r./. App Ceram. Technol, 7/ 3/263-275(2010) International Journal o Applied p p Ceramic TECHNOLOGY ceramic Product D Single-and Multilayered Interphases in SiC/SiC Composites Exposed to Severe Environmental Conditions An overview Roger R. Naslain, Rene J -F. Pailler, and Jacques L. Lamon Laboratory for ThermoStructural Composites(LCTS), University of Bordeaux, 33600 Pessac, france Pyrocarbon(PyC), the common interphase for SiC/SiC, is not stable under severe environmental conditions. It could be replaced by boron nitride more resistant to oxidation but poorly compatible with nuclear applications. Other materials, such as ternary carbides seem promising but their use in SiC/SiC has not been demonstrated. The most efficient way to improve the behavior of Py C interphase in severe environments is to replace part of PyC by a material displaying a better compatibility, such as SiC itself. Issues related to the design and behavior of layered interphases are reviewed with a view to demonstrate their interest in high-temperature nuclear reactors. Introduction residual thermal stresses. The interphase protects fibers against chemical reactions that could occur during pre The interphase plays a key role in the behavior of cessing and use of CMCs in aggressive environments ceramic matrix composites(CMCs). It prevents the early It has been postulated that the best failure of the fibers, matrix microcracks being arrested terials for SiC/SiC might be those with a layered structure, or/and deflected parallel to fiber axis(the so-called the layers being parallel to fiber surface, weakly bonded to mechanical fuse"function). It also transfers load from one another but strongly adherent to fibers. Pyrocar the fibers to the matrix and eventually releases part of bon(PyC) was thought to be the best interphase material for sic/sic in terms of their mechanical behavior -5 Unfortunately, Py C is oxidation prone even at low This paper was presented in part ar the 33rd Annual International Conference on Advanced temperatures with the result that PyC interphase can be consumed and the FM coupling nce, that C 200) The American Ceramic Society interphase becomes the weak point of SiC/SiC whe
Single- and Multilayered Interphases in SiC/SiC Composites Exposed to Severe Environmental Conditions: An Overview Roger R. Naslain,* Rene´ J.-F. Pailler, and Jacques L. Lamon Laboratory for ThermoStructural Composites (LCTS), University of Bordeaux, 33600 Pessac, France Pyrocarbon (PyC), the common interphase for SiC/SiC, is not stable under severe environmental conditions. It could be replaced by boron nitride more resistant to oxidation but poorly compatible with nuclear applications. Other materials, such as ternary carbides seem promising but their use in SiC/SiC has not been demonstrated. The most efficient way to improve the behavior of PyC interphase in severe environments is to replace part of PyC by a material displaying a better compatibility, such as SiC itself. Issues related to the design and behavior of layered interphases are reviewed with a view to demonstrate their interest in high-temperature nuclear reactors. Introduction The interphase plays a key role in the behavior of ceramic matrix composites (CMCs). It prevents the early failure of the fibers, matrix microcracks being arrested or/and deflected parallel to fiber axis (the so-called ‘‘mechanical fuse’’ function). It also transfers load from the fibers to the matrix and eventually releases part of residual thermal stresses. The interphase protects fibers against chemical reactions that could occur during processing and use of CMCs in aggressive environments. It has been postulated, that the best interphase materials for SiC/SiC might be those with a layered structure, the layers being parallel to fiber surface, weakly bonded to one another but strongly adherent to fibers.1–4 Pyrocarbon (PyC) was thought to be the best interphase material for SiC/SiC in terms of their mechanical behavior.1–5 Unfortunately, PyC is oxidation prone even at low temperatures with the result that PyC interphase can be consumed and the FM coupling degraded. Hence, that interphase becomes the weak point of SiC/SiC when Int. J. Appl. Ceram. Technol., 7 [3] 263–275 (2010) DOI:10.1111/j.1744-7402.2009.02424.x Ceramic Product Development and Commercialization This paper was presented in part at the 33rd Annual International Conference on Advanced Ceramics and Composites, Daytona Beach, FL, January 18–23, 2009 *naslain@lcts.u-bordeaux1.fr r 2009 The American Ceramic Society
International Journal of Applied Ceramic Technolog-Naslain, Pailler and Lamon Vol. 7, No 3, 2010 used in oxidizing atmospheres(gas turbines). Within the isobaric and P for pressure pulsed) from propane or same conceptual framework, two main alternatives have prop been proposed: boron nitride(BN)and (X-Y)n multi layers (MD). The former displays a structure similar to Fiber/PyC-Interphase Bonding that of graphite while being more oxidation res sistant In the latter, with X= PyC or BN, Y=SiC, and n The second requirement is a strong bonding be- 1-10, Part of the oxidation prone X constituent is tween the fber and the PyC interphase. Achieving replaced by a material Y exhibiting a better oxidation such a strong bonding is first a matter of surface chem- resistance, such as SiC itself. 8- Furthermore, this istry. The surface of desized Si-C-O fibers(Nicalon concept of material multilayering has been extended to the matrix, yielding"self-healing"composites with and free carbon. As a result, there is in SiC/SiC(CVI)a outstanding lifetimes in oxidizing atmospheres. 10,13 thin and irregular dual layer of hous silica and More recently, SiC/SiC composites have beer carbon which introduces a weak link near the fiber sur- envisaged as structural materials in high-temperature face. 4.4 The fibers should be pretreated to clean (HT) nuclear reactors. The interphase appears again as a their surface and achieve a strong fiber/interphase(Fi) possible weak point, PyC being known to undergo bonding. Another example is the SiC+C fibers(Hi- anisotropic volume change when exposed to neutrons, Nicalon or HN, Nippon Carbon, Tokyo, Japan), whose whereas BN undergoes nuclear reactions. 5-7 microstructure is not fully stabilized after processing and is presently pursued following two similar routes: (i)use whose surface may also contain some oxygen. During nd (ii) MLs, to minimize the effect of neutron irradiation. 5-21 CVI processing, fibers undergo a postshrinkage weak ening the FI interface. Again, the fibers should be pre- The aim of the present overview is to recall the basis treated. Finally, stoichiometric SiC fibers(Hi-Nicalon of the layered interphase concept, to discuss its appli type S(HNS)or Tyranno SA (TSA, Ube Industries, cation to SiC/SiC exposed to oxidizing environment Yamaguchi, Japan)), fabricated at higher temperatures and tentatively, to neutron irradiation. are assumed to be dimensionally stable at composite processing temperatures. Further, their surface consists Py C Single-Layer Interphase: The Reference of free carbon(resulting from SiC decomposition) Hence, their bonding with PyC interphase is expected PyC has a structure similar to that of graphite but to be relatively strong Finally, the roughness of fiber, which is low for or less distorted and stacked with rotational disorder. Nicalon and Hi-Nicalon but significant for stoichiomet Our layered interphase concept requires that: (i)the ric fibers, adds a mechanical contribution to the fi bond layers should be oriented parallel to fiber surface and (i ing in a transition zone where the nanometric grapl the bonding between the fber and the PyC interphase lay should be strong enough. Otherwise, debonding/crack deflection would occur at the fiber surface exposing the SiC (Nicalon)/Py C/SiC: A Case History fiber to mechanical damage and to the atmosphere.3.4 Studies on SiC/PyC/SiC (CVI) fabricated with Py c Texture Nicalon or Hi-Nicalon fibers(the as-received or pre- treated), clearly show the positive effect of FM-interfa- PyC displays a variety of microtexture and anisot- cial design on material properties. As shown in Fig. I in polarized light (extinction angle, Ae)or/and trans- PyC interphase(samples posites with a single 500 nm ropy, which can be characterized by optical microscopy tensile curves for the con and D) exhibit extended non- mission electron microscopy(TEM)(L2, N parameters): linear domains related to damaging phenomena, with the larger the Ae, the higher the anisotropy. The pre high failure strains. How vever. co ferred PyC for an interphase is rough laminar (Ae>18) fibers(sample D)is much stronge This RL-PyC has a tendency to grow with graphene The shapes of the curves are different: continuously layers parallel to fiber surface. It is usually deposite convex for the composite with pretreated fibers(sample D) CVI(I-CVI or P-CVI, where I stands for isothe and with a plateau-like feature for that with the as-
used in oxidizing atmospheres (gas turbines). Within the same conceptual framework, two main alternatives have been proposed: boron nitride (BN) and (X–Y)n multilayers (ML). The former displays a structure similar to that of graphite while being more oxidation resistant.5–7 In the latter, with X 5 PyC or BN, Y 5 SiC, and n 5 1–10, part of the oxidation prone X constituent is replaced by a material Y exhibiting a better oxidation resistance, such as SiC itself.1–4,8–12 Furthermore, this concept of material multilayering has been extended to the matrix, yielding ‘‘self-healing’’ composites with outstanding lifetimes in oxidizing atmospheres.10,13,14 More recently, SiC/SiC composites have been envisaged as structural materials in high-temperature (HT) nuclear reactors. The interphase appears again as a possible weak point, PyC being known to undergo anisotropic volume change when exposed to neutrons, whereas BN undergoes nuclear reactions.15–17 Research is presently pursued following two similar routes: (i) use of thin single PyC layers and (ii) use of (PyC–SiC)n MLs, to minimize the effect of neutron irradiation.15–21 The aim of the present overview is to recall the basis of the layered interphase concept, to discuss its application to SiC/SiC exposed to oxidizing environment and tentatively, to neutron irradiation. PyC Single-Layer Interphase: The Reference PyC has a structure similar to that of graphite but the elementary graphene layers are of limited size more or less distorted and stacked with rotational disorder. Our layered interphase concept requires that: (i) the layers should be oriented parallel to fiber surface and (ii) the bonding between the fiber and the PyC interphase should be strong enough. Otherwise, debonding/crack deflection would occur at the fiber surface exposing the fiber to mechanical damage and to the atmosphere.3,4 PyC Texture PyC displays a variety of microtexture and anisotropy, which can be characterized by optical microscopy in polarized light (extinction angle, Ae) or/and transmission electron microscopy (TEM) (L2, N parameters): the larger the Ae, the higher the anisotropy.22 The preferred PyC for an interphase is rough laminar (Ae4181) This RL-PyC has a tendency to grow with graphene layers parallel to fiber surface. It is usually deposited by CVI (I-CVI or P-CVI, where I stands for isothermal/ isobaric and P for pressure pulsed) from propane or propylene.1,5,9,23 Fiber/PyC-Interphase Bonding The second requirement is a strong bonding between the fiber and the PyC interphase.3,4 Achieving such a strong bonding is first a matter of surface chemistry. The surface of desized Si–C–O fibers (Nicalon, Nippon Carbon, Tokyo, Japan) is enriched in oxygen and free carbon. As a result, there is in SiC/SiC (CVI) a thin and irregular dual layer of amorphous silica and carbon which introduces a weak link near the fiber surface.1,3–6,24,25 The fibers should be pretreated to clean their surface and achieve a strong fiber/interphase (FI) bonding. Another example is the SiC1C fibers (HiNicalon or HN, Nippon Carbon, Tokyo, Japan), whose microstructure is not fully stabilized after processing and whose surface may also contain some oxygen. During CVI processing, fibers undergo a postshrinkage weakening the FI interface. Again, the fibers should be pretreated. Finally, stoichiometric SiC fibers (Hi-Nicalon type S (HNS) or Tyranno SA (TSA, Ube Industries, Yamaguchi, Japan)), fabricated at higher temperatures are assumed to be dimensionally stable at composite processing temperatures. Further, their surface consists of free carbon (resulting from SiC decomposition).26 Hence, their bonding with PyC interphase is expected to be relatively strong. Finally, the roughness of fiber, which is low for Nicalon and Hi-Nicalon but significant for stoichiometric fibers, adds a mechanical contribution to the FI bonding in a transition zone where the nanometric graphene layers stacks become progressively parallel to fiber axis.1,7 SiC (Nicalon)/PyC/SiC: A Case History Studies on SiC/PyC/SiC (CVI) fabricated with Nicalon or Hi-Nicalon fibers (the as-received or pretreated), clearly show the positive effect of FM-interfacial design on material properties.1–4 As shown in Fig. 1, tensile curves for the composites with a single 500 nm PyC interphase (samples I and J) exhibit extended nonlinear domains related to damaging phenomena, with high failure strains. However, composite with pretreated fibers (sample J) is much stronger. The shapes of the curves are different: continuously convex for the composite with pretreated fibers (sample J) and with a plateau-like feature for that with the as- 264 International Journal of Applied Ceramic Technology—Naslain, Pailler and Lamon Vol. 7, No. 3, 2010
wwceramics. org/ACT Layered Interphases in SiC/SiC Composites I,J: Pycsoo M, N: PyC reated Nicalon as-received Nicalon J- Stronger F\-bonding I- weak FM-bondiag Fig 1. Tensile curves at ambient of 2D-SiC/SiC (CV) fabricated from Nicalon fibers with different interphases: single P)yC layers or (ByC-SiC) ML interphases(adapted from Droillard'y received fibers (sample I). Matrix microcracks are compressive TRS and lower interfacial shear stress deflected near fiber surface (I-inset, right) for the latter, Experimental studies have shown that the mechanical at the level of the weak carbon/ silica interface and within properties. 5.28 Dupel et through an optimum for e(PyC)=few the PyC interphase(in a diffused manner and over short 100 nm. at distance, J-inset, left) for the former. Here, the PyC inter- tensile properties of ID-SiC(NCG)Py C/SiC (P-CVD phase does play its role of mechanical fuse. These two inicomposites were optima for e(PyC)=220 nm.The microcrack deflection modes correspond to very different calculated radial TRS in the interphase was compressive FM coupling(weak in I and stronger in D. Interestingly, (enhancing thus the FM bonding) at low e(PyC)values toughness of the composites fabricated with pretreated and tensile(favoring FM debonding) for e(PyC)> fibers as well as their fatigue resistance in tensile cyclic 400 nm. loading are higher when it is the Py C interphase, which is In sic/sic fabricated with stoichiometric fibers the active mechanical fuse(strong bonding). 22/Similar the situation is different because the CTEs of the main results have been reported for composites fabricated from constituents are now similar but the fiber surface is Hi-Nicalon fibers and Pyc interphas highly crystalline and rough. The mechanical properties of 2D-SiC/PyC/SiC( CVI)are either constant or Influence of Py C-Interphase Thickness slightly dependent on interphase thickness, when increases from 25 to PyC-layer thickness, e(PyC), has an infuence on be expected from fiber roughness, the interfacial fric- the mechanical properties of SiC/Pyc/SiC ce tional stress tf is high composites with Ty through thermal residual stresses(TRS) and fiber surfac fibers than for those with Hi-Nicalon fibers and decreases as e(PyC Increases In SiC/PyC/SiC(CV) fabricated at≈1000°C using Nicalon( Ceramic Grade, NCG)or Hi-Nicalon Crack Deflection Modeling fibers, the coefficient of thermal expansion(CTE)of the fibers is lower than that of the matrix, which results in Crack deflection at an interface in brittle materials compressive radial TRS at the FM interface and has been modeled. ,3 Recently, Pompidou and reinforces FM bonding. Further, the fiber surface is Lamon 3. 34 have proposed a model, derived from the very smooth. Hence, increasing e(PyC) relaxes radial approach of Cook and Gordon, which is applicable to
received fibers (sample I). Matrix microcracks are deflected near fiber surface (I-inset, right) for the latter, at the level of the weak carbon/silica interface and within the PyC interphase (in a diffused manner and over short distance, J-inset, left) for the former. Here, the PyC interphase does play its role of mechanical fuse. These two microcrack deflection modes correspond to very different FM coupling (weak in I and stronger in J). Interestingly, toughness of the composites fabricated with pretreated fibers as well as their fatigue resistance in tensile cyclic loading are higher when it is the PyC interphase, which is the active mechanical fuse (strong bonding).1,2,27 Similar results have been reported for composites fabricated from Hi-Nicalon fibers and PyC interphase. Influence of PyC-Interphase Thickness PyC-layer thickness, e(PyC), has an influence on the mechanical properties of SiC/PyC/SiC composites, through thermal residual stresses (TRS) and fiber surface roughness. In SiC/PyC/SiC (CVI) fabricated at 10001C using Nicalon (Ceramic Grade, NCG) or Hi-Nicalon fibers, the coefficient of thermal expansion (CTE) of the fibers is lower than that of the matrix, which results in compressive radial TRS at the FM interface and reinforces FM bonding. Further, the fiber surface is very smooth. Hence, increasing e(PyC) relaxes radial compressive TRS and lower interfacial shear stress. Experimental studies have shown that the mechanical properties go through an optimum for e(PyC) 5 few 100 nm.11,15,28 Dupel et al. 29 have reported that the tensile properties of 1D-SiC (NCG)/PyC/SiC (P-CVI) minicomposites were optima for e(PyC) 5 220 nm. The calculated radial TRS in the interphase was compressive (enhancing thus the FM bonding) at low e(PyC) values and tensile (favoring FM debonding) for e(PyC)4 400 nm. In SiC/SiC fabricated with stoichiometric fibers, the situation is different because the CTEs of the main constituents are now similar but the fiber surface is highly crystalline and rough. The mechanical properties of 2D-SiC/PyC/SiC (CVI) are either constant or slightly dependent on interphase thickness, when e(PyC) increases from 25 to 250 nm.17,30 As it could be expected from fiber roughness, the interfacial frictional stress tf is higher for composites with Tyranno fibers than for those with Hi-Nicalon fibers and decreases as e(PyC) increases.12 Crack Deflection Modeling Crack deflection at an interface in brittle materials has been modeled.31,32 Recently, Pompidou and Lamon33,34 have proposed a model, derived from the approach of Cook and Gordon,32 which is applicable to Fig. 1. Tensile curves at ambient of 2D-SiC/SiC (CVI) fabricated from Nicalon fibers with different interphases: single PyC layers or (PyC–SiC)n ML interphases (adapted from Droillard1 ). www.ceramics.org/ACT Layered Interphases in SiC/SiC Composites 265
266 International Journal of Applied Ceramic Technolog-Naslain, Pailler and Lamon Vol. 7, No 3, 2010 The model has been applied to SiC/PyCSiC to examine crack deflection probability at a given interface or within the interphase, to show the influence of fiber pretreatment and to discuss the effect of graphene-layer SIC/Pyc orientation in the interphase Deflection at first interface 0.8 (PyC /SiCm)is very unlikely because for the related E2/EI ratio (N 0.07), the value of oi should be extremely low(Fig. 2) in accordance with experiments By contrast, deflection at second interface( SiCpPyCi)is PyC/Pyc most likely because for the corresponding E2/E, ratio (10 for Hi-Nicalon) the width of the debonding domain (debonding potential) is very large. This is the most frequently observed case(weak FM bonding). If the fiber has been pretreated to strengthen the FM bonding, the representative point may move above the master curve,with debonding no longer occurring at that interface. However, it may take place within the PyC Fig.2. Values of of of ratio provided by the master curve for for a EyE, value of 1, a still significant deflection p us fber/matrix and fberlinterphase/matrix systems(cracked tential (Fin awep s a f now urf e nene layers are lamn al (material I)cited second)(adapted from Pompidou and treated Nicalon fiber deposited perpendicular to fber surface, crack deflection within the PyC interphase becomes no longer possible composites with single- or multilayered interphases. because the ratio opyc//opyci N 2. 17 is well above When a crack of tip radius p is placed in an elastic the master curve for E2/E1=1 medium and subjected to a uniaxial tension Oz(in a direction z perpendicular to crack plane), it generates a Layered Interphases for SiC/SiC Exposed to multiaxial stress field near crack tip whose orr compo- Oxidizing Atmosphere nent(in radial direction)is maximum at a distance on the order of p(omx=o(r=p)). If an interface is PyC is oxidation prone even at temperature as low as placed perpendicular to primary crack extension direc tion near crack tip, a secondary local crack may nucleat 500C, its oxidation resulting in the formation of gas- eous oxides(active oxidation) and degradation of FM at that interface if om.>o:, where o; is the interface coupling 35.36 Two approaches have been selected to solve debonding stress. Deflection results from coalescence of this problem relying on self-healing (or self-sealing)mech- both cracks. 2 When applied to a microcomposite anisms by condensed oxides(passive oxidation). The first loaded in tension along fiber axis, debonding would occur when of/of s omax / oma(with r>D, where of whereas in the second, part of PyC is replaced, in the the failure stress of iber and, l the distance so-called Ml interphases, by Sic or tic to reduce the between crack tip and interface (or ligament), thickness of each elementary PyC sublayer to a few 10 nm om and omax were computed and their ratio plotted and to favor self-healing phenomena? versus Youngs moduli ratio E2/E1, as shown in Fig. 2 The domain under this master curve corresponds to the Boron-Doped PyCInterpbase debonding situation and that above to conditions where debonding cannot occur. The curve exhibits a The addition of boron to Py C increases its graphitic mum corresponding to the highest debonding potential. character at low B concentration and improves its Conversely, when E2/En decreases and tends to zero, oxidation resistance at high B levels by blocking the debonding becomes quite impossible. But, the crack can so-called active sites and forming a Auid oxide(B2O3)in be arrested. Failure of the reinforcing material depends a temperature range(500-900oC), where the growth on its strength versus stress operating: of versus omx. kinetics of silica is still too slow
composites with single- or multilayered interphases. When a crack of tip radius r is placed in an elastic medium and subjected to a uniaxial tension szz (in a direction z perpendicular to crack plane), it generates a multiaxial stress field near crack tip whose srr component (in radial direction) is maximum at a distance on the order of r (smax rr ¼ srrðr ¼ rÞ). If an interface is placed perpendicular to primary crack extension direction near crack tip, a secondary local crack may nucleate at that interface if smax rr > sc i , where si c is the interface debonding stress. Deflection results from coalescence of both cracks.32 When applied to a microcomposite loaded in tension along fiber axis, debonding would occur when sc i =sc f smax rr =smax zz (with r4l), where sf c is the failure stress of the fiber and, l, the distance between crack tip and interface (or ligament),33,34 smax rr and smax zz were computed and their ratio plotted versus Young’s moduli ratio E2/E1, as shown in Fig. 2. The domain under this master curve corresponds to the debonding situation and that above to conditions where debonding cannot occur. The curve exhibits a maximum corresponding to the highest debonding potential. Conversely, when E2/E1 decreases and tends to zero, debonding becomes quite impossible. But, the crack can be arrested. Failure of the reinforcing material depends on its strength versus stress operating: sf c versus smax zz . The model has been applied to SiC/PyC/SiC to examine crack deflection probability at a given interface or within the interphase, to show the influence of fiber pretreatment and to discuss the effect of graphene-layer orientation in the interphase. Deflection at first interface (PyCi/SiCm) is very unlikely because for the related E2/E1 ratio ( 0.07), the value of si c should be extremely low (Fig. 2) in accordance with experiments. By contrast, deflection at second interface (SiCf/PyCi) is most likely because for the corresponding E2/E1 ratio (10 for Hi-Nicalon) the width of the debonding domain (debonding potential) is very large. This is the most frequently observed case (weak FM bonding). If the fiber has been pretreated to strengthen the FM bonding, the representative point may move above the master curve, with debonding no longer occurring at that interface. However, it may take place within the PyC interphase, that is, at a PyCi/PyCi interface that shows for a E2/E1 value of 1, a still significant deflection potential (Fig. 2). This is the situation observed, for pretreated Nicalon fiber.1–4 If now the graphene layers are deposited perpendicular to fiber surface, crack deflection within the PyC interphase becomes no longer possible because the ratio sc PyC===sc PyC? 2:17 is well above the master curve for E2/E1 5 1. Layered Interphases for SiC/SiC Exposed to Oxidizing Atmosphere PyC is oxidation prone even at temperature as low as 5001C, its oxidation resulting in the formation of gaseous oxides (active oxidation) and degradation of FM coupling.35,36 Two approaches have been selected to solve this problem relying on self-healing (or self-sealing) mechanisms by condensed oxides (passive oxidation). The first one is based on single-layer interphases containing boron whereas in the second, part of PyC is replaced, in the so-called ML interphases, by SiC or TiC to reduce the thickness of each elementary PyC sublayer to a few 10 nm and to favor self-healing phenomena.9,10 Boron-Doped PyC Interphase The addition of boron to PyC increases its graphitic character at low B concentration and improves its oxidation resistance at high B levels by blocking the so-called active sites and forming a fluid oxide (B2O3) in a temperature range (500–9001C), where the growth kinetics of silica is still too slow.37 Fig. 2. Values of sc i =sc 2 ratio provided by the master curve for various fiber/matrix and fiber/interphase/matrix systems (cracked material (material 1) cited second) (adapted from Pompidou and Lamon34). 266 International Journal of Applied Ceramic Technology—Naslain, Pailler and Lamon Vol. 7, No. 3, 2010
Layered Interphases in SiC/SiC Composites Jacques et al. have studied the influence of B-doped X> PyC interphase on the oxidation resistance of lD-SiC/SiC 必 matrix s fiber (CVD)microcomposites with pretreated Nicalon fibers Their interphases contain up to 30 at. %B. They showed as expected, that the microtexture of the PyC interphase was significantly improved at low B addition, 8 at. %B, amorphous). More importantly, lifetime in tensile static fatigue(beyond proportional limit PL) in air at 700C was dramatically improved as the B content was raised, the best results being observed for graded composition terphase. Crack deflection and failure occur with che interphase at a location where the interphase micro- texture and graphene-layer orientation were optimal (at 112 ≈8a%B) BN-inte BN Interphases Hi-Nicalon fiber tow moving in a temperature grader The use of BN interphase in SiC/SiC raises sev Ts1100C: T2=1150C 13=1250C) at medium residence time(v=2.5 mb)(adapted from Jacques et al. 12) problems, which still remain imperfectly solved. They include the occurrence of corrosion by precursor and the chemical reactivity of BN with oxygen and moisture hen prepared at low temperature. SiCm/BN interface, these two scenarios corresponding to the"inside"and"outside"debonding reported by Mo F scher et al, in related experiments. In the case of NH3: BF3-NH3 precursor has the advantage of outside" debonding, both the interfacial shear stress yielding crystallized Bn deposits at relatively low tem- and tensile failure stress were lower but the lifetime in rature.' Unfortunately, it involves gaseous species tensile static fatigue at 700C in dry or wet air was dra- (BFs and HF), which are corrosive for Sic-based fibers matically improved(crack deflection occurring far from and alter their strength (as received Nicalon and fiber surface) Nicalon). Conversely, this precursor is compatible with carbon substrates and it could be used to deposit BN Interphases as Deposited by CVI from BClg-NHs- Bn on fibers with a carbon layer surface (pretreated or H2: BCl3-NH3-H2 precursor is usually preferred be- stoichiometric fibers). However, an extra carbon layer cause it is much less corrosive. -/ In principle,BN often remains between the SiC fiber and the BN coating, could be deposited at temperature as low as 700%C. which could be the weakest link in the interfacial zone. However, under such mild conditions, it is nanoporous, ne way to solve the corrosion problem and to play poorly organized and highl ly reactive. Hence, the pro- with the mechanical fuse location could be to deposit cessing temperature should be increased. In the BN in a temperature gradient (TG-CVI). Jacques case of complex fber architectures(nD-preforms), BN et al. have fabricated ID-SiC/BN/SiC minicomposites can be deposited at the highest temperature compatible with a radial crystallinity gradient by simply passing a with the ICVI process(N 1100%C)and further an Hi-Nicalon tow through a three-temperature zones fur- nealed at a temperature corresponding to the up nace Under optimized conditions, in terms of fiber pro- limit of the thermal stability dor main of the fibers.a gression speed, the FI bonding was strong (crack alternative is to deposit Bn on fber tows, which can be deflection occurring within BN interphase(Fig. 3) done at higher temperature(1400-1600%),partic and both interfacial shear stress and tensile stress were larly for stoichiometric SiC fibers. As an example, BN high. At lower fiber speed, crack deflection occurred at deposited on a TSA tow at 1580%C was reported to be fiber surface (as a result of some surface crystallization) nearly stoichiometric, with an impurity content hereas for higher fiber speed it was observed at th <5 at. %, highly crystallized and textured. 49
Jacques et al. 38 have studied the influence of B-doped PyC interphase on the oxidation resistance of 1D-SiC/SiC (CVI) microcomposites with pretreated Nicalon fibers. Their interphases contain up to 30 at.% B. They showed, as expected, that the microtexture of the PyC interphase was significantly improved at low B addition, 8 at.% B, and degraded beyond this value (the interphase becoming amorphous). More importantly, lifetime in tensile static fatigue (beyond proportional limit PL) in air at 7001C was dramatically improved as the B content was raised, the best results being observed for graded composition interphase. Crack deflection and failure occur within the interphase at a location where the interphase microtexture and graphene-layer orientation were optimal (at 8 at.% B). BN Interphases The use of BN interphase in SiC/SiC raises several problems, which still remain imperfectly solved. They include the occurrence of corrosion by precursor and the chemical reactivity of BN with oxygen and moisture when prepared at low temperature. BN Interphases as Deposited by CVI from BF3– NH3: BF3–NH3 precursor has the advantage of yielding crystallized BN deposits at relatively low temperature.6,39 Unfortunately, it involves gaseous species (BF3 and HF), which are corrosive for SiC-based fibers and alter their strength (as received Nicalon and HiNicalon).40 Conversely, this precursor is compatible with carbon substrates and it could be used to deposit BN on fibers with a carbon layer surface (pretreated or stoichiometric fibers).41 However, an extra carbon layer often remains between the SiC fiber and the BN coating, which could be the weakest link in the interfacial zone. One way to solve the corrosion problem and to play with the mechanical fuse location could be to deposit BN in a temperature gradient (TG-CVI). Jacques et al. 42 have fabricated 1D-SiC/BN/SiC minicomposites with a radial crystallinity gradient by simply passing a Hi-Nicalon tow through a three-temperature zones furnace. Under optimized conditions, in terms of fiber progression speed, the FI bonding was strong (crack deflection occurring within BN interphase (Fig. 3)) and both interfacial shear stress and tensile stress were high. At lower fiber speed, crack deflection occurred at fiber surface (as a result of some surface crystallization) whereas for higher fiber speed it was observed at the SiCm/BN interface, these two scenarios corresponding to the ‘‘inside’’ and ‘‘outside’’ debonding reported by Morscher et al.,43 in related experiments. In the case of ‘‘outside’’ debonding, both the interfacial shear stress and tensile failure stress were lower but the lifetime in tensile static fatigue at 7001C in dry or wet air was dramatically improved (crack deflection occurring far from fiber surface).42 BN Interphases as Deposited by CVI from BCl3–NH3– H2: BCl3–NH3–H2 precursor is usually preferred because it is much less corrosive.44–47 In principle, BN could be deposited at temperature as low as 7001C. However, under such mild conditions, it is nanoporous, poorly organized and highly reactive. Hence, the processing temperature should be increased.5,44–47 In the case of complex fiber architectures (nD-preforms), BN can be deposited at the highest temperature compatible with the ICVI process ( 11001C) and further annealed at a temperature corresponding to the upper limit of the thermal stability domain of the fibers. An alternative is to deposit BN on fiber tows, which can be done at higher temperature (1400–16001C), particularly for stoichiometric SiC fibers.48 As an example, BN deposited on a TSA tow at 15801C was reported to be nearly stoichiometric, with an impurity content o5 at.%, highly crystallized and textured.49 Fig. 3. BN interphase deposited from BCl3–NH3–H2 on Hi-Nicalon fiber tow moving in a temperature gradient (T1r11001C; T2 5 11501C; T3 5 12501C) at medium residence time (v 5 2.5 m/h) (adapted from Jacques et al.42). www.ceramics.org/ACT Layered Interphases in SiC/SiC Composites 267
International Journal of Applied Ceramic Technolog-Naslain, Pailler and Lamon Vol. 7, No 3, 2010 interphase matrIx matrix deflected crack fiber 300nm Fig 4. ID-SiC(HN)BN/SiC (CVI) minicomposite with BN interphase deposited from tris(dimethylamino)borane: matrix crack deflected within the BN interphase, as seen by transmission electron microscopy(BFmode) at low(a)and higb()magnifications(adapted from acqu Another efficient way to improve the oxidation resis- reported that such a treatment increased the interfacial of BN is to dope the bn precursor with a silane shear stress, without achieving the high values required (such as The resulting posits for crack deflection within the interphase. Finally, with a Si content ranging from 15 to 40 wt%, display crack deflection according to these different scenarios oxidation rates(at 1200C in oxygen or N 1500.C in can be modeled, as already discussed for the composites air)two to three orders of magnitude lower than those for with PyC interphases(see Fig. 2)33.34 undoped BN. Further, their use in SiC/SiC signifi cantly improves the resistance of the composite to oxy- BN Interphases as Deposited by CvI from On lics: Another way to reduce corrosion during bn dep temperatures.48.5ob However, such Si-doped BN is amor- sition is through the use of halogen-free organometallic precursors. S. Jacques and colleagues have deposited phous and hence at variance with the requirements of our bN interphases from tris(dimethylamino) borane, layered interphase concept. Interestingly, it has been s gested to use dual BN-based interphases containing one BIN(CH3)2]3 in H2-NH3 Aow, on Hi-Nicalon fibers layer of textured pure BN (acting as mechanical fuse)and (NH, being used to avoid the occurrence of free carbon in phase for SiC/Sic in gas turbine application. ar intra one layer of BN(Si) for improving oxidation the coating). The tensile curves of their minicomposite characteristic of SiC/SiC with relatively strong FM bond tance ing, high matrix crack density at failure and high inter- weak, owing to the occurrence of addi g is relatively acial shear stress(here, t=230 MPa). The latter is,to our knowledge, the highest value reported for SiC/BN/ of silica orland carbon at the fber bn coating orland at Sic composites. Finally, crack deflection did occur within the SiC matrix/bn coating interfaces. #-47.51.52 Matrix crack deflection often occurs at these weak interfaces quirement of our layered interphase concept. (mostly at the former)and not within the bn inter- ed on Sic fibers with In situ phase. Carbon is also present as single layer in compos Generated BN Surface: The preceeding sections suggest tes fabricated with pretreated fibers(Nicalon and that SiC fibers with a carbon surface may not be the Nicalon) or stoichiometric fibers. It can be chemi- nost appropriate for BN deposition. An altern cally removed before BN deposition. LeGallet et al have would be to use SiC fibers with a BN surface 54-56
Another efficient way to improve the oxidation resistance of BN is to dope the BN precursor with a silane (such as HSiCl3).48,50a–c The resulting BN(Si) deposits with a Si content ranging from 15 to 40 wt%, display oxidation rates (at 12001C in oxygen or 15001C in air) two to three orders of magnitude lower than those for undoped BN.50a Further, their use in SiC/SiC signifi- cantly improves the resistance of the composite to oxygen- and water-containing atmospheres at intermediate temperatures.48,50b However, such Si-doped BN is amorphous and hence at variance with the requirements of our layered interphase concept. Interestingly, it has been suggested to use dual BN-based interphases containing one layer of textured pure BN (acting as mechanical fuse) and one layer of BN(Si) for improving oxidation resistance.50c Finally, BN(Si) has been envisaged as interphase for SiC/SiC in gas turbine application.50d Generally speaking, the FM bonding is relatively weak, owing to the occurrence of additional thin layers of silica or/and carbon at the fiber/BN coating or/and at the SiC matrix/BN coating interfaces.44–47,51,52 Matrix crack deflection often occurs at these weak interfaces (mostly at the former) and not within the BN interphase. Carbon is also present as single layer in composites fabricated with pretreated fibers (Nicalon and Hi-Nicalon) or stoichiometric fibers. It can be chemically removed before BN deposition. LeGallet et al. have reported that such a treatment increased the interfacial shear stress, without achieving the high values required for crack deflection within the interphase.52 Finally, crack deflection according to these different scenarios can be modeled, as already discussed for the composites with PyC interphases (see Fig. 2).33,34 BN Interphases as Deposited by CVI from Organometallics: Another way to reduce corrosion during BN deposition is through the use of halogen-free organometallic precursors.53 S. Jacques and colleagues have deposited BN interphases from tris(dimethylamino) borane, B[N(CH3)2]3 in H2–NH3 flow, on Hi-Nicalon fibers (NH3 being used to avoid the occurrence of free carbon in the coating). The tensile curves of their minicomposites is characteristic of SiC/SiC with relatively strong FM bonding, high matrix crack density at failure and high interfacial shear stress (here, t 5 230 MPa). The latter is, to our knowledge, the highest value reported for SiC/BN/ SiC composites. Finally, crack deflection did occur within the BN interphase (Fig. 4), in accordance with the requirement of our layered interphase concept. BN Interphases Deposited on SiC Fibers with In Situ Generated BN Surface: The preceeding sections suggest that SiC fibers with a carbon surface may not be the most appropriate for BN deposition. An alternative would be to use SiC fibers with a BN surface.54–56 Fig. 4. 1D-SiC (HN)/BN/SiC (CVI) minicomposite with BN interphase deposited from tris(dimethylamino)borane: matrix crack deflected within the BN interphase, as seen by transmission electron microscopy (BF mode) at low (a) and high (b) magnifications (adapted from Jacques et al.53). 268 International Journal of Applied Ceramic Technology—Naslain, Pailler and Lamon Vol. 7, No. 3, 2010
wwceramics. org/ACT Layered Interphases in SiC/SiC Composites A straightforward approach is to start with a stoic- st Pyc-1 layer biometric SiC fiber containing some boron(acting narily as sintering aid), as shown by Sacks and Brennan. When such a fiber is treated at high tem perature in an N-containing atmosphere, B atoms diffuse radially from fiber core to react with nitrogen yielding a strongly adherent BN layer at fiber surface (typically, 100-200 nm thick). Further, diffusion in BN being anisotropic, the bn coating tends to grow with Bn layers perpendicular to fiber surface and hence strongly bonded to the fiber. If now a BN interphase is deposited on such a substrate by CVI, the BN layers would have a tendency to be oriented, after some gime, parallel to the fiber o-called Sylramic-iBN and Super Sylramic-iBN fibers may have been developed on the basis of some related In the composites, crack deflection Fig. 5. ID-SiC (HN/SiC (CVI) with(lyC-SiCIo ML would occur either within the BN interphase or at the interphase: transmission electron microscopic image of a matrix BN-SiCm interface(outside debonding). These features microcrack deflected within the ML interphase(adapted from these composites at high temperatures dizing ranging from 3 to 100 nm for PyC and 10 to 500 nm for environmen Sic while the number of PyC-SiC sequences is in the ML C-D Interp bases funge of 3-10. The first material deposited on fiber sur- face is usually pyc but it could also be SiC in an attempt to strengthen the FI bonding. 1> Both MLs with con- ML interphases, (X-Y)m extend the concept of stant sublayer thickness or graded sublayer thickness(on layered interphase from the atomic to the nanometer PyC or/and SiC) have been used. scales, the interphase being now a stack of films of Replacing a Pyc single interphase by a(Pyc-sic)m different materials X and Y, and the X-Y elementary ML interphase does not change markedly tensile properties, sequence repeated n times. Their main advantage is that as shown for composites fabricated with Nicalon fibers they can be highly tailored. 42. 78 As an example, the (Fig. 1). Tensile curves fall into two groups oxidation resistance of Sic/SiC could be improved by depending on whether the fibers are pretreated(strong replacing PyC or BN single-layer interphases(100- FM bonding) or not(weak FM bonding). Similar con- 200 nm thick) by(PyC-SiC), or( BN-SiC), ML inter- clusion can be drawn for SiC/SiC with Hi-Nicalon phases in which the thickness of the oxidation-prone fibers" or TSA stoichiometric fibers. Crack deflect- PyC or BN mechanical fuse is reduced to a few nan tion occurs at the FI (or/and interphase/matrix) interfa meters. This design criterion is based on oxygen gas when the FM bonding is weak (e-g, for the phase diffusion consideration"> and formation of heal- as-received Nicalon or Hi-Nicalon fibers) and withi ng condensed oxides(silica or/and boria). s ML inter- the ML interphase when it is strong enough(treated are deposited by CVI (switching from X to Y fibers). In this latter case, a matrix microcrack exhibits rs). A key requirement a dual propagation mode across the ml interphase (Fig. 5)with an overall propagation path significantly (PyC-SiC)n ML Interphases: Since the pioneering work The lifetime of SiC/SiC with ML interphase under of Droillard and colleagues, (PyC-SiC), ML interphas- load, at high temperature in air is improved with respect es have been used in a variety of SiC/SiC. -6>Depending to their counterparts with single PyC interphas on CVI conditions, SiC sublayers are either microcrystal lized(with rough SiC/PyC interfaces)or nanocrystallized Other(X-Yn ML Interphases: At least two other M h smooth SiC/PyC interfaces. Sublayer thickness interphases,(BN-SiC)n and(PyC-TiC)m have be
A straightforward approach is to start with a stoichiometric SiC fiber containing some boron (acting primarily as sintering aid),26 as shown by Sacks and Brennan.54 When such a fiber is treated at high temperature in an N-containing atmosphere, B atoms diffuse radially from fiber core to react with nitrogen, yielding a strongly adherent BN layer at fiber surface (typically, 100–200 nm thick). Further, diffusion in BN being anisotropic, the BN coating tends to grow with BN layers perpendicular to fiber surface and hence strongly bonded to the fiber. If now a BN interphase is deposited on such a substrate by CVI, the BN layers would have a tendency to be oriented, after some transition regime, parallel to the fiber surface.25,27 The so-called Sylramic-iBN and Super Sylramic-iBN fibers may have been developed on the basis of some related mechanism.55 In the composites, crack deflection would occur either within the BN interphase or at the BN–SiCm interface (outside debonding). These features could explain the good mechanical properties of these composites at high temperatures in oxidizing environment.43,55,57 ML (X–Y)n Interphases ML interphases, (X–Y)n, extend the concept of layered interphase from the atomic to the nanometer scales, the interphase being now a stack of films of different materials X and Y, and the X–Y elementary sequence repeated n times. Their main advantage is that they can be highly tailored.1,4,5,58 As an example, the oxidation resistance of SiC/SiC could be improved by replacing PyC or BN single-layer interphases (100– 200 nm thick) by (PyC–SiC)n or (BN–SiC)n ML interphases in which the thickness of the oxidation-prone PyC or BN mechanical fuse is reduced to a few nanometers. This design criterion is based on oxygen gas phase diffusion consideration35 and formation of healing condensed oxides (silica or/and boria).13 ML interphases are deposited by CVI (switching from X to Y gaseous precursors).58 A key requirement is again a strong bonding between fiber surface and interphase. (PyC–SiC)n ML Interphases: Since the pioneering work of Droillard and colleagues,1,2 (PyC–SiC)nML interphases have been used in a variety of SiC/SiC.59–65 Depending on CVI conditions, SiC sublayers are either microcrystallized (with rough SiC/PyC interfaces) or nanocrystallized with smooth SiC/PyC interfaces. Sublayer thickness is ranging from 3 to 100 nm for PyC and 10 to 500 nm for SiC while the number of PyC–SiC sequences is in the range of 3–10. The first material deposited on fiber surface is usually PyC but it could also be SiC in an attempt to strengthen the FI bonding.64,65 Both MLs with constant sublayer thickness or graded sublayer thickness (on PyC or/and SiC) have been used.1,65 Replacing a PyC single interphase by a (PyC–SiC)n ML interphase does not change markedly tensile properties, as shown for composites fabricated with Nicalon fibers (Fig. 1).1 Tensile curves fall into two groups depending on whether the fibers are pretreated (strong FM bonding) or not (weak FM bonding). Similar conclusion can be drawn for SiC/SiC with Hi-Nicalon fibers59,60 or TSA stoichiometric fibers.65 Crack deflection occurs at the FI (or/and interphase/matrix) interface when the FM bonding is weak (e.g., for the as-received Nicalon or Hi-Nicalon fibers) and within the ML interphase when it is strong enough (treated fibers). In this latter case, a matrix microcrack exhibits a dual propagation mode across the ML interphase (Fig. 5) with an overall propagation path significantly lengthened.59,60 The lifetime of SiC/SiC with ML interphase under load, at high temperature in air is improved with respect to their counterparts with single PyC interphase.60,66 Other (X–Y)n ML Interphases: At least two other ML interphases, (BN–SiC)n and (PyC–TiC)n, have been Fig. 5. 1D-SiC (HN)/SiC (CVI) with (PyC–SiC)10 ML interphase: transmission electron microscopic image of a matrix microcrack deflected within the ML interphase (adapted from Bertrand et al.60). www.ceramics.org/ACT Layered Interphases in SiC/SiC Composites 269
270 International Journal of Applied Ceramic Technolog-Naslain, Pailler and Lamon Vol. 7, No 3, 2010 studied also with a view to improve the oxidation resis- ible at high temperatures with oxide fibers and matrices tance. The potential advantage of the former lies in the in alumina-based oxide-oxide CMCs. They have been fact that B atoms are now present in the interphase envisaged recently as potential interphase in SiC/SiC which favors crack healing at intermediate temperature mainly on the basis of their oxidation resistance. 0b-d Unfortunately, experiments with pretreated Hi-Nicalon Firstly, the interlayer bonding in such oxides is not as fibers have shown that the fber/BNi bond is relatively weak as it is in PyC or BN. Secondly, the deposition weak, crack deflection occurring at that interface and of monazite by sol-gel is not as straightforward as the not within the ML interphase. Nevertheless, the lifetime CVi deposition of PyC or Bn (particularly in nD-fiber in air at 700C under load of minicomposites fabricated preforms commonly used in real application at plant with Hi-Nicalon and a(Bn4o-SiC25)10 ML interphase level). Further, monazite has been reported to be stable, was significantly improved. with respect to carbon and SiC, only in relatively narrow (PyC-TiC), ML interphases can also improve the temperature/oxygen partial pressure conditions and oxidation resistance of SiC/SiC, although titanium ox- could be reduced during SiC/SiC Processing ides are not commonly regarded as healing oxides. Such performed at low Po, and long duration (CVD) interphases have been deposited on the as-received Monazite may inhibit to some extent the oxidation of sic Hi-Nicalon, according to a combination of conven- fibers but the inhibition process seems to depend on the tional P-CVI(for PyC) and reactive P-CVI(for occurrence of dopants(e.g, Al or Zr). Finally, other TiC). When the amount of TiC is low, each sublay refractory oxides, such as zirconia, have also been consid er consists of a PyC film reinforced with nanometric ered but they do not have layered crystal structures and TiC particles, which results in a strong FM bondi hence do not fall into our layered interphase concept. The lifetime of such minicomposites, under load in air Finally, porous interphases (also referred to as at 700%C, is much higher(> 300 h)than that(20 h)for " pseudo-porous")consist of a mixture, at the nanometer their counterparts with PyC single-layer interphase. One scale, of a refractory material, such as SiC, with(fugitive) of the reasons, which could explain such unexpected carbon. However, such interphases do not display a behavior might be a strong interfacial bonding marked anisotropic texture nor protect the fibers in an environment and may undergo some sintering Miscellaneous Interp base Materials when exposed for a long time at high temperature A few additional materials have been identified as Interphases in SiC/SiC for HT Nuclear Reactors rential interphase materials for SiC/SiC but with lim- ited success up to now. This is the case for the ternary SiC/SiC are potential structural materials for both carbides(MAX phases), such as Ti3 SiC2 or Ti3AlC2 fission and fusion hT nuclear reactors.5-17.73 This new which display layered crystal structures. However, the and extremely demanding application raises specific deposition by CVD/CVI is difficult. Further, they constraints on the fibers, the matrix and the interphases tend to grow with the layers perpendicular to the sub strate surface and their ability to arrest/deflect a matrix Sic/sic environment in ht Nuclear reactors crack in a CMC has not been formally established. Oxides with layered crystal structures have been The environment that would see SiC/SiC considered as potential interphases in CMCs. 0ad This cooled fast reactors is not so different from that they is the case for the easily cleavable phyllosilicates(and presently experience in advanced gas turbines, in terms related phyllosiloxides), such as mica fluorphlogopite of temperature, gas pressure, and lifetime. However KMg3(AISi3)O1oF2 (and related K(Mg2AlSigO12), they are not expected to see permanently oxidizing whose weak interlayer bonding is somewhat similar to atmosphere importantly, they would be co that in PyC or BN. a However, their deposition uously exposed to intense irradiation by fast neutrons, (according to a multistep I process) is difficult, a-particles, and electromagnetic radiations. In HT their thermal stability limited and their compatibility advanced fission reactors, SiC/SiC will be exposed to with SiC and Sic-CVi can be questioned. Other layered moderately energetic neutrons(N 2 Mev) but at tem oxides such as rare-earth ortho-phosphates(e. g, mona- peratures that could be higher than about 1200%C, te LapO4)are much more refractory and are compat- hereas in Tokamak fusion reactor blankets, they will
studied also with a view to improve the oxidation resistance. The potential advantage of the former lies in the fact that B atoms are now present in the interphase, which favors crack healing at intermediate temperatures. Unfortunately, experiments with pretreated Hi-Nicalon fibers have shown that the fiber/BN1 bond is relatively weak, crack deflection occurring at that interface and not within the ML interphase. Nevertheless, the lifetime in air at 7001C under load of minicomposites fabricated with Hi-Nicalon and a (BN40–SiC25)10 ML interphase was significantly improved.9 (PyC–TiC)n ML interphases can also improve the oxidation resistance of SiC/SiC, although titanium oxides are not commonly regarded as healing oxides. Such interphases have been deposited on the as-received Hi-Nicalon, according to a combination of conventional P-CVI (for PyC) and reactive P-CVI (for TiC).67 When the amount of TiC is low, each sublayer consists of a PyC film reinforced with nanometric TiC particles, which results in a strong FM bonding. The lifetime of such minicomposites, under load in air at 7001C, is much higher (4300 h) than that (20 h) for their counterparts with PyC single-layer interphase. One of the reasons, which could explain such unexpected behavior might be a strong interfacial bonding. Miscellaneous Interphase Materials A few additional materials have been identified as potential interphase materials for SiC/SiC but with limited success up to now. This is the case for the ternary carbides (MAX phases), such as Ti3SiC2 or Ti3AlC2, which display layered crystal structures. However, their deposition by CVD/CVI is difficult.68 Further, they tend to grow with the layers perpendicular to the substrate surface and their ability to arrest/deflect a matrix crack in a CMC has not been formally established.69 Oxides with layered crystal structures have been considered as potential interphases in CMCs.70a–d This is the case for the easily cleavable phyllosilicates (and related phyllosiloxides), such as mica fluorphlogopite KMg3(AlSi3)O10F2 (and related K(Mg2Al)Si4O12), whose weak interlayer bonding is somewhat similar to that in PyC or BN.70a However, their deposition (according to a multistep sol–gel process) is difficult, their thermal stability limited and their compatibility with SiC and SiC-CVI can be questioned. Other layered oxides such as rare-earth ortho-phosphates (e.g., monazite LaPO4) are much more refractory and are compatible at high temperatures with oxide fibers and matrices in alumina-based oxide–oxide CMCs. They have been envisaged recently as potential interphase in SiC/SiC mainly on the basis of their oxidation resistance.70b–d Firstly, the interlayer bonding in such oxides is not as weak as it is in PyC or BN.70b Secondly, the deposition of monazite by sol–gel is not as straightforward as the CVI deposition of PyC or BN (particularly in nD-fiber preforms commonly used in real application at plant level). Further, monazite has been reported to be stable, with respect to carbon and SiC, only in relatively narrow temperature/oxygen partial pressure conditions and could be reduced during SiC/SiC processing (usually performed at low PO2 and long duration (CVI)).70c Monazite may inhibit to some extent the oxidation of SiC fibers but the inhibition process seems to depend on the occurrence of dopants (e.g., Al or Zr).70d Finally, other refractory oxides, such as zirconia, have also been considered but they do not have layered crystal structures and hence do not fall into our layered interphase concept.71 Finally, porous interphases (also referred to as ‘‘pseudo-porous’’) consist of a mixture, at the nanometer scale, of a refractory material, such as SiC, with (fugitive) carbon.72 However, such interphases do not display a marked anisotropic texture nor protect the fibers in an oxidizing environment and may undergo some sintering when exposed for a long time at high temperature. Interphases in SiC/SiC for HT Nuclear Reactors SiC/SiC are potential structural materials for both fission and fusion HT nuclear reactors.15–17,73 This new and extremely demanding application raises specific constraints on the fibers, the matrix and the interphases. SiC/SiC Environment in HT Nuclear Reactors The environment that would see SiC/SiC in e.g. gas cooled fast reactors is not so different from that they presently experience in advanced gas turbines, in terms of temperature, gas pressure, and lifetime. However, they are not expected to see permanently oxidizing atmospheres. More importantly, they would be continuously exposed to intense irradiation by fast neutrons, a-particles, and electromagnetic radiations. In HT advanced fission reactors, SiC/SiC will be exposed to moderately energetic neutrons ( 2 MeV) but at temperatures that could be higher than about 12001C, whereas in Tokamak fusion reactor blankets, they will 270 International Journal of Applied Ceramic Technology—Naslain, Pailler and Lamon Vol. 7, No. 3, 2010
wwceramics. org/ACT Layered Interphases in SiC/SiC Composites be irradiated by much more energetic neutrons Pyc-based interphases have been up to now the in- (14.1 Mev) formed during the deuterium/titrium fu terphases of choice. However, they raise a problem of sion reaction but likely at somewhat lower temperatures. anisotropic dimensional change under neutron irradia- Such severe irradiation conditions are known to tion strongly depending on their degree of crystalliza- their constituents, on the basis of literature da 2a change(in a more or less dramatic manner)the structure tion. Graphite and highly oriented pyrolytic graphite of materials and hence their properties, as briefly undergo a moderate shrinkage along the a-axis, that is discussed in the following sections for SiC/Sic arallel to grap layer, and a significant swelling along the perpendicular c-axis. This anisotropy is strong at low temperature/high irradiation dose but it decreases Irradiation of Silicon Carbide ature is raised, The behavior of turbostratic PyC is more complex. Although similar to Monolithic SiC undergoes a moderate swelling that of graphite parallel to graphene layers, it first when irradiated by neutrons, as the result of am- shrinks in a perpendicular direction at low irradiation orphization or point defect formation at low temper dose and then swells. . Because in a PyC-interphase a minimum(0.2-0.4 vol% for a dose of modify residual thermal stresses(particularly in radial 1-8 dpa)at 1100-1200C, then increases to reach direction)and alter the FM bonding. Hence, the inter 1. vol% at 1600 C 7.73a, b SiC matrix when deposited phase may be again the vulnerable constituent of SiC/ by CVI is pure, well crystallized and assumed to behave Sic when exposed in a prolonged manner to neutron as monolithic SiC irradiation Potential solutions to overcome this diffi- ect neutro fibers strongly depends on their composition and struc- for SiC/SiC Exposed to Oxidizing Atmosphere. 16 17.80 ture. On the one hand, stoichiometric fibers, which are Finally, pseudo-porous SiC interphases might also be well crystallized and with a small impurity content, also an alternative as previously mentioned. o/o However, behave like monolithic SiC. On the other hand, Si-C- their dimensional stability under neutron irradiation is O(Nicalon type)an not well known prepared at lower temperature and poorly crystallized undergo a permanent shrinkage. 74 As a result of this volume change mismatch upon Irradiated sicisic neutron irradiation between SiC(CVD) matrix and 1st/ 2nd SiC fiber generations, debonding at FM interface A compilation of strength data(mostly from Alex- usually occurs with mechanical properties degradation. ural tests)reported by different authors, for a variety of This key feature explains why stoichiometric SiC fibers are SiC/PyC/SiC composites that have been neutron irra- ic to be used in nudlear reactors diated(in a broad temperature range: 200-1000"C) suggests that: (i)there is no loss in strength up to an Q Neutron Irradiation of interphase Materials: Boron irradiation dose of 10 dpa for materials fabricated with itride is poorly compatible with nuclear reactor envi- stoichiometric SiC fibers, but conversely, (ii)the ronment.///Firstly, 5 B isotope(present at a level of strength drops by a 60% almost linearly when irradi- a 20 at.% in natural boron) has an extremely high ation dose increases up to 10 dpa for those with Si-C-O neutron capture cross section. Hence, the use of Bn (Nicalon)or SiC+C(Hi-Nicalon)fibers. ( 6.73. This interphase would suppose that it is deposited from 5B- result is consistent with the dimensional change (per riched gaseous precursor. Secondly, BN when neu- manent shrinkage)reported for lese two latter fibers tron irradiated, undergoes nuclear reactions producing which lowers the FM bonding and load transfer. Hence, gaseous species(helium)and radioactive long-life the analysis of the effects of neutron irradiation and in cies(such as 6C). Further, because in a nuclear reactor terphase design on the mechanical properties of Sic/SiC the atmosphere is, in principle, not oxidizing bn should be pursued for composites with stoichiometric interphases are, for all these reasons, not presently fibers, using exclusively tensile tests and analysis of mi- d in this field of application crostructure-strength relations with appropriate models
be irradiated by much more energetic neutrons (14.1 MeV) formed during the deuterium/titrium fusion reaction but likely at somewhat lower temperatures. Such severe irradiation conditions are known to change (in a more or less dramatic manner) the structure of materials and hence their properties, as briefly discussed in the following sections for SiC/SiC and their constituents, on the basis of literature data. Irradiation of Silicon Carbide Monolithic SiC undergoes a moderate swelling when irradiated by neutrons, as the result of amorphization or point defect formation at low temperature and cavity and dislocation loops formation at high temperature. It first decreases as temperature is raised, passes through a minimum (0.2–0.4 vol% for a dose of 1–8 dpa) at 1100–12001C, then increases to reach 1.5 vol% at 16001C.17,73a,b SiC matrix when deposited by CVI is pure, well crystallized and assumed to behave as monolithic SiC. The effect of neutron irradiation on SiC-based fibers strongly depends on their composition and structure. On the one hand, stoichiometric fibers, which are well crystallized and with a small impurity content, also behave like monolithic SiC. On the other hand, Si–C– O (Nicalon type) and SiC1C (Hi-Nicalon) fibers, prepared at lower temperature and poorly crystallized, undergo a permanent shrinkage.74 As a result of this volume change mismatch upon neutron irradiation between SiC (CVI) matrix and 1st/ 2nd SiC fiber generations, debonding at FM interface usually occurs with mechanical properties degradation.73,75 This key feature explains why stoichiometric SiC fibers are preferred for SiC/SiC to be used in nuclear reactors. Neutron Irradiation of Interphase Materials: Boron nitride is poorly compatible with nuclear reactor environment.76,77 Firstly, 10 5 B isotope (present at a level of 20 at.% in natural boron) has an extremely high neutron capture cross section. Hence, the use of BN interphase would suppose that it is deposited from 11 5 Benriched gaseous precursor. Secondly, BN when neutron irradiated, undergoes nuclear reactions producing gaseous species (helium) and radioactive long-life species (such as 14 6 C). Further, because in a nuclear reactor the atmosphere is, in principle, not oxidizing, BN interphases are, for all these reasons, not presently used in this field of application. PyC-based interphases have been up to now the interphases of choice. However, they raise a problem of anisotropic dimensional change under neutron irradiation strongly depending on their degree of crystallization. Graphite and highly oriented pyrolytic graphite undergo a moderate shrinkage along the a-axis, that is, parallel to graphene layer, and a significant swelling along the perpendicular c-axis. This anisotropy is strong at low temperature/high irradiation dose but it decreases as irradiation temperature is raised.78 The behavior of turbostratic PyC is more complex. Although similar to that of graphite parallel to graphene layers, it first shrinks in a perpendicular direction at low irradiation dose and then swells.21,79 Because in a PyC-interphase graphene layers are preferably oriented parallel to the fiber surface, this dimensional change anisotropy may modify residual thermal stresses (particularly in radial direction) and alter the FM bonding. Hence, the interphase may be again the vulnerable constituent of SiC/ SiC when exposed in a prolonged manner to neutron irradiation. Potential solutions to overcome this diffi- culty are those already discussed in ‘Layered Interphases for SiC/SiC Exposed to Oxidizing Atmosphere.’16,17,80 Finally, pseudo-porous SiC interphases might also be an alternative as previously mentioned.16,72,80 However, their dimensional stability under neutron irradiation is not well known. Irradiated SiC/SiC A compilation of strength data (mostly from flexural tests) reported by different authors, for a variety of SiC/PyC/SiC composites that have been neutron irradiated (in a broad temperature range: 200–10001C) suggests that: (i) there is no loss in strength up to an irradiation dose of 10 dpa for materials fabricated with stoichiometric SiC fibers, but conversely, (ii) the strength drops by 60% almost linearly when irradiation dose increases up to 10 dpa for those with Si–C–O (Nicalon) or SiC1C (Hi-Nicalon) fibers.16,73,81 This result is consistent with the dimensional change (permanent shrinkage) reported for these two latter fibers, which lowers the FM bonding and load transfer. Hence, the analysis of the effects of neutron irradiation and interphase design on the mechanical properties of SiC/SiC should be pursued for composites with stoichiometric fibers, using exclusively tensile tests and analysis of microstructure–strength relations with appropriate models. www.ceramics.org/ACT Layered Interphases in SiC/SiC Composites 271
International Journal of Applied Ceramic Technolog-Naslain, Pailler and Lamon Vol. 7, No 3, 2010 mg TSA (where the fm bonding seems to be higher and not to markedly depend on Py C thickness) Unirrad Replacing the PyC single-layer interphase by a (PyC-SiC)n ML interphase (with e(PyC)=20 sic)=100 nm and n= 5) yields relatively brittle composites, with extremely limited nonlinear domain Dsc件 HNS)/Pyc/sic and very low strain at failure( a.1%). Their tensile curves after neutron irradiation(N 8 dpa; 800C)ar similar to those of their unirradiated counterparts. Fi Strain, E(%) nally, composites with"pseudo-porous"SiC interphase- es have also been irradiated and mechanically tested. 6. 83 Fig. 6. Tensile curves of Sic/pyC/SiC composites with Hi-Nicalon type S SiC fibers and thick py C interphase, before and after neutron irradiation, (adapted from Ozawa et al) Concluding Remarks 1. The interphases in SiC/SiC are ideally materials with a layered structure, in which the layers are parallel to Figure 6 shows tensile curve for SiC(HNS)/PyC/ the fiber surface, weakly bonded to one another bu SiC(CVI)composites recorded at ambient after neu- strongly adherent to the fiber. Matrix crack deflection tron irradiation at 1000@C 9-21,82For such a composite occurs within the interphase in a diffused manner and with a thick PyC single-layer interphase, the featu res of over short distances the tensile curve after irradiation show a weakening of 2. Anisotropic PyC could be regarded as the inter- the FM bonding(with a plateau-like shape, broad hys- phase of choice. It is deposited by CVI with graphene teresis loops, and some residual strain after unloading) layers parallel to fber surface. Achieving a strong bond Further, the strain to failure is high(and close to that of ing between the interphase and the fiber is not straight ry tow under tension) as opposed to that of the unir- forward, and it may require a fiber pretreatment. The radiated material(which seems to exhibit a premature optimal Py C thickness depends on both residual thermal failure). Irradiation at higher dose(up to a 8 dpa at stresses and fiber roughness. SiC/SiC with optimized 800C)of composites with still thicker PyC interphase PyC interphase displays high load transfer and good (a 700 nm)did not change markedly the tensile behav- mechanical properties under static or cyclic loadings ior after irradiation. Because the thickness of the Pyc in a broad temperature range. Unfortunately, PyC is interphase is here extremely large(vs 100-200 nm in oxidation prone even at low temperatures most SiC/SiC), the change in tensile behavior could be 3. BN-layered interphase is more resistant to oxida- tentatively attributed to an evolution of the PyC nano- tion. However, its deposition by CVI with optimal texture during irradiation. structure and bonding to the fiber is not straight 2D-SiC(HNS)/PyC/SiC(CVn)cor th forward. BFs-NH3 ereas much thinner PyC interphase(50-60 nm), display after BCl3-NH3-H2 requires high temperature to achieve neutron irradiation (750C with dose up to 12 dpa),a high crystallinity and corrosion resistance. Corrosion very different behavior. Their tensile curves(not shown problem could be solved through the use of TG-CVI or in Fig. 6), before and after irradiation are very similar, halogen-free organometallic precursor. Finally, achieving with convex curvature up to failure, relatively narrow strong fiber/BN bonding remains a key issue. One way to hysteresis loops and limited residual strain after unload- solve this difficulty might be to use pretreated ng. These features suggest a relatively strong FM bonding (stoichiometric) fibers with a BN surface and little evolution of the interfacial zone during irradi- 4. Another way to improve oxidation resistance of ation. However, in both cases, the strain to failure is low SiC/SiC is to reduce the thickness of PyC interphase and comparatively to that of the fiber. Hence, reducing the to play with self-healing phenomena. SiC/SiC with thickness of the PyC interphase in composites fabricated (X-Y)n interphases (with X= PyC or BN and Y= SiC with Hi-Nicalon S fibers seems to enhance the stability of or TiC) displays improved lifetime under load in oxidizing the FM-interphase bonding. It is noteworthy that such a atmospheres. The concept of Ml material associated conclusion cannot be drawn for those produced from self-healing phenomena
Figure 6 shows tensile curve for SiC (HNS)/PyC/ SiC (CVI) composites recorded at ambient after neutron irradiation at 10001C.19–21,82 For such a composite with a thick PyC single-layer interphase, the features of the tensile curve after irradiation show a weakening of the FM bonding (with a plateau-like shape, broad hysteresis loops, and some residual strain after unloading). Further, the strain to failure is high (and close to that of dry tow under tension) as opposed to that of the unirradiated material (which seems to exhibit a premature failure).82 Irradiation at higher dose (up to 8 dpa at 8001C) of composites with still thicker PyC interphase ( 700 nm) did not change markedly the tensile behavior after irradiation.20 Because the thickness of the PyC interphase is here extremely large (vs 100–200 nm in most SiC/SiC), the change in tensile behavior could be tentatively attributed to an evolution of the PyC nanotexture during irradiation. 2D-SiC (HNS)/PyC/SiC (CVI) composites, with much thinner PyC interphase (50–60 nm), display after neutron irradiation (7501C with dose up to 12 dpa), a very different behavior.19 Their tensile curves (not shown in Fig. 6), before and after irradiation are very similar, with convex curvature up to failure, relatively narrow hysteresis loops and limited residual strain after unloading. These features suggest a relatively strong FM bonding and little evolution of the interfacial zone during irradiation. However, in both cases, the strain to failure is low comparatively to that of the fiber. Hence, reducing the thickness of the PyC interphase in composites fabricated with Hi-Nicalon S fibers seems to enhance the stability of the FM-interphase bonding. It is noteworthy that such a conclusion cannot be drawn for those produced from TSA (where the FM bonding seems to be higher and not to markedly depend on PyC thickness).17 Replacing the PyC single-layer interphase by a (PyC–SiC)n ML interphase (with e(PyC) 5 20 nm, e(SiC) 5 100 nm and n 5 5) yields relatively brittle composites, with extremely limited nonlinear domain and very low strain at failure ( 0.1%). Their tensile curves after neutron irradiation ( 8 dpa; 8001C) are similar to those of their unirradiated counterparts.20 Finally, composites with ‘‘pseudo-porous’’ SiC interphases have also been irradiated and mechanically tested.16,83 Concluding Remarks 1. The interphases in SiC/SiC are ideally materials with a layered structure, in which the layers are parallel to the fiber surface, weakly bonded to one another but strongly adherent to the fiber. Matrix crack deflection occurs within the interphase in a diffused manner and over short distances. 2. Anisotropic PyC could be regarded as the interphase of choice. It is deposited by CVI with graphene layers parallel to fiber surface. Achieving a strong bonding between the interphase and the fiber is not straightforward, and it may require a fiber pretreatment. The optimal PyC thickness depends on both residual thermal stresses and fiber roughness. SiC/SiC with optimized PyC interphase displays high load transfer and good mechanical properties under static or cyclic loadings in a broad temperature range. Unfortunately, PyC is oxidation prone even at low temperatures. 3. BN-layered interphase is more resistant to oxidation. However, its deposition by CVI with optimal structure and bonding to the fiber is not straightforward. BF3–NH3 is a corrosive precursor whereas BCl3–NH3–H2 requires high temperature to achieve high crystallinity and corrosion resistance. Corrosion problem could be solved through the use of TG-CVI or halogen-free organometallic precursor. Finally, achieving a strong fiber/BN bonding remains a key issue. One way to solve this difficulty might be to use pretreated (stoichiometric) fibers with a BN surface. 4. Another way to improve oxidation resistance of SiC/SiC is to reduce the thickness of PyC interphase and to play with self-healing phenomena. SiC/SiC with (X–Y)n interphases (with X 5 PyC or BN and Y 5 SiC or TiC) displays improved lifetime under load in oxidizing atmospheres. The concept of ML material associated with self-healing phenomena is still more efficient when Fig. 6. Tensile curves of SiC/PyC/SiC composites with Hi-Nicalon type S SiC fibers and thick PyC interphase, before and after neutron irradiation, (adapted from Ozawa et al.82). 272 International Journal of Applied Ceramic Technology—Naslain, Pailler and Lamon Vol. 7, No. 3, 2010