Availableonlineatwww.sciencedirect.com DIRECTO COMPOSITES SCIENCE SCIENCE AND TECHNOLOGY ELSEVIER Composites Science and Technology 64(2004)155-170 www.elsevier.com/locate/compscitech Review article Design, preparation and properties of non-oxide CMCs for application in engines and nuclear reactors: an overview* R. Naslain* aboratoire des Composites Thermostructuraux, University Bordeaux I, Domaine Universitaire, 3 Allee de La boetie, 33600 Pessac, france Received 30 April 2003; accepted 3 June 200 Dedicated to Professor Anthony G. Evans Sic-based ceramic matrix composites, consisting of carbon or SiC fibers embedded in a Sic-matrix, are tough ceramics when the fiber/matrix bonding is properly optimized through the use of a thin interphase. They are fabricated according to different proces- sing routes(chemical vapor infiltration, polymer impregnation/pyrolysis, liquid silicon infiltration or slurry impregnation /hot pressing) each of them displaying advantages and drawbacks which are briefly discussed. Sic-matrix composites are highly tailor- ble materials in terms of fiber-type(carbon fibers of Sic-based fibers such as Si-C-o. SiC +C or quasi-stoichiometric SiC rein- forcements), interphase (pyrocarbon or hexagonal BN, as well as(Pyc-SiC)n or (BN-SiC)n multilayered interphases), matrix (simple Sic or matrices with improved oxidation resistance, such as self-healing matrices)and coatings( SiC or engineered multi layered coatings). The potential of Sic-matrix composites for application in advanced aerojet engines(after-burner hot section), gas turbine of electrical power/steam cogeneration(combustion chamber) and inner wall of the plasma chamber of nuclear fusion reaction, all of them corresponding to very severe conditions is discussed. C 2003 Elsevier Ltd. All rights reserved Keywords: A. Ceramic-matrix composites(CMC); SiC-matrix composites: B Interphases Contents 1. Introduction 2. Processing 156 2.1. The gas phase route 156 2.2. The liquid phase routes 2.3. The ceramic route 2.4. Hybrid processes 3. Material design 3. 1. Fibers .160 3.2. Interphases 3.3. Matrices 3. 4. Coat w Presented at Multifunctional Materials and Structures: Present Status and Future Perspectives a Symposium in Honor of A.G. Evans on the occasion of his 60th birthday, Max-Planck Institute fur Metallforschung, Stuttgart, 16-20 March 2003 *Tel:+33-5-5684-4706;fax:+33-5-5684-1225 E-Imail address: admin(a Icts. u-bordeaux fr(R. Naslain). .3538/S- see front matter C 2003 Elsevier Ltd. All rights reserved 10.1016/S0266-3538(03)00230=6
Review Article Design, preparation and properties of non-oxide CMCs for application in engines and nuclear reactors: an overview§ R. Naslain* Laboratoire des Composites Thermostructuraux, University Bordeaux 1, Domaine Universitaire, 3 Alle´e de La Boe´tie, 33600 Pessac, France Received 30 April 2003; accepted 3 June 2003 Dedicated to Professor Anthony G. Evans Abstract SiC-based ceramic matrix composites, consisting of carbon or SiC fibers embedded in a SiC-matrix, are tough ceramics when the fiber/matrix bonding is properly optimized through the use of a thin interphase. They are fabricated according to different processing routes (chemical vapor infiltration, polymer impregnation/pyrolysis, liquid silicon infiltration or slurry impregnation/hot pressing) each of them displaying advantages and drawbacks which are briefly discussed. SiC-matrix composites are highly tailorable materials in terms of fiber-type (carbon fibers of SiC-based fibers such as Si–C–O, SiC+C or quasi-stoichiometric SiC reinforcements), interphase (pyrocarbon or hexagonal BN, as well as (PyC–SiC)n or (BN–SiC)n multilayered interphases), matrix (simple SiC or matrices with improved oxidation resistance, such as self-healing matrices) and coatings (SiC or engineered multilayered coatings). The potential of SiC-matrix composites for application in advanced aerojet engines (after-burner hot section), gas turbine of electrical power/steam cogeneration (combustion chamber) and inner wall of the plasma chamber of nuclear fusion reaction, all of them corresponding to very severe conditions is discussed. # 2003 Elsevier Ltd. All rights reserved. Keywords: A. Ceramic-matrix composites (CMC); SiC-matrix composites; B. Interphases 0266-3538/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0266-3538(03)00230-6 Composites Science and Technology 64 (2004) 155–170 www.elsevier.com/locate/compscitech Contents 1. Introduction ............................................................................................................................................................................... 156 2. Processing................................................................................................................................................................................... 156 2.1. The gas phase route........................................................................................................................................................... 156 2.2. The liquid phase routes ..................................................................................................................................................... 158 2.3. The ceramic route.............................................................................................................................................................. 158 2.4. Hybrid processes ............................................................................................................................................................... 159 3. Material design........................................................................................................................................................................... 160 3.1. Fibers................................................................................................................................................................................. 160 3.2. Interphases ........................................................................................................................................................................ 161 3.3. Matrices............................................................................................................................................................................. 162 3.4. Coatings............................................................................................................................................................................. 163 § Presented at Multifunctional Materials and Structures: Present Status and Future Perspectives a Symposium in Honor of A.G. Evans on the occasion of his 60th birthday, Max-Planck Institute fu¨r Metallforschung, Stuttgart, 16–20 March 2003. * Tel.: +33-5-5684-4706; fax: +33-5-5684-1225. E-mail address: admin@lcts.u-bordeaux.fr (R. Naslain)
R. Naslain/ Composites Science and Technology 64(2004)155-170 4. Examples of potential application 4.1. Aerojet engines and stationary gas turbines ++,,+++ 4.2. Nuclear fusion reactors 5. Summary. l67 References 1.Introduction ing atmospheres(combustion gas, for example) which is required of the applications previously Non-oxide CMCs, i.e. mainly those consisting of a tioned, being of several hundreds or thousands hours Sic-based matrix reinforced with either carbon or and even more. This problem has been addressed via the Sic-fibers and which will be referred to as C/SiC and design of innovative self-healing interphases and matri- SiC/SiC composites, have been extensively studied dur- s ough the use of specific coatings, with the ing the last two decades since their discovery in the mid result that the durability of these CMCs in sever seventies [1-3]. These tough ceramics have the potential environments is now good enough for applications in for being used up to about 1500oC, as structural mate- aeronautic engines [7, 8 rials, in different fields including advanced engines, gas The use of SiC/SiC composites in high temperature turbines for power/steam co-generation, heat exchan- nuclear reactors of the future, in place of monolithic gers, heat treatment and materials growth furnaces, as SiC, such as the first wall, blanket and divertor of well as nuclear reactors of the future nuclear fusion reactor, is another very challenging The main advantage of CMCs with respect to their potential application requiring among others a high monolithic counterparts lies in the fact that they are thermal conductivity, an excellent hermeticity with tough although their constituents are intrinsically brit- respect to gases(cooling gas fluids or gaseous species tle. This key property is achieved through a proper formed by nuclear reactions) and low residual radio- design of the fiber/matrix(FM) interface arresting and activity. Preliminary research in this field appears to be deflecting cracks formed under load in the brittle matrix very encouraging [9, 10 and preventing the early failure of the fibrous reinfor The aim of the present contribution is to give an cement [4. In its classical treatment, crack deflection is overview of the state of the art in the material design, controlled via the deposition of a thin layer of a com- processing and properties control of Sic-matrix during pliant material with a low shear strength, on the fiber the last few years and underlying the weak points which surface, referred to as the interphase and acting as a still require a significant effort of research mechanical fuse (to protect the fiber). It has been pos- tulated that the best interphase materials might be those with a layered crystal structure, such as pyrocarbon 2. Processing (PyC) or hexagonal boron nitride (hex-BN), or a layered microstructure, such as (PyC-SiC)n or Sic-matrix composites are processed according to (BN-SiC)n, the layers being deposited parallel to the (1)a gas phase route, also referred to as chemical vap fiber surface and the interphase strongly bonded to the infiltration(Cvi),(2)a liquid phase route including the fiber [5, 6]. From a mechanical standpoint, these CMCs polymer impregnation/pyrolysis(PIP)and liquid silicon are damageable elastic materials, i.e. when loaded at a infiltration (LSi)also called (reactive)melt infiltration high enough level, microcracking and FM-debonding (RMI or MD) processes, as well as (3)a ceramic route. occur, which are responsible for a stiffness lowering and i.e. a technique combining the impregnation of the non-linear stress-strain behavior. On the one hand, reinforcement with a slurry and a sintering step at high these damaging phenomena are beneficial since they are temperature and high pressure. Each of these routes has at the origin of the non-brittle character of these advantages and draw backs that will be briefly discussed ceramics On the other hand, they are detrimental since they favor the in-depth diffusion of oxygen towards the 2. 1. The gas phase route oxidation-prone interphase and fibers which in turn may embrittle the composites In the gas phase route, the different constituents of Hence, an important challenge has been to improve the composite, i.e. the interphase, the matrix and the the oxidation resistance of these non-oxide CMCs, the external coating, are successively deposited from gas lifetime under load at high temperatures and in oxidiz- eous precursors at moderate temperatures (900-1 100C)
1. Introduction Non-oxide CMCs, i.e. mainly those consisting of a SiC-based matrix reinforced with either carbon or SiC-fibers and which will be referred to as C/SiC and SiC/SiC composites, have been extensively studied during the last two decades since their discovery in the mid seventies [1–3]. These tough ceramics have the potential for being used up to about 1500 C, as structural materials, in different fields including advanced engines, gas turbines for power/steam co-generation, heat exchangers, heat treatment and materials growth furnaces, as well as nuclear reactors of the future. The main advantage of CMCs with respect to their monolithic counterparts lies in the fact that they are tough although their constituents are intrinsically brittle. This key property is achieved through a proper design of the fiber/matrix (FM) interface arresting and deflecting cracks formed under load in the brittle matrix and preventing the early failure of the fibrous reinforcement [4]. In its classical treatment, crack deflection is controlled via the deposition of a thin layer of a compliant material with a low shear strength, on the fiber surface, referred to as the interphase and acting as a mechanical fuse (to protect the fiber). It has been postulated that the best interphase materials might be those with a layered crystal structure, such as pyrocarbon (PyC) or hexagonal boron nitride (hex-BN), or a layered microstructure, such as (PyC–SiC)n or (BN–SiC)n, the layers being deposited parallel to the fiber surface and the interphase strongly bonded to the fiber [5,6]. From a mechanical standpoint, these CMCs are damageable elastic materials, i.e. when loaded at a high enough level, microcracking and FM-debonding occur, which are responsible for a stiffness lowering and non-linear stress–strain behavior. On the one hand, these damaging phenomena are beneficial since they are at the origin of the non-brittle character of these ceramics. On the other hand, they are detrimental since they favor the in-depth diffusion of oxygen towards the oxidation-prone interphase and fibers which in turn may embrittle the composites. Hence, an important challenge has been to improve the oxidation resistance of these non-oxide CMCs, the lifetime under load at high temperatures and in oxidizing atmospheres (combustion gas, for example) which is required in some of the applications previously mentioned, being of several hundreds or thousands hours and even more. This problem has been addressed via the design of innovative self-healing interphases and matrices and through the use of specific coatings, with the result that the durability of these CMCs in severe environments is now good enough for applications in aeronautic engines [7,8]. The use of SiC/SiC composites in high temperature nuclear reactors of the future, in place of monolithic SiC, such as the first wall, blanket and divertor of nuclear fusion reactor, is another very challenging potential application requiring among others a high thermal conductivity, an excellent hermeticity with respect to gases (cooling gas fluids or gaseous species formed by nuclear reactions) and low residual radioactivity. Preliminary research in this field appears to be very encouraging [9,10]. The aim of the present contribution is to give an overview of the state of the art in the material design, processing and properties control of SiC-matrix during the last few years and underlying the weak points which still require a significant effort of research. 2. Processing SiC-matrix composites are processed according to: (1) a gas phase route, also referred to as chemical vapor infiltration (CVI), (2) a liquid phase route including the polymer impregnation/pyrolysis (PIP) and liquid silicon infiltration (LSI) also called (reactive) melt infiltration (RMI or MI) processes, as well as (3) a ceramic route, i.e. a technique combining the impregnation of the reinforcement with a slurry and a sintering step at high temperature and high pressure. Each of these routes has advantages and drawbacks that will be briefly discussed. 2.1. The gas phase route In the gas phase route, the different constituents of the composite, i.e. the interphase, the matrix and the external coating, are successively deposited from gaseous precursors at moderate temperatures (900–1100 C) 4. Examples of potential application.............................................................................................................................................. 163 4.1. Aerojet engines and stationary gas turbines......................................................................................................................163 4.2. Nuclear fusion reactors ..................................................................................................................................................... 165 5. Summary .................................................................................................................................................................................... 167 References ....................................................................................................................................................................................... 168 156 R. Naslain / Composites Science and Technology 64 (2004) 155–170
in/ Composites Science and Technology 64(2004)155-170 and under reduced pressures (or sometimes at the yielding near-net-shape parts(Fig. 1). On the other atmospheric pressure). The starting material is a porous hand, I-CvI is a relatively slow technique since it has to nD-fiber preform(with usually n=2 or 3), self standing be performed at low deposition rate in order to avoid a or maintained with a tooling (at least at the beginning of too rapid sealing of the pore entrance by the deposit the densification process). During the densification steps The densification rate is improved by applying a pres- (CVI-steps), the interphase and then the SiC-matrix, are sure gradient to the preform, i.e. by replacing the deposited on the fiber surface, within the pore network slow diffusion mass transfer by the much faster con of the preform, according to the following overall vection mass transfer within the pore network (PG- equations(written for the main constituents of a Sic- CVI) or by introducing an inverse temperature gra matrix composite dient (TG-CVI) as mentioned above or by combining both of them(F-CVI with F standing for forced) CH3SiCl3g (1) [13-15]. Another efficient way to increase the densifi- cation rate is to immerse the heated fiber preform in a boiling liquid precursor at reflux (calefaction pro- 2CxHy(g) Lx C(s)+ yH (2) cess)[16]. In the so-called pressure-pulsed CVI-pro- cess(P-CVD), the fiber preform is filled and evacuated BXa+ NH BN(s)+3HX(g) periodically, with a residence time of the gaseous pre in the preform of the order of a few seconds Although P-CVI has been first presented as a way to with X=F CI shorten the overall densification duration of the pre- form [17] its main interest may rather be in its ability to The key point is to maintain the preform porosity yield highly engineered interphases or matrices througl open until the end of the densification process the sequencial use of several precursors [Eqs. (1)and(2) [1, 2, 11, 12]. This is achieved by keeping the pore entran- or Eqs. (1)and (3), for example], as further discussed in ces at a low enough temperature, i.e. by applying an the materials design section [18]. However, increasing inverse temperature gradient to the preform, or by per- the deposition rate is often at the expense of the flex- nodical surface machining(to re-open the pore entrances ibility of the process. Finally, the CvI-process whatever as they become sealed by the deposit). The CVI-process its version results in composites which display sig- displays a number of advantages but also a few draw- nificant residual porosity(typically, backs. It yields deposits with a high purity and well- open) and hence a relatively low thermal conductivity controlled composition and microstructure, as further (although SiC is intrinsically an excellent heat con- discussed in the materials design section. In its most ductor) and a poor hermeticity with respect to gas and common version, the I-CVI technique(I standing for liquid fluids. Despite its drawbacks, CVI is a matured isothermal /isobaric), it is a highly flexible process inas- enough processing route for SiC-matrix composites, much as a large number of preforms, eventually of dif- which has been already transferred to the plant level ferent shapes and sizes can be treated simultaneously [ 19] MTS r preform with DSiC/c/SiC vacuum Z Fig. 1. Schematic of the I-CVI process for the fabrication of C/Sic or SiC/SiC composites from a nD-fiber preform and gaseous precursors
and under reduced pressures (or sometimes at the atmospheric pressure). The starting material is a porous nD-fiber preform (with usually n=2or 3), self standing or maintained with a tooling (at least at the beginning of the densification process). During the densification steps (CVI-steps), the interphase and then the SiC-matrix, are deposited on the fiber surface, within the pore network of the preform, according to the following overall equations (written for the main constituents of a SiCmatrix composite): CH3SiCl3ð Þ g ! H2 SiCð Þs þ 3HClð Þ g ð1Þ 2CxHy gð Þ ! 2xCð Þs þ yH2ð Þ g ð2Þ BX3ð Þ g þ NH3ð Þ g ! BNð Þs þ 3HXð Þ g with X ¼ F; Cl ð3Þ The key point is to maintain the preform porosity open until the end of the densification process [1,2,11,12]. This is achieved by keeping the pore entrances at a low enough temperature, i.e. by applying an inverse temperature gradient to the preform, or by periodical surface machining (to re-open the pore entrances as they become sealed by the deposit). The CVI-process displays a number of advantages but also a few drawbacks. It yields deposits with a high purity and wellcontrolled composition and microstructure, as further discussed in the materials design section. In its most common version, the I-CVI technique (I standing for isothermal/isobaric), it is a highly flexible process inasmuch as a large number of preforms, eventually of different shapes and sizes can be treated simultaneously yielding near-net-shape parts (Fig. 1). On the other hand, I-CVI is a relatively slow technique since it has to be performed at low deposition rate in order to avoid a too rapid sealing of the pore entrance by the deposit. The densification rate is improved by applying a pressure gradient to the preform, i.e. by replacing the slow diffusion mass transfer by the much faster convection mass transfer within the pore network (PGCVI) or by introducing an inverse temperature gradient (TG-CVI) as mentioned above or by combining both of them (F-CVI with F standing for forced) [13–15]. Another efficient way to increase the densifi- cation rate is to immerse the heated fiber preform in a boiling liquid precursor at reflux (calefaction process) [16]. In the so-called pressure-pulsed CVI-process (P-CVI), the fiber preform is filled and evacuated periodically, with a residence time of the gaseous precursor in the preform of the order of a few seconds. Although P-CVI has been first presented as a way to shorten the overall densification duration of the preform [17], its main interest may rather be in its ability to yield highly engineered interphases or matrices through the sequencial use of several precursors [Eqs. (1) and (2) or Eqs. (1) and (3), for example], as further discussed in the materials design section [18]. However, increasing the deposition rate is often at the expense of the flexibility of the process. Finally, the CVI-process whatever its version results in composites which display significant residual porosity (typically, 10–15%, mainly open) and hence a relatively low thermal conductivity (although SiC is intrinsically an excellent heat conductor) and a poor hermeticity with respect to gas and liquid fluids. Despite its drawbacks, CVI is a matured enough processing route for SiC-matrix composites, which has been already transferred to the plant level [19]. Fig. 1. Schematic of the I-CVI process for the fabrication of C/SiC or SiC/SiC composites from a nD-fiber preform and gaseous precursors. R. Naslain / Composites Science and Technology 64 (2004) 155–170 157
R. Naslain/ Composites Science and Technology 64(2004)155-170 e route 2D-fabrics or even lD-fiber tows. i.e. to the ceramic route that will be discussed later on [24-27 There are two different liquid phase routes for the In the LSI-process, also referred to as the rMi or MI fabrication of Sic-matrix composites depending on process(depending on whether the infiltration is reactive hether the precursor is a Si-C based polymer, such as a or not), a porous nD-fiber preform(with n=1, 2 or 3)is polycarbosila ne or PCS(PIP-process)or liq uld silicon first consolidated with a carbon deposit by Cvi [Eq(2) pure or alloyed (Lsi or rMi process) or by PIP utilizing in this latter case a liquid carbon pre- In the PIP-process, a fiber preform, which can be a cursor such as a phenolic resin or a pitch(Fig. 2). In a 3D-preform similar to those used in CVI or more sim- second step, the residual open porosity is filled with liquid ply a stack of 2D-fabrics or lD-plies, is impregnated silicon(mp=1410 C)or with a liquid silicon-based alloy, with a Si-C precursor (in the molten state or in solu- which climbs by capillary forces in the pore network [28- tion in an organic solvent), e.g. under vacuum by resin 31]. Liquid silicon and its related alloys spontaneously wet transfer moulding(RTM), a technique commonly used carbon, with which they react according to the following for polymer matrix composites [20-23]. Different pre- equation, written for pure silicon cursors are employed such as polycarbosilanes(PCS)or C)+Sio, SiCs display a low enough viscosity to flow in the pore net- with an evolution of heat and a volume expansion. work of the fiber preform and a high ceramic yield After Despite its apparent simplicity and short processing uring to render the precursor infusible, which is achieved time, the RMi-process raises some difficulties. Firstly hermally or under radiation (y-rays or E-beam), the the infiltration temperature is relatively high(typically, green body is pyrolyzed at a temperature ranging from 1400-1600oC)which means that only the fibers with a 1000 to 1200C(this relatively low temperature being high thermal stability, namely HM carbon fibers or compatible with the use of fibers of limited thermal quasi-stoichiometric SiC fibers prepared at high tem- stability). Assuming that the precursor is a Yajima's peratures [such as Tyranno SA(from Ube Industries, type PCs, the pyrolysis results in a matrix which is a Japan) or Sylramic fibers(from Dow Corning, USA) SiC+C mixture or pure SiC depending on the nature of can be employed. Secondly, liquid silicon is a corrosive he atmosphere, according to the following overall medium with respect to PyC or hex-BN interphases as well as to the fibers themselves. Hence, specific inter [(CH3)SiH-CH21, nSiC+nC+3nH, phases acting both as mechanical fuse(crack deflection) (4) and diffusion barrier, such as dual hex-BN/SiC inter- [(CH3)SiH-CH21- nSiC +nCH +H, phases should be used. Further and as in CVI, the pores entrances should remain open until the end of the den sification process(which is here very fast compared to the ceramic yield being 89.6% and 68.9%, respectively. CVI)requiring specific care in the management of the Actually, the ceramic yield is in between, part of the liquid silicon flow to and in the fiber preform. Finally, carbon being lost as gaseous species even when the s ce, wnIch limits its refractoriness and creep resistance. To the matrix formed by RMi often contains free silicon olysis is conducted under an inert atmosphere. Hend there is a significant shrinkage during pyrolysis. The minimize the free silicon content, different treatments pyrolytic residue is porous and the porosity largely op have been suggested (silicon vaporization at high tem- [the gaseous species formed according to Eqs. (4)or(5) perature under vacuum, leaching treatment or use of have to escape from the composite, creating porosity]. silicon-metal alloys instead of pure silicon, the alloying In order to achieve a high enough densifiction level, element entrapping the silicon in excess in a refractory several impregnation/pyrolysis sequences(typically 6 to silicide, such as MoSi2)[29, 32]. On the other hand, the 10, and even more) have therefore to be performed RMI-process displays some important advantages: it is which is time consuming and costly. One way to reduce a fast densification technique and it yields composites the number of PI-P cycles is to load the liquid polymer with almost no residual open porosity(and hence with precursor with a filler, that is a powder with a fine an excellent hermeticity with respect to gas and liquid granulometry, which can be pure silicon carbide or a fluids)as well as a high thermal conductivity. It is used, mixture of SiC with additives, e.g. a boron-bearing spe- in a complementary manner, with either CVI or PIP, to cies such as boron carbide to entrap oxygen at medium fill the residual porosity inherent to those techniques temperatures in service conditions, as discussed in the materials design section. However, loading the liquid 2.3. The ceramic route precursor with a powder considerably increases its visc- osity and may render impossible the complete impreg. In the ceramic route, the matrix precursor is a slurry, nation of a complex nD-fiber preform. In such a case, i.e. a stable suspension of a p-sic powder in a liquid one has to move to more simple fiber arrangements: which also contains sintering additives and a fugitive
2.2. The liquid phase routes There are two different liquid phase routes for the fabrication of SiC-matrix composites depending on whether the precursor is a Si-C based polymer, such as a polycarbosilane or PCS (PIP-process) or liquid silicon, pure or alloyed (LSI or RMI process). In the PIP-process, a fiber preform, which can be a 3D-preform similar to those used in CVI or more simply a stack of 2D-fabrics or 1D-plies, is impregnated with a Si–C precursor (in the molten state or in solution in an organic solvent), e.g. under vacuum by resin transfer moulding (RTM), a technique commonly used for polymer matrix composites [20–23]. Different precursors are employed such as polycarbosilanes (PCS) or poly (vinylsilanes). The precursor should wet the fibers, display a low enough viscosity to flow in the pore network of the fiber preform and a high ceramic yield. After curing to render the precursor infusible, which is achieved thermally or under radiation (g-rays or E-beam), the green body is pyrolyzed at a temperature ranging from 1000 to 1200 C (this relatively low temperature being compatible with the use of fibers of limited thermal stability). Assuming that the precursor is a Yajima’s type PCS, the pyrolysis results in a matrix which is a SiC+C mixture or pure SiC depending on the nature of the atmosphere, according to the following overall equations: ½ ð Þ CH3 SiH CH2 n ! Ar nSiC þ nC þ 3nH% 2 ð4Þ ½ ! ð Þ CH3 SiH CH2 H2 nSiC þ nCH% 4 þ H% 2 ð5Þ the ceramic yield being 89.6% and 68.9%, respectively. Actually, the ceramic yield is in between, part of the carbon being lost as gaseous species even when the pyrolysis is conducted under an inert atmosphere. Hence, there is a significant shrinkage during pyrolysis. The pyrolytic residue is porous and the porosity largely open [the gaseous species formed according to Eqs. (4) or (5) have to escape from the composite, creating porosity]. In order to achieve a high enough densifiction level, several impregnation/pyrolysis sequences (typically 6 to 10, and even more) have therefore to be performed which is time consuming and costly. One way to reduce the number of PI-P cycles is to load the liquid polymer precursor with a filler, that is a powder with a fine granulometry, which can be pure silicon carbide or a mixture of SiC with additives, e.g. a boron-bearing species such as boron carbide to entrap oxygen at medium temperatures in service conditions, as discussed in the materials design section. However, loading the liquid precursor with a powder considerably increases its viscosity and may render impossible the complete impregnation of a complex nD-fiber preform. In such a case, one has to move to more simple fiber arrangements: 2D-fabrics or even 1D-fiber tows, i.e. to the ceramic route that will be discussed later on [24–27]. In the LSI-process, also referred to as the RMI or MI process (depending on whether the infiltration is reactive or not), a porous nD-fiber preform (with n=1, 2or 3) is first consolidated with a carbon deposit by CVI [Eq. (2)] or by PIP utilizing in this latter case a liquid carbon precursor such as a phenolic resin or a pitch (Fig. 2). In a second step, the residual open porosity is filled with liquid silicon (mp=1410 C) or with a liquid silicon-based alloy, which climbs by capillary forces in the pore network [28– 31]. Liquid silicon and its related alloys spontaneously wet carbon, with which they react according to the following equation, written for pure silicon: Cð Þs þ Sið Þl ! SiCð Þs ð6Þ with an evolution of heat and a volume expansion. Despite its apparent simplicity and short processing time, the RMI-process raises some difficulties. Firstly, the infiltration temperature is relatively high (typically, 1400–1600 C) which means that only the fibers with a high thermal stability, namely HM carbon fibers or quasi-stoichiometric SiC fibers prepared at high temperatures [such as Tyranno SA (from Ube Industries, Japan) or Sylramic fibers (from Dow Corning, USA)] can be employed. Secondly, liquid silicon is a corrosive medium with respect to PyC or hex-BN interphases as well as to the fibers themselves. Hence, specific interphases acting both as mechanical fuse (crack deflection) and diffusion barrier, such as dual hex-BN/SiC interphases should be used. Further and as in CVI, the pores entrances should remain open until the end of the densification process (which is here very fast compared to CVI) requiring specific care in the management of the liquid silicon flow to and in the fiber preform. Finally, the matrix formed by RMI often contains free silicon which limits its refractoriness and creep resistance. To minimize the free silicon content, different treatments have been suggested (silicon vaporization at high temperature under vacuum, leaching treatment or use of silicon–metal alloys instead of pure silicon, the alloying element entrapping the silicon in excess in a refractory silicide, such as MoSi2) [29,32]. On the other hand, the RMI-process displays some important advantages: it is a fast densification technique and it yields composites with almost no residual open porosity (and hence with an excellent hermeticity with respect to gas and liquid fluids) as well as a high thermal conductivity. It is used, in a complementary manner, with either CVI or PIP, to fill the residual porosity inherent to those techniques. 2.3. The ceramic route In the ceramic route, the matrix precursor is a slurry, i.e. a stable suspension of a b-SiC powder in a liquid which also contains sintering additives and a fugitive 158 R. Naslain / Composites Science and Technology 64 (2004) 155–170
R. Naslain/ Composites Science and Technology 64(2004)155-170 888 =皇 → filtrate 4001450cto 882 si+C→Sc) g a Sic-matrix composite by the liquid silicon infiltration process:(a) preparation of a 2D-fiber preform by a prepreg route b)consolidation of the material by pyrolysis and(c)infiltration of liquid silicon, according to Corman et al. [30] binder. The reinforcement, e. g. a continuous fiber tow cesses are sometimes used to opitimize the densification (uncoated or coated with an appropriate interphase)is of fiber preforms or/and the microstructure of the com- impregnated with the slurry and wound on a drum, posites. Firstly, CvI (in its I-CVi or P-CVI versions)is yielding a 1D-prepreg-type intermediate product. After the method of choice for the deposition of simple or drying, the layers are stacked in the die of a unidirec- highly engineered interphases whatever the technique(s) tional press and the composite sintered at high tem- further employed for the infiltration of the Sic-matrix perature under pressure [33] since it yields deposits of relatively uniform composi- It is well known that the sintering of SiC powder is tion, structure and thickness, even with preforms of difficult and requires very high temperatures, even in the complex fiber architecture. Secondly and as previously presence of sintering aids. Furthermore, since it is per- mentioned, LSI (in its MI or RMi versions)is utilized to formed here under pressure (to achieve low residual fill the residual open porosity of composites prepared porosity), the combined effect of high temperature and either by CVI or PIP, in order to increase both the high pressure was considered for a long time as a source thermal conductivity and hermeticity of the composites of too severe fiber degradation and this route more or Finally, more complex flow charts have been suggested less disregarded. However, it has been shown recently to densify nD-preforms combining, for example, an that the use of nanosized p-Sic powder(with a particu- impregnation with a Sic powder slurry (to introduce late size of N30 nm) and oxide sintering additives rapidly a significant amount of Sic in the preform), a few (Al,O3 and Y2O3)forming transient eutectics with Sio at relatively low temperature, considerably helps the sintering of the SiC-matrix which was effective at Coated fiber fabric M1780oC and 15-20 MPa. Further, the HM carbon fibers utilized in the first experiments [33] could be eplaced by the quasi-stoichiometric SiC fibers whose high thermal stability is compatible with these some- impregnation what mild sintering conditions. SiC/PyC/SiC compo- sites fabricated by this So-called NITE process(for nano filtration and transient eutectics) display a very low residual porosity, high mechanical and thermal proper ties as well as an excellent hermeticity with respect to gaseous cooling fluids such as helium(Fig 3)[34-36]. It Pressure then appears that this process could be attractive if the fiber volume fraction (presently: 20%) could be increased, the relatively thick PyC interphase(800 nm) replaced by a more appropriate material and the 1D-fiber architecture by a multidirectional fiber arrangement Products 2. 4. Hybrid processes Fig. 3. Flow chart of the NITE-process for the fabrication of sic Each of the processes discussed previously displaying matrix composites from nanosize Sic particle slurry with oxide sin advantages and drawbacks, hybrid (or combined) pro- tering additives and pressure sintering, according to Katoh et al. [36]
binder. The reinforcement, e.g. a continuous fiber tow (uncoated or coated with an appropriate interphase) is impregnated with the slurry and wound on a drum, yielding a 1D-prepreg-type intermediate product. After drying, the layers are stacked in the die of a unidirectional press and the composite sintered at high temperature under pressure [33]. It is well known that the sintering of SiC powder is difficult and requires very high temperatures, even in the presence of sintering aids. Furthermore, since it is performed here under pressure (to achieve low residual porosity), the combined effect of high temperature and high pressure was considered for a long time as a source of too severe fiber degradation and this route more or less disregarded. However, it has been shown recently that the use of nanosized b-SiC powder (with a particulate size of 30 nm) and oxide sintering additives (Al2O3 and Y2O3) forming transient eutectics with SiO2 at relatively low temperature, considerably helps the sintering of the SiC-matrix which was effective at 1780 C and 15–20 MPa. Further, the HM carbon fibers utilized in the first experiments [33] could be replaced by the quasi-stoichiometric SiC fibers whose high thermal stability is compatible with these somewhat mild sintering conditions. SiC/PyC/SiC composites fabricated by this so-called NITE process (for nano infiltration and transient eutectics) display a very low residual porosity, high mechanical and thermal properties as well as an excellent hermeticity with respect to gaseous cooling fluids such as helium (Fig. 3) [34–36]. It then appears that this process could be attractive if the fiber volume fraction (presently: 20%) could be increased, the relatively thick PyC interphase (800 nm) replaced by a more appropriate material and the 1D-fiber architecture by a multidirectional fiber arrangement. 2.4. Hybrid processes Each of the processes discussed previously displaying advantages and drawbacks, hybrid (or combined) processes are sometimes used to opitimize the densification of fiber preforms or/and the microstructure of the composites. Firstly, CVI (in its I-CVI or P-CVI versions) is the method of choice for the deposition of simple or highly engineered interphases whatever the technique(s) further employed for the infiltration of the SiC-matrix, since it yields deposits of relatively uniform composition, structure and thickness, even with preforms of complex fiber architecture. Secondly and as previously mentioned, LSI (in its MI or RMI versions) is utilized to fill the residual open porosity of composites prepared either by CVI or PIP, in order to increase both the thermal conductivity and hermeticity of the composites. Finally, more complex flow charts have been suggested to densify nD-preforms combining, for example, an impregnation with a SiC powder slurry (to introduce rapidly a significant amount of SiC in the preform), a few Fig. 3. Flow chart of the NITE-process for the fabrication of SiCmatrix composites from nanosize SiC particle slurry with oxide sintering additives and pressure sintering, according to Katoh et al. [36]. Fig. 2. Fabrication of a SiC-matrix composite by the liquid silicon infiltration process: (a) preparation of a 2D-fiber preform by a prepreg route from a matrix slurry, (b) consolidation of the material by pyrolysis and (c) infiltration of liquid silicon, according to Corman et al. [30]. R. Naslain / Composites Science and Technology 64 (2004) 155–170 159
R. Naslain/ Composites Science and Technology 64(2004)155-170 PI-P sequences with a polymer precursor(to consolidate the Sic particulates deposited from the slurry)and a 3.5 SA FIber deposition of SiC from the gas phase by P-CVI [37, 38 2 ( LorM Tyranno(zMI) 3. Material design In terms of material design, the objective is (1)to Nicalon(NL201) achieve the best mechanical behavior in static and cyclic Hi-Nicalon loading, particularly at high temperatures and (2)to improve oxidation resistance (SiC/Sic being intrinsi 100012001400160018002000 cally oxidation-prone)and durability under load in corrosive environments, such as fuel combustion gas he mechanical behavior of SiC-matrix composites is Fig 4. Variations of the room temperature failure strength of differ mostly controlled by the fibers and the interphase ent SiC-based fibers as a function of their heat treatment temperature whereas the oxidation resistance and durability are Tyranno (Lox M and ZMD)and Nicalon(NL 201)are SiCO(M) depending upon the properties of the fibers, the inter- fibers, Hi-Nicalon is a SiC +C fiber and SA fiber is a quasi-stoichio- phase, the matrix and the external coating. These dif- metric SiC fiber, according to Ishikawa et al.[42] Terent points will be briefly discussed that of pure SiC (Ex400 GPa), their strain at failure 3.. Fibers relatively high (ER=1. 4%)and more importantly, they decompose beyond 1100-1200oC with a strength Generally speaking, two kinds of fibers are used to degradation [40]. Hence, SiC/SiC composites fabri- inforce a SiC-matrix with a view to achieve high cated with these fibers should be processed by low The former are available with a wide range of mechan- limited in temperature (Table l PIP)and their use is ical and thermal properties at high temperatures, and at SiC-based fibers of second generation (such as Hi- a relatively low cost(at least for those also used for the Nicalon) are oxygen-free fibers consisting of a mixture reinforcement of polymer matrices). Further, they are of Sic-nanocrystals(a5 nm in mean size) and free car chemically compatible with Sic almost up to its bon [C/Si (at) ratio=1.39]. They do not undergo decomposition temperature (N2500C). Conversely, decomposition at high temperature since they do not their coefficient of thermal expansion( CTE) is aniso- contain significant amount of SiC.O, phase. They creep tropic (it is very low and even negative along the axis at moderate temperatures (1200C) but their creep but large and positive radially) and different from that resistance is improved (1400C)if they have been of SiC. As a result, C/SiC composites already exhibit a submitted to a heat treatment at 1400-1600C which microcracked matrix in the as-prepared state, these stabilizes the fiber microstructure[40, 41]. The SiC fibers microcracks facilitating the in-depth oxygen diffusion of third generation(such as Hi-Nicalon S, Tyranno $A when exposed to oxidizing atmospheres [39]. More or Sylramic) are oxygen-free and quasi-stoichiometric importantly, carbon fibers undergo active oxidation at [with C/Si (at)=1.00-1. 08][40, 42, 43]. Further, their very low temperature(450C)and could be used only grain size is relatively large(20-200 nm)and their ther in composites where they are perfectly protected again mal stability is excellent, since they are prepared at very oxidation. Finally, the HR carbon fibers(which are the high temperature(1600-2000C) As a result, their use most attractive in terms of cost and availability), have in SiC/SiC composites is compatible with all the fabri- to be properly stabilized through a high temperature cation processes depicted in Section 2, including those treatment(hTT) when used to reinforce SiC at high temperatures. Conversely, the Sic fibers of third The situation is a priori better with SiC fibers since all generation are very stiff (E N400 GPa)and their wea- the problems related to FM-compatibility at high tem- veability is low. Further, their somewhat low strain at peratures are solved and since Sic displays a much bet- failure(0.6-0.8%)limits the extension of the non-linear ter oxidation resistance than carbon. However, all the stress-strain domain partly responsible for the non-brit Sic-based fibers are not pure Sic and most of them are tle character of SiC/Sic composites. Finally, these very expensive( Fig. 4). The early Si-C-O fibers [such as advanced Sic fibers are costly. Obviously, there is still a the Nicalon(from Nippon Carbon, Japan) fiber), refer- need for low cost Sic-based fibers exhibiting moderate red to as Sic-based fibers of first generation, consist of stiffness and relatively high failure strength as well as Sic-nanocrystals(1-2 nm in size) and free carbon high thermal stability and creep resistance(Table 1) embedded in an amorphous SiCrOy matrix. As a nterestingly, a new ceramic fiber, corresponding to result, their stiffness(E-220 GPa)is much lower than the overall formula SiBN3 C, has been recently proposed
PI-P sequences with a polymer precursor (to consolidate the SiC particulates deposited from the slurry) and a deposition of SiC from the gas phase by P-CVI [37,38]. 3. Material design In terms of material design, the objective is (1) to achieve the best mechanical behavior in static and cyclic loading, particularly at high temperatures and (2) to improve oxidation resistance (SiC/SiC being intrinsically oxidation-prone) and durability under load in corrosive environments, such as fuel combustion gas. The mechanical behavior of SiC-matrix composites is mostly controlled by the fibers and the interphase whereas the oxidation resistance and durability are depending upon the properties of the fibers, the interphase, the matrix and the external coating. These different points will be briefly discussed. 3.1. Fibers Generally speaking, two kinds of fibers are used to reinforce a SiC-matrix with a view to achieve high toughness and reliabilty: carbon and SiC-based fibers. The former are available with a wide range of mechanical and thermal properties at high temperatures, and at a relatively low cost (at least for those also used for the reinforcement of polymer matrices). Further, they are chemically compatible with SiC almost up to its decomposition temperature (2500 C). Conversely, their coefficient of thermal expansion (CTE) is anisotropic (it is very low and even negative along the axis but large and positive radially) and different from that of SiC. As a result, C/SiC composites already exhibit a microcracked matrix in the as-prepared state, these microcracks facilitating the in-depth oxygen diffusion when exposed to oxidizing atmospheres [39]. More importantly, carbon fibers undergo active oxidation at very low temperature (450 C) and could be used only in composites where they are perfectly protected again oxidation. Finally, the HR carbon fibers (which are the most attractive in terms of cost and availability), have to be properly stabilized through a high temperature treatment (HTT) when used to reinforce SiC. The situation is a priori better with SiC fibers since all the problems related to FM-compatibility at high temperatures are solved and since SiC displays a much better oxidation resistance than carbon. However, all the SiC-based fibers are not pure SiC and most of them are very expensive (Fig. 4). The early Si–C–O fibers [such as the Nicalon (from Nippon Carbon, Japan) fiber], referred to as SiC-based fibers of first generation, consist of SiC-nanocrystals (1–2nm in size) and free carbon embedded in an amorphous SiCxOy matrix. As a result, their stiffness (E=220 GPa) is much lower than that of pure SiC (E400 GPa), their strain at failure relatively high ("R=1.4%) and more importantly, they decompose beyond 1100–1200 C with a strength degradation [40]. Hence, SiC/SiC composites fabricated with these fibers should be processed by low temperature techniques (CVI or PIP) and their use is limited in temperature (Table 1). SiC-based fibers of second generation (such as HiNicalon) are oxygen-free fibers consisting of a mixture of SiC-nanocrystals (5 nm in mean size) and free carbon [C/Si (at) ratio=1.39]. They do not undergo decomposition at high temperature since they do not contain significant amount of SiCxOy phase. They creep at moderate temperatures (1200 C) but their creep resistance is improved (1400 C) if they have been submitted to a heat treatment at 1400–1600 C which stabilizes the fiber microstructure [40,41]. The SiC fibers of third generation (such as Hi-Nicalon S, Tyranno SA or Sylramic) are oxygen-free and quasi-stoichiometric [with C/Si (at)=1.00–1.08] [40,42,43]. Further, their grain size is relatively large (20–200 nm) and their thermal stability is excellent, since they are prepared at very high temperature (1600–2000 C). As a result, their use in SiC/SiC composites is compatible with all the fabrication processes depicted in Section 2, including those at high temperatures. Conversely, the SiC fibers of third generation are very stiff (E 400 GPa) and their weaveability is low. Further, their somewhat low strain at failure (0.6–0.8%) limits the extension of the non-linear stress-strain domain partly responsible for the non-brittle character of SiC/SiC composites. Finally, these advanced SiC fibers are costly. Obviously, there is still a need for low cost SiC-based fibers exhibiting moderate stiffness and relatively high failure strength as well as high thermal stability and creep resistance (Table 1). Interestingly, a new ceramic fiber, corresponding to the overall formula SiBN3C, has been recently proposed Fig. 4. Variations of the room temperature failure strength of different SiC-based fibers as a function of their heat treatment temperature: Tyranno (Lox M and ZMI) and Nicalon (NL 201) are Si–C–O (M) fibers, Hi-Nicalon is a SiC+C fiber and SA fiber is a quasi-stoichiometric SiC fiber, according to Ishikawa et al. [42]. 160 R. Naslain / Composites Science and Technology 64 (2004) 155–170
R. Naslain/ Composites Science and Technology 64 (2004)155-170 Table I Properties of Sic-based and related fibers from literature sources(properties may vary from lot to lot and some of the fibers are still experimental) Properties Nicalon Hi-Nical Hi- Nical Sylramic Carborundum SHC-O) on si-c SA SIC(AI) SIC(B. T1) SIB-C-N Y-SiC Diameter (um) 8-10 8-14 Density(g/cm) 2.55 3.0-3.1 18-2.0 Tensile strength(GPa) Young s modulus(GPa) 386-40 0.4 CTE(10-6K- 3.1-3.2 3.0-3. Thermal conduct(W/mK) 1.4-3.0 40-45 Resistivity(S2 cm 0.1 ≈1470 ≈1500 Depending on processing conditions. bAt25°C c Temperature at which m=0.5 in BSR-test. [44]. It is prepared from a polymer precursor and it dis-(interphases with a layered crystal structure)or/and plays surprising properties. Firstly, it remains amor along the interfaces between the elementary layers phous up to extremely high temperatures(1700-1900oC, (interphases with a layered microstructure)and in both depending on the atmosphere) and hence it exhibits a case roughly parallel to the fiber surface (mode high failure strength(3-4 GPa). Its weaveability is good I-mode II deflection mode). The last condition, i.e owing to its small diameter(8-14 um)and relatively low that requiring a strong bonding between the interphase stifness (E-180-350 GPa). Surprisingly, its creep and the fiber is often underestimated although it is resistance is similar to that of the best Sic polycrystal essential. If it is not satisfied, FM-debonding will pre- line fibers, although it is actually amorphous(Table 1). ferentially occurred at the fiber surface, sometimes over Finally, it may have a good compatibility with boron very long distances, with the result that the load trans nitride(one of the best interphase materials for SiC/Sic fer function is largely lost and the fibers exposed to the composites) and it displays an oxidation resistance bet- ter than those of Si, N4 and SiC. Unfortunately, fibers from the Si-B-C-N quarternary system are still experi mental and their cost. relative to that of other advanced Sic fibers is not known 3.2. Interp The interphase is a thin film of a compliant material with a low shear strength(typically, 0.I-l um in thick- ness), which is deposited on the fiber surface prior to the infiltration of the matrix and whose main function (b) is to arrest or/and deflect the matrix microcracks hence protecting the fibers from an early failure by notch effect(mechanical fuse function). In addition, the interphase has a load transfer function(as in any composite) and may act as diffusion barrier during composite processing, when necessary. Although different kinds of interphase concepts have been suggested 145, it has been postulated that the best interphase materials might be those with a layered crys- tal structure(Pyrocarbon, hexagonal-BN)or a layered microstructure [(PyC-SiCn or(BN-SiC)nl, the layers being deposited parallel to the fiber surface and weakly Fig. 5. D bonded to one another, and the interphase being composites: (a) single layer pyrocarbon or hexagonal BN strongly bonded to the fiber surface( Fig. 5)[46]. When (b) porous SiC single layer interphase, (c)multilayered(X- phase, with X=PyC or BN and y=SiC (schematic), (d) these conditions are properly satisfied, crack deflection tion in a multilayered(Pyc-SiC)o interphase, according to Naslain occurs within the interphase either along atomic planes (a-c)[46] and Bertrand et al. [50]
[44]. It is prepared from a polymer precursor and it displays surprising properties. Firstly, it remains amorphous up to extremely high temperatures (1700–1900 C, depending on the atmosphere) and hence it exhibits a high failure strength (3–4 GPa). Its weaveability is good owing to its small diameter (8–14 mm) and relatively low stiffness (E=180–350 GPa). Surprisingly, its creep resistance is similar to that of the best SiC polycrystalline fibers, although it is actually amorphous (Table 1). Finally, it may have a good compatibility with boron nitride (one of the best interphase materials for SiC/SiC composites) and it displays an oxidation resistance better than those of Si3N4 and SiC. Unfortunately, fibers from the Si–B–C–N quarternary system are still experimental and their cost, relative to that of other advanced SiC fibers, is not known. 3.2. Interphases The interphase is a thin film of a compliant material with a low shear strength (typically, 0.1–1 mm in thickness), which is deposited on the fiber surface prior to the infiltration of the matrix and whose main function is to arrest or/and deflect the matrix microcracks, hence protecting the fibers from an early failure by notch effect (mechanical fuse function). In addition, the interphase has a load transfer function (as in any fiber composite) and may act as diffusion barrier during composite processing, when necessary. Although different kinds of interphase concepts have been suggested [45], it has been postulated that the best interphase materials might be those with a layered crystal structure (Pyrocarbon, hexagonal-BN) or a layered microstructure [(PyC–SiC)n or (BN–SiC)n], the layers being deposited parallel to the fiber surface and weakly bonded to one another, and the interphase being strongly bonded to the fiber surface (Fig. 5) [46]. When these conditions are properly satisfied, crack deflection occurs within the interphase either along atomic planes (interphases with a layered crystal structure) or/and along the interfaces between the elementary layers (interphases with a layered microstructure) and in both case roughly parallel to the fiber surface (mode I!mode II deflection mode). The last condition, i.e. that requiring a strong bonding between the interphase and the fiber is often underestimated although it is essential. If it is not satisfied, FM-debonding will preferentially occurred at the fiber surface, sometimes over very long distances, with the result that the load transfer function is largely lost and the fibers exposed to the Table 1 Properties of SiC-based and related fibers from literature sources (properties may vary from lot to lot and some of the fibers are still experimental) Properties Nicalon (Si–C–O) Hi-Nical on Si–C Hi-Nical on S SiC Tyranno SA SiC (Al) Sylramic SiC (B, Ti) Bayer Si–B–C–N Carborundum a-SiC Diameter (mm) 14 14 128–10 10 8–14 28 Density (g/cm3 ) 2.55 2.74 3.10 3.0–3.1 3.0–3.1 1.8–2.0 3.1 Tensile strength (GPa) 3.0 2.8 2.6 2.8–3.0 3.0–3.2 3–4a 1.4 Young’s modulus (GPa) 220 270 390–420 390–420 386–400 180–350a 420 Failure strain (%) 1.4 1.0 0.6 0.7 0.6–0.8 0.7–1.5a 0.4 CTE (106 K1 ) 3.1–3.23.3–3.5 – – 4.0–5.4 3.0–3.5a 4.5 Thermal conduct (W/mK)b 1.4–3.0 5.0–7.8 18 65 40–45 – – Resistivity ( cm) 103 –104 1.4 0.1 – – – – Creep parameter (T0.5) c 1110 1240 1470 1500 1400 1500 1550 a Depending on processing conditions. b At 25 C. c Temperature at which m=0.5 in BSR-test. Fig. 5. Different simple or engineered interphases used in SiC-matrix composites: (a) single layer pyrocarbon or hexagonal BN interphases, (b) porous SiC single layer interphase, (c) multilayered (XY)n interphase, with X=PyC or BN and Y=SiC (schematic), (d) crack deflection in a multilayered (PyC–SiC)10 interphase, according to Naslain (a–c) [46] and Bertrand et al. [50]. R. Naslain / Composites Science and Technology 64 (2004) 155–170 161
R. Naslain/ Composites Science and Technology 64(2004)155-170 atmosphere, e.g. to oxygen diffusing along the crack carbon or hex-BN. Furthermore, these interphases network. Classically, the fiber /interphase bonding can display a self-healing behavior when exposed to an be strengthened through thermal or/and chemical oxidizing atmosphere: the fluid oxide phase formed by treatments of the fibers(often proprietary and whose the oxidation of the bn, B4C or SiC layers filling the details are not known)in order to modify their surface nanosized cracks resulting from crack deflection in composition or/and morphology. A recent example has the interphase. Finally, in a fast neutron environment been given for the hex-BN interphase. This layered high temperatures, multilayers with a limited overall crystal structure interphase is often poorly bonded to thickness of pyrocarbon or hex- BN are more stable than Sic-based fibers when deposited by CVD/CVI [Eq (3)], relatively thick and homogeneous PyC or hex-BN single a thin film of carbon or/and silica being present at the layer interphases [51, 52 surface of Si-C-O fibers(Nicalon, for example) which actually acts as mechanical fuse in SiC/SiC composites 3. 3. Matrices [47]. Conversely, when it is formed by CVR(chemical vapor reaction), i.e. by reaction of a boron-containing The choice of silicon carbide as a matrix is based on fiber(such as the Sylramic fiber) with a nitrogen-con its high melting point (2500C), excellent mechanical taining atmosphere, the BN-interphase is thin and properties at high temperatures related to its covalent adherent to the fiber, with the result that the corre- character, relatively good oxidation resistance up to sponding SiC/BN SiC composites display a much better about 1500C in oxygen-rich atmospheres and stability behavior in corrosive environments (here the combus- in fast neutron environments. Also, silicon carbide can ion gas of a burner rig), the fibers being protected by be easily deposited in a fiber preform by a variety of he adherent BN-coating [48, 49]. When the fiber/inter- techniques, as discussed in Section 2. However, when phase is strong, crack deflection and debonding occur in used as a matrix in either C/SiC or SiC/Sic composites the interphase itself, often in a very diffuse manner(the (which are from a mechanical standpoint, so-called matrix microcrack giving rise to an infinity of highly inverse composites with r ef), silicon carbide branched nanosized cracks of very low opening, with undergoes multiple microcracking when loaded in ten he result that the load function of the interphase is sion beyond a relatively low stress level (100-200 MPa) least partly maintained and the fibers not directly [4. These microcracks, whose density and opening exposed to the atmosphere. The resulting Sic/Sic com- depend on the fiber architecture, FM-bonding and osites are simultaneously tough and strong, resistant to applied load, facilitate the in-depth diffusion of oxygen fatigue and to the environment. towards the oxidation-prone interphases and fibers, The number of interphase materials with a layered when the composite is exposed to an oxidizing atmo crystal structure being very limited and the best of them sphere at medium or high temperatures. (pyrocarbon and hex-BN) being further oxidation- Hence, the objective of material design is here to prone, the concept of layered interphases has been engineer or tailor the Sic-matrix in order to impede or extended from the crystal structure to the micro- at least to slow down oxygen diffusion in the material structure levels [46]. Now the interphase is an artificially and to increase its durability in corrosive environments produced(X-Yn multilayer where X is a crack deflect The guideline is (1)to introduce in the Sic-matrix, ele- ng material (such as pyrocarbon or hex-BN) and Y a ments(such as boron)forming fluid oxide phases in a material forming a fluid oxide phase when exposed to wide temperature range that could fill the cracks, ren- an oxidizing atmosphere(such as Sic or B4C)(Fig. 5c, dering the whole material self-healing and (2)to reduce d ). In the elemental X-Y sequence, the thickness of the the crack opening through a proper design of the layers is typically a few nanometers or a few 10 nm, FM-interfacial zone, as discussed in Section 3. 2. Boron whereas n the number of elemental X-Y sequences oxide phases are efficient at relatively low temperatures which is repeated to build the interphase is in the range (500-1000C)whereas silica-rich phases are more appro- of 2-30[46, 50].(PyC-SiC)n and(BN-SiC)n multilayered priate at high temperatures(1000-1500oC) In both cases, interphases are examples of highly engineered nano one assumes(and this is generally true in most applica- objects produced by P-CVI, the design parameters being tions) that the oxygen partial pressure in the environment the thickness of the elemental layers and the number of is high enough(passive oxidation regime) to form con- elemental sequences which can be controlled through densed covering and protective oxides, according to the the deposition kinetics and both the number and dura- following equations written for SiC and for a commonly tion of the pressure pulses for each precursor [see Eqs. used boron-bearing additive, boron carbide (H3.(X-Yn interphases display several advantages Firstly, the number of crack deflection possibilities is SiC(s)+202(g) SiO(s)+CO increased, since numerous X/r interfaces are present Secondly, the overall amount of oxidation-prone mate- rial is decreased, e.g. Sic partly replacing either pyr- SiC(+3/202g) SiO2(s 1)+ CO(g) (7)
atmosphere, e.g. to oxygen diffusing along the crack network. Classically, the fiber/interphase bonding can be strengthened through thermal or/and chemical treatments of the fibers (often proprietary and whose details are not known) in order to modify their surface composition or/and morphology. A recent example has been given for the hex-BN interphase. This layered crystal structure interphase is often poorly bonded to SiC-based fibers when deposited by CVD/CVI [Eq. (3)], a thin film of carbon or/and silica being present at the surface of Si–C–O fibers (Nicalon, for example) which actually acts as mechanical fuse in SiC/SiC composites [47]. Conversely, when it is formed by CVR (chemical vapor reaction), i.e. by reaction of a boron-containing fiber (such as the Sylramic fiber) with a nitrogen-containing atmosphere, the BN-interphase is thin and adherent to the fiber, with the result that the corresponding SiC/BN/SiC composites display a much better behavior in corrosive environments (here the combustion gas of a burner rig), the fibers being protected by the adherent BN-coating [48,49]. When the fiber/interphase is strong, crack deflection and debonding occur in the interphase itself, often in a very diffuse manner (the matrix microcrack giving rise to an infinity of highly branched nanosized cracks of very low opening, with the result that the load function of the interphase is at least partly maintained and the fibers not directly exposed to the atmosphere. The resulting SiC/SiC composites are simultaneously tough and strong, resistant to fatigue and to the environment. The number of interphase materials with a layered crystal structure being very limited and the best of them (pyrocarbon and hex-BN) being further oxidationprone, the concept of layered interphases has been extended from the crystal structure to the microstructure levels [46]. Now the interphase is an artificially produced (XY)n multilayer where X is a crack deflecting material (such as pyrocarbon or hex-BN) and Y a material forming a fluid oxide phase when exposed to an oxidizing atmosphere (such as SiC or B4C) (Fig. 5c, d). In the elemental XY sequence, the thickness of the layers is typically a few nanometers or a few 10 nm, whereas n the number of elemental XY sequences which is repeated to build the interphase is in the range of 2–30 [46,50]. (PyC–SiC)n and (BN–SiC)n multilayered interphases are examples of highly engineered nano objects produced by P-CVI, the design parameters being the thickness of the elemental layers and the number of elemental sequences which can be controlled through the deposition kinetics and both the number and duration of the pressure pulses for each precursor [see Eqs. (1)–(3)]. (XY)n interphases display several advantages. Firstly, the number of crack deflection possibilities is increased, since numerous X/Y interfaces are present. Secondly, the overall amount of oxidation-prone material is decreased, e.g. SiC partly replacing either pyrocarbon or hex-BN. Furthermore, these interphases display a self-healing behavior when exposed to an oxidizing atmosphere: the fluid oxide phase formed by the oxidation of the BN, B4C or SiC layers filling the nanosized cracks resulting from crack deflection in the interphase. Finally, in a fast neutron environment at high temperatures, multilayers with a limited overall thickness of pyrocarbon or hex-BN are more stable than relatively thick and homogeneous PyC or hex-BN single layer interphases [51,52]. 3.3. Matrices The choice of silicon carbide as a matrix is based on its high melting point (2500 C), excellent mechanical properties at high temperatures related to its covalent character, relatively good oxidation resistance up to about 1500 C in oxygen-rich atmospheres and stability in fast neutron environments. Also, silicon carbide can be easily deposited in a fiber preform by a variety of techniques, as discussed in Section 2. However, when used as a matrix in either C/SiC or SiC/SiC composites (which are from a mechanical standpoint, so-called inverse composites with "R m < "R f ), silicon carbide undergoes multiple microcracking when loaded in tension beyond a relatively low stress level (100–200 MPa) [4]. These microcracks, whose density and opening depend on the fiber architecture, FM-bonding and applied load, facilitate the in-depth diffusion of oxygen towards the oxidation-prone interphases and fibers, when the composite is exposed to an oxidizing atmosphere at medium or high temperatures. Hence, the objective of material design is here to engineer or tailor the SiC-matrix in order to impede or at least to slow down oxygen diffusion in the material and to increase its durability in corrosive environments. The guideline is (1) to introduce in the SiC-matrix, elements (such as boron) forming fluid oxide phases in a wide temperature range that could fill the cracks, rendering the whole material self-healing and (2) to reduce the crack opening through a proper design of the FM-interfacial zone, as discussed in Section 3.2. Boron oxide phases are efficient at relatively low temperatures (500–1000 C) whereas silica-rich phases are more appropriate at high temperatures (1000–1500 C). In both cases, one assumes (and this is generally true in most applications) that the oxygen partial pressure in the environment is high enough (passive oxidation regime) to form condensed covering and protective oxides, according to the following equations written for SiC and for a commonly used boron-bearing additive, boron carbide: SiCð Þs þ 2O2ð Þ g ! SiO2ð Þs þ CO2ð Þ g ð7Þ SiCð Þs þ 3=2O2ð Þ g ! SiO2ð Þ s;l þ COð Þ g ð70 Þ 162 R. Naslain / Composites Science and Technology 64 (2004) 155–170
B C(s)+402(g) 2B2 O3(s. 1)+ COz(g) B4C(s)+7/202(g)- 2B2O3(s, 1+ CO(g) (8) Coatings are often applied to C/SiC and SiC/SiC composites to still improve their durability in corrosive At intermediate temperatures, both Sic and BC environments, the most commonly employed being undergo oxidation yielding a SiOr-B2O3 phase whose relatively thick SiC layer deposited by CVD to seal the viscosity depends on the sio2/ B2O3 ratio and the oxi- residual open porosity inherent to the CvI and PIP dation kinetics, and hence on the matrix design, at least processes. SiC coatings are already microcracked when to some extent deposited on a C/SiC composite due to CtE mismatch A simple and commonly used way to introduce a or they undergo microcracking when loaded beyond the B-bearing phase in a Sic-matrix is to employ a hybrid proportional limit in SiC/SiC composites. However, processing technique(see Section 2), i.e. a multistep they behave relatively well, particularly at high tem- infiltration approach, combining for example an peratures in oxidizing atmospheres if the material is not impregnation of a fiber preform with a slurry consisting cycled in temperature or/and under load, since a pro- of a suspension of boron-based particulates(such as tective silica scale is formed [39]. When this condition is B4C, SiB or boron itself) alone or mixed with Sic par not fulfilled, they do not provide an efficient prote iculates in a solution of polycarbosilane, with one or particularly at medium temperatures, microcracks being several CVI (or LSI) step(s) for the infiltration of the continuously formed which may not be filled rapidly Sic-matrix [ 37, 38 enough by silica(whose growth kinetics is very slow below A more elegant way, in terms of material design and 900oC), a feature resulting in an in-depth diffusion of flexibility, is to introduce the B-bearing phase as layers oxygen and a degradation of the interphase and fibers in the Sic-matrix, which can be done at different scales Under such frequently encountered conditions, multi- by I-CVI or P-CVI(Fig 6). One possibility among oth layered self-heal ing coatings, designed as previously depic ers, is to build a multilayered Sic-based matrix com ed for interphases and matrices can be envisaged [55] bining concentric layers(around single fibers, groups of Finally, in specific environments such as high pressure fibers or fiber yarns, in the preform) of crack arresters fuel combustion gas with a relatively high partial pres (such as pyrocarbon or hex -BN)and of binary or ternary sure of water vapor, silica is no longer protective and phases from the B-C, Si-C or Si-B-C systems forming silicon carbide undergoes a recession. In such a case, fluid phases in oxidizing atmospheres that could fill the oxide based coatings referred to as environmental bar matrix cracks as soon as they are initiated and propagate rier coatings (EBC) have been designed and tested under applied load in the material. Under such condi- These coatings are usually multilayered combining for tions, oxygen is entrapped in the oxide phases as it dif- examples: (1)a first layer of silicon, a second and inter fuses in the crack network and eventually does not reach mediate layer of mullite and a third external layer of the oxidation- prone fibers and FM-interfacial zone, baryum-strontium-alumino silicates(BSAS)[56] or(2) ncreasing thus the material durability [7, 53, 54 an inner layer of silicon carbide and an outer layer of yttrium silicate [57]. These layers are deposited either by PVd or cvd as well as from slurries 4. Examples of potential application Potential application of Sic-matrix composites in two ifferent fields, namely aerojet engines and stationary gas turbines for electrical power/steam cogeneration, on the one hand. and nuclear fusion reactors. on the other hand, will be briefly discussed. These application fields correspond to very severe service conditions(high tem peratures and corrosive environments in both cases, well as radiation exposure in nuclear reactors) and long 0817 X1,7018W035 lifetime requirement (from a few hundreds to a few ten housand hours) Fig. 6. Multilayered self-healing matrix deposited by l-CVI: each elementary sequence contains crack deflector(mechanical fuse) and layers forming fluid oxide phases by reaction with oxygen diffusing in 4.1. Aerojet engines and stationary gas turbines the crack network. Such sequence are either around elementary fibers (simple arrow), group of fibers(double arrow) or fiber yarn(triple The hot parts of aerojet engines and stationary gas arrow), according to Lamouroux et al. [71- turbines are presently fabricated with heavy cobalt or
B4Cð Þs þ 4O2ð Þ g ! 2B2O3ð Þ s;l þ CO2ð Þ g ð8Þ B4Cð Þs þ 7=2O2ð Þ g ! 2B2O3ð Þ s;l þ COð Þ g ð80 Þ At intermediate temperatures, both SiC and B4C undergo oxidation yielding a SiO2–B2O3 phase whose viscosity depends on the SiO2/B2O3 ratio and the oxidation kinetics, and hence on the matrix design, at least to some extent. A simple and commonly used way to introduce a B-bearing phase in a SiC-matrix is to employ a hybrid processing technique (see Section 2), i.e. a multistep infiltration approach, combining for example an impregnation of a fiber preform with a slurry consisting of a suspension of boron-based particulates (such as B4C, SiB6 or boron itself) alone or mixed with SiC particulates in a solution of polycarbosilane, with one or several CVI (or LSI) step(s) for the infiltration of the SiC-matrix [37,38]. A more elegant way, in terms of material design and flexibility, is to introduce the B-bearing phase as layers in the SiC-matrix, which can be done at different scales by I-CVI or P-CVI (Fig. 6). One possibility among others, is to build a multilayered SiC-based matrix combining concentric layers (around single fibers, groups of fibers or fiber yarns, in the preform) of crack arresters (such as pyrocarbon or hex-BN) and of binary or ternary phases from the B–C, Si–C or Si–B–C systems forming fluid phases in oxidizing atmospheres that could fill the matrix cracks as soon as they are initiated and propagate under applied load in the material. Under such conditions, oxygen is entrapped in the oxide phases as it diffuses in the crack network and eventually does not reach the oxidation-prone fibers and FM-interfacial zone, increasing thus the material durability [7,53,54]. 3.4. Coatings Coatings are often applied to C/SiC and SiC/SiC composites to still improve their durability in corrosive environments, the most commonly employed being a relatively thick SiC layer deposited by CVD to seal the residual open porosity inherent to the CVI and PIP processes. SiC coatings are already microcracked when deposited on a C/SiC composite due to CTE mismatch, or they undergo microcracking when loaded beyond the proportional limit in SiC/SiC composites. However, they behave relatively well, particularly at high temperatures in oxidizing atmospheres if the material is not cycled in temperature or/and under load, since a protective silica scale is formed [39]. When this condition is not fulfilled, they do not provide an efficient protection, particularly at medium temperatures, microcracks being continuously formed which may not be filled rapidly enough by silica (whose growth kinetics is very slow below 900 C), a feature resulting in an in-depth diffusion of oxygen and a degradation of the interphase and fibers. Under such frequently encountered conditions, multilayered self-healing coatings, designed as previously depicted for interphases and matrices can be envisaged [55]. Finally, in specific environments such as high pressure fuel combustion gas with a relatively high partial pressure of water vapor, silica is no longer protective and silicon carbide undergoes a recession. In such a case, oxide based coatings referred to as environmental barrier coatings (EBC) have been designed and tested. These coatings are usually multilayered combining for examples: (1) a first layer of silicon, a second and intermediate layer of mullite and a third external layer of baryum-strontium-alumino silicates (BSAS) [56] or (2) an inner layer of silicon carbide and an outer layer of yttrium silicate [57]. These layers are deposited either by PVD or CVD as well as from slurries. 4. Examples of potential application Potential application of SiC-matrix composites in two different fields, namely aerojet engines and stationary gas turbines for electrical power/steam cogeneration, on the one hand, and nuclear fusion reactors, on the other hand, will be briefly discussed. These application fields correspond to very severe service conditions (high temperatures and corrosive environments in both cases, as well as radiation exposure in nuclear reactors) and long lifetime requirement (from a few hundreds to a few ten thousand hours). 4.1. Aerojet engines and stationary gas turbines The hot parts of aerojet engines and stationary gas turbines are presently fabricated with heavy cobalt or Fig. 6. Multilayered self-healing matrix deposited by I-CVI: each elementary sequence contains crack deflector (mechanical fuse) and layers forming fluid oxide phases by reaction with oxygen diffusing in the crack network. Such sequence are either around elementary fibers (simple arrow), group of fibers (double arrow) or fiber yarn (triple arrow), according to Lamouroux et al. [7]. R. Naslain / Composites Science and Technology 64 (2004) 155–170 163
R. Naslain/ Composites Science and Technology 64(2004)155-170 nickel base superalloys(p 88-9 g/cm ). It is admitted be said that the after-burner components of an aerojet that these highly performant materials(from a mechan- engine can be potentially produced with Sic-based ical standpoint) have reached their temperature limit matrix composites on the basis of the presently available (e1100oC): they are already used at temperatures close scientific knowledge and technology (Fig. 7). The to their melting point(solidus of the phase diagram), required durability can be achieved through the use of sometimes as single crystals(rotor blades)and they have oxidation resistant interphases (stabilized BN, for to be cooled which complicates the design of the parts. example, self-healing matrices or/and multilayered Replacing the superalloys by light, tough, refractory coatings, however, economical consideration(such as and creep resistant SiC-matrix composites will permit a the fiber cost) is still an issue significant increase of service temperature and hence an The application of SiC-matrix composites to the increase of the engine efficiency, a reduction of the No combustion chamber is more demanding in terms of ser CO emission(through an optimization of the fuel/air vice temperature, corrosion resistance and durability. It ratio), a simplification of the part design and a weight has been studied mainly in America and Japan, within saving(typically, 30-50%). However, their use still rises the scope of programs of research/development on gas a number of questions dealing with their durability, turbines of co-generation [30, 60-62, 64-66]. The first reliability, manufacture, design and cost. Presently, the results are encouraging since SiC/Sic combustor liners application of Sic-matrix composites which are envi- have experienced without failure several tens of thou saged in this field is limited mostly to non rotating parts sands of hours field-testing at temperatures ranging including combustor liners, after-burner components from 1000 to 1250C in a water vapor rich high pres (exhaust cone and flame holder) and exhaust nozzles sure environment. The combustor for a 4-5 Mw gas (outer and inner flaps) in military aerojet engines, as turbine consisted of two concentric cylindrical ceramic well as combustor liners of large size in stationary gas liners of large size (outer liner: 0.=75 cm, length turbine for electrical power/steam cogeneration Lo=20 cm, thickness eo=3-4 mm; inner liner: oi=33 [9,30,58-62] m, Li=20 cm, ei=2-3 mm)in a metal housing with an Flaps for exhaust nozzles of military aerojet engines alumina-based fiber thermal insulation between the have been bench-and in-flight-tested for more than one outer liner and the metallic housing(Fig. 8). Several decade. Outer flaps for the M53-2 and M88-2 materials were successively tested including SiC(Nica SNECMA aerojet engines have been designed and fab- lon)/enhanced SiC(CVI), SiC (Hi-Nicalon )/enchanced ricated with nD-C/SiC composites(n=2 or 3)since they SiC(CVI)and SiC(Hi-Nicalon/ SiC(RMI)composites. are exposed to moderate temperatures(650-700 C). with either a pyrocarbon or a hex-BN interphases. The They correspond to a weight saving of about 50% with external coating was first a simple Sic coating and then respect to their superalloy counterpart. This weight an oxide based environmental barrier coating(EBC)when iving at the very rear part of the engine has a notice- it became clear that the former underwent unacceptable able incidence on the location of the gravity center of recession, as previously discussed [60, 62, 64]. Although the engine and the manoeuvrability of the fighter. The durability under these extremely severe conditions still first outer flaps were in-flight tested as early as 1989 and remains an issue, preliminary tests have established the are now in volume production for the M88-2 engines of the Rafale fighter. They are as far as we know the first actual application of CMCs in this field. Inner flaps seeing a temperature of 850C have been designed and fabricated with 2D-SiC(Nicalon )/SiC composites and bench-and in- fight-tested successfully but are still in development [8, 58]. Finally, divergent flap inserts fc the General Electric F1l0 turbofan engine have also been designed and fabricated with different CMCs, the service temperature being similarly of 800-900 oC Among the different fiber/matrix compositions tested the best results were obtained with 2D-Nicalon/BN/SIC and 2D-Nicalon/carbon (inhibited) materials which showed no loss of properties after A100 h engine ground testing [63]. Other after-burner components, uch as fame holder and exhaust cone experiencing Fig. 7. After-burner parts in a military ngine already fabri- temperatures in the range 800-1100 and 800-950oC cated with SiC-matrix composites, ac respectively, have been designed, manufactured and (a)exhaust cone, (b)outer flap,(c)inner me hold bench-tested, in a preliminary study. They consisted of I The enhanced SiC matrix is a matrix that contains tion Sic (Nicalon)/SiC composites [58]. To conclude, it can
nickel base superalloys ( 8–9 g/cm3 ). It is admitted that these highly performant materials (from a mechanical standpoint) have reached their temperature limit (1100 C): they are already used at temperatures close to their melting point (solidus of the phase diagram), sometimes as single crystals (rotor blades) and they have to be cooled which complicates the design of the parts. Replacing the superalloys by light, tough, refractory and creep resistant SiC-matrix composites will permit a significant increase of service temperature and hence an increase of the engine efficiency, a reduction of the NOx/ CO emission (through an optimization of the fuel/air ratio), a simplification of the part design and a weight saving (typically, 30–50%). However, their use still rises a number of questions dealing with their durability, reliability, manufacture, design and cost. Presently, the application of SiC-matrix composites which are envisaged in this field is limited mostly to non rotating parts including combustor liners, after-burner components (exhaust cone and flame holder) and exhaust nozzles (outer and inner flaps) in military aerojet engines, as well as combustor liners of large size in stationary gas turbine for electrical power/steam cogeneration [19,30,58–62]. Flaps for exhaust nozzles of military aerojet engines have been bench- and in-flight-tested for more than one decade. Outer flaps for the M53-2and M88-2 SNECMA aerojet engines have been designed and fabricated with nD-C/SiC composites (n=2or 3) since they are exposed to moderate temperatures (650–700 C). They correspond to a weight saving of about 50% with respect to their superalloy counterpart. This weight saving at the very rear part of the engine has a noticeable incidence on the location of the gravity center of the engine and the manoeuvrability of the fighter. The first outer flaps were in-flight tested as early as 1989 and are now in volume production for the M88-2engines of the Rafale fighter. They are as far as we know the first actual application of CMCs in this field. Inner flaps seeing a temperature of 850 C have been designed and fabricated with 2D-SiC(Nicalon)/SiC composites and bench- and in-flight-tested successfully but are still in development [8,58]. Finally, divergent flap inserts for the General Electric F110 turbofan engine have also been designed and fabricated with different CMCs, the service temperature being similarly of 800–900 C. Among the different fiber/matrix compositions tested, the best results were obtained with 2D-Nicalon/BN/SiC and 2D-Nicalon/carbon (inhibited) materials which showed no loss of properties after 100 h engine ground testing [63]. Other after-burner components, such as flame holder and exhaust cone experiencing temperatures in the range 800–1100 and 800–950 C respectively, have been designed, manufactured and bench-tested, in a preliminary study. They consisted of SiC (Nicalon)/SiC composites [58]. To conclude, it can be said that the after-burner components of an aerojet engine can be potentially produced with SiC-based matrix composites on the basis of the presently available scientific knowledge and technology (Fig. 7). The required durability can be achieved through the use of oxidation resistant interphases (stabilized BN, for example), self-healing matrices or/and multilayered coatings, however, economical consideration (such as the fiber cost) is still an issue. The application of SiC-matrix composites to the combustion chamber is more demanding in terms of service temperature, corrosion resistance and durability. It has been studied mainly in America and Japan, within the scope of programs of research/development on gas turbines of co-generation [30,60–62,64–66]. The first results are encouraging since SiC/SiC combustor liners have experienced without failure several tens of thousands of hours field-testing at temperatures ranging from 1000 to 1250 C in a water vapor rich high pressure environment. The combustor for a 4–5 MW gas turbine consisted of two concentric cylindrical ceramic liners of large size (outer liner: 1o=75 cm, length Lo=20 cm, thickness eo=3–4 mm; inner liner: 1i=33 cm, Li=20 cm, ei=2–3 mm) in a metal housing with an alumina-based fiber thermal insulation between the outer liner and the metallic housing (Fig. 8). Several materials were successively tested including SiC (Nicalon)/enhanced SiC (CVI), SiC (Hi-Nicalon)/enchanced1 SiC (CVI) and SiC (Hi-Nicalon/SiC (RMI) composites, with either a pyrocarbon or a hex-BN interphases. The external coating was first a simple SiC coating and then an oxide based environmental barrier coating (EBC) when it became clear that the former underwent unacceptable recession, as previously discussed [60,62,64]. Although durability under these extremely severe conditions still remains an issue, preliminary tests have established the Fig. 7. After-burner parts in a military aerojet engine already fabricated with SiC-matrix composites, according to Christin [19]: (a) exhaust cone, (b) outer flap, (c) inner flap, (d) flame holder. 1 The enhanced SiC matrix is a matrix that contains an oxidation inhibitor. 164 R. Naslain / Composites Science and Technology 64 (2004) 155–170