and manufacturing ELSEVIER Composites: Part A 30(1999)463 Evaluating the effect of oxygen content in BN interfacial coatings on the stability of SiC/BN/SiC composites K.L. More,, K.S. Ailey,R.A. Lowden, H.T. Lin Metals and Ceramics Division, Oak Ridge National Laboratory, Building 4515, MS 6064. PO Box 2008, Oak Ridge, TN 37931-6064, USA Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC 27695, USA Boron nitride was studied as a fiber-matrix interface coating for Nicalon/SiC composites. The effect of initial O-impurity content within the as-processed bN coatings on the long-term interface stability was investigated at elevated temperatures in flowing oxygen. two types of Nicalon"/SiC composites were used for this study; one composite had a bn coating with 2%oxygen(low-O BN) and another co had BN with an oxygen concentration 11%(high-O BN) in the as-processed state. The high-o BN is actually most representative of BN coatings available commercially. The bn coatings in both the high-o and low-O BN containing composites were structurally similar. The samples used here were thinned to 200 um before oxidation and the final preparation for electron microscopy examination of the interface ion was done after the reactions were completed Thin samples were used to simulate maximum corrosion effects that would occur at the surface of an actual part during service. Ech sample was exposed to flowing oxygen at temperatures as high as 950oC for times up to 400 h After each oxidation experiment, the bN coatings were examined by TEM to quantify the extent of any reaction which occurred at either fiber/BN and BN/SiC matrix interfaces. At 950C for 100 h, there were no interface microstructural changes observed in the low-o bn but there was extensive silica formation at the fiber/BN interfaces in the high-o BN. After 400 h at 950 C, large voids formed at the fiber/BN interface in the high-O BN sample only. Oxygen present within the initial BN coating contributed significantly to the degradation of the interfacial properties of the composite. Several techniques, including transmission electron microscopy (TEM), Auger electron spectroscopy (AES), energy-dispersive spectrometry(EDS), and electron energy-loss spectroscopy(EEls)were used to characterize changes in structure and chemistry of the fiber-matrix interface region and to elucidate and quantify composite degradation mechanisms. c 1999 Elsevier Science ltd. all rights reserved Keywords: A Ceramic matrix composites( CMCs); BN; Oxidation; Characterization 1. Introduction been investigated. bn has received much attention recentl for use in Sic-based composite systems, since it has The fibers and matrix play major roles in determining the improved oxidation resistance compared to carbon [5,6] final properties of a composite, however, it is the fiber- However, there are concerns regarding the long-term stabi matrix interface that has significant influence on the fracture lity of many commercially available Bn coatings. Several behavior and mechanical properties of fiber-reinforced recent studies have addressed the issue of the stability of BN composites. Interface coatings are commonly applied to in oxygen-and water-containing environments [7-9 not only to protect the fibers during matrix processing, but It has also become important to fully evaluate the 'type also to control interfacial forces and provide protection for of bn being used as an interface coating since there are the fibers during use in corrosive environments at elevated many parameters that will influence the bn stability at temperatures [1, 2]. Thus, the development of a stable and different use temperatures. Obviously, all BN coatings are chemically compatible fiber coating is necessary in order for inherently different and understanding the atomic level fiber-reinforced composites to be used in critical applica- structure and chemistry of the bn will provide invaluable tions. Carbon was the most commonly used interlayer in the information for composite life-prediction. The crystal struc- past, but as a result of its poor oxidation resistance at ture and chemical composition of the BN will depend on the elevated temperatures [3, 4], other interface materials have deposition parameters used, i.e., gas precursors and deposi tion temperature will influence such factors as BN stoichio- Corresponding author. Tel. 1-423-5747788: fax: 1-423- metry, degree of crystallinity, and impurity content in the 5744913 BN. All of these factors play a role in the long-term (/99/.see front matter e 1999 Elsevier Science Ltd. All rights reserved 59-835X(98)00135-3
Evaluating the effect of oxygen content in BN interfacial coatings on the stability of SiC/BN/SiC composites K.L. Morea,*, K.S. Aileyb , R.A. Lowdena , H.T. Lina a Metals and Ceramics Division, Oak Ridge National Laboratory, Building 4515, MS 6064, PO Box 2008, Oak Ridge, TN 37931-6064, USA b Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC 27695, USA Abstract Boron nitride was studied as a fiber–matrix interface coating for Nicalone/SiC composites. The effect of initial O-impurity content within the as-processed BN coatings on the long-term interface stability was investigated at elevated temperatures in flowing oxygen. Two types of Nicalone/SiC composites were used for this study; one composite had a BN coating with , 2% oxygen (low-O BN) and another composite had BN with an oxygen concentration . 11% (high-O BN) in the as-processed state. The high-O BN is actually most representative of BN coatings available commercially. The BN coatings in both the high-O and low-O BN containing composites were structurally similar. The samples used here were thinned to , 200 mm before oxidation and the final preparation for electron microscopy examination of the interface region was done after the reactions were completed. Thin samples were used to simulate maximum corrosion effects that would occur at the surface of an actual part during service. Ech sample was exposed to flowing oxygen at temperatures as high as 9508C for times up to 400 h. After each oxidation experiment, the BN coatings were examined by TEM to quantify the extent of any reaction which occurred at either the fiber/BN and BN/SiC matrix interfaces. At 9508C for 100 h, there were no interface microstructural changes observed in the low-O BN but there was extensive silica formation at the fiber/BN interfaces in the high-O BN. After 400 h at 9508C, large voids formed at the fiber/BN interface in the high-O BN sample only. Oxygen present within the initial BN coating contributed significantly to the degradation of the interfacial properties of the composite. Several techniques, including transmission electron microscopy (TEM), Auger electron spectroscopy (AES), energy-dispersive spectrometry (EDS), and electron energy-loss spectroscopy (EELS) were used to characterize changes in structure and chemistry of the fiber–matrix interface region and to elucidate and quantify composite degradation mechanisms. q 1999 Elsevier Science Ltd. All rights reserved. Keywords: A. Ceramic matrix composites (CMCs); BN; Oxidation; Characterization 1. Introduction The fibers and matrix play major roles in determining the final properties of a composite, however, it is the fiber– matrix interface that has significant influence on the fracture behavior and mechanical properties of fiber-reinforced composites. Interface coatings are commonly applied to not only to protect the fibers during matrix processing, but also to control interfacial forces and provide protection for the fibers during use in corrosive environments at elevated temperatures [1,2]. Thus, the development of a stable and chemically compatible fiber coating is necessary in order for fiber-reinforced composites to be used in critical applications. Carbon was the most commonly used interlayer in the past, but as a result of its poor oxidation resistance at elevated temperatures [3,4], other interface materials have been investigated. BN has received much attention recently for use in SiC-based composite systems, since it has improved oxidation resistance compared to carbon [5,6]. However, there are concerns regarding the long-term stability of many commercially available BN coatings. Several recent studies have addressed the issue of the stability of BN in oxygen- and water-containing environments [7–9]. It has also become important to fully evaluate the ‘type’ of BN being used as an interface coating since there are many parameters that will influence the BN stability at different use temperatures. Obviously, all BN coatings are inherently different and understanding the atomic level structure and chemistry of the BN will provide invaluable information for composite life-prediction. The crystal structure and chemical composition of the BN will depend on the deposition parameters used, i.e., gas precursors and deposition temperature will influence such factors as BN stoichiometry, degree of crystallinity, and impurity content in the BN. All of these factors play a role in the long-term Composites: Part A 30 (1999) 463–470 1359-835X/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S1359-835X(98)00135-3 * Corresponding author. Tel.: 1 1-423-5747788; fax: 1 1-423- 5744913
K.L. More et al. /Composites: Part A 30 (1999)463-470 multiple layers of ceramic-grade Nicalon plain-weave fabric were stacked and rotated in a0 30 sequence in a graphite holder;(2)layers were compressed in the holder cavity to produce a preform having a nominal fiber loading of-40 vol %;()sizing was removed from fibers through multiple acetone washes; (4) preform was coated with BN deposited via chemical vapor deposition from gas mixtures (a) containing BCl3, NH,, and H2 at a temperature of 1100C and a reactor pressure of 5 kPa for 2 h, which resulted in an optimum coating thickness of -0.42 um [10]; and(5)the preforms were densified with SiC using the forced-flot thermal-gradient chemical vapor infiltration(FCVi) process which has been described in detail previously [11]. The final composite was 85-90% of theoretical density. Two slightly different Nicalon"/BN/SiC composites, both prepared as described above, were compared in this study; one compo- site had a BN interface coating with an oxygen content during oxidation >ll at. %(high-O BN) and the other composite had a BN Remove these interface coating with <2 at. oxygen(low-O BN). The composites with the high-O BN were made 2 years before the composites with the low-O BN. The high oxygen content in the bn was attributed to outgassing of moisture in the FCVI furnace. The furnace was completely cleaned and, shortly thereafter, BN having extremely low oxygen contents was successfully grown In order to determine the short- and long-term chemical (c)TEM preparation after oxidation and structural stability of composites with different BN interfacial coatings, both composite samples (low-O and Fig 1. Nicalon" BN SiC composite sample geometry used igh-O BN) were subjected to a series of identical, high- experiments.(a) Optical image showing the orientation of fib temperature oxidation exposures. Thin sections (sized center of 2x 2 mm sample.(b)Schematic of 200 um thick saI specifically to make TEM specimens after the exposures section illustrating preferred orientation of fibers.(c)Schem were used for this study instead of bulk pieces in order to sample as in(b) illustrating final preparation step for TEM maximize the effects of oxidation and to better simulate worst-case operating conditions at the surface of a real part stability/compatibility of any one particular BN coating, and during service in an oxidizing environment. Small pieces 2 X2 mm, were cut from each composite and mechanically the degradation mechanisms related to ground and polished to a final thickness of 200 um. Each the interface coating, it is necessary to characterize the BN coatings before and after testing/exposures using a combi specimen was also cut so that several fiber tows were nation of techniques oriented with the long axis of the fibers perpendicular to In an attempt to determine and separate various micro- the specimen surface, as shown in Fig. 1(a). In this way, structural parameters that contribute to the degradation of the ends/surfaces of both the fibers and bn coatings were fiber-reinforced composites containing BN interfacial coat- exposed during oxidation, as illustrated in the corresponding ings, two different BN interfacial coatings in Nicalon"/SiC schematic in Fig. 1(b). After oxidation, each exposed composites were evaluated for oxidative stability. The bn sample was dimpled in the center of one of the end-on coatings were structurally similar and differed only in their fiber tows and ion-milled to perforation for subsequent as-processed oxygen contents. The chemistry and micro- microstructural examination of the interface regions in structures of each BN coating were ully characterized before cross-section using TEN 1(c)). The final TEM and after oxidation using several analytical techniques specimen preparation pling and ion-milling Kinetics of the interface reactions have been calculated and also remove surface ox formed during the the microstructural results are being used to interpret the exposures The 200 um thick specimens were exposed to dry oxygen mechanical properties determined for each composite flowing at 100 ml/min in a compact reaction chamber specifically designed for reacting TEM specimens [121 2. Experimental procedure Reactions were conducted on two composite specimens simultaneously, one with the low-O BN and one with the Composites for this study were fabricated as follows: (1) high-O BN, at temperatures of 425, 600 and 950C for time
stability/compatibility of any one particular BN coating, and in order to elucidate the degradation mechanisms related to the interface coating, it is necessary to characterize the BN coatings before and after testing/exposures using a combination of techniques. In an attempt to determine and separate various microstructural parameters that contribute to the degradation of fiber-reinforced composites containing BN interfacial coatings, two different BN interfacial coatings in Nicalone/SiC composites were evaluated for oxidative stability. The BN coatings were structurally similar and differed only in their as-processed oxygen contents. The chemistry and microstructures of each BN coating were fully characterized before and after oxidation using several analytical techniques. Kinetics of the interface reactions have been calculated and the microstructural results are being used to interpret the mechanical properties determined for each composite. 2. Experimental procedure Composites for this study were fabricated as follows: (1) multiple layers of ceramic-grade Nicalone plain-weave fabric were stacked and rotated in a 0 ^ 308 sequence in a graphite holder; (2) layers were compressed in the holder cavity to produce a preform having a nominal fiber loading of ,40 vol.%; (3) sizing was removed from fibers through multiple acetone washes; (4) preform was coated with BN deposited via chemical vapor deposition from gas mixtures containing BCl3, NH3, and H2 at a temperature of 11008C and a reactor pressure of 5 kPa for 2 h, which resulted in an optimum coating thickness of ,0.42 mm [10]; and (5) the preforms were densified with SiC using the forced-flow, thermal-gradient chemical vapor infiltration (FCVI) process which has been described in detail previously [11]. The final composite was 85–90% of theoretical density. Two slightly different Nicalone/BN/SiC composites, both prepared as described above, were compared in this study; one composite had a BN interface coating with an oxygen content .11 at.% (high-O BN) and the other composite had a BN interface coating with ,2 at.% oxygen (low-O BN). The composites with the high-O BN were made 2 years before the composites with the low-O BN. The high oxygen content in the BN was attributed to outgassing of moisture in the FCVI furnace. The furnace was completely cleaned and, shortly thereafter, BN having extremely low oxygen contents was successfully grown. In order to determine the short- and long-term chemical and structural stability of composites with different BN interfacial coatings, both composite samples (low-O and high-O BN) were subjected to a series of identical, hightemperature oxidation exposures. Thin sections (sized specifically to make TEM specimens after the exposures) were used for this study instead of bulk pieces in order to ‘maximize’ the effects of oxidation and to better simulate worst-case operating conditions at the surface of a real part during service in an oxidizing environment. Small pieces, 2 × 2 mm, were cut from each composite and mechanically ground and polished to a final thickness of , 200 mm. Each specimen was also cut so that several fiber tows were oriented with the long axis of the fibers perpendicular to the specimen surface, as shown in Fig. 1(a). In this way, the ends/surfaces of both the fibers and BN coatings were exposed during oxidation, as illustrated in the corresponding schematic in Fig. 1(b). After oxidation, each exposed sample was dimpled in the center of one of the end-on fiber tows and ion-milled to perforation for subsequent microstructural examination of the interface regions in cross-section using TEM (see Fig. 1(c)). The final TEM specimen preparation steps (dimpling and ion-milling) also remove surface oxides that formed during the exposures. The 200 mm thick specimens were exposed to dry oxygen flowing at 100 ml/min in a compact ‘reaction chamber’ specifically designed for reacting TEM specimens [12]. Reactions were conducted on two composite specimens simultaneously, one with the low-O BN and one with the high-O BN, at temperatures of 425, 600 and 9508C for time 464 K.L. More et al. / Composites: Part A 30 (1999) 463–470 Fig. 1. Nicalone BN SiC composite sample geometry used for oxidation experiments. (a) Optical image showing the orientation of fiber tows in center of 2 × 2 mm sample. (b) Schematic of 200 mm thick sample crosssection illustrating preferred orientation of fibers. (c) Schematic of same sample as in (b) illustrating final preparation step for TEM examination after oxidation
K.L. More et al. /Composites: Part A 30(1999)463-470 than the highly ordered graphitic BN(the ideal value of CVI SiC d (002)=3.33 A) when the 'mean d(002) value falls in between that of Bn, and Bng, the hexagonal BN structure is referred to as meso-graphitic BN. Clearly, both bn coat ings(high-O and low-O)had a meso-graphitic structure with individual bn crystallites having varying ranges of atomic order. The statistical variations between the mean d(002)value and range of d(002)values was quite different BN layer for each BN, as shown in by the plot in Fig. 4, the low-O BN had a slightly more graphitic component than the high-O BN. This effect was most pronounced near the surface of the fiber, where the highest oxygen content was found in the Nicalon 0.15 a thin silica layer(<5 nm thick) was identified between Fig. 2. TEM image of a Bn interface coatin the bn and the Nicalon fiber in both composites, as shown Fig. 5. This layer formed during the initial stages of BN deposition at 1100C, at these temperatures and in the ntervals of 1, 10, 100 and 400 h. Each specimen was presence of small amounts of O2(possibly from outgassing prepared as described above for examination in a Hitachi of moisture from walls of furnace), limited decomposition HF 2000 FEG/TEM operated at 200 kV and equipped with a of the Nicalon surface is likely. The oxygen content in the Noran energy-dispersive spectrometer(EDS)and a Gatan BN coatings was measured using both AES and EElS and parallel-detection electron energy-loss spectrometer qualitatively compared by EDS. For the AEs analyses, (PEELS) for compositional analysis of the small composite bars were fractured in situ and depth regions. Analytical TEM was used for the identification of profiles were conducted through several BN coatings the reaction products formed during each exposure and to (primarily in areas where fibers had pulled out and left document structural and chemical changes in the BN and the Bn in the trough). Auger analysis showed that the n/ Nicalon fibers as a function of exposure time end tempera- B< I in the high-O BN coatings and near stoichiometric ture Reaction kinetics associated with the isothermal expo- in the low-O BN coatings. The high-O BN had significant sures were evaluated by measuring the thicknesses of the amounts of oxygen, ranging from 10 to 14 at. % and also reaction products formed within the interfacial region silicon(2-5 at. %) The low-O BN had very low levels of oxygen; the oxygen content was al ways less than 2 at. % and no silicon was found. The oxygen content in the high- 3. Results and discussion O BN was much greater(2x) close to the fiber surface (within -0. 1 um)in the as-processed state The bn coatings in both composites(low-O and high-O A series of TEM images comparing the interface mi BN)were-0. 4 um thick and were structurally similar. A structure observed after oxidation at 950c for the low-o Fig. 2 and representative hign -resolution hg is shown in BN and the high-O BN are shown in Fig. 6(a)-(d).Afte shown in Fig 3. The BN was hexagonal and nano-crystal- he BN-Nicalon interfaces in the high-O BN composite, line in both composites, with a nominal crystallite size of whereas no changes were observed for the low-O BN <5 nm. The degree of order in the hexagonal Bn lattice composites after 100 h(compare Fig. 6(a) and Fig. 6(b)) (graphitization index) can be determined by measuring the More severe degradation occurred at the BN-Nicalon lattice interlayer spacing, d(002)[13]. These measurements interfaces in the high-O BN composite than in the low-O are most often accomplished on bulk BN using X-ray BN composite after oxidation for 400 h, as shown in Fi diffraction(XRD). Since the amount of BN present in the 6(c),(d). A stable 25 nm thick Sio2 glass layer was formed actual composite samples used in this work was extremely after 400 h at the BN-Nicalon interfaces in the low-O small and the crystallite size was also very small, XRD BN. Two glass phases formed during the 400 h oxidation could not be used. Instead, d(002) values were measured at the BN-Nicaloninterface in the high-O BN; the initial m individual crystallites in high-resolution TEM images, SiO2 glass(labeled'S in Fig. 6(d)), the thickness of which such as those labeled in Fig 3. Area ' in Fig 3 represents appeared to remain stable from 100 to 400 h, and a second a crystallite of turbostratic BN (characterized by havin glass which formed between the SiO2 and the bn and was only two-dimensional ordering and a rotation of the layers), found by eELs to be a borosilicate glass( the exact glass BN, where the measured value of d(002)= 3.63 A, and composition is unknown-labeled B in Fig. 6(d)).The Area'B' represents a crystallite of graphitic BN (highly most drastic change at the interface, however, was the ordered in three dimensions), BNg, where d(002 formation of voids within the borosilicate glass phase at a 3.40A. Turbostratic BN will have a larger value of d(002) majority of the interfaces in the high-o BN. These voids
intervals of 1, 10, 100 and 400 h. Each specimen was prepared as described above for examination in a Hitachi HF 2000 FEG/TEM operated at 200 kV and equipped with a Noran energy-dispersive spectrometer (EDS) and a Gatan parallel-detection electron energy-loss spectrometer (PEELS) for compositional analysis of the interfacial regions. Analytical TEM was used for the identification of the reaction products formed during each exposure and to document structural and chemical changes in the BN and the Nicalone fibers as a function of exposure time end temperature. Reaction kinetics associated with the isothermal exposures were evaluated by measuring the thicknesses of the reaction products formed within the interfacial region. 3. Results and discussion The BN coatings in both composites (low-O and high-O BN) were ,0.4 mm thick and were structurally similar. A TEM image of a typical BN interface coating is shown in Fig. 2 and representative high-resolution TEM images are shown in Fig. 3. The BN was hexagonal and nano-crystalline in both composites, with a nominal crystallite size of ,5 nm. The degree of order in the hexagonal BN lattice (graphitization index) can be determined by measuring the lattice interlayer spacing, d(002) [13]. These measurements are most often accomplished on bulk BN using X-ray diffraction (XRD). Since the amount of BN present in the actual composite samples used in this work was extremely small and the crystallite size was also very small, XRD could not be used. Instead, d(002) values were measured from individual crystallites in high-resolution TEM images, such as those labeled in Fig. 3. Area ‘A’ in Fig. 3 represents a crystallite of turbostratic BN (characterized by having only two-dimensional ordering and a rotation of the layers), BNt, where the measured value of d(002) 3.63 A˚ , and Area ‘B’ represents a crystallite of graphitic BN (highly ordered in three dimensions), BNg, where d(002) 3.40 A˚ . Turbostratic BN will have a larger value of d(002) than the highly ordered graphitic BN (the ideal value of d(002) 3.33 A˚ ) when the ‘mean’ d(002) value falls in between that of BNt and BNg, the hexagonal BN structure is referred to as meso-graphitic BN. Clearly, both BN coatings (high-O and low-O) had a meso-graphitic structure with individual BN crystallites having varying ranges of atomic order. The statistical variations between the mean d(002) value and range of d(002) values was quite different for each BN, as shown in by the plot in Fig. 4; the low-O BN had a slightly more graphitic component than the high-O BN. This effect was most pronounced near the surface of the fiber, where the highest oxygen content was found in the high-O BN. A thin silica layer (,5 nm thick) was identified between the BN and the Nicalone fiber in both composites, as shown in Fig. 5. This layer formed during the initial stages of BN deposition at 11008C; at these temperatures and in the presence of small amounts of O2 (possibly from outgassing of moisture from walls of furnace), limited decomposition of the Nicalon surface is likely. The oxygen content in the BN coatings was measured using both AES and EELS and qualitatively compared by EDS. For the AES analyses, small composite bars were fractured in situ and depth profiles were conducted through several BN coatings (primarily in areas where fibers had pulled out and left BN in the ‘trough’). Auger analysis showed that the N/ B , 1 in the high-O BN coatings and near stoichiometric in the low-O BN coatings. The high-O BN had significant amounts of oxygen, ranging from 10 to 14 at.%, and also silicon (2–5 at.%). The low-O BN had very low levels of oxygen; the oxygen content was always less than 2 at.%, and no silicon was found. The oxygen content in the highO BN was much greater (2×) close to the fiber surface (within ,0.1 mm) in the as-processed state. A series of TEM images comparing the interface microstructure observed after oxidation at 9508C for the low-O BN and the high-O BN are shown in Fig. 6(a)–(d). After 100 h at 9508C, a 20 nm thick silica layer was identified at the BN–Nicalone interfaces in the high-O BN composite, whereas no changes were observed for the low-O BN composites after 100 h (compare Fig. 6(a) and Fig. 6(b)). More severe degradation occurred at the BN–Nicalone interfaces in the high-O BN composite than in the low-O BN composite after oxidation for 400 h, as shown in Fig. 6(c),(d). A stable 25 nm thick SiO2 glass layer was formed after 400 h at the BN–Nicalone interfaces in the low-O BN. Two glass phases formed during the 400 h oxidation at the BN–Nicalone interface in the high-O BN; the initial SiO2 glass (labeled ‘S’ in Fig. 6(d)), the thickness of which appeared to remain stable from 100 to 400 h, and a second glass which formed between the SiO2 and the BN and was found by EELS to be a borosilicate glass (the exact glass composition is unknown—labeled ‘B’ in Fig. 6(d)). The most drastic change at the interface, however, was the formation of voids within the borosilicate glass phase at a majority of the interfaces in the high-O BN. These voids K.L. More et al. / Composites: Part A 30 (1999) 463–470 465 Fig. 2. TEM image of a BN interface coating in Nicalone–BN–SiC composite
K.L. More et al./Composites: Part 4 30(1999)463-470 守 282g82<日2乙35日028
466 K.L. More et al. / Composites: Part A 30 (1999) 463–470 Fig. 3. High-resolution TEM images of crystalline structure of individual BN grains in coating (‘A’ is turbostratic BN and ‘B’ is meso-graphitic BN)
K.L. More et al. /Composites: Part 4 30(1999)463-470 oxygen in the Nicalon-BN system is restricted, SiO2 will form before either B2O3 or a borosilicate glass at the interfaces. When a composite is subjected to a mechanical load, microcracking occurs and oxygen will rapidly diffuse to many interior interfaces in the composite. However, this will also result in rapid oxidation of the exposed surfaces causing crack closure, and the amount of oxygen that actu- 3.7 ally reaches the majority of interfaces will still be limited The conditions described by Sheldon et al. are being simu- lated in our experiments by using thin, unstressed samples he amount of oxygen reaching the Nicalon-BN inter- faces is limited to the oxygen that can reach the interfaces either from the environment, which is controlled by the rate of diffusion of oxygen through BN via pipe/grain boundary 3.3 diffusion, or by the diffusion of oxygen already present in High-O BN LOW-O BN the bn to the Nicalon-BN interface. Clearly, the diffu- sion of oxygen that already exists within the Bn lattice will Fig. 4. Statistical variation in d(002)between high-O and low-O BN inter- contribute more significantly to the formation of reaction face coatings. products, particularly when no stresses(causing micro- ks)are present, as the experimen nce given In were elongated in shape and were found around the entire this paper shows. Sheldon et al. also predicted the formation circumference of more than 50% of the fibers within the of borosilicate solid solutions in the later stages of the same single tow examined by TEM. When the oxidation was oxidation process after the formation of a Sio2 layer. A carried out at the intermediate temperature of 600"C for SiO2-B2O3 glass layer does form in the high-O BN after 00 h, interface reaction products were also observed for oxidation exposures at 950C for times exceeding 100 h In the high-O BN composites only, as shown in Fig. 7 fact, a borosilicate glass layer also forms after long exposure (compare to Fig. 6(d)). At 600C, a similar glass phase times at 600C(see Fig. 6(d)and Fig. 7). Voids that form segregation also occurred (labeled'S and 'B in Fig. 7) within the borosilicate glass layer are due to the trapping and the onset of void formation between the silica and boro- of volatile by-products, such as N, BO, etc, resulting from silicate glasses was observed after the 400 h exposure. The the oxidation of Bn to form B2O3 and reaction of the B2O3 voids here were much smaller in size but also more numer with SiO, to form the borosilicate glass. Voids can form in ous. No silica formed in the low-O BN composites at the this layer since the viscosity of the silicate glass is signifi BN-Nicalon interfaces at 600C; the BN-Nicalon cantly reduced due to the presence of boron. The SiO2-B2O interfaces in this sample remained stable at 600C, at least phase diagram [17 shows that, at 950C, even small after 400 h exposures amounts of boron can result in a glass phase with reduced The formation of a silica layer between BN and ceramic- viscosity and lower melting point. At 600C, the amount of grade Nicalon fibers has been observed by other research- boron must be higher to cause the same results. Of course ers after oxidation [6, 14, 15]. Sheldon et al. 16] used ther- the presence of a reduced-viscosity, low-melting point glass modynamic calculations to show that if the amount of at the interfaces will be detrimental to the mechanical of the Oxygen can 'existin hexagonal BN in several different forms; as B2O3, as a borosilicate glass (when silicon is present), and as a BN O, phase, or as a combination of these phases. Brozek and Hubacek [18] suggested that free B2O3 is present in crystalline BN in addition to oxygen in constitutionally bonded water present at the surfaces of eSiO individual crystallites. Evidence for the presence of a BNO, hase in CVD BN was given by Guimon et al. [19] . X-ray electron spectroscopy (XPS) was used to show that signifi cant amounts of oxygen were present in the bn and the b films were nonstoichiometric with n/b< XPs results 10 nm indicated the possibility of a metastable ternary ph BN,OY identified from a boron binding energy intermediate between pure BN and B2O,. It was suggested that this phase Fig. 5. TEM image showing thin silica layer (<5 nm) present at BN- resulted from the substitution of oxygen for nitrogen on Nicalon interface after composite processing. the bn lattice. The results from the work by Guimon are
were elongated in shape and were found around the entire circumference of more than 50% of the fibers within the single tow examined by TEM. When the oxidation was carried out at the intermediate temperature of 6008C for 400 h, interface reaction products were also observed for the high-O BN composites only, as shown in Fig. 7 (compare to Fig. 6(d)). At 6008C, a similar glass phase segregation also occurred (labeled ‘S’ and ‘B’ in Fig. 7) and the onset of void formation between the silica and borosilicate glasses was observed after the 400 h exposure. The voids here were much smaller in size but also more numerous. No silica formed in the low-O BN composites at the BN–Nicalone interfaces at 6008C; the BN–Nicalone interfaces in this sample remained stable at 6008C, at least after 400 h exposures. The formation of a silica layer between BN and ceramicgrade Nicalone fibers has been observed by other researchers after oxidation [6,14,15]. Sheldon et al. [16] used thermodynamic calculations to show that if the amount of oxygen in the Nicalone–BN system is restricted, SiO2 will form before either B2O3 or a borosilicate glass at the interfaces. When a composite is subjected to a mechanical load, microcracking occurs and oxygen will rapidly diffuse to many interior interfaces in the composite. However, this will also result in rapid oxidation of the exposed surfaces causing crack closure, and the amount of oxygen that actually reaches the majority of interfaces will still be limited. The conditions described by Sheldon et al. are being simulated in our experiments by using thin, unstressed samples; the amount of oxygen reaching the Nicalone–BN interfaces is limited to the oxygen that can reach the interfaces either from the environment, which is controlled by the rate of diffusion of oxygen through BN via pipe/grain boundary diffusion, or by the diffusion of oxygen already present in the BN to the Nicalone–BN interface. Clearly, the diffusion of oxygen that already exists within the BN lattice will contribute more significantly to the formation of reaction products, particularly when no stresses (causing microcracks) are present, as the experimental evidence given in this paper shows. Sheldon et al. also predicted the formation of borosilicate solid solutions in the later stages of the same oxidation process after the formation of a SiO2 layer. A SiO2–B2O3 glass layer does form in the high-O BN after oxidation exposures at 9508C for times exceeding 100 h. In fact, a borosilicate glass layer also forms after long exposure times at 6008C (see Fig. 6(d) and Fig. 7). Voids that form within the borosilicate glass layer are due to the ‘trapping’ of volatile by-products, such as N, BO, etc., resulting from the oxidation of BN to form B2O3 and reaction of the B2O3 with SiO2 to form the borosilicate glass. Voids can form in this layer since the viscosity of the silicate glass is signifi- cantly reduced due to the presence of boron. The SiO2–B2O3 phase diagram [17] shows that, at 9508C, even small amounts of boron can result in a glass phase with reduced viscosity and lower melting point. At 6008C, the amount of boron must be higher to cause the same results. Of course, the presence of a reduced-viscosity, low-melting point glass at the interfaces will be detrimental to the mechanical properties of the composite. Oxygen can ‘exist’ in hexagonal BN in several different forms; as B2O3, as a borosilicate glass (when silicon is present), and as a BNxOy phase, or as a combination of these phases. Brozek and Hubacek [18] suggested that free B2O3 is present in crystalline BN in addition to oxygen in constitutionally bonded water present at the surfaces of individual crystallites. Evidence for the presence of a BNxOy phase in CVD BN was given by Guimon et al. [19]. X-ray electron spectroscopy (XPS) was used to show that signifi- cant amounts of oxygen were present in the BN and the BN films were nonstoichiometric with N/B , 1. XPS results indicated the possibility of a metastable ternary phase, BNxOy, identified from a boron binding energy intermediate between pure BN and B2O3. It was suggested that this phase resulted from the substitution of oxygen for nitrogen on the BN lattice. The results from the work by Guimon are K.L. More et al. / Composites: Part A 30 (1999) 463–470 467 Fig. 4. Statistical variation in d(002) between high-O and low-O BN interface coatings. Fig. 5. TEM image showing thin silica layer (,5 nm) present at BN– Nicalone interface after composite processing
K.L. More et al. /Composites: Part A 30 (1999)463-470 SIoF 10 nm Sio 25 nm fiber 25 nm Fig. 6. TEM images showing the microstructural evolution at Nicalon"-BN interfaces after o at 950C: (a) low-o Bn after 100 h; (b) high-O Bn after 100 h;(c) low-O BN after 400 h;(d) high-O BN after 400 h(silica layer labeled"S and borosilicate glass layer labeledB) &Onsistent with the results found in our work for the high-o BN-Nicalon" interface from the high-O BN after 400h BN coatings: the bn d(002)spacings measured indicated a at 950C in argon is shown in Fig 8. There was no reaction meso-graphitic structure(more turbostratic than the low-o at the BN-nicalon interfaces for the low-O Bn after BN); the coatings had oxygen levels greater than 11 at 400 h in Ar; however, the interface reactions for the high (higher close to the Nicalon surface); the high-O BN coat- O BN were very similar to those observed for the high-O BN ings had a N/B <l. Thus, there is clear evidence for the oxidized for 400 h at 950C (compare Fig 8 and Fig. 6(d)) presence of a similar metastable ternary BN,O, phase in the The role of the oxygen within the BN, therefore, was clearly high-o bn coatings. Auger results also indicated that the silicon found within the high-O BN coating was bonded with oxygen indicating the possibility of a small amount of"free borosilicate glass phase surrounding the bn grains Oxygen is clearly dissolved in the bn structure regardless of the form the oxygen takes. The thermodynamic driving force is for the oxygen to move from within the BN coating to the bn-nicalon interface where reactions with sic will occur. The presence of the dissolved oxygen and oxygen- containing phases in the high-O BN will certainly contribute to the formation of the initial SiO, and subsequent sio B,O3 observed in the later stages of oxidation In order to fully evaluate the role of the oxygen within the bn (as opposed to the contribution of oxygen from the n environment), the composite samples were also annealed in argon under the same time and temperature conditions Fig. 7. TEM image of Nicalon-BN interface in high-o bn after 400 h at used for the oxidation experiments. A TEM image of a 600C (silica layer labeled'S'and borosilicate glass layer labeledB)
consistent with the results found in our work for the high-O BN coatings: the BN d(002) spacings measured indicated a meso-graphitic structure (more turbostratic than the low-O BN); the coatings had oxygen levels greater than 11 at.% (higher close to the Nicalon surface); the high-O BN coatings had a N/B , 1. Thus, there is clear evidence for the presence of a similar metastable ternary BNxOy phase in the high-O BN coatings. Auger results also indicated that the silicon found within the high-O BN coating was bonded with oxygen indicating the possibility of a small amount of ‘free’ borosilicate glass phase surrounding the BN grains. Oxygen is clearly dissolved in the BN structure regardless of the form the oxygen takes. The thermodynamic driving force is for the oxygen to move from within the BN coating to the BN–Nicalon interface where reactions with SiC will occur. The presence of the dissolved oxygen and oxygencontaining phases in the high-O BN will certainly contribute to the formation of the initial SiO2 and subsequent SiO2– B2O3 observed in the later stages of oxidation. In order to fully evaluate the role of the oxygen within the BN (as opposed to the contribution of oxygen from the environment), the composite samples were also annealed in argon under the same time and temperature conditions used for the oxidation experiments. A TEM image of a BN–Nicalone interface from the high-O BN after 400 h at 9508C in argon is shown in Fig. 8. There was no reaction at the BN–Nicalone interfaces for the low-O BN after 400 h in Ar; however, the interface reactions for the highO BN were very similar to those observed for the high-O BN oxidized for 400 h at 9508C (compare Fig. 8 and Fig. 6(d)). The role of the oxygen within the BN, therefore, was clearly 468 K.L. More et al. / Composites: Part A 30 (1999) 463–470 Fig. 6. TEM images showing the microstructural evolution at Nicalone–BN interfaces after oxidation at 9508C: (a) low-O BN after 100 h; (b) high-O BN after 100 h; (c) low-O BN after 400 h; (d) high-O BN after 400 h (silica layer labeled ‘S’ and borosilicate glass layer labeled ‘B’). Fig. 7. TEM image of Nicalone–BN interface in high-O BN after 400 h at 6008C (silica layer labeled ‘S’ and borosilicate glass layer labeled ‘B’)
K.L. More et al. /Composites: Part A 30(1999)463-470 amples clearly indicate that the interface properties BN degrade significantly with increased oxygen impurity content in the bn and this effect should only intensify with the appli of formation of the silica on the surface of the Nicalon fiber) is not only limited to strength of the fiber; oxidized Nicalon fibers have signifi cantly reduced strengths as compared to the as-processed fiber. The chemistry of the fiber surface changes drastically and localized stress concentrations are formed at the fibe fiber 25m surface, changing their fracture strength and behavior. The oxygen, in essence, attacks the fiber surface causing fiber degradation. The fiber coating should protect the fiber Fig8. TEM image of Nicalon-BN interface in high-O BN after 400 h at surface from degradation and, clearly, the high-O BN is a 950° C in argon. source of oxygen which acts to degrade the fiber properties the more significant contributor to the BN-Nicalon"inter- face degradation, as opposed to the minimal role of oxygen.Conclusions diffused from the environment Mechanical properties evaluation of the two composite Microstructural characterization was conducte systems compared in this study are currently ongoing, but lon"/SiC composite materials having either a low some preliminary results are very encouraging. The graphs a high-O BN interfacial coating. The fiber-matrix shown in Fig. 9 are a comparison of the stress rupture beha- in these composites were fully evaluated before vior in air at 600C for the Nicalon/SiC composites with oxidation in order to determine the role of oxygen present low-O BN and high-O BN. These results are also compared within the as-processed BN in the degradation of the to earlier data obtained under similar conditions for Nica- composite's mechanical properties. Oxidation at the BN lon SiC with a carbon interface. Both the BN-containing Nicalon interfaces in the high-O bn progresses as SiC/SiC composites showed improved mechanical proper- follows: a SiO2 layer forms between the Nicalon fibe ties over the composites with a carbon interface, but more and the Bn; after the silica layer reaches a stable thickness, importantly, the low-O BN SiC/SiC consistently showed progressive internal oxidation of the Bn leads to the forma- superior performance at all stress levels when compared tion of an additional SiO2-B2O3 glass layer; gaseous reac with the high-O BN SiC/SiC composites at 600C. The tion by-products are accommodated by void formation interfaces in these samples have not been characterized, within the reduced-viscosity, lower-melting point boro- but the microstructural changes observed in the unstressed silicate glass. The interfaces in the high-O BN showed 1000 F6oPcint ar xure G100 CVI811/C (600C)- C-interfac V1618/BN (600 C)- High-O BN CVI1074/BN (600C)- LOw-O BN 0.0010.010.1 01001000104 Time to Failure, h Fig 9. Stress rupture behavior at 600C for both high-O and low-O BN-containing composites compared to similar composites having a carbon interfacial coating. Note that the low-O BN SiC/SiC composites consistently showed superior performance at all stress levels
the more significant contributor to the BN–Nicalone interface degradation, as opposed to the minimal role of oxygen diffused from the environment. Mechanical properties evaluation of the two composite systems compared in this study are currently ongoing, but some preliminary results are very encouraging. The graphs shown in Fig. 9 are a comparison of the stress rupture behavior in air at 6008C for the Nicalon/SiC composites with low-O BN and high-O BN. These results are also compared to earlier data obtained under similar conditions for Nicalone/SiC with a carbon interface. Both the BN-containing SiC/SiC composites showed improved mechanical properties over the composites with a carbon interface, but more importantly, the low-O BN SiC/SiC consistently showed superior performance at all stress levels when compared with the high-O BN SiC/SiC composites at 6008C. The interfaces in these samples have not been characterized, but the microstructural changes observed in the unstressed samples clearly indicate that the interface properties degrade significantly with increased oxygen impurity content in the BN and this effect should only intensify with the application of stress. Of course, the interface degradation (especially the formation of the silica on the surface of the Nicalon fiber) is not only limited to the ‘interface’ but clearly effects the strength of the fiber; oxidized Nicalon fibers have signifi- cantly reduced strengths as compared to the as-processed fiber. The chemistry of the fiber surface changes drastically and localized stress concentrations are formed at the fiber surface, changing their fracture strength and behavior. The oxygen, in essence, attacks the fiber surface causing fiber degradation. The fiber coating should protect the fiber surface from degradation and, clearly, the high-O BN is a source of oxygen which acts to degrade the fiber properties. 4. Conclusions Microstructural characterization was conducted on Nicalone/SiC composite materials having either a low-O BN or a high-O BN interfacial coating. The fiber–matrix interfaces in these composites were fully evaluated before and after oxidation in order to determine the role of oxygen present within the as-processed BN in the degradation of the composite’s mechanical properties. Oxidation at the BN– Nicalone interfaces in the high-O BN progresses as follows: a SiO2 layer forms between the Nicalone fiber and the BN; after the silica layer reaches a stable thickness, progressive internal oxidation of the BN leads to the formation of an additional SiO2–B2O3 glass layer; gaseous reaction by-products are accommodated by void formation within the reduced-viscosity, lower-melting point borosilicate glass. The interfaces in the high-O BN showed K.L. More et al. / Composites: Part A 30 (1999) 463–470 469 Fig. 9. Stress rupture behavior at 6008C for both high-O and low-O BN-containing composites compared to similar composites having a carbon interfacial coating. Note that the low-O BN SiC/SiC composites consistently showed superior performance at all stress levels. Fig. 8. TEM image of Nicalone–BN interface in high-O BN after 400 h at 9508C in argon
K.L. More et al. /Composites: Part A 30(1999)463-470 milar degradation, but to a lesser extent, at 600C. [4] Filipuzzi L, Camus G, Naslain R, Thebault J Oxidation mechanisms Even at elevated temperatures, very little oxygen from and kinetics of lD-SiC/C/SiC composite materials: I. An experimen the environment actually makes its way to the interior tal approach. J Am Ceramic Soc 1994: 77: 459-466. of the composite via the BN; most of the degradation at 5 Lowden RA, Schwarz OJ, More KL. Ceramic Eng Sci Proc 1993;14(7-8)375-384 the interfaces is caused by the oxygen already incorpo [6 Naslain R, Dugne O, Guette A. Boron nitride interphase in ceramic- rated in the Bn lattice during processing. The interfaces matrix composites. J Am Ceramic Soc 1991: 74(1): 2482-2488. in the low-O BN were not degraded as much as those in [7] Matsuda T Stability to moisture for chemically vapour deposited BN the high-O BN, although some formation of silica was Mater sci198924:2353-2358 observed after 400 h at 950C in the low-O BN. Clearly [8 Baskaran $, Halloran Jw. Fibrous monolithic ceramics: Ill, mechan- ical properties and oxidation behavior of the SiC/BN system. J Am he main conclusion to be drawn from this work is that Ceramic Soc199477(5):1249-1255 relatively protective BN interface coatings can be [9 Morscher GN. Tensile stress rupture of SiC/SiCm Minicomposites achieved when the oxygen content is very low carbon and Bn interphases at elevated temperatures in air. J Am Ceramic Soc 1997: 80(8): 2029-2042 [10 Lowden RA, More KL, Schwarz OJ BN interface coatings for Nica- Acknowledgements lon/SiC composites(in preparation [11 Besmann TM, Lowden RA, Stinton DP, Starr TL. A method for the Research sponsored by the US Department of Energy id chemical vapor infiltration of ceramic composites. J Phy 9895(50( Colloque C5):229-239 Assistant Secretary for Energy Efficiency and Renewable [12] Allard LF, Ailey KS, Datye AK, Bigelow WC. An ex-situ reactor with Energy, the Office of Transportation Technologies, as pa anaerobic specimen transfer capabilities for TEM studies of reactive of the High Temperature Materials Laboratory Fellowship pecimens. Microscopy Microanalysis 1997: 3(2):595-5 program, and the Office of Industrial Technologies, Indus- [3] Thomas ), Weston NE, Connor TE. Turbostratic boron nitride, ther- trial Energy Efficiency Division, as part of the Continuous Fiber Ceramic Composites Program, both managed by Soc1963;84(24):4619-4622. 14 Sun EY, Nutt SR, Brennan JJ. Interfacial microstructure and chem- ockheed Martin Energy Research Corp. under contract try of Sic/Bn dual-coated nicalon-fiber-reinforced glass-ceramic Imber DE-ACo5-96OR22464 matrix composites. J Am Ceramic Soc 1994, 77(6): 1329-1339 15 Prouhet S, Camus G, Labrugere C, Guette A Mechanical character ization of Sic(O) fiber/SiC(CVI)matrix composites with a BN-inter References ase. J Am Ceramic Soc 1994: 77(3): 649-656 16 Sheldon BW, Sun EY, Nutt SR, Brennan JJ. Oxidation of BN-coated [1 Mah T, Mendiratta MG, Katz AP, Mazdiyasni Ks. Ceramic Bull SiC fibers in ceramic matrix composites. J Am Ceramic Soc 1987;66:304-308 996;792):539-543 [2] Lehman RL. In: Wachtman JB, editors. Treatise on materials science 17 Rocket TJ, Foster WR. J Am Ceramic Soc 1965: 48(2): 78 and technology. Structural ceramics, vol. 29. New York: Academic [18] Brozek V, Hubacek M. A contribution to the crystallochemistry of ss1989 BN. J Solid State Chem 1992: 100: 120-129 [3]Lowden RA Fiber coatings and the mechanical properties of a fiber- [19] Guimon C, Gonbeau D, Pfister-Guillouzo G, Dugne O, Guette A einforced ceramic composite. In: Sacks MD, editor. Ceramic trans- Naslain R, Lahaye M. XPS study of BN thin films deposited by ctions-advanced composite materials, vol. 19. Westerville, OH: CVD on Sic plane substrates. Surf Interface Analysis 1990; 16 The American Ceramic Society, 1991: 619-630 440-445
similar degradation, but to a lesser extent, at 6008C. Even at elevated temperatures, very little oxygen from the environment actually makes its way to the interior of the composite via the BN; most of the degradation at the interfaces is caused by the oxygen already incorporated in the BN lattice during processing. The interfaces in the low-O BN were not degraded as much as those in the high-O BN, although some formation of silica was observed after 400 h at 9508C in the low-O BN. Clearly, the main conclusion to be drawn from this work is that relatively protective BN interface coatings can be achieved when the oxygen content is very low. Acknowledgements Research sponsored by the US Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, the Office of Transportation Technologies, as part of the High Temperature Materials Laboratory Fellowship program, and the Office of Industrial Technologies, Industrial Energy Efficiency Division, as part of the Continuous Fiber Ceramic Composites Program, both managed by Lockheed Martin Energy Research Corp. under contract number DE-AC05-96OR22464. References [1] Mah T, Mendiratta MG, Katz AP, Mazdiyasni KS. Ceramic Bull 1987;66:304–308. [2] Lehman RL. In: Wachtman JB, editors. Treatise on materials science and technology. Structural ceramics, vol. 29. New York: Academic Press, 1989. [3] Lowden RA. Fiber coatings and the mechanical properties of a fiberreinforced ceramic composite. In: Sacks MD, editor. Ceramic transactions—advanced composite materials, vol. 19. Westerville, OH: The American Ceramic Society, 1991:619–630. [4] Filipuzzi L, Camus G, Naslain R, Thebault J. Oxidation mechanisms and kinetics of 1D-SiC/C/SiC composite materials: I. An experimental approach. J Am Ceramic Soc 1994;77:459–466. [5] Lowden RA, Schwarz OJ, More KL. Ceramic Eng Sci Proc 1993;14(7–8):375–384. [6] Naslain R, Dugne O, Guette A. Boron nitride interphase in ceramic– matrix composites. J Am Ceramic Soc 1991;74(1):2482–2488. [7] Matsuda T. Stability to moisture for chemically vapour deposited BN. J Mater Sci 1989;24:2353–2358. [8] Baskaran S, Halloran JW. Fibrous monolithic ceramics: III, mechanical properties and oxidation behavior of the SiC/BN system. J Am Ceramic Soc 1994;77(5):1249–1255. [9] Morscher GN. Tensile stress rupture of SiCf/SiCm Minicomposites with carbon and BN interphases at elevated temperatures in air. J Am Ceramic Soc 1997;80(8):2029–2042. [10] Lowden RA, More KL, Schwarz OJ. BN interface coatings for Nicalon/SiC composites (in preparation). [11] Besmann TM, Lowden RA, Stinton DP, Starr TL. A method for the rapid chemical vapor infiltration of ceramic composites. J Phys 1989;5(50 (Colloque C5)):229–239. [12] Allard LF, Ailey KS, Datye AK, Bigelow WC. An ex-situ reactor with anaerobic specimen transfer capabilities for TEM studies of reactive specimens. Microscopy Microanalysis 1997;3(2):595–596. [13] Thomas J, Weston NE, Connor TE. Turbostratic boron nitride, thermal transformation to ordered-layer-lattice boron nitride. J Am Chem Soc 1963;84(24):4619–4622. [14] Sun EY, Nutt SR, Brennan JJ. Interfacial microstructure and chemistry of SiC/BN dual-coated nicalon–fiber-reinforced glass–ceramic matrix composites. J Am Ceramic Soc 1994;77(6):1329–1339. [15] Prouhet S, Camus G, Labrugere C, Guette A. Mechanical characterization of SiC(O) fiber/SiC (CVI) matrix composites with a BN-interphase. J Am Ceramic Soc 1994;77(3):649–656. [16] Sheldon BW, Sun EY, Nutt SR, Brennan JJ. Oxidation of BN-coated SiC fibers in ceramic matrix composites. J Am Ceramic Soc 1996;79(2):539–543. [17] Rocket TJ, Foster WR. J Am Ceramic Soc 1965;48(2):78. [18] Brozek V, Hubacek M. A contribution to the crystallochemistry of BN. J Solid State Chem 1992;100:120–129. [19] Guimon C, Gonbeau D, Pfister-Guillouzo G, Dugne O, Guette A, Naslain R, Lahaye M. XPS study of BN thin films deposited by CVD on SiC plane substrates. Surf Interface Analysis 1990;16: 440–445. 470 K.L. More et al. / Composites: Part A 30 (1999) 463–470