾sumed with the formation of silica at the BN/fiber interface. At the BNjSiC interface, the initial free carbon has been oxidized or transformed in silica and silicon carbide. These silica or silica plus sili￾con carbide formation clearly leads to strong interfacial bondings. When compared to room temperature properties, a worse mechanical beha￾vior can therefore result. Obviously, the evolution of the interfacial zones at a given temperature depends on time but also on the ability for oxygen to reach the oxidizible layers. At this temperature, a sealing of the Sic/matrix cracks is considered to be negligible, especially far from the BN interphase and when a dynamic loading is applied. Thus, the failures of the com￾posites appear all the later since the applied stress is low as the total crack opening is directly corre￾lated to strain. On the other hand, the total amount of crack opening depends on both cracks width and cracks density. A higher density of microcracking might increase the oxidation rate of the interfacial sublayers if we suppose that the gaseous diffusion within the radial cracks is not rate-limiting (a reasonable hypothesis at such mod￾erate temperature of 600°C). In that case, more interfacial zones are damaged when the cracks density increases. Because the crack density decrea￾ses when the BN interphase thickness decreases (at least in the range 0.24.7pm) the better behavior under static loading of the composites that include the thinnest BN interphase could be explained. For the same reasons, also when considering the very thin carbon sublayers (when compared to the pyrocarbon interphases) and some possible protec￾tion by oxide formers, SiC/BN/SiC composites are more oxidation resistant than SiC/C/SiC materials under static loading (Fig. 2). In contrast, the composites with a pyrocarbon interphase exhibit higher resistance to dynamic fatigue in tension-tension, at 600°C in air. In that case, the mechanical behavior depends on the slid￾ing and wear resistance, besides the stability of each constituent in the oxidizing environment and the ability for oxygen to diffuse into the composite. Even if it seems that the materials with a BN interphase are less altered by oxidation, it has been shown’* that the Sic/C/Sic composites present a better dynamic fatigue resistance at room tem￾perature. This has been related to the much higher interfacial shear resistance, twice to five times greater than those of materials with a BN inter￾phase. With regard to the influence of the BN thickness, thinner interphases lead to lower sliding resistance and less developed microcracking inside the 0” tows at room temperature. Then, it could be supposed that the fiber/matrix bonding changes slower during mechanical tests at 600°C than those of thicker interphases. The result is a better reten￾tion of the initial mechanical properties.These hypotheses are based principally on characteristics evaluated at room temperature, especially the slid￾ing resistance and the crack spacing. A more thor￾ough study should consider the coupled influence of the temperature with the interphase thickness, the influence of the stress field in the fiber/matrix region, and the environmental effects on the inter￾facial properties. 5 Conclusion The large potentialities of ceramic composites require the preservation of their thermomech￾anical properties for long duration, in oxidative environments, under high temperature fatigue conditions. This implies that the fibers, the matrix and the interfacial properties such as load transfer, wear resistance and crack deflection could be maintained. Boron nitride interphases appeared to be more efficient than pyrocarbon ones under static fatigue at 600°C in air. Although boron nitride, deposited by ICVI from BCls-NHs-HZ mixtures, seems to be partially oxidized even at this moderate tempera￾ture, its resistance towards corrosion is still higher than carbon. As a result of this oxidation, a Si￾B-O glassy phase has been found in the cracks