International y ournal of Applied Ceramic TechnologyNaslain Vol.2,No.2,2005 filtration), or finally the so-called ceramic or slurry diffusion barrier. Further, the matrix is rarely pure SiC routes(SI-HP: slurry infiltration and hot processing) but a mixture of Sic and free silicon(free silicon low each displaying advantages and drawbacks. generally ering its refractoriness and creep resistance), however, speaking, the matrix should be homogeneously distrib- the content of the latter can be limited if liquid silicon is uted in the preform with limited residual porosity and replaced by a suitable silicon alloy. On the other hand, the FM-bonding well controlled with no significant fb- RMI is a fast densification technique and the corre- er degradation. Further, the process should be flexible sponding composites are near net shape with low resid with limited handling and yield near net shape com- ual porosity(Vp <5%) posites, in order to lower production cost I-CVI and RMi are the that display, from In the CVI-process, the interphase, the matrix, and our viewpoint, the best potential in terms of cost and the seal-coating(used to seal the open residual porosity volume production. Further, they are complementary, and enhance the oxidation resistance)are successively de- i.e., the residual porosity of CVI-composites, at a suit posited from gaseous precursors. In conventional CVI able state of densification, can be filled via an RMI-step (referred to as I-CVI, I standing for isothermal/isobaric), Conversely, the PIP-process, which is also a low-temper there are no temperature/pressure gradients in the fiber ature technique, is lengthy since several time-consuming low-temperature (typically, PI/P sequences( from 6 to 10) to achieve 900-1100C), low-pressure(<100 kPa) process, yielding an able densification near net shape composites with limited fiber degradar significant residual porosity and implies considerable and materials of high microstructural quality. It is also a handling. It can also be combined with RMI, as previ- highly flexible process, a large number of preforms(whic ously mentioned. Fin HP is both could be different in size and shapes) being treated l800° for SiC) and a high-pressure(≈25MPa) multaneously with limited handling, in large infiltration process, which is only compatible with fibers of hi furnaces. All these features justify that I-CVI has been thermal stability(carbon or stoichiometric SiC fibers rapidly transferred from the laboratory to the plant levels. with a risk of fiber degradation. It has been improved Conversely, in I-CVI, the densification rate is relatively through the use of nanometric SiC particles slurry and slow and the residual porosity is significant(typically, 10- additives(Al2O3, Y2O3)forming a liquid phase at 15%). The densification rate can be actually improved by sintering temperature(see, e.g., the NITE-process) pplying to the preform a temperature gradient TG of the main advantages of SI-HP lies in CVD), a pressure gradient(P-CVI), or both(as in forced that it is a fast densification process, yielding composites or F-CVD), but it is at the expense of fexibility(some with almost no residual porosity and hence a high ther fixturing being necessary for each preform to create the mal conductivity. However, its extension to large multi gradient(s). It can also be improved by performing in- directional fiber preforms seems to be problematic termediate surface machining(to re-open the porosity) From this brief analysis, it appears that none of the but that requires additional handling and raises the fab- existing processes is perfect and that hybrid techniques rication cost. Residual porosity(which is detrimental to combining two approaches, such as PIP/RMI or CVI hermal conductivity and oxidation resistance) is usually RMI, might presently be the most appropriate choice; sealed by depositing on the external surface of the com- each step could still be improved in order to gain in posites a suitable coating at the end of the process reproducibility and cost, at plant level In the RMi (or more simply, MI)process, the fiber form is first consolidated with carbon( deposited on the coated fibers, e. g, by PIp)and then impregnated with liquid silicon(or an Si alloy), silicon reacting exo- thermally with the carbon to form in situ the SiC-based The choice of a suitable reinforcement, for a given matrix. RMI is a hT (1400-1600.C)and liquid matrix, is dictated by several considerations including silicon is a highly reactive medium. Hence, it can be FM compatibility, mechanical or/and thermal proper used only with fibers of high thermal stability(carbon or ties, chemical compatibility with the high service tem- oxygen-free SiC-based fibers)protected with a suitable perature, density, and cost. Covalent nonoxide fibers interphase, e. g, dual pyrocarbon/SiC or boron nitride(carbon and oxygen-free SiC fibers) display the best HT (BN)/SiC interphases where the SiC-sublayer acts as a mechanical properties and can be good heat conductorsinfiltration), or finally the so-called ceramic or slurry routes (SI–HP: slurry infiltration and hot processing), each displaying advantages and drawbacks. Generally speaking, the matrix should be homogeneously distrib￾uted in the preform with limited residual porosity and the FM-bonding well controlled with no significant fib￾er degradation. Further, the process should be flexible with limited handling and yield near net shape com￾posites, in order to lower production cost. In the CVI-process, the interphase, the matrix, and the seal-coating (used to seal the open residual porosity and enhance the oxidation resistance) are successively de￾posited from gaseous precursors. In conventional CVI (referred to as I-CVI, I standing for isothermal/isobaric), there are no temperature/pressure gradients in the fiber preform.1,2,7 I-CVI is a low-temperature (typically, 900–11001C), low-pressure (o100 kPa) process, yielding near net shape composites with limited fiber degradation and materials of high microstructural quality. It is also a highly flexible process, a large number of preforms (which could be different in size and shapes) being treated si￾multaneously with limited handling, in large infiltration furnaces. All these features justify that I-CVI has been rapidly transferred from the laboratory to the plant levels. Conversely, in I-CVI, the densification rate is relatively slow and the residual porosity is significant (typically, 10– 15%). The densification rate can be actually improved by applying to the preform a temperature gradient (TG￾CVI), a pressure gradient (P-CVI), or both (as in forced or F-CVI), but it is at the expense of flexibility (some fixturing being necessary for each preform to create the gradient(s).8 It can also be improved by performing in￾termediate surface machining (to re-open the porosity) but that requires additional handling and raises the fab￾rication cost. Residual porosity (which is detrimental to thermal conductivity and oxidation resistance) is usually sealed by depositing on the external surface of the com￾posites a suitable coating at the end of the process. In the RMI (or more simply, MI) process, the fiber preform is first consolidated with carbon (deposited on the coated fibers, e.g., by PIP) and then impregnated with liquid silicon (or an Si alloy), silicon reacting exo￾thermally with the carbon to form in situ the SiC-based matrix. RMI is a HT process (1400–16001C) and liquid silicon is a highly reactive medium. Hence, it can be used only with fibers of high thermal stability (carbon or oxygen-free SiC-based fibers) protected with a suitable interphase, e.g., dual pyrocarbon/SiC or boron nitride (BN)/SiC interphases where the SiC-sublayer acts as a diffusion barrier.9 Further, the matrix is rarely pure SiC but a mixture of SiC and free silicon (free silicon low￾ering its refractoriness and creep resistance), however, the content of the latter can be limited if liquid silicon is replaced by a suitable silicon alloy. On the other hand, RMI is a fast densification technique and the corre￾sponding composites are near net shape with low resid￾ual porosity (Vpo5%). I-CVI and RMI are the processes that display, from our viewpoint, the best potential in terms of cost and volume production. Further, they are complementary, i.e., the residual porosity of CVI-composites, at a suit￾able state of densification, can be filled via an RMI-step. Conversely, the PIP-process, which is also a low-temper￾ature technique, is lengthy since several time-consuming PI/P sequences (from 6 to 10) are necessary to achieve an acceptable densification. It yields composites with a significant residual porosity and implies considerable handling. It can also be combined with RMI, as previ￾ously mentioned. Finally, SI–HP is both a HT (1700– 18001C for SiC) and a high-pressure (
25 MPa) process, which is only compatible with fibers of high thermal stability (carbon or stoichiometric SiC fibers) with a risk of fiber degradation.10 It has been improved through the use of nanometric SiC particles slurry and additives (Al2O3, Y2O3) forming a liquid phase at sintering temperature (see, e.g., the NITE-process).11 One of the main advantages of SI–HP lies in the fact that it is a fast densification process, yielding composites with almost no residual porosity and hence a high ther￾mal conductivity. However, its extension to large multi￾directional fiber preforms seems to be problematic. From this brief analysis, it appears that none of the existing processes is perfect and that hybrid techniques combining two approaches, such as PIP/RMI or CVI/ RMI, might presently be the most appropriate choice; each step could still be improved in order to gain in reproducibility and cost, at plant level. Material Design The choice of a suitable reinforcement, for a given matrix, is dictated by several considerations including FM compatibility, mechanical or/and thermal proper￾ties, chemical compatibility with the high service tem￾perature, density, and cost. Covalent nonoxide fibers (carbon and oxygen-free SiC fibers) display the best HT mechanical properties and can be good heat conductors 76 International Journal of Applied Ceramic Technology—Naslain Vol. 2, No. 2, 2005