Availableonlineatwww.sciencedirect.com Science Direct E噩≈RS ELSEVIER Journal of the European Ceramic Society 28(2008)447-453 www.elsevier.comlocate/jeurceramsoc Interface engineering in mullite fiber/mullite matrix composites KK. Chawla k Department of Materials Science and Engineering, University of Alabama at Birmingham, Birmingham, AL 35294-4461, USA Available online 6 April 2007 Abstract Mullite fiber/mullite matrix composites are attractive because of their inherent oxidation resistance at high temperatures. Mullite has better creep resistance than alumina. However. chemical int between oxides are often very severe; with the result no gain is made over monolithic mullite in terms of toughness. Even in the absence of chemical bonding, a strong mechanical bond component may be present. This originates from radial compressive stress due to thermal expansion mismatch and/or the surface roughness of interface. Thus, the microstructure and behavior of the interface region are the key factors in obtaining an effective control of damage in composites and enhancement of toughness. This body of work on mullite/mullite composites shows the feasibility of producing fully dense tough oxide/oxide composites by interface engineering. Coatings such as Bn alone or Sic/BN double coating function effectively for mullite fiber/mullite matrix composites in that they provide a nonbrittle fracture and increased work of fracture at room temperature. It would appear that for use at high temperatures in air, one needs to identify structural analogs of bN among oxides o 2007 Elsevier Ltd. All rights reserved. Keywords:Fibers:Mullite;Composites:Fracture:Toughness and toughening Introduction nber fallure Mullite fiber/mullite matrix composites form an impor Mullite-based composites are attractive because of their inher It has been amply suggested that reinforcement with con- tinuous fibers such as carbon. alumina. silicon carbide. and ent oxidation resistance at high temperatures and their favorable mullite fibers in brittle matrices can result in toughening 3 thermal shock and damage tolerance properties. In fiber rein- forced ceramic matrix composites, the local response of the It is not necessary for all these failure mechanisms to oper- ate simultaneously for a given fiber/matrix system and often fiber/matrix interface during fracture is of great importance in many composite systems only one or two of these tough- the key factors in obtaining an effective control of damage Interface engineering approach involves incorporation of fiber in composites and enhancement of toughness. When a crack coatings that can bring into play energy absorbing processes noves through a matrix containing unidirectional fibers, a such as crack deflection and fiber pullout, and thus make the variety of failure mechanisms may come into play as shown in Fig. 1 composite damage-tolerant. It would appear that the debond- ing of the fiber/matrix interface is a prerequisite for phenomena such as crack deflection, crack bridging by fibers, and fiber ● matrix fracture pullo interfacial debonding at the crack tip, followed by crack deflection. interfacial debonding in the crack wake, followed by crack 2. Criteria for interfacial debonding deflection: frictional sliding between the fiber and matrix There are two main criteria for interfacial bonding both of hem are difficult to use in practice. We indicate a third one that is relatively simple and involves only radial stress (i.e, normal Tel:+12059759725;fax:+12059348485 to the fiber/matrix interface)component. A brief description of E-mail address. chawla(uab.edu these criteria follows. 0955-2219/S-see front matter o 2007 Elsevier Ltd. All rights reserved. doi: 10. 1016/j-jeurceramsoc. 2007.03.008
Available online at www.sciencedirect.com Journal of the European Ceramic Society 28 (2008) 447–453 Interface engineering in mullite fiber/mullite matrix composites K.K. Chawla ∗ Department of Materials Science and Engineering, University of Alabama at Birmingham, Birmingham, AL 35294-4461, USA Available online 6 April 2007 Abstract Mullite fiber/mullite matrix composites are attractive because of their inherent oxidation resistance at high temperatures. Mullite has better creep resistance than alumina. However, chemical interactions between oxides are often very severe; with the result no gain is made over monolithic mullite in terms of toughness. Even in the absence of chemical bonding, a strong mechanical bond component may be present. This originates from radial compressive stress due to thermal expansion mismatch and/or the surface roughness of interface. Thus, the microstructure and behavior of the interface region are the key factors in obtaining an effective control of damage in composites and enhancement of toughness. This body of work on mullite/mullite composites shows the feasibility of producing fully dense, tough oxide/oxide composites by interface engineering. Coatings such as BN alone or SiC/BN double coating function effectively for mullite fiber/mullite matrix composites in that they provide a nonbrittle fracture and increased work of fracture at room temperature. It would appear that for use at high temperatures in air, one needs to identify structural analogs of BN among oxides. © 2007 Elsevier Ltd. All rights reserved. Keywords: Fibers; Mullite; Composites; Fracture; Toughness and toughening 1. Introduction Mullite fiber/mullite matrix composites form an important subpart of the oxide fiber/oxide matrix compsites.1–3 Mullite-based composites are attractive because of their inherent oxidation resistance at high temperatures and their favorable thermal shock and damage tolerance properties. In fiber reinforced ceramic matrix composites, the local response of the fiber/matrix interface during fracture is of great importance. The microstructure and behavior of the interface region are the key factors in obtaining an effective control of damage in composites and enhancement of toughness. When a crack moves through a matrix containing unidirectional fibers, a variety of failure mechanisms may come into play as shown in Fig. 1: • matrix fracture; • interfacial debonding at the crack tip, followed by crack deflection; • interfacial debonding in the crack wake, followed by crack deflection; • frictional sliding between the fiber and matrix; ∗ Tel.: +1 205 975 9725; fax: +1 205 934 8485. E-mail address: kchawla@uab.edu. • fiber failure; • fiber pullout. It has been amply suggested that reinforcement with continuous fibers such as carbon, alumina, silicon carbide, and mullite fibers in brittle matrices can result in toughening.3 It is not necessary for all these failure mechanisms to operate simultaneously for a given fiber/matrix system and often in many composite systems only one or two of these toughness contributions will dominate the total fracture toughness. Interface engineering approach involves incorporation of fiber coatings that can bring into play energy absorbing processes such as crack deflection and fiber pullout, and thus make the composite damage-tolerant. It would appear that the debonding of the fiber/matrix interface is a prerequisite for phenomena such as crack deflection, crack bridging by fibers, and fiber pullout. 2. Criteria for interfacial debonding There are two main criteria for interfacial bonding. Both of them are difficult to use in practice. We indicate a third one that is relatively simple and involves only radial stress (i.e., normal to the fiber/matrix interface) component. A brief description of these criteria follows: 0955-2219/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2007.03.008
K.K. Chawla/Journal of the European Ceramic Sociery 28(2008)447-453 Fiber pullout Debonding where a is the elastic mismatch parameter and is defined as E2-E, where Er where e is the elastic modulus and v is the poissons ratio for each material of interest For debonding and sliding to occur, the interfacial energy, Ti must not exceed an upper bound relative Main crack direction to the fracture energy of the second material, T2. Unfortunately, reliable Ti and T2 are usually not readily available for many systems, making this criterion difficult or impossible to use when exploring new material systems. 3. Importance of interfacial characteristics for debonding Fig. 1. A schematic of different failure mechanisms that may when a crack moves through a matrix containing unidirectional fibers. a rough fiber/matrix interface results in strong mechanical keying, which can prevent interfacial debonding and fiber pull out. A smooth interface. on the other hand. results in weak 2.1. Strength-based criterion keying, which is conducive to fiber pullout. Fig. 2 shows a schematic of a periodic roughness at the fiber/matrix inter- Debonding of the fiber/matrix interface appears to be a pre- face. Many investigators have found that interfacial roughness requisite for phenomena such as crack deflection, crack bridging has a pronounced effect on the interfacial sliding stress. The by fibers, and fiber pullout. Cook and Gordon first proposed radial strain at the fiber/matrix interface consists of two parts crack deflection or the formation of a secondary crack at a weak one is due to the thermal mismatch between the fiber and interface.For the case of a fiber/matrix system having identi- matrix, which can be either tensile or compressive, and the cal elastic constants (i. e, identical materials), Cook and Gordon, other one comes from the roughness induced clamping, which estimated the strength of the interface necessary to cause a diver- is always in compression or zero. The important point to note sion of the crack from its original direction. For any given crack, here is that even when the coefficients of thermal expansion either a triaxial state of stress (plane strain)or a biaxial stress of the coating, fiber, and matrix are such that a radial ten- plane stress) is present at the crack tip The applied principal sile stress exists at the fiber/coating interface, fiber pullout stress component, Cy, has a very high value at the crack tip, may not occur because of a strong mechanical bonding due which decreases sharply with distance from the crack tip as it to a roughness induced clamping at the fiber/matrix. Thus, in must be because the surface of a crack is a free surface. ox, the CMCs having an extremely rough interface, the pullout would stress component normal to the fiber/matrix interface, is 0 at the not be expected, and they would likely fail like monolithic crack tip. o then rises quickly to a maximum a short distance materials away from the tip and then quickly decreases with distance in The two sources that contribute to the radial stress component a fashion similar to that seen with ay. If the fiber/matrix tensile acting normal to the fiber/matrix interface are interface strength is less than the 1 maximum value of r fracture should occur at the interface ahead of the approaching crack tip. (i) thermal mismatch induced radial compressive stress: Cook and Gordon estimated that if the interface had strength of (ii) mechanical gripping induced by the fibersurface roughness about 1/5 or less than that of the main stress component, y, it will debond in front of the crack tip. Again, it should be noted In ceramic matrix composites, interfacial roughness induced that they studied a system with identical component interface stresses, especially the radial stress, will affect the interface debonding, the sliding friction of debonded fibers, as 2.2. Energy-based criterion shown in Fig. l, and the fiber pullout length. Fiber pullout is one of the important energy dissipating fracture processes in fiber reinforced ceramic or glass matrix composites. An absence An energy-based criterion has been proposed for interna- of strong chemical bond and a purely mechanical bond at the ial debonding by He and Hutchinson. If Ti is the interfacial fiber/matrix interface is highly desirable for the fiber pullout (debond)energy and r2 is the fracture energy of the second to occur. Even when the coefficients of thermal expansion of material or fiber in Mode I then interfacial debonding and slid- the coating, fiber, and matrix are such that a radial tensile stress ing will occur rather than brittle cracking through the fiber, when exists at the fiber/coating interface after cooling from anelevated the following inequality is satisfied: processing temperature, fiber pullout may not occur because of a strong mechanical bonding due to a roughness induced n≤()2fora=0 (1) clamping at the fiber/matrix interface. The radial stress result ing from the surface roughness of the fiber during fiber pullout
448 K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 Fig. 1. A schematic of different failure mechanisms that may come into play when a crack moves through a matrix containing unidirectional fibers. 2.1. Strength-based criterion Debonding of the fiber/matrix interface appears to be a prerequisite for phenomena such as crack deflection, crack bridging by fibers, and fiber pullout. Cook and Gordon first proposed crack deflection or the formation of a secondary crack at a weak interface.4 For the case of a fiber/matrix system having identical elastic constants (i.e., identical materials), Cook and Gordon, estimated the strength of the interface necessary to cause a diversion of the crack from its original direction. For any given crack, either a triaxial state of stress (plane strain) or a biaxial stress (plane stress) is present at the crack tip. The applied principal stress component, σy, has a very high value at the crack tip, which decreases sharply with distance from the crack tip as it must be because the surface of a crack is a free surface. σx, the stress component normal to the fiber/matrix interface, is 0 at the crack tip. σx then rises quickly to a maximum a short distance away from the tip and then quickly decreases with distance in a fashion similar to that seen with σy. If the fiber/matrix tensile interface strength is less than the maximum value of σx, fracture should occur at the interface ahead of the approaching crack tip. Cook and Gordon estimated that if the interface had strength of about 1/5 or less than that of the main stress component, σy, it will debond in front of the crack tip. Again, it should be noted that they studied a system with identical components. 2.2. Energy-based criterion An energy-based criterion has been proposed for interfacial debonding by He and Hutchinson.5 If Γ i is the interfacial (debond) energy and Γ 2 is the fracture energy of the second material or fiber in Mode I, then interfacial debonding and sliding will occur rather than brittle cracking through the fiber, when the following inequality is satisfied: Γi ≤ 1 4 Γ2 for α = 0 (1) where α is the elastic mismatch parameter and is defined as: α = E 2 − E 1 E 2 − E 1 , where E = E (1 − υ2) (2) where E is the elastic modulus and υ is the Poisson’s ratio for each material of interest. For debonding and sliding to occur, the interfacial energy, Γ i must not exceed an upper bound relative to the fracture energy of the second material, Γ 2. Unfortunately, reliable Γ i and Γ 2 are usually not readily available for many systems, making this criterion difficult or impossible to use when exploring new material systems. 3. Importance of interfacial characteristics for debonding A rough fiber/matrix interface results in strong mechanical keying, which can prevent interfacial debonding and fiber pullout. A smooth interface, on the other hand, results in weak keying, which is conducive to fiber pullout. Fig. 2 shows a schematic of a periodic roughness at the fiber/matrix interface. Many investigators have found that interfacial roughness has a pronounced effect on the interfacial sliding stress. The radial strain at the fiber/matrix interface consists of two parts: one is due to the thermal mismatch between the fiber and matrix, which can be either tensile or compressive, and the other one comes from the roughness induced clamping, which is always in compression or zero. The important point to note here is that even when the coefficients of thermal expansion of the coating, fiber, and matrix are such that a radial tensile stress exists at the fiber/coating interface, fiber pullout may not occur because of a strong mechanical bonding due to a roughness induced clamping at the fiber/matrix. Thus, in CMCs having an extremely rough interface, the pullout would not be expected, and they would likely fail like monolithic materials. The two sources that contribute to the radial stress component acting normal to the fiber/matrix interface are: (i) thermal mismatch induced radial compressive stress; (ii) mechanical gripping induced by the fiber surface roughness. In ceramic matrix composites, interfacial roughness induced interface stresses, especially the radial stress, will affect the interface debonding, the sliding friction of debonded fibers, as shown in Fig. 1, and the fiber pullout length. Fiber pullout is one of the important energy dissipating fracture processes in fiber reinforced ceramic or glass matrix composites. An absence of strong chemical bond and a purely mechanical bond at the fiber/matrix interface is highly desirable for the fiber pullout to occur. Even when the coefficients of thermal expansion of the coating, fiber, and matrix are such that a radial tensile stress exists at the fiber/coating interface after cooling from an elevated processing temperature, fiber pullout may not occur because of a strong mechanical bonding due to a roughness induced clamping at the fiber/matrix interface. The radial stress resulting from the surface roughness of the fiber during fiber pullout,
K.K. Chawla/ Journal of the European Ceramic Society 28(2008)447-453 n Fig. 2. Schematic of periodic interfacial roughness. The fiber and matrix are indicated by f and m, respectively. A is the amplitude of the interfacial roughness. should be added algebraically to the radial thermal stress, 1.e the effect of surface roughness of the fiber along the debonded interface on the radial stress at the interface needs to be con- sidered together with the thermal stresses. The sense of the two e surface ro ughness of the avail- able oxide fibers can be varied significantly. In ceramic matrix composites, roughness-induced interfacial gripping, especially in the radial direction. will affect the interface debondin the sliding friction of debonded fibers, and the fiber pullout The objective of this paper is to show that by applying the principle of interfacial engineering in mullite fiber rein- forced mullite matrix composites, one can get fully dense but tough, damage tolerant composites. Thus, approaches involving a porous matrix or interlace are not considered here. Thin film coatings on fibers are used to modify and control the interface behavior in CMCs. In particular, BN can be a weak interphase between the mullite fiber and mullite matrix because of its graphite-type layer structure. 6- A major disadvantage of BN coatings is their poor oxidation resistance at high temperatures There are two ways around this problem. One possible way is to e a thick bn coating so a portion of the coating can be sacri- ficed during processing. The other possibility is to protect the B 2.5gm coating it by a second coating with better oxidation resistance, such as SiC. i.e.. use sic/bn double coating. That still retains a onoxide interphase in an oxide fiber/oxide matrix composite. It should be recognized here that both bn and sic/bn coatings are Fig 3. A mullite fiber/mullite matrix composite processed without an interphase nonoxides and thus susceptible to oxidation at high temperatures coating. The etched cross-section shows that mullite fiber in the center was lost after processing
K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 449 Fig. 2. Schematic of periodic interfacial roughness. The fiber and matrix are indicated by f and m, respectively. A is the amplitude of the interfacial roughness. should be added algebraically to the radial thermal stress, i.e., the effect of surface roughness of the fiber along the debonded interface on the radial stress at the interface needs to be considered together with the thermal stresses. The sense of the two can be the same or opposite. The surface roughness of the available oxide fibers can be varied significantly. In ceramic matrix composites, roughness-induced interfacial gripping, especially in the radial direction, will affect the interface debonding, the sliding friction of debonded fibers, and the fiber pullout length.1,3 The objective of this paper is to show that by applying the principle of interfacial engineering in mullite fiber reinforced mullite matrix composites, one can get fully dense but tough, damage tolerant composites. Thus, approaches involving a porous matrix or interlace are not considered here. Thin film coatings on fibers are used to modify and control the interface behavior in CMCs. In particular, BN can be a weak interphase between the mullite fiber and mullite matrix because of its graphite-type layer structure.6–8 A major disadvantage of BN coatings is their poor oxidation resistance at high temperatures. There are two ways around this problem. One possible way is to use a thick BN coating so a portion of the coating can be sacri- ficed during processing. The other possibility is to protect the BN coating it by a second coating with better oxidation resistance, such as SiC, i.e., use SiC/BN double coating. That still retains a nonoxide interphase in an oxide fiber/oxide matrix composite. It should be recognized here that both BN and SiC/BN coatings are nonoxides and thus susceptible to oxidation at high temperatures in air. Fig. 3. A mullite fiber/mullite matrix composite processed without an interphase coating. The etched cross-section shows that mullite fiber in the center was lost after processing
450 K.K. Chawla/Journal of the European Ceramic Sociery 28(2008)447-453 70 wt %o Al2O3, 30wt.% SiO, i.e., compared to Nextel 480 boria is Two of the most common interphases used in fiber com- 6Osites are carbon(C)and boron nitride(BN). These materials toughness. The graphitic form of c bon has easily cleavable basal planes. This makes it ideal as a weak interface for crack deflection. The problem with carbon is that it is readily oxidized at temperatures above 400Cin air. Boron nitride also has a layer structure like that of graphite Boron nitride, although a nonoxide is more oxidation resis- tant than carbon but its oxidation resistance is a function of its Double coatings of BN (inner)/SiC(outer)double layer were 200n produced by CVD An interfacial testing system with a fat-bottomed, diamond indenter was used to obtain the interface characteristics. Three Sic BN point flexural tests were done to examine the load-displacement characteristics of the composites composite showing the mullite matrix(M)and Mullite fiber/mullite matrix composites were made via a mullite fiber (F)with sle coating of sic and bn in between sol-gel route. Mullite gels can be single-phase or diphasic according to the scale of component mixing, depending upon 4. Materials and experimental procedure the nature of the alumina and silica precursors. Single-phase 3M Company has developed a series of oxide fibers. This Nextel 550/Mullite(HP) series of fibers, called the Nextel fibers, mainly consists of alumina and mullite-type fibers. Nextel ceramic oxide fibers Sic/BN coated Uncoated are continuous and polycrystalline. These fibers are typically transparent, nonporous, and have a diameter of 10-12 um. Two mullite-based fibers from this series. Nextel 480 and Nextel 550 were used. Nextel 480 fibers have 70 wt %o Al2 O3, 28 wt% SiO nd 2 wt %o B2O3 and are currently not produced. BN coatings on Nextel 480 were produced by CVD. Nextel 550 fibers have 0 0.1 0.2 0.3 Displacement, mm extel 480/Mullite(HP) (b)400 0.00.10203040.50.6 Fig. 6.(a)Stress vs displacement for a mullite fiber(Nextel 550)mullite matrix 5am the double coating shows a damage-tolerant behavior.(b) Stress vs displacement for a mullite fiber(Nextel 480)/mullite matrix comp h no coating and two composites with different coating thicknesses of BN. The uncoated composite Fig. 5. An indentation induced crack deflection at the BN coating in the as-made and the composite with 0.3 um BN coating show catastrophic, brittle fracture while the composite with 1 um bn coating shows a damage tolerant behavior
450 K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 Fig. 4. A cross-section of the composite showing the mullite matrix (M) and mullite fiber (F) with the double coating of SiC and BN in between. 4. Materials and experimental procedure 3M Company has developed a series of oxide fibers.9 This series of fibers, called the Nextel fibers, mainly consists of alumina and mullite-type fibers. Nextel ceramic oxide fibers are continuous and polycrystalline. These fibers are typically transparent, nonporous, and have a diameter of 10–12m. Two mullite-based fibers from this series, Nextel 480 and Nextel 550 were used. Nextel 480 fibers have 70 wt.% Al2O3, 28 wt.% SiO2, and 2 wt.% B2O3 and are currently not produced. BN coatings on Nextel 480 were produced by CVD. Nextel 550 fibers have Fig. 5. An indentation induced crack deflection at the BN coating in the as-made composite. 70 wt.% Al2O3, 30 wt.% SiO2, i.e., compared to Nextel 480, boria is missing. Two of the most common interphases used in fiber composites are carbon (C) and boron nitride (BN). These materials have intrinsic low fracture toughness. The graphitic form of carbon has easily cleavable basal planes. This makes it ideal as a weak interface for crack deflection. The problem with carbon is that it is readily oxidized at temperatures above 400 ◦C in air. Boron nitride also has a layer structure like that of graphite. Boron nitride, although a nonoxide, is more oxidation resistant than carbon, but its oxidation resistance is a function of its stoichiometry. Double coatings of BN (inner)/SiC (outer) double layer were produced by CVD. An interfacial testing system with a flat-bottomed, diamond indenter was used to obtain the interface characteristics. Threepoint flexural tests were done to examine the load-displacement characteristics of the composites. Mullite fiber/mullite matrix composites were made via a sol–gel route. Mullite gels can be single-phase or diphasic according to the scale of component mixing, depending upon the nature of the alumina and silica precursors. Single-phase Fig. 6. (a) Stress vs. displacement for a mullite fiber (Nextel 550)/mullite matrix composite with no coating and double coating of SiC/BN. The composite with the double coating shows a damage-tolerant behavior. (b) Stress vs. displacement for a mullite fiber (Nextel 480)/mullite matrix composite with no coating and two composites with different coating thicknesses of BN. The uncoated composite and the composite with 0.3m BN coating show catastrophic, brittle fracture while the composite with 1m BN coating shows a damage tolerant behavior.
K.K. Chawla/ Journal of the European Ceramic Society 28(2008)447-453 gels are prepared from aluminum and silicon alkoxides or salts; 5. Results and discussion they have molecular-scale mixing because of a polymerized oxide network formed by hydrolytic condensation. Diphasic As expected, no fiber pullout was observed in the uncoated gels involve mixing of sols of boehmite and silica or mixing of mullite/mullite composites as shown in Fig. 3. A cross-section one colloidal component with alkoxide or salt of other. The two of the composite showing the mullite matrix(M) and mullite routes are quite different, mainly because of the different scales fiber(F) with the double coating of SiC and Bn in between of component mixing. The single-phase and diphasic gels show is shown in Fig. 4. We used two types of coatings, a thicker different types of mullite crystallization behavior during heat- Bn (1 um) and a BN/SiC double coating. The thicker coating ing. Single-phase gels have a very short interdiffusion distance allows for a part of the coating to be sacrificed by oxidation because of the molecular-scale mixing, and therefore, mullite during processing. The objective of using an SiC coating was crystallization can occur at temperatures as low as 1000C In to provide oxidation protection to Bn during processing. The diphasic gels, however, the diffusion distance is much longer, efficacy of BN coating to deflect an oncoming crack is shown in so mullite crystallization does not occur until above 1250C. Fig. 5. The crack, introduced by means of an indentation, can be Retardation of mullite crystallization in the diphasic gels pro- seen to deflect at the bn coating and go around the fiber rather vides a useful processing window o This is a key point. With than penetrate it. In the BN interphase, which is isostructural diphasic gels, >95%o of theoretical density (TD)can be obtained to C, the orientation of the basal planes parallel to the substrate through one of the following two methods. surface is attributed to low surface energy perpendicular to the basal planes. The orientation of(0002)BN basal planes parallel Sintering times of more than I h between 1200 and 1300oC. to the fiber surfaces is the favorable for the damage tolerance with carefully controlled heating of composites properties. It enables easy sliding along these Sintering for I h with a low heating rate(2 CImin)and a high planes thus producing a weak fiber/matrix interface In both cases, thick bn coating or Sic/bn double compaction pressure(441 MPa)for a green bod a noncatastrophic failure mode was observed. The stress- displacement curves obtained in a three-point bend tests for Compared to this, in single-phase gels, crystalline mullite composites containing interfacial coatings of BN and SiC/BN forms at very low temperatures, which makes densification diffi- coated as well as uncoated composites are shown in Fig 6a and cult because of the high degree of covalent bonding in crystalline b. Fig 7a shows the fracture surfaces of the composites contain- mullite. The result is that the densities obtained at the same ing l um BN-coated fibers and while Fig 7b shows the fracture hot pressing temperature are much lower than with the viscous- surface of a composite containing SiC/BN-coated fibers In both phase processing cases, the phenomenon of fiber pullout occurred; which led to a (a) 25m 100m Fig. 7. Fiber pullout(a)in thick BN-coated and(b) SiC/BN-coated mullite fiber/mullite matrix composite
K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 451 gels are prepared from aluminum and silicon alkoxides or salts; they have molecular-scale mixing because of a polymerizedoxide network formed by hydrolytic condensation. Diphasic gels involve mixing of sols of boehmite and silica or mixing of one colloidal component with alkoxide or salt of other. The two routes are quite different, mainly because of the different scales of component mixing. The single-phase and diphasic gels show different types of mullite crystallization behavior during heating. Single-phase gels have a very short interdiffusion distance because of the molecular-scale mixing, and therefore, mullite crystallization can occur at temperatures as low as 1000 ◦C. In diphasic gels, however, the diffusion distance is much longer, so mullite crystallization does not occur until above 1250 ◦C. Retardation of mullite crystallization in the diphasic gels provides a useful processing window.10 This is a key point. With diphasic gels, >95% of theoretical density (TD) can be obtained through one of the following two methods: - Sintering times of more than 1 h between 1200 and 1300 ◦C, with carefully controlled heating; - Sintering for 1 h with a low heating rate (2◦ C/min) and a high compaction pressure (441 MPa) for a green body. Compared to this, in single-phase gels, crystalline mullite forms at very low temperatures, which makes densification diffi- cult because of the high degree of covalent bonding in crystalline mullite. The result is that the densities obtained at the same hot pressing temperature are much lower than with the viscousphase processing. 5. Results and discussion As expected, no fiber pullout was observed in the uncoated mullite/mullite composites as shown in Fig. 3. A cross-section of the composite showing the mullite matrix (M) and mullite fiber (F) with the double coating of SiC and BN in between is shown in Fig. 4. We used two types of coatings, a thicker BN (1m) and a BN/SiC double coating. The thicker coating allows for a part of the coating to be sacrificed by oxidation during processing. The objective of using an SiC coating was to provide oxidation protection to BN during processing. The efficacy of BN coating to deflect an oncoming crack is shown in Fig. 5. The crack, introduced by means of an indentation, can be seen to deflect at the BN coating and go around the fiber rather than penetrate it. In the BN interphase, which is isostructural to C, the orientation of the basal planes parallel to the substrate surface is attributed to low surface energy perpendicular to the basal planes. The orientation of (0 0 0 2) BN basal planes parallel to the fiber surfaces is the favorable for the damage tolerance of composites properties.11 It enables easy sliding along these planes thus producing a weak fiber/matrix interface. In both cases, thick BN coating or SiC/BN double coating, a noncatastrophic failure mode was observed. The stressdisplacement curves obtained in a three-point bend tests for composites containing interfacial coatings of BN and SiC/BNcoated as well as uncoated composites are shown in Fig. 6a and b. Fig. 7a shows the fracture surfaces of the composites containing 1m BN-coated fibers and while Fig. 7b shows the fracture surface of a composite containing SiC/BN-coated fibers. In both cases, the phenomenon of fiber pullout occurred; which led to a Fig. 7. Fiber pullout (a) in thick BN-coated and (b) SiC/BN-coated mullite fiber/mullite matrix composite.
452 K.K. Chawla Journal of the European Ceramic Sociery 28(2008)447-453 Coatings such as bn alone or sic/bn double coati on effectively for mullite fiber/mullite matrix composites in I um that they impart damage tolerant characteristics to mullite fiber/mullite matrix composites. However, both BN and sic/BN BN coatings are nonoxides, and thus are susceptible to oxidation at high temperatures in air. Of course, SiC is more resistant to oxi dation than BN. What this points to is the need for oxide analogs of BN that would function as interphase materials at high temper atures. Beta-aluminas or micas are possible candidate materials. Layered oxides having B-alumina and magnetoplumbite struc- tures should be explored as possible easy cleavage coatings in mullite-based composites 6. Conclusions Chemical interactions between oxides are often severe such that selection of possible mullite-based composites or oxide fiber/oxide matrix composites, in general, is limited Even in the absence of chemical bonding a strong mechanical bond component may be present. This originates from radial M compressive stress due to thermal expansion mismatch and/or the surface roughness of interface. Coatings such as bn alone or sicbn double coating func Fig 8. Fracture surface of mullite fiber, Nextel 550(F)with double coating of tion effectively for mullite fiber/mullite matrix composites i SiC/BNinamullite matrix(M). The interface between BN and SiC is highlighted that they provide a nonbrittle fracture and increased work of with a dotted line. Note the sliding occurred along the fiber/BN coating. The fracture at room temperature SiC coating remained bonded to the mullite matrix Need oxide analogs of bn that would function as inter phase materials at high temperatures. Beta-aluminas or micas are possible candidate materials. Layered oxides having B- higher work of fracture in the coated fiber composites than in the alumina and magnetoplumbite structures should be explored uncoated fiber composite. The BN/SiC double coating system as possible easy cleavage coatings in mullite-based compos consists of 0.08 um BN and0. 16 um SiC-layers. In this case BN is rather turbostractic, and does not undergo any microstruc- tural change up to 1300C. An interesting high magnification Acknowledgments micrograph of the fracture surface of mullite fiber, Nextel 550 with double coating of SiC/BN in a mullite matrix(M)is The work presented in this paper is summary of work done shown in Fig 8. The interface between BN and SiC is high- over more than a decade and involves contributions from many lighted with a dotted line. Note the sliding occurred along the students, Post-Docs, and colleagues. In particular, I wish to fiber/BN coating. The SiC coating remained bonded to the mul- acknowledge the important part played by Professor Dr. H lite matrix. The outer thin SiC layer of the double coating system Schneider during a period of very fertile collaboration. I wish to improves the oxidation resistance of Bn up to 1200-1300C in thank the following agencies that sponsored parts of this work spite of a partial oxidation of SiC to SiO A word of caution is in order here. The application of a coat- US Office of Naval Research ing on fiber can affect its strength. If the thermal mismatch can US Department of Energy, Office of Transportation Tech- result in a residual tensile strength in the fiber, it will weaker nologies, through the High Temperature Materials Laboratory the fiber. The tensile strength of Nextel 480 fiber increased User Program and HTML Faculty Fellowship Program Oak with increasing BN coating thickness until 0. 2 um For coatings Ridge National Lab greater than 0.2 um in thickness, the fiber strength decreased. German Aerospace Research Establishment(DLR) There are two factors affecting the strength of the coated fiber North Atlantic Treaty Organization(NATO) simultaneously. One is the smooth boron nitride coa heals the surface defects of such fiber and thus, contributes to References rength enhancement. The second one is the volume fraction of the coating material which is a weak material. When coating 1. Chawla, KK, Coffin, C and Xu, Z.R., Int Mater. Rev, 2000, 45, 165 thickness is below 0.2 um, the first factor plays a dominant role 2. Schneider, H. and Komarneni, S,ed, Mullite. Wiley-VCH, Weinheim in the strength enhancement. When coating thickness is above 3. Chawla, K.K., Ceramic Matrix Composites(2nded. ) Kluwer, 0.2 um, the second factor becomes dominant to reduce the fiber 4. Cook, J and Gordon, J E. Proc R Soc. Lond., 1964, A228. strength 5. He, M -Y and Hutchinson, J w, Int J Solids Struct, 1989. 25. 1053-
452 K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 Fig. 8. Fracture surface of mullite fiber, Nextel 550 (F) with double coating of SiC/BN in a mullite matrix (M). The interface between BN and SiC is highlighted with a dotted line. Note the sliding occurred along the fiber/BN coating. The SiC coating remained bonded to the mullite matrix. higher work of fracture in the coated fiber composites than in the uncoated fiber composite. The BN/SiC double coating system consists of 0.08m BN and 0.16m SiC-layers. In this case BN is rather turbostractic,11 and does not undergo any microstructural change up to 1300 ◦C. An interesting high magnification micrograph of the fracture surface of mullite fiber, Nextel 550 (F) with double coating of SiC/BN in a mullite matrix (M) is shown in Fig. 8. The interface between BN and SiC is highlighted with a dotted line. Note the sliding occurred along the fiber/BN coating. The SiC coating remained bonded to the mullite matrix. The outer thin SiC layer of the double coating system improves the oxidation resistance of BN up to 1200–1300 ◦C in spite of a partial oxidation of SiC to SiO2. A word of caution is in order here. The application of a coating on fiber can affect its strength. If the thermal mismatch can result in a residual tensile strength in the fiber, it will weaken the fiber. The tensile strength of Nextel 480 fiber increased with increasing BN coating thickness until 0.2m. For coatings greater than 0.2m in thickness, the fiber strength decreased. There are two factors affecting the strength of the coated fiber simultaneously. One is the smooth boron nitride coating which heals the surface defects of such fiber, and thus, contributes to strength enhancement. The second one is the volume fraction of the coating material which is a weak material. When coating thickness is below 0.2 m, the first factor plays a dominant role in the strength enhancement. When coating thickness is above 0.2m, the second factor becomes dominant to reduce the fiber strength. Coatings such as BN alone or SiC/BN double coating function effectively for mullite fiber/mullite matrix composites in that they impart damage tolerant characteristics to mullite fiber/mullite matrix composites. However, both BN and SiC/BN coatings are nonoxides, and thus are susceptible to oxidation at high temperatures in air. Of course, SiC is more resistant to oxidation than BN. What this points to is the need for oxide analogs of BN that would function as interphase materials at high temperatures. Beta-aluminas or micas are possible candidate materials. Layered oxides having -alumina and magnetoplumbite structures should be explored as possible easy cleavage coatings in mullite-based composites. 6. Conclusions • Chemical interactions between oxides are often severe such that selection of possible mullite-based composites or oxide fiber/oxide matrix composites, in general, is limited. • Even in the absence of chemical bonding, a strong mechanical bond component may be present. This originates from radial compressive stress due to thermal expansion mismatch and/or the surface roughness of interface. • Coatings such as BN alone or SiC/BN double coating function effectively for mullite fiber/mullite matrix composites in that they provide a nonbrittle fracture and increased work of fracture at room temperature. • Need oxide analogs of BN that would function as interphase materials at high temperatures. Beta-aluminas or micas are possible candidate materials. Layered oxides having - alumina and magnetoplumbite structures should be explored as possible easy cleavage coatings in mullite-based composites. Acknowledgments The work presented in this paper is summary of work done over more than a decade and involves contributions from many students, Post-Docs, and colleagues. In particular, I wish to acknowledge the important part played by Professor Dr. H. Schneider during a period of very fertile collaboration. I wish to thank the following agencies that sponsored parts of this work: • US Office of Naval Research. • US Department of Energy, Office of Transportation Technologies, through the High Temperature Materials Laboratory User Program and HTML Faculty Fellowship Program Oak Ridge National Lab. • German Aerospace Research Establishment (DLR) • North Atlantic Treaty Organization (NATO). References 1. Chawla, K. K., Coffin, C. and Xu, Z. R., Int. Mater. Rev., 2000, 45, 165. 2. Schneider, H. and Komarneni, S., ed., Mullite. Wiley-VCH, Weinheim, 2005. 3. Chawla, K. K.,Ceramic Matrix Composites(2nd ed.). Kluwer, Boston, 1998. 4. Cook, J. and Gordon, J. E., Proc. R. Soc. Lond., 1964, A228, 508. 5. He, M.-Y. and Hutchinson, J. W., Int. J. Solids Struct., 1989, 25, 1053–1067.
K Chawla/Journal of the European Ceramic Society 28(2008)447-453 6.Rice,Rw., US Patent4,642,271(1987). 9. Chawla, KK, Fibrous Materials. Cambridge University Press, Cambridge, 7. Thomas J. Watson. N. E. and O Conner. TE. J. Am. Chem. Soc. 1963 10. Chawla, K.K., Xu, Z.R., Ha, J.S., Schmuicker, M. and Schneider, H, Appl. 8. Wyckoff, R. w. G, Crystal Structures, vol. 1(2nd ed. ) R.E. Kreiger Pub- Comp. Mater, 1997, 4, 263 lishing Co, Malabar, FL, 1982 1. Chawla, K. K, Xu Z.R. and Ha. J.-S, J. Euro Ceram Soc., 1996. 16. 293
K.K. Chawla / Journal of the European Ceramic Society 28 (2008) 447–453 453 6. Rice, R.W., US Patent 4,642,271 (1987). 7. Thomas, J., Watson, N. E. and O’Conner, T. E., J. Am. Chem. Soc., 1963, 84, 4619. 8. Wyckoff, R. W. G., Crystal Structures, vol. 1 (2nd ed.). R.E. Kreiger Publishing Co, Malabar, FL, 1982. 9. Chawla, K. K., Fibrous Materials. Cambridge University Press, Cambridge, UK, 1998. 10. Chawla, K. K., Xu, Z. R., Ha, J. S., Schmucker, M. and Schneider, H., ¨ Appl. Comp. Mater., 1997, 4, 263. 11. Chawla, K. K., Xu, Z. R. and Ha, J.-S., J. Euro. Ceram. Soc., 1996, 16, 293