wwceramics. org/ACT (depending on their microstructure). Further, they are plex and contradictory functions. First, it should arrest ght and some of them are available in large quantity at and deflect the ma atrix microcr a relatively low cost(carbon fibers). Obviously, they are the interphase debonding energy, I, is low relative to the the reinforcement of choice for nonoxide matrices, e.g., failure energy of the fiber, Ib a generally accepted crite- since amic equilibrium with rion being I /T <1/4. This is the so-called mechanical carbon at high temperature. Also, there exist well-iden- fuse function(the interphase protecting the fiber from an tified nonoxide interphases (pyrocarbon and boron ni- early failure). Second, the interphase may act as a diffu tride)compatible with both carbon and SiC within a sion barrier(as previously mentioned for the rMi proc- wide temperature range. Unfortunately, nonoxide CMCs ess) and relax partly thermal residual stresses. It has been re oxidation prone and their long exposure to oxidizing recently postulated that the best interphase materials res efficient against oxida- might be those with a layered crystal structure or micro- tion(PAO). Oxide-based CMCs are by essence inert in structure, the layers being deposited, parallel to the fiber oxidizing atmospheres and migh ght appear more attrac- surface, weakly bonded to one another(for low T but tive. However, most oxide-based fibers(containing a- strongly adherent to the fiber surface(to avoid debonding alumina, mullite, or zirconia) display poor HT mecha at the fiber surface). In SiC-matrix composites, the best al properties(they suffer from grain growth and creep interphase material from a mechanical standpoint beyond about 1000-1100C). Further, there on(Fig. 1a). 1 Unfor ly no stable oxide interphase formally equivalent to tunately, pyrocarbon is intrisically oxidation-prone at pyrocarbon or BN, i. e, a layered oxide with a low shear temperatures as low as 500%C. BN is an interesting al- at col deposited on fibers ternative since it has a similar layered crystal structure and though there are few oxide interphase materials, such a better oxidation resistance, its oxidation starting at monazite or hibonite, that could deflect matrix cracks in about 800C and yielding a Auid B2O3 oxide known oxide-oxide composites but not as easily as their not for its healing properties. However, its formation on a oxide counterparts). Also, oxide- based CMCs are in- SiC fiber is not straightforward. When deposited at low sulating materials and their density can be slightly temperature by CVD/CVI, it is amorphous or poorly higher than that of C-or SiC-based composites. Final- crystallized and hence sensitive to moisture. Its crystalli ly, fibers should exhibit a good weavability, which sup- zation by heat treatment is often limited by the thermal poses a low enough diameter(typically, 10 um or less) stability of the fibers and the bonding with the fibers is when their stiffness is high(this is the case for most poor. An interesting alternative might be to form a arbon fibers but not for all stoichiometric SiC fibers) more adherent bn coating by annealing a SiC fiber con- and preferably a high failure strain. To conclude, non- taining some boron(used as a sintering aid) in a nitriding oxide CMCs(C/SiC or SiC/SiC)are presently preferred atmosphere at high temperature. SiC/SiC composites for most structural applications even though their use with such an in situ formed BN interphase have been in oxidizing atmospheres raises a difficult problem of reported to be more oxidation resistant than those with durability. BN interphase deposited by CVD/CVl. The choice of a concept of damage tolerance is a k Since the number of thermally stable materials with step in the design of CMCs In SiC-matrix composites, layered structures is limited, the concept of layered inter damage tolerance is achieved through a weakening of the phase has been further extended to materials with a lay FM-bonding(controlled by an interphase), which allows ered microstructure at the nanometer scale, i.e., to (X-Y) the matrix microcracks to be deflected by the FM-inter- multilayers. Such interphases offer a much higher design based on the use of a highly porous matrix (and no in- pulsed CVI (or P-CVT) being the ers by, e.g,pressure- terphase), is known. Its use might be appropriate in ox overall thickness of the interphase, the thicknesses of the X ide/oxide composites since both constituents are inert and Y sublayers, the number of X-Y sequences, n, and Conversely, it might be the X/Bonding. As an example, in(PyC-SiC)n or(BN- lematic in nonoxide CMCs since a porous matrix will SiC)m, the amount of oxidation-prone mechanical favor fiber oxidation and lower thermal conductivity. (-= PyC or BN) can be strongly reduced(the thickness The design of the interphase in SiC-matrix compos- of X-layers being a few nanometers, typically 3-20 nm) ites is not straightforward since with the result that the durability of the composites(depending on their microstructure). Further, they are light and some of them are available in large quantity at a relatively low cost (carbon fibers). Obviously, they are the reinforcement of choice for nonoxide matrices, e.g., SiC, since SiC is in thermodynamic equilibrium with carbon at high temperature. Also, there exist well-iden￾tified nonoxide interphases (pyrocarbon and boron ni￾tride) compatible with both carbon and SiC within a wide temperature range. Unfortunately, nonoxide CMCs are oxidation prone and their long exposure to oxidizing atmospheres requires efficient protection against oxida￾tion (PAO). Oxide-based CMCs are by essence inert in oxidizing atmospheres and might appear more attrac￾tive. However, most oxide-based fibers (containing a￾alumina, mullite, or zirconia) display poor HT mechan￾ical properties (they suffer from grain growth and creep beyond about 1000–11001C). Further, there is present￾ly no stable oxide interphase formally equivalent to pyrocarbon or BN, i.e., a layered oxide with a low shear strength that could be easily deposited on fibers (al￾though there are few oxide interphase materials, such as monazite or hibonite, that could deflect matrix cracks in oxide–oxide composites but not as easily as their non￾oxide counterparts).12 Also, oxide-based CMCs are in￾sulating materials and their density can be slightly higher than that of C- or SiC-based composites. Final￾ly, fibers should exhibit a good weavability, which sup￾poses a low enough diameter (typically, 10 mm or less) when their stiffness is high (this is the case for most carbon fibers but not for all stoichiometric SiC fibers) and preferably a high failure strain. To conclude, non￾oxide CMCs (C/SiC or SiC/SiC) are presently preferred for most structural applications even though their use in oxidizing atmospheres raises a difficult problem of durability. The choice of a concept of damage tolerance is a key step in the design of CMCs. In SiC-matrix composites, damage tolerance is achieved through a weakening of the FM-bonding (controlled by an interphase), which allows the matrix microcracks to be deflected by the FM-inter￾faces. However, another concept of damage tolerance, based on the use of a highly porous matrix (and no in￾terphase), is known. Its use might be appropriate in ox￾ide/oxide composites since both constituents are inert in oxidizing atmospheres.13,14 Conversely, it might be prob￾lematic in nonoxide CMCs since a porous matrix will favor fiber oxidation and lower thermal conductivity. The design of the interphase in SiC-matrix compos￾ites is not straightforward since the interphase has com￾plex and contradictory functions.4 First, it should arrest and deflect the matrix microcracks, which supposes that the interphase debonding energy, Gi , is low relative to the failure energy of the fiber, Gf, a generally accepted crite￾rion being Gi /Gf o1/4.5 This is the so-called mechanical fuse function (the interphase protecting the fiber from an early failure). Second, the interphase may act as a diffu￾sion barrier (as previously mentioned for the RMI proc￾ess) and relax partly thermal residual stresses. It has been recently postulated that the best interphase materials might be those with a layered crystal structure or micro￾structure, the layers being deposited, parallel to the fiber surface, weakly bonded to one another (for low Gi ) but strongly adherent to the fiber surface (to avoid debonding at the fiber surface).4 In SiC-matrix composites, the best interphase material from a mechanical standpoint is probably an anisotropic pyrocarbon (Fig. 1a).4,15 Unfor￾tunately, pyrocarbon is intrisically oxidation-prone at temperatures as low as 5001C. BN is an interesting al￾ternative since it has a similar layered crystal structure and a better oxidation resistance, its oxidation starting at about 8001C and yielding a fluid B2O3 oxide known for its healing properties. However, its formation on a SiC fiber is not straightforward. When deposited at low temperature by CVD/CVI, it is amorphous or poorly crystallized and hence sensitive to moisture. Its crystalli￾zation by heat treatment is often limited by the thermal stability of the fibers and the bonding with the fibers is poor. 16 An interesting alternative might be to form a more adherent BN coating by annealing a SiC fiber con￾taining some boron (used as a sintering aid) in a nitriding atmosphere at high temperature. SiC/SiC composites with such an in situ formed BN interphase have been reported to be more oxidation resistant than those with a BN interphase deposited by CVD/CVI.17 Since the number of thermally stable materials with layered structures is limited, the concept of layered inter￾phase has been further extended to materials with a lay￾ered microstructure at the nanometer scale, i.e., to (X–Y)n multilayers. 4 Such interphases offer a much higher design flexibility, the adjustable parameters by, e.g., pressure￾pulsed CVI (or P-CVI) being the nature of X and Y, the overall thickness of the interphase, the thicknesses of the X and Y sublayers, the number of X–Y sequences, n, and the X/Y bonding. As an example, in (PyC–SiC)n or (BN– SiC)n, the amount of oxidation-prone mechanical fuse (X 5 PyC or BN) can be strongly reduced (the thickness of X-layers being a few nanometers, typically 3–20 nm) with the result that the durability of the composites in www.ceramics.org/ACT SiC-Matrix Composites: Application 77