J. Aum Ceram. Sac, 85 [7 1815-22(2002) ournal Developing Interfacial Carbon-Boron-Silicon Coatings for Silicon Nitride- Fiber-Reinforced Composites for Improved Oxidation Resistance Kiyoshi Sato, f Hiroki Morozumi, Osamu Funayama, Hiroshi Kaya, and Takeshi Isoda Tonen General Sekiya, Iruma-gun, Saitama 356-8500, Japan C-B-Si coatings were formed on a Sin fiber using chemical deposition(CVD). That work has demonst vapor deposition and embedded in a Si-N-C matrix using oxidation resistance on single-filament layer was anticipated to form borosilicate glass containing only-a microcomposite oxygen-diffusion s. Two types of C-B-Si coatings were A Bn and oxide layer has been investigated elsewhere as a tested on the fiber-matrix interface, and they improved the ubstitute for the carbon layer. h-Bn has a layered crystal multilayered:a crystalline sublayer composed of B-Si-C was between its crystal layers. A BN layer is useful for controlling SiC-fiber-reinforced Sic and SiC-fiber-reinforced glass. 2The second coating was a graphitelike carbon layer containing a oxidation starting temperature of BN is only 100 K higher than small amount of boron and silicon. The carbon(sub)layer of both coatings weakened the fiber-matrix bonding, giving the carbon, but BN is expected to form B, O,, which seals th composites a high flexural strength(l1 GPa). The composites ygen-diffusion passes. A composite with a Bn layer shows retained 60%0-70% of their initial strength, even after oxida oxidation resistance to >1100 K, but it experiences embrittlement tion at 1523 K for 100 h. The mechanism for improved oxidation at intermediate temperatures, 900-1 100 K, because of the active resistance was discussed through the microstructure of the xidation of E nterface, morphology of the fracture surface, and oxygen An interfacial oxide coating offers the considerable advantage distribution on a cross section of the oxidized composite. of causing no oxidation degradation. However, an oxide interface is not adequate for a non-oxide fiber, because the oxide layer diffuses oxygen to the fiber. Recently, all-oxide composites . Introduction have been actively studied because of their stability under an oxidizing atmosphere. -However, among the disadvantages of F ER-REINFORCED ceramic composites show promise for over. all-oxide composites is high-temperature creep above -1500 K, which results from an intrinsic property of ordinal oxides. Thus, the bonding between fiber and matrix is important to giving a the problem of improving the oxidation resistance of ceramic composite strength and toughness, because weak bonding prevents a matrix crack from propagating to a fiber. A carbon layer has composites has remained unsolved. The present study focused on a non-oxide composite, because the carbon layer oxidizes above -700 K in air, limiting the the high strength of such a composite at high temperature improves oxidation resistance of the composite , Two approaches have oxidation resistance. A C-B-Si coating was produced using CVD been investigated as a solution to this problem:()preventin ea at oxidation of the carbon layer; and(ii)replacing the carbon lay the C-B-Si layer would form a borosilicate glass, sealing the with an oxidation-resistant layer oxygen-diffusion passes, as had the BN,and boron-containing Various studies have investigated preventing oxidation of the carbon layers, and(ii) controlling the B: Si ratio in the C-B-Si carbon layer. The addition of boron in a matrix or an interface layer would change the softening temperature and viscosity of the layer, to form a B2O3 or borosilicate-glass seal at low tempera- borosilicate glass, improving the performance of the seal. The tures, has been studied. The oxygen-diffusion passes that must be ternary system C-B-Si was selected to form a carbon-rich phase, ealed are the matrix cracks and interfacial gaps resulting from the which would cause fiber-matrix debonding. A Si-B-N layer could oxidation loss of the carbon layer. The addition of a boride powde function as C-B-Si if a h-BN sublayer was formed, but a nitride to the matrix partially prevents oxidatio coating was more difficult than a carbon coating to fabricate usi Jacques et al. lo have produced an interface layer of 20 mol% CVD. a polymer-impregnation and pyrolysis(PIP)method- boron-containing carbon on a SiC fiber using chemical vapor vas u sed in the present study to fabricate the composite. The advantage of the PIP method was that it could make use of shaping techniques from the industrial process for fiber-reinforced plastics T.A. Parthasarathry-contributing editor However, improving the oxidation resistance of PlP composites was difficult because of open pores and matrix cracks, formed by pyrolysis shrinkage of the polymer, that diffused oxygen into th composite. The present method of providing oxidation resistance Manuscript No. 188719. Received February 25, 2000; approved September 10, to PIP composites should be effective on other porous composite orted by the Ministry of Economy, Trade, and Industry, and work conducted such as reaction-bonded composites and pore-free composites such as chemical vapor infiltrated composites dence should be addressed. Now with Advanced This paper first describes the properties of a C-B-Si coating on Materials International Co, Fuji, Shizuoka 416-0946, Japan. Kanagawa 234-0192, Japan. echnical Center, Nissan Motor Co, Ltd, Atsugi, Si3N4 fiber. Results for the investigation of composites with C-B-Si interfacial layer are described in terms of the fiber-matrix Formerly Tonen Corporation interface microstructure, mechanical properties, and oxidation 1815
Developing Interfacial Carbon-Boron-Silicon Coatings for Silicon Nitride-Fiber-Reinforced Composites for Improved Oxidation Resistance Kiyoshi Sato,* ,† Hiroki Morozumi,‡ Osamu Funayama,* Hiroshi Kaya, and Takeshi Isoda Tonen General Sekiya,§ Iruma-gun, Saitama 356-8500, Japan C-B-Si coatings were formed on a Si3N4 fiber using chemical vapor deposition and embedded in a Si-N-C matrix using polymer impregnation and pyrolysis. The boron-containing layer was anticipated to form borosilicate glass and seal oxygen-diffusion passes. Two types of C-B-Si coatings were tested on the fiber–matrix interface, and they improved the oxidation resistance of the composite. The first coating was multilayered: a crystalline sublayer composed of B-Si-C was sandwiched between two graphitelike carbon sublayers. The second coating was a graphitelike carbon layer containing a small amount of boron and silicon. The carbon (sub)layer of both coatings weakened the fiber–matrix bonding, giving the composites a high flexural strength (1.1 GPa). The composites retained 60%–70% of their initial strength, even after oxidation at 1523 K for 100 h. The mechanism for improved oxidation resistance was discussed through the microstructure of the interface, morphology of the fracture surface, and oxygen distribution on a cross section of the oxidized composite. I. Introduction FIBER-REINFORCED ceramic composites show promise for overcoming the brittleness of monolithic ceramics.1–3 Control of the bonding between fiber and matrix is important to giving a composite strength and toughness, because weak bonding prevents a matrix crack from propagating to a fiber.4,5 A carbon layer has been applied at the fiber–matrix interface for this purpose,1–5 but the carbon layer oxidizes above 700 K in air, limiting the oxidation resistance of the composite.6,7 Two approaches have been investigated as a solution to this problem: (i) preventing oxidation of the carbon layer; and (ii) replacing the carbon layer with an oxidation-resistant layer. Various studies have investigated preventing oxidation of the carbon layer. The addition of boron in a matrix or an interface layer, to form a B2O3 or borosilicate-glass seal at low temperatures, has been studied. The oxygen-diffusion passes that must be sealed are the matrix cracks and interfacial gaps resulting from the oxidation loss of the carbon layer. The addition of a boride powder to the matrix partially prevents oxidation of the carbon interface.8,9 Jacques et al.10 have produced an interface layer of 20 mol% boron-containing carbon on a SiC fiber using chemical vapor deposition (CVD). That work has demonstrated the advantage of oxidation resistance on single-filament-reinforced composites only—a microcomposite. A BN and oxide layer has been investigated elsewhere as a substitute for the carbon layer. h-BN has a layered crystal structure, similar to graphite, which causes slip and debonding between its crystal layers. A BN layer is useful for controlling adhesion between the fiber and matrix in many composites, such as SiC-fiber-reinforced SiC11 and SiC-fiber-reinforced glass.12 The oxidation starting temperature of BN is only 100 K higher than carbon, but BN is expected to form B2O3, which seals the oxygen-diffusion passes. A composite with a BN layer shows oxidation resistance to 1100 K, but it experiences embrittlement at intermediate temperatures, 900–1100 K, because of the active oxidation of BN.13,14 An interfacial oxide coating offers the considerable advantage of causing no oxidation degradation.15–17 However, an oxide interface is not adequate for a non-oxide fiber, because the oxide layer diffuses oxygen to the fiber.18 Recently, all-oxide composites have been actively studied because of their stability under an oxidizing atmosphere.19–22 However, among the disadvantages of all-oxide composites is high-temperature creep above 1500 K, which results from an intrinsic property of ordinal oxides. Thus, the problem of improving the oxidation resistance of ceramic composites has remained unsolved. The present study focused on a non-oxide composite, because the high strength of such a composite at high temperature improves oxidation resistance. A C-B-Si coating was produced using CVD and applied at the fiber–matrix interface. We anticipated that (i) the C-B-Si layer would form a borosilicate glass, sealing the oxygen-diffusion passes, as had the BN11,12 and boron-containing carbon10 layers, and (ii) controlling the B:Si ratio in the C-B-Si layer would change the softening temperature and viscosity of the borosilicate glass, improving the performance of the seal. The ternary system C-B-Si was selected to form a carbon-rich phase, which would cause fiber–matrix debonding. A Si-B-N layer could function as C-B-Si if a h-BN sublayer was formed, but a nitride coating was more difficult than a carbon coating to fabricate using CVD. A polymer-impregnation and pyrolysis (PIP) method23–26 was used in the present study to fabricate the composite. The advantage of the PIP method was that it could make use of shaping techniques from the industrial process for fiber-reinforced plastics. However, improving the oxidation resistance of PIP composites was difficult because of open pores and matrix cracks, formed by pyrolysis shrinkage of the polymer, that diffused oxygen into the composite. The present method of providing oxidation resistance to PIP composites should be effective on other porous composites, such as reaction-bonded composites and pore-free composites, such as chemical vapor infiltrated composites. This paper first describes the properties of a C-B-Si coating on a Si3N4 fiber. Results for the investigation of composites with a C-B-Si interfacial layer are described in terms of the fiber–matrix interface microstructure, mechanical properties, and oxidation T. A. Parthasarathy—contributing editor Manuscript No. 188719. Received February 25, 2000; approved September 10, 2001. Supported by the Ministry of Economy, Trade, and Industry, and work conducted by the Petroleum Energy Center. *Member, American Ceramic Society. † Author to whom correspondence should be addressed. Now with Advanced Materials International Co., Fuji, Shizuoka 416-0946, Japan. ‡ Present address: Nissan Technical Center, Nissan Motor Co., Ltd., Atsugi, Kanagawa 234-0192, Japan. § Formerly Tonen Corporation. J. Am. Ceram. Soc., 85 [7] 1815–22 (2002) 1815 journal
l816 Journal of the American Ceramic Society-Sato et al. Vol. 85. No. 7 Bobbin Exhaust Precursor gas Exhaust Oiling roller chamber Deposition chamber Winder N2 Bobbin Fig. 1. Schematic drawing of the apparatus for fiber coating resistance of the composites Properties of the SiC-fiber-reinforced reinforcement and methylhydrosilazane(MHS; NN710, Tonen composite produced when a C-B-Si layer is applied to a Sic fiber Corp, Tokyo, Japan)as a matrix precursor. Details of the are evaluated fabrication process and properties of MHS are described else where 30, MHS was a random copolymer mposed of -SiH,NHI and-SiMeHNHH units and converted to an amor- IL. Experimental Procedure phous Si-N-C by pyrolysis at 1200 K in a nitrogen-gas or inert (I Preparation and Analysis of coatings atmosphere. The chemical composition of the pyrolyzed MHS was Two fibers were used as reinforcements in the present study: () (in mol%)43 silicon, 38 nitrogen, 18 carbon, and 1 oxygen, with an amorphous Si,Na fib developed by Tonen Corporation a 1: 1 ratio of -SiH,NH-ISiMeHNHh and pyrolysis was and(ii) a commercial SiC fiber(Hinicalon, Nippon Carbon C performed in nitrogen gas at 1623 K 30, Unidirectionally rein- Tokyo, Japan). The Si3 N4 fiber was a strand composed of 1000 forced composites were fabricated by the following process: () fabrication of unidirectionally fiber-aligned prepreg; (ii) stacking single filaments. The diameter of the filament was 10 um. The and curing of the prepreg sheets: (i) pyrolysis of a cured sample chemical composition of the fiber was(in mass%)60 silicon, 37 at 1623 K: and (iv) densification of the sample by seven cycles of nitrogen, <1 carbon, and <3 oxygen, The fabrication method and basic character of the fiber are described elsewhere 27,28 impregnation and pyrolysis. Details of the process are described The C-B-Si coating was formed using the CVD apparatus elsewhere 30-32 described in Fig. 1, which consisted of a strand feeder, cleaning The mechanical properties of the composites were evaluated chamber, deposition chamber, and strand winder. The cleaning a terms of flexural strength and interlaminar shear strength (ILSS)at deposition chambers were tubular fumaces with an inner diameter room temperature. The test pieces, which were cut from the of 60 mm and lengths of I and 3 m, respectively. A strand of fiber composite panels, measured 4 mm X 40 mm X 3 mm for the was fed from a bobbin to the cleaning chamber at 33 mm/s and flexural test and 4 mm x 12 mm x 2 mm for the ilss test. the heated to 1073 K under a nitrogen-gas atmosphere, where the longitudinal direction of the test pieces was aligned with the fiber sizing on the fiber thermally decomposed. The desized strand was orientation. The flexural strength was measured using the three. oated with C-B-Si by the reaction of BCl3, SiCla, CHa, and H under atmospheric pressure, in the deposition chamber. Nitroge The span and testing speed of the ILSs test were 8 mm and 8.3 was used as the carrier gas. Finally, the coated strand was resized um/s, respectively. Both tests were completed five times. The with a polyether and wound onto another bobbin in the winder volume content of fiber (e was calculated from the dimensions of The single-filament strengths of the uncoated fiber and the the composite panel and the amount of fiber used. The bulk density oated fiber were tested. according to ASTM standard. 1 at room was calculated from the weight and dimensions of the test piece temperature(25 mm gauge length, 8.3 um/s testing speed, 25 used for the flexural-strength test. The true density of the compos- tests). Cross-section areas of the filaments, needed for strength ite was measured using a pycnometer at 303 K, using n-butanol a calculations. were measured from observations of the fracture the medium, on a sample crushed under No 80 mesh. surface using scanning electron microscopy (SEM). Auger elec- For the oxidation test on the composites, the test pieces measuring tron spectroscopy (AES, Model No. PHI650, ULVAC, Ltd 4 mm X 40 mm x 3 mm were placed inside a tube fumace, heated okyo, Japan), with accelerating voltage of 3 kv and sample to a given temperature at a rate of 0.167 K/s, and maintained at that current of 3 nA, was used to examine the elemental depth profiles temperature for a given time under a dry-air flow of 74 mmol/s. Some of the coatings. The etching rate, determined using argon sputter samples were exposed at the same temperature and time, under a ing, was 0.27 nm/s, with SiO, as a standard nitrogen-gas flow of 74 as references The deposits in the deposition chamber were analyzed to The microstructure of the fiber-matrix interface was investi investigate the synthesis mechanism of the multilayered coating ated using transmission electron microscopy(TEM; Model No. Graphite sheets were placed inside the chamber, along its tubular JEM3010, JEOL) using the following techniques: (i)observation wall, before the coating operation. The graphite sheets were on a bright-field image; (ii)crystal-structure analysis by nanobeam emoved from the furnace after the operation was complete and electron diffraction (NBED)with a 5 nm beam, and (iii)elemental analyzed using electron probe microanalysis(EPMA; Model No analysis by energy dispersive spectroscopy(EDS), with a 5 nm JXA8600MX, JEOL, Tokyo, Japan) and X-ray diffractometry beam. aes depth profiles were investigated on the surfaces of a (XRD; Model No. RINT 1400, Rigaku Co, Ltd, Tokyo, Japan) ullout fiber and on the matrices from which a fiber had debonded after fracture. The measuring conditions for aes were the same as those for the coated fiber (2) Preparation and Analysis of the Composite Composite panels measuring 100 mm X 100 mm x 3 mm were fabricated using the PIP process, with the coated fiber as a IlL. Results Modulus Single-Filam Materials, ASTM Designation D-3379- Society for Testing and Materials, West Conshohocken, P. deposition conditions of the coatings are shown in Table I. A
resistance of the composites. Properties of the SiC-fiber-reinforced composite produced when a C-B-Si layer is applied to a SiC fiber are evaluated. II. Experimental Procedure (1) Preparation and Analysis of Coatings Two fibers were used as reinforcements in the present study: (i) an amorphous Si3N4 fiber27,28 developed by Tonen Corporation; and (ii) a commercial SiC fiber29 (Hinicalon, Nippon Carbon Co., Tokyo, Japan). The Si3N4 fiber was a strand composed of 1000 single filaments. The diameter of the filament was 10 m. The chemical composition of the fiber was (in mass%) 60 silicon, 37 nitrogen, 1 carbon, and 3 oxygen. The fabrication method and basic character of the fiber are described elsewhere.27,28 The C-B-Si coating was formed using the CVD apparatus described in Fig. 1, which consisted of a strand feeder, cleaning chamber, deposition chamber, and strand winder. The cleaning and deposition chambers were tubular furnaces with an inner diameter of 60 mm and lengths of 1 and 3 m, respectively. A strand of fiber was fed from a bobbin to the cleaning chamber at 33 mm/s and heated to 1073 K under a nitrogen-gas atmosphere, where the sizing on the fiber thermally decomposed. The desized strand was coated with C-B-Si by the reaction of BCl3, SiCl4, CH4, and H2, under atmospheric pressure, in the deposition chamber. Nitrogen was used as the carrier gas. Finally, the coated strand was resized with a polyether and wound onto another bobbin in the winder. The single-filament strengths of the uncoated fiber and the coated fiber were tested, according to ASTM standard,¶ at room temperature (25 mm gauge length, 8.3 m/s testing speed, 25 tests). Cross-section areas of the filaments, needed for strength calculations, were measured from observations of the fracture surface using scanning electron microscopy (SEM). Auger electron spectroscopy (AES; Model No. PHI650, ULVAC, Ltd., Tokyo, Japan), with accelerating voltage of 3 kV and sample current of 3 nA, was used to examine the elemental depth profiles of the coatings. The etching rate, determined using argon sputtering, was 0.27 nm/s, with SiO2 as a standard. The deposits in the deposition chamber were analyzed to investigate the synthesis mechanism of the multilayered coating. Graphite sheets were placed inside the chamber, along its tubular wall, before the coating operation. The graphite sheets were removed from the furnace after the operation was complete and analyzed using electron probe microanalysis (EPMA; Model No. JXA8600MX, JEOL, Tokyo, Japan) and X-ray diffractometry (XRD; Model No. RINT 1400, Rigaku Co., Ltd., Tokyo, Japan). (2) Preparation and Analysis of the Composite Composite panels measuring 100 mm 100 mm 3 mm were fabricated using the PIP process, with the coated fiber as a reinforcement and methylhydrosilazane (MHS; NN710, Tonen Corp., Tokyo, Japan) as a matrix precursor. Details of the fabrication process and properties of MHS are described elsewhere.30,31 MHS was a random copolymer, composed of –[SiH2NH]– and –[SiMeHNH]– units and converted to an amorphous Si-N-C by pyrolysis at 1200 K in a nitrogen-gas or inert atmosphere. The chemical composition of the pyrolyzed MHS was (in mol%) 43 silicon, 38 nitrogen, 18 carbon, and 1 oxygen, with a 1:1 ratio of –[SiH2NH]–:–[SiMeHNH]–, and pyrolysis was performed in nitrogen gas at 1623 K.30,31 Unidirectionally reinforced composites were fabricated by the following process: (i) fabrication of unidirectionally fiber-aligned prepreg; (ii) stacking and curing of the prepreg sheets; (iii) pyrolysis of a cured sample at 1623 K; and (iv) densification of the sample by seven cycles of impregnation and pyrolysis. Details of the process are described elsewhere.30–32 The mechanical properties of the composites were evaluated in terms of flexural strength and interlaminar shear strength (ILSS) at room temperature. The test pieces, which were cut from the composite panels, measured 4 mm 40 mm 3 mm for the flexural test and 4 mm 12 mm 2 mm for the ILSS test. The longitudinal direction of the test pieces was aligned with the fiber orientation. The flexural strength was measured using the threepoint bend test, with a 30 mm span and a 8.3 m/s testing speed. The span and testing speed of the ILSS test were 8 mm and 8.3 m/s, respectively. Both tests were completed five times. The volume content of fiber (Vf ) was calculated from the dimensions of the composite panel and the amount of fiber used. The bulk density was calculated from the weight and dimensions of the test piece used for the flexural-strength test. The true density of the composite was measured using a pycnometer at 303 K, using n-butanol as the medium, on a sample crushed under No. 80 mesh. For the oxidation test on the composites, the test pieces measuring 4 mm 40 mm 3 mm were placed inside a tube furnace, heated to a given temperature at a rate of 0.167 K/s, and maintained at that temperature for a given time under a dry-air flow of 74 mmol/s. Some samples were exposed at the same temperature and time, under a nitrogen-gas flow of 74 mmol/s, as references. The microstructure of the fiber–matrix interface was investigated using transmission electron microscopy (TEM; Model No. JEM3010, JEOL) using the following techniques: (i) observation on a bright-field image; (ii) crystal-structure analysis by nanobeam electron diffraction (NBED) with a 5 nm beam; and (iii) elemental analysis by energy dispersive spectroscopy (EDS), with a 5 nm beam. AES depth profiles were investigated on the surfaces of a pullout fiber and on the matrices from which a fiber had debonded after fracture. The measuring conditions for AES were the same as those for the coated fiber. III. Results (1) Coatings Two types of fiber coating, coating I and II, were effective in the improvement of the oxidation resistance of composites. The deposition conditions of the coatings are shown in Table I. A ¶ “Standard Test Method for Tensile Strength and Young’s Modulus for HighModulus Single-Filament Materials,” ASTM Designation D-3379–75. American Society for Testing and Materials, West Conshohocken, PA. Fig. 1. Schematic drawing of the apparatus for fiber coating. 1816 Journal of the American Ceramic Society—Sato et al. Vol. 85, No. 7
July 2002 C-B-Si Coatings for S, Fiber-Reinforced Composites for Improved Oxidation Resistance l817 Table. Fabrication Conditions and Thicknesses of C-B-Si Coatings ating condition Thickness of Gas-flow rate(mmols-) Coating BCl3 SiCl CH4 H, N I0.298 0741.124461448 40-120 I0.149 0740.005.581448 R0.000 3720.004.171448100-200 (a) Sputter time()0.81.0 ference condition, R, included a carbon coating, the properties of which were described in a previous paper. Figure 2 shows depth profiles of coatings I and Il Coating thickness, determined by aES analysis, is shown in Table I Coating I had a boron-containing sublayer Coating Il contained a small amount of boron. However, the aes depth profile had the following problems, so that only relative changes in elemental composition were valid: (i)disagree 百E8 -- ment of the composition of the Si3 N4 fiber with the chemical analytical value;27,28 and(ii)miscalculation of 15 mol% of boron on the Si3N4 fiber, nevertheless nondetection of a boron peak in the AES spectra. These disagreements were explained by inaccu- facies of the default-calculation parameters and misreading of the (b)0002。0406 081.0 spectrum background by the calculation program. Therefore, the Sputter time(ks) depth profile on Fig. 2 was revised using the AES spectra as follows. Fig. 2. Depth profile of apparent elemental composition on(a)coating (1) For coating I, the surface consisted of carbon only, and boron, silicon, and nitrogen were detected at depths of 15, 30, and 60 nm, respectively. Boron had the maximum value, at a depth of 60 nm. The concentrations of silicon and nitrogen increased gradually and reached saturation at a depth of 120 nm, which the fiber side to the matrix side. The L2 presented the image of a nded to a ber interfa crystal lattice with 0.26 nm interlayer spacing. The NBED pattern (2) For coating Il, the surface consisted of carbon, and of the L2 layer was obscure spots, suggesting that L2 was silicon, and nitrogen were detected at depths of 15, 30, and 30 nm composed of disordered crystallites. The NBED patterns of bright respectively. Boron did not mark the obvious maximum value. layers LI and l3 lacked a lattice image and showed obscure rings, Silicon and nitrogen increased gradually, reaching saturation at a which polarized the brightness toward the direction of the fiber depth of 80 nm surface. Therefore, LI and L3 were composed of crystallites with The deposits in the CVd chamber after the preparation of very low crystallinity, oriented toward the fiber surface. EDS was coating I were examined using EPMA and XRD. The deposit sed on the fiber-matrix interface to obtain elemental information were >10 um thick, i.e., thick enough to make the peaks of the oron,although definitely contained in the sample, was not substrate graphite negligible on EPMA. Figure 3 shows changes in detected because of limitations in the present equipment. The peak ak intensity for each element along the longitudinal direction of intensities of carbon and silicon on the interface were about three the chamber. The temperature distribution in the chamber was mes and one-forth, respectively, that on the Si-N-C matrix; measured, using a thermocouple inserted into the chamber, under therefore, the interface consisted mainly of carbon and a small the flow of a carrier gas only. No deposit was observed below.8 amount of silicon. m(0. 8 m upstream from the center of the furnace). At-08 m boron and carbon were first detected. The boron attained a maximum value at-0.75 m. The carbon intensity exhibited a minimum at the boron maximum and then increased gradually on the uited a maximum at-03 m. Peaks of B,C were detected from 1500 ≌ peaks of 3C-SiC were detected from.5 to-0.1 m. The existence of carbon in the deposit was not confirmed using XRD. Because 0.8 the intrusion depth of the X-rays was greater than the thickness of 2 POB the deposits, the diffraction peaks of the substrate graphite over 06 lapped that of the deposit. The single-filament strength of the coated fibers is shown in 90.4 Table Il. The strength of the as-fabricated fiber was scattered among the fabrication lots of the fiber. Thus, a direct comparison of the strengths of the coated fibers was inadequate; the strength retention ratio was used. The retention ratio of each fiber was 77%121% 0.0 (2) Interface between Fiber and Matrin Position from center of furnace(m) The composite reinforced with Si3N4 fiber coated with coatin I is described here as composite I. Figure 4(a)shows the TEM Fig. 3. Change of relative intensity of boron, carbon, and silicon peaks image of the fiber-matrix interface of non-oxidized composite I EMPA analysis on the deposits in the coating chamber after fabrication coating I. Relative intensity he ratio of the peak intensity of ar The interface of this composite had a layered structure, composed lement on the deposit to that on the simple body of the element under the of a bright layer 5 nm thick (LI), a dark layer 10-15 m thick same measuring conditions. Origin of the horizontal axis is positioned on (2), and a bright layer 10-15 nm thick(3), in that order, from the center of the coating chamber. Negative direction is the upstream side
reference condition, R, included a carbon coating, the properties of which were described in a previous paper.33 Figure 2 shows depth profiles of coatings I and II. Coating thickness, determined by AES analysis, is shown in Table I. Coating I had a boron-containing sublayer. Coating II contained a small amount of boron. However, the AES depth profile had the following problems, so that only relative changes in elemental composition were valid: (i) disagreement of the composition of the Si3N4 fiber with the chemical analytical value;27,28 and (ii) miscalculation of 15 mol% of boron on the Si3N4 fiber, nevertheless nondetection of a boron peak in the AES spectra. These disagreements were explained by inaccuracies of the default-calculation parameters and misreading of the spectrum background by the calculation program. Therefore, the depth profile on Fig. 2 was revised using the AES spectra as follows. (1) For coating I, the surface consisted of carbon only, and boron, silicon, and nitrogen were detected at depths of 15, 30, and 60 nm, respectively. Boron had the maximum value, at a depth of 60 nm. The concentrations of silicon and nitrogen increased gradually and reached saturation at a depth of 120 nm, which corresponded to a coating–fiber interface. (2) For coating II, the surface consisted of carbon, and boron, silicon, and nitrogen were detected at depths of 15, 30, and 30 nm, respectively. Boron did not mark the obvious maximum value. Silicon and nitrogen increased gradually, reaching saturation at a depth of 80 nm. The deposits in the CVD chamber after the preparation of coating I were examined using EPMA and XRD. The deposits were 10 m thick, i.e., thick enough to make the peaks of the substrate graphite negligible on EPMA. Figure 3 shows changes in peak intensity for each element along the longitudinal direction of the chamber. The temperature distribution in the chamber was measured, using a thermocouple inserted into the chamber, under the flow of a carrier gas only. No deposit was observed below –0.8 m (0.8 m upstream from the center of the furnace). At –0.8 m, boron and carbon were first detected. The boron attained a maximum value at –0.75 m. The carbon intensity exhibited a minimum at the boron maximum and then increased gradually on the downstream side. Silicon was detected from –0.7 m and exhibited a maximum at –0.3 m. Peaks of B4C were detected from –0.7 to –0.6 m from XRD analysis of the deposits. Very weak peaks of 3C-SiC were detected from –0.5 to –0.1 m. The existence of carbon in the deposit was not confirmed using XRD. Because the intrusion depth of the X-rays was greater than the thickness of the deposits, the diffraction peaks of the substrate graphite overlapped that of the deposit. The single-filament strength of the coated fibers is shown in Table II. The strength of the as-fabricated fiber was scattered among the fabrication lots of the fiber. Thus, a direct comparison of the strengths of the coated fibers was inadequate; the strength retention ratio was used. The retention ratio of each fiber was 77%–121%. (2) Interface between Fiber and Matrix The composite reinforced with Si3N4 fiber coated with coating I is described here as composite I. Figure 4(a) shows the TEM image of the fiber–matrix interface of non-oxidized composite I. The interface of this composite had a layered structure, composed of a bright layer 5 nm thick (L1), a dark layer 10–15 nm thick (L2), and a bright layer 10–15 nm thick (L3), in that order, from the fiber side to the matrix side. The L2 presented the image of a crystal lattice with 0.26 nm interlayer spacing. The NBED pattern of the L2 layer was obscure spots, suggesting that L2 was composed of disordered crystallites. The NBED patterns of bright layers L1 and L3 lacked a lattice image and showed obscure rings, which polarized the brightness toward the direction of the fiber surface. Therefore, L1 and L3 were composed of crystallites with very low crystallinity, oriented toward the fiber surface. EDS was used on the fiber–matrix interface to obtain elemental information. Boron, although definitely contained in the sample, was not detected because of limitations in the present equipment. The peak intensities of carbon and silicon on the interface were about three times and one-forth, respectively, that on the Si-N-C matrix; therefore, the interface consisted mainly of carbon and a small amount of silicon. Table I. Fabrication Conditions and Thicknesses of C-B-Si Coatings Coating Coating condition Thickness of coating on Si3N4 fiber (nm) Gas-flow rate (mmols 1 ) Temperature BCl (K) 3 SiCl4 CH4 H2 N2 I 0.298 0.149 0.074 1.12 4.46 1448 40–120 II 0.149 0.223 0.074 0.00 5.58 1448 30–80 R 0.000 0.372 0.372 0.00 4.17 1448 100–200 Fig. 2. Depth profile of apparent elemental composition on (a) coating I and (b) coating II using AES. Fig. 3. Change of relative intensity of boron, carbon, and silicon peaks by EMPA analysis on the deposits in the coating chamber after fabrication of coating I. Relative intensity was the ratio of the peak intensity of an element on the deposit to that on the simple body of the element under the same measuring conditions. Origin of the horizontal axis is positioned on the center of the coating chamber. Negative direction is the upstream side. July 2002 C-B-Si Coatings for S3N4-Fiber-Reinforced Composites for Improved Oxidation Resistance 1817
l818 Journal of the American Ceramic Society-Sato et al. Vol. 85. No. 7 Table ll. Strength and Elastic Modulus of Row Fibers and Coated Fibers As-fabricated fiber Coated fiber Strength(GPa) E(GPa) Strength(GPa) E(GPa) Fiber 1.71 0.38 0.56 157 2.66 0.38 91 37 2.56 0.65 216 96 TsD is standard deviation Figure 4(b)shows a TEM image of the fiber-matrix interface of carbon Silicon and nitrogen were detected at a depth from 15 nm, non-oxidized composite Il. The interface was a monolayer 20 nm increased slowly, and reached saturation at a depth of 50 nm, where thick, with no crystal lattice, and it exhibited an obscure ring the Si-N-C matrix was located. Seven of eight samples showed result pattern under NBED. Under EDS, the peak intensities of carbon similar to those in Fig. 5(b), the one exception had a boron-containin and silicon on the interface were, respectively, about three times surface layer. In the case of composite Il, the pullout fiber had a and one-third that on the si-N-C matrix; therefore, the interface surface layer 80-110 nm thick, which consisted of carbon and a small consisted of carbon and a small amount of silicon amount of boron. The surface layer of the fractured matrix consisted Figure 4(c) shows the fiber-matrix interface of composite I mainly of carbon and was 45 vol%), nonbrittle intensity of oxygen was about twice that on LI and L2: on the fracture, and high strength. Table IV shows the strengths of the matrix(M) and the fiber (f), no oxygen was detected. opposites after oxidation Composites I and ll retained high strength, Figure 5 shows the AES spectra of the fractured surfaces of 0.8-0.6 GPa, and maintained their nonbrittle fracture, even after composite I. The AES spectra(Fig. 5)are shown instead of AE xidation at 1523 K for 100 h. The fracture surfaces of oxidized lepth profiles to illustrate the depth distribution of boron, because the composites I and Il had a region 0.1-0.4 mm wide around their aES depth profile cannot show boron content, as explained for Fig. 2 periphery, in which fibers failed along a matrix crack plane. The Figure 5(a) is the typical spectra of the surface of a pullout fiber. The center region of the fracture surface showed many pullout fibers fiber surface consisted mainly of carbon and a small amount of Reference composite r showed a large degradation in strength, 0.2 nitrogen and boron. Silicon was detected from a depth of 15 nm. GPa, and brittle fracture after oxidation. No pullout fiber was Silicon and nitrogen gradually increased, reaching saturation at a observed Heat exposure in a nitrogen-gas flow at 1523 K for 10 h depth of 80 nm because of the proximity of the fiber. Boron was caused no change of the strengths for all composites; therefore, the detected at depth 5 to 80 nm. Four pullout fibers were strength degradation using the oxidation test resulted from the mined, and all revealed similar results. Figure 5(b)is the typic oxidatio AES spectra of the fracture surface of the matrix from which a fiber Figure 6 shows the oxygen-concentration map of a cross section of had pulled out. The surface of the fractured matrix consisted of composites I, Il, and R after oxidation at 1523 K for 100 h. The matrix M B L3 L1 F 2 (a 10nm 5m(c) 10nm Fig 4. TEM image of the fiber-matrix interface of the as-fabricated composites(a)I and(b) ll, and(c)of the oxidized composite I at 1523 K for 10 h. M and F indicate matrix and fiber, respectively. L, LI, L2, and L3 indicate sublayers on the interf and B indicate an oxidized sublayer and a bubble
Figure 4(b) shows a TEM image of the fiber–matrix interface of non-oxidized composite II. The interface was a monolayer 20 nm thick, with no crystal lattice, and it exhibited an obscure ring pattern under NBED. Under EDS, the peak intensities of carbon and silicon on the interface were, respectively, about three times and one-third that on the Si-N-C matrix; therefore, the interface consisted of carbon and a small amount of silicon. Figure 4(c) shows the fiber–matrix interface of composite I oxidized at 1523 K for 10 h. The observed point located just under the oxygen-sealing layer is discussed later. The fiber, L1, and L2 exhibited no change with oxidation. L3 and the matrix showed obvious changes: (i) bright bubbles (B in Fig. 4(c)), where L3 was located before oxidation; and (ii) a layer that formed at the fiber side of the matrix (O in Fig. 4(c)). Under EDS, mainly carbon and silicon were detected on L1 and L2. On L3 and O, the peak intensity of oxygen was about twice that on L1 and L2; on the matrix (M) and the fiber (F), no oxygen was detected. Figure 5 shows the AES spectra of the fractured surfaces of composite I. The AES spectra (Fig. 5) are shown instead of AES depth profiles to illustrate the depth distribution of boron, because the AES depth profile cannot show boron content, as explained for Fig. 2. Figure 5(a) is the typical spectra of the surface of a pullout fiber. The fiber surface consisted mainly of carbon and a small amount of nitrogen and boron. Silicon was detected from a depth of 15 nm. Silicon and nitrogen gradually increased, reaching saturation at a depth of 80 nm because of the proximity of the fiber. Boron was detected at depths from 15 to 80 nm. Four pullout fibers were examined, and all revealed similar results. Figure 5(b) is the typical AES spectra of the fracture surface of the matrix from which a fiber had pulled out. The surface of the fractured matrix consisted of carbon. Silicon and nitrogen were detected at a depth from 15 nm, increased slowly, and reached saturation at a depth of 50 nm, where the Si-N-C matrix was located. Seven of eight samples showed results similar to those in Fig. 5(b); the one exception had a boron-containing surface layer. In the case of composite II, the pullout fiber had a surface layer 80–110 nm thick, which consisted of carbon and a small amount of boron. The surface layer of the fractured matrix consisted mainly of carbon and was 15 nm thick. No boron-containing layer was found on the matrix surface. (3) Oxidation Resistance of Composites The mechanical properties of the as-fabricated composites are shown in Table III. All the composites had dense matrices (total porosity of 11 vol%), high fiber content (Vf 45 vol%), nonbrittle fracture, and high strength. Table IV shows the strengths of the composites after oxidation. Composites I and II retained high strength, 0.8–0.6 GPa, and maintained their nonbrittle fracture, even after oxidation at 1523 K for 100 h. The fracture surfaces of oxidized composites I and II had a region 0.1–0.4 mm wide around their periphery, in which fibers failed along a matrix crack plane. The center region of the fracture surface showed many pullout fibers. Reference composite R showed a large degradation in strength, 0.2 GPa, and brittle fracture after oxidation. No pullout fiber was observed. Heat exposure in a nitrogen-gas flow at 1523 K for 10 h caused no change of the strengths for all composites; therefore, the strength degradation using the oxidation test resulted from the oxidation. Figure 6 shows the oxygen-concentration map of a cross section of composites I, II, and R after oxidation at 1523 K for 100 h. The matrix Table II. Strength and Elastic Modulus of Row Fibers and Coated Fibers Fiber Coating As-fabricated fiber Coated fiber Strength retention rate (%) Strength (GPa) E (GPa) Strength (GPa) E (GPa) Average SD† Average SD† Average SD† Average SD† Si3N4 I 2.07 0.50 182 24 1.59 0.70 175 27 77 Si3N4 II 1.78 0.54 173 16 1.90 0.61 151 10 106 Si3N4 R 1.71 0.38 169 11 2.06 0.56 157 5 121 SiC I 2.66 0.38 291 37 2.56 0.65 216 23 96 † SD is standard deviation. Fig. 4. TEM image of the fiber–matrix interface of the as-fabricated composites (a) I and (b) II, and (c) of the oxidized composite I at 1523 K for 10 h. M and F indicate matrix and fiber, respectively. L, L1, L2, and L3 indicate sublayers on the interface. O and B indicate an oxidized sublayer and a bubble. 1818 Journal of the American Ceramic Society—Sato et al. Vol. 85, No. 7
July 2002 C-B-Si Coatings for S,N/Fiber-Reinforced Composites for Improved Oxidation Resistance l819 C N t=0. 9ks e t=0. 6ks e t=0.3ks- u山 t=Oks 300 53 155 200 355 405 Kinetic energy (ev t=0. 9ks e t=0. 6ks e t=0.3ks r t=oks 250 300 480 530 200 355 Kinetic energy (ev) Fig. 5. Ime ge of the f ivative AES spectra on the surfaces of a(a) pullout fiber and(b)matrix from which a fiber was debonded of the fractured petra measurement was 60 s, which corresponded to -15 nm in dept posite R, of which the interface was a referential carbon layer, To evaluate the effect of C-B-Si coating for another fiber, lly oxidized to its center region. On the other hand, oxidation of coating I was applied on Sic fiber. The results are shown on composites I and Il was limited to a region 0. 1-0. 4 mm wide around Tables II-IV. The mechanical properties of the coated fiber and the periphery of the sample. The oxidized region corresponded as-fabricated composite were equal to the Si3 N4 fiber. The strengt roughly to the region that showed the flat surface after fracture. The of oxidized composite at 1523 K for 100 h was 51% of as- oxidation of composites I, I, and r proceeded microscopically on the fabricated composite. The strength after oxidation was lower than arts of the matrix adjacent to the fiber-matrix interfaces or the matrix the Si,Na fiber-reinforced composite, but higher than the case acks, and no oxidation occurred on the fibers. where carbon coating was used Table Ill. Mechanical Properties of Fabricated Composites Porosity lexural strength(GPa) LSS(MPa)° Fiber (vol% Average 11 44 0.03 V is volume faction of fiber. SD is standard deviation. ILSS is interlaminar shear strength
of composite R, of which the interface was a referential carbon layer, was fully oxidized to its center region. On the other hand, oxidation of composites I and II was limited to a region 0.1–0.4 mm wide around the periphery of the sample. The oxidized region corresponded roughly to the region that showed the flat surface after fracture. The oxidation of composites I, II, and R proceeded microscopically on the parts of the matrix adjacent to the fiber–matrix interfaces or the matrix cracks, and no oxidation occurred on the fibers. To evaluate the effect of C-B-Si coating for another fiber, coating I was applied on SiC fiber. The results are shown on Tables II–IV. The mechanical properties of the coated fiber and as-fabricated composite were equal to the Si3N4 fiber. The strength of oxidized composite at 1523 K for 100 h was 51% of asfabricated composite. The strength after oxidation was lower than the Si3N4-fiber-reinforced composite, but higher than the case where carbon coating was used. Table III. Mechanical Properties of Fabricated Composites Fiber Coating Vf (vol%)† Density (Mg/m3 ) Porosity (vol%) Flexural strength (GPa) ILSS (MPa)§ Average SD‡ Average SD‡ Si3N4 I 67 2.31 11 1.07 0.08 44 3 Si3N4 II 65 2.51 3 1.12 0.07 50 3 Si3N4 R 62 2.35 8 1.11 0.06 56 16 SiC I 57 2.55 4 1.07 0.03 93 7 † Vf is volume faction of fiber. ‡ SD is standard deviation. § ILSS is interlaminar shear strength. Fig. 5. Depth change of the first derivative AES spectra on the surfaces of a (a) pullout fiber and (b) matrix from which a fiber was debonded of the fractured composite I. Time interval of each spectra measurement was 60 s, which corresponded to 15 nm in depth. July 2002 C-B-Si Coatings for S3N4-Fiber-Reinforced Composites for Improved Oxidation Resistance 1819
1820 Journal of the American Ceramic Society-Sato et al. Vol. 85. No. 7 Table IV. Flexural Strength of Composites after Oxidation EDS, which was applied simultaneously with TEM, was insuffI After oxidation at 1523 K for After oxidation at 1523 K for cient for lightweight elements, such as boron. If it is confirmed that the fiber-matrix interface became coated with almost no change Flexural strength during the PIP composite fabrication, the chemical composition of (GPa) the coating applies to the fiber-matrix interface. To confirm the Retention Coating Average SD rate(%) Average SD rate ( Retention difference between the coating and the fiber-matrix interface, the AES result of the coating was compared with that of the fracture surface of the fiber-matrix interface as follows First. the boron- Ⅱ0.680.13610.670.0260 containing sublayer, the second sublayer of the coating, was found I0.590.03550 on the pullout fiber only. Second, the carbon-rich layer, the third sublayer of the coating, was found on the pullout fiber and the tSD is standard deviation bRittle fracture was shown fractured matrix surface. These results showed that(i)the layered structure of the coating changed little during the pip process and that(ii) the fracture of the fiber-matrix interface proceeded in the IV Discussion outer la of the interface, which corresponded to the third (I) Structure of Coatings and Fiber-Matrix Interface ublayer of the coating. The d terence between the coati g the CVd reactor. in which thickness of 40-120 nm(Table D), revealed by AEs analysis, an fibers were fed continuously Therefore. it was expected that the the interface thickness of 30-50 nm(Fig. 4), obtained from TEM elemental depth profile of the coating would roughly agree with observation, would present a problem if the coating were to the chemical composition of the deposit, which changed along the became the fiber-matrix interface with no changes. However, th horizontal direction of the CVD reactor. Actually, the depth profile thickness obtained using two te observations was considered within the range of eight measurements using AES deposit(Fig. 3)on the concentration change of boron and carbon Therefore, the sublayers of the coating and the fiber-matrix AES and EPMA results showed coating I consisted of three interface corresponded to each other as follows. L3 was the third sublayers(from the fiber side): (i) first sublayer without boron; (i) sublayer of the coating, a carbon layer with a small amount of second sublayer of boron, silicon, and carbon; and(iii) third silicon and a graphitelike structure. L2 was the second sublayer of sublayer mainly of carbon. The weak peaks of B.C and Sic, the coating, composed of crystallites, with a 0. 26 nm lattice detected using XRD on the deposit, corresponded to the second interlayer spacing, and it consisted of carbon, boron, and silicon sublayer therefore, boron and silicon were determined to be The L2 interlayer spacing was similar to the 0.25 nm of the closest ontained as B.c and Sic. packing plane of a-or B-SiC, the 0.26 nm of the(104) plane of TEM observation of the fiber-matrix interface of composite I B.C, and the 0.24 nm of the(021) plane of B. C. BC, and Sic that revealed a three-layered structure. However, the chemical compo- were detected using XRD on the deposits in the coat sition of each sublayer was uncertain, because the sensitivity of as mentioned before. Therefore, the crystallite seemed to be Sic Z=.5 4z=05 4z=02 42e03 (d) Fig. 6. Oxygen distribution maps of (b) composite R(carbon coating), (c)composite I, and(d)composite Il using EPMA. (a) SEM image of the composite R at the same position as(b).(e) Distance from the surface of the composites
IV. Discussion (1) Structure of Coatings and Fiber–Matrix Interface The coating was prepared using the CVD reactor, in which fibers were fed continuously. Therefore, it was expected that the elemental depth profile of the coating would roughly agree with the chemical composition of the deposit, which changed along the horizontal direction of the CVD reactor. Actually, the depth profile of coating I (Fig. 2) agreed generally with the EPMA result of the deposit (Fig. 3) on the concentration change of boron and carbon. AES and EPMA results showed coating I consisted of three sublayers (from the fiber side): (i) first sublayer without boron; (ii) second sublayer of boron, silicon, and carbon; and (iii) third sublayer mainly of carbon. The weak peaks of B4C and SiC, detected using XRD on the deposit, corresponded to the second sublayer; therefore, boron and silicon were determined to be contained as B4C and SiC. TEM observation of the fiber–matrix interface of composite I revealed a three-layered structure. However, the chemical composition of each sublayer was uncertain, because the sensitivity of EDS, which was applied simultaneously with TEM, was insufficient for lightweight elements, such as boron. If it is confirmed that the fiber–matrix interface became coated with almost no change during the PIP composite fabrication, the chemical composition of the coating applies to the fiber–matrix interface. To confirm the difference between the coating and the fiber–matrix interface, the AES result of the coating was compared with that of the fracture surface of the fiber–matrix interface as follows. First, the boroncontaining sublayer, the second sublayer of the coating, was found on the pullout fiber only. Second, the carbon-rich layer, the third sublayer of the coating, was found on the pullout fiber and the fractured matrix surface. These results showed that (i) the layered structure of the coating changed little during the PIP process and that (ii) the fracture of the fiber–matrix interface proceeded in the outer layer of the interface, which corresponded to the third sublayer of the coating. The difference between the coating thickness of 40–120 nm (Table I), revealed by AES analysis, and the interface thickness of 30–50 nm (Fig. 4), obtained from TEM observation, would present a problem if the coating were to became the fiber–matrix interface with no changes. However, the thickness obtained using two TEM observations was considered within the range of eight measurements using AES. Therefore, the sublayers of the coating and the fiber–matrix interface corresponded to each other as follows. L3 was the third sublayer of the coating, a carbon layer with a small amount of silicon and a graphitelike structure. L2 was the second sublayer of the coating, composed of crystallites, with a 0.26 nm lattice interlayer spacing, and it consisted of carbon, boron, and silicon. The L2 interlayer spacing was similar to the 0.25 nm of the closest packing plane of - or -SiC, the 0.26 nm of the 104 plane of B4C, and the 0.24 nm of the (021) plane of B4C. B4C, and SiC that were detected using XRD on the deposits in the coating apparatus as mentioned before. Therefore, the crystallite seemed to be SiC Table IV. Flexural Strength of Composites after Oxidation Fiber Coating After oxidation at 1523 K for 10 h After oxidation at 1523 K for 100 h Flexural strength (GPa) Retention rate (%) Flexural strength (GPa) Retention Average SD rate (%) † Average SD† Si3N4 I 0.92 0.15 86 0.83 0.08 77 Si3N4 II 0.68 0.13 61 0.67 0.02 60 Si3N4 R 0.27‡ 0.05 25 0.20‡ 0.03 18 SiC I 0.59 0.03 55 0.51 0.03 48 † SD is standard deviation. ‡ Brittle fracture was shown. Fig. 6. Oxygen distribution maps of (b) composite R (carbon coating), (c) composite I, and (d) composite II using EPMA. (a) SEM image of the composite R at the same position as (b). (e) Distance z from the surface of the composites. 1820 Journal of the American Ceramic Society—Sato et al. Vol. 85, No. 7
July 2002 C-B-Si Coatings for S,N/Fiber-Reinforced Composites for Improved Oxidation Resistance l821 and/or B.C, however, the crystal phase could not be identified supplied from the outside. More et al showed a borosilicate- because of the obscurity of the electron diffraction pattern of L2 -matrix interface of th LI was the first sublayer of the coating, but its composition wa oxidized SiC/BN/SiC composite. The interface-layer morphology uncertain. LI was estimated to be a graphitelike carbon, because of was similar to that of the present oxidized composite(Fig. 4(c)) the similarity of its TEM image and electron diffraction pattern to These similarities also suggested the formation of borosilicate those for L3 glass at the interfa In the case of coating Il, TEM observation and eds analysis o the interface showed that the fiber-matrix interface of this com posite was a monolayer 20 nm thick, consisting mainly of carbon, with a small amount of silicon, and having a weak crystal Newly developed C-B-Si interfacial coatings were orientation parallel to the fiber surface. AES analysis of the coated the fiber-matrix interfaces of a Si,Na fiber-reinforced fiber and the fracture surface of the interface revealed that the to improve the oxidation resistance of the composite, The Si3N4 coating structure remained even after the PIP process, as in coating fiber was coated with the C-B-Si layer using CVD and embedded I. When the information of the AES analysis was added to the in the Si-N-C matrix by a PIP process. Two types of C-B-Si TEM observation, the interface was determined to be a graphit coatings enhanced the oxidation resistance of the PlP composites, like carbon laver containing a small amount of boron and silico although the matrix had many cracks, resulting from pyrolysis with an outer carbon-rich sublayer. shrinkage of the precursor, that allowed the easy permeation of The fracture of the fiber-matrix interface of composites I and Il oxygen. The first coating, coating I, formed a multilayered proceeded on the matrix side of the interface layer, which fiber-matrix interface, which consisted of three sublayers: a corresponded to the outer sublayer of the fiber coating. The crystalline sublayer containing boron, silicon, and carbon was debonding on the outer surface of the coating was necessary for the sandwiched between two graphitelike carbon layers. The second coati ing to be the pip at the inner part of the fiber coating, the fiber coating was lost from face, which consisted of a graphitelike carbon layer containing a the fiber surface after cyclic impregnation of PIP. small amount of boron and silicon. Debonding between the fiber and the matrix occurred at the carbon(sub)layer for both of th composites and gave the composites a flexural strength as high as (2) Mechanism for Improving Oxidation Resistance 1. 1 GPa. The composites retained 77%(coating I)and 60% (coating If) of their original strength, even after oxidation at 1523 The high strength of the composites was obtained by weakened K for 360 ks. Coating I was also effective in the improvement of iber-matrix bonding by the carbon sublayer for the monolayered the oxidation resistance of a SiC-fiber-reinforced composite terface(composite ID) and the multilayered interface(composite D), as shown by AES analysis of the fracture surfaces. The carbon The mechanism by which oxidation resistance was improved sublayer prevented the propagation of a matrix crack through the hypothesized as follows. The carbon(sub )layer was easily oxi- fiber as a conventional carbon interface I-5 dized near the surface of the composite. Simultaneously, the The mechanism for d oxidation resistance boron- and silicon-containing(sub)layer (the center crystall ivolves the microstructure of the interface, the morphology of the sublayer for coating I and the graphitelike carbon layer itself fo fracture surface, and the oxygen distribution on a cross section of coating ID) formed borosilicate glass, the periphery of which sealed the oxidized composite. The carbon sublayer adjacent to the the matrix cracks. As a result, the 0. 1-0.3 nm wide periphery of surface of the sample oxidizes easily if the sample is exposed the coml pne fiber and the matrix with the borosilicate glass,but the of the c-B-Si interface over a conventional carbon interface is the inside of the composite was unoxidized and showed many long formation of borosilicate glass. The center crystalline sublayer pullout fibers on its fracture surface supplies boron and silicon in the case of coating I, and the No borosilicate layers, the existence of which would have verified boron-containing graphitelike carbon layer supplies boron in the the mechanism of oxidation resistance, were directly detected at the case of coating Il Boron-rich borosilicate glass melts at K 4 interface by microanalytical techniques, such as TEM and EDS near the starting temperature for the oxidation of carbon. Borosil because of equipment limitations for the detection of boron. However, cate glass seals the matrix cracks and the fiber-matrix gaps the proposed mechanism adequately explained the morphology of the resulting from oxidation loss of the carbon sublayers and prevents fracture surfaces and the oxygen concentration distributions of the the permeation of oxygen into the composite cross sections of the oxidized composites If few matrix cracks exist in a composite, oxygen permeation is stopped within the thin oxygen-sealing layer around the composite References However, a composite fabricated using the PIP process has many E. Fitzer and R. Gadow."Fiber-Reinforced Silicon Carbide. Am. Cera. Soc natrix cracks. Therefore, full suppression of oxygen permeation Ball,652]326-35(1986) equires a thick oxygen-sealing layer around the composite. In the oxygen-sealing layer, the borosilicate glass bonds matrix crac M. Prewo, "Fiber-Reinforced Ceramics: New Opportunities for Composite esul flat fracture Irface at the pe Materials," Am Ceran. Soc. Bull., 68[2] 395-400(19 On the other hand, all the C-B-Si interface layers remain unoxi R.J.Kerans, R. S. Hay, N. J. Pagano, and T, A. Parthasarathy,""The Role of the dized at the inside of the composite, causing much fiber pullout on Fiber-Matrix Interface in Ceramic Composites, Am. Ceram. Soc. Bull., 68 [2] 4212(89 No borosilicate glass, the direct evidence of an antioxidation J.J. Brennan,"Eflects of Bischoff, O. Sbaizero, M. Rhule, A G. Evans, D B. Marshall, and on the Properties of Fiber-Reinforced Ceramics, mechanism, was detected using TEM or EDS in the present J. Am. Ceram Soc., 73 6]1691-99(1990) samples because of detection limitations of the equipment. How K. M. Prewo, "Fatigue and Stress Rupture of Silicon Carbide-Fiber-Reinforced ever, the phenomenon is well supported by the hypothesis that L. Filipuzzi, G. Camus R. Naslain, and J, Th ""Oxidation Mechanisms and borosilicate glass seals the oxygen-diffusion passes. In an attempt Kinetics of ID-SiC/C/SiC Composite Materials: L, An Experimental Approach, to determine whether borosilicate glass and B2O3 could form by the oxidation of the other boron-containing interfacial layer. T. E. Steyer, F. w. Zok, and D. P. Walls, "Stress Rupture of an Enhanced Sheldon et al.>conducted thermodynamic calculations on the Soc, 81 [8]2140-46(1998) system consisting of a SiC fiber, a Bn interface, and an SiC sS. Zhu, M. Mizuno, Y. Nagano, J. Cao, Y. Kagawa, and H. Kaya, "Creep and matrix. Lee et al. 7 observed fiber-matrix interfaces for the same system. Those researchers showed that B,O3 and/or borosilicate eroc819269-7198 S. Jacques, A. Guette, F. Langlais, R. Naslain, and S. Goujard,"High- glass formed when the amount of oxygen was high or oxygen was Temperature Lifetime in Air of SiC/C(BySiC Microcomposites Prepared by
and/or B4C; however, the crystal phase could not be identified because of the obscurity of the electron diffraction pattern of L2. L1 was the first sublayer of the coating, but its composition was uncertain. L1 was estimated to be a graphitelike carbon, because of the similarity of its TEM image and electron diffraction pattern to those for L3. In the case of coating II, TEM observation and EDS analysis of the interface showed that the fiber–matrix interface of this composite was a monolayer 20 nm thick, consisting mainly of carbon, with a small amount of silicon, and having a weak crystal orientation parallel to the fiber surface. AES analysis of the coated fiber and the fracture surface of the interface revealed that the coating structure remained even after the PIP process, as in coating I. When the information of the AES analysis was added to the TEM observation, the interface was determined to be a graphitelike carbon layer containing a small amount of boron and silicon with an outer carbon-rich sublayer. The fracture of the fiber–matrix interface of composites I and II proceeded on the matrix side of the interface layer, which corresponded to the outer sublayer of the fiber coating. The debonding on the outer surface of the coating was necessary for the coating to be applied on the PIP composite. If debonding occurred at the inner part of the fiber coating, the fiber coating was lost from the fiber surface after cyclic impregnation of PIP.33 (2) Mechanism for Improving Oxidation Resistance The high strength of the composites was obtained by weakened fiber–matrix bonding by the carbon sublayer for the monolayered interface (composite II) and the multilayered interface (composite I), as shown by AES analysis of the fracture surfaces. The carbon sublayer prevented the propagation of a matrix crack through the fiber as a conventional carbon interface.1–5 The mechanism for improved oxidation resistance apparently involves the microstructure of the interface, the morphology of the fracture surface, and the oxygen distribution on a cross section of the oxidized composite. The carbon sublayer adjacent to the surface of the sample oxidizes easily if the sample is exposed under an oxidizing atmosphere at high temperature. The advantage of the C-B-Si interface over a conventional carbon interface is the formation of borosilicate glass. The center crystalline sublayer supplies boron and silicon in the case of coating I, and the boron-containing graphitelike carbon layer supplies boron in the case of coating II. Boron-rich borosilicate glass melts at 700 K,34 near the starting temperature for the oxidation of carbon. Borosilicate glass seals the matrix cracks and the fiber–matrix gaps resulting from oxidation loss of the carbon sublayers and prevents the permeation of oxygen into the composite. If few matrix cracks exist in a composite, oxygen permeation is stopped within the thin oxygen-sealing layer around the composite. However, a composite fabricated using the PIP process has many matrix cracks. Therefore, full suppression of oxygen permeation requires a thick oxygen-sealing layer around the composite. In the oxygen-sealing layer, the borosilicate glass bonds matrix cracks and fiber–matrix interfaces. This hard bonding of the interfaces results in a flat fracture surface at the periphery of the composite. On the other hand, all the C-B-Si interface layers remain unoxidized at the inside of the composite, causing much fiber pullout on the fracture surface. No borosilicate glass, the direct evidence of an antioxidation mechanism, was detected using TEM or EDS in the present samples because of detection limitations of the equipment. However, the phenomenon is well supported by the hypothesis that borosilicate glass seals the oxygen-diffusion passes. In an attempt to determine whether borosilicate glass and B2O3 could form by the oxidation of the other boron-containing interfacial layer, Sheldon et al.35 conducted thermodynamic calculations on the system consisting of a SiC fiber, a BN interface, and an SiC matrix. Lee et al.17 observed fiber–matrix interfaces for the same system. Those researchers showed that B2O3 and/or borosilicate glass formed when the amount of oxygen was high or oxygen was supplied from the outside. More et al.36 showed a borosilicateglass layer with bubbles on the fiber–matrix interface of the oxidized SiC/BN/SiC composite. The interface-layer morphology was similar to that of the present oxidized composite (Fig. 4(c)). These similarities also suggested the formation of borosilicate glass at the interface. V. Conclusions Newly developed C-B-Si interfacial coatings were applied at the fiber–matrix interfaces of a Si3N4-fiber-reinforced composite to improve the oxidation resistance of the composite. The Si3N4 fiber was coated with the C-B-Si layer using CVD and embedded in the Si-N-C matrix by a PIP process. Two types of C-B-Si coatings enhanced the oxidation resistance of the PIP composites, although the matrix had many cracks, resulting from pyrolysis shrinkage of the precursor, that allowed the easy permeation of oxygen. The first coating, coating I, formed a multilayered fiber–matrix interface, which consisted of three sublayers: a crystalline sublayer containing boron, silicon, and carbon was sandwiched between two graphitelike carbon layers. The second coating, coating II, formed a morphologically monolayered interface, which consisted of a graphitelike carbon layer containing a small amount of boron and silicon. Debonding between the fiber and the matrix occurred at the carbon (sub)layer for both of the composites and gave the composites a flexural strength as high as 1.1 GPa. The composites retained 77% (coating I) and 60% (coating II) of their original strength, even after oxidation at 1523 K for 360 ks. Coating I was also effective in the improvement of the oxidation resistance of a SiC-fiber-reinforced composite. The mechanism by which oxidation resistance was improved is hypothesized as follows. The carbon (sub)layer was easily oxidized near the surface of the composite. Simultaneously, the boron- and silicon-containing (sub)layer (the center crystalline sublayer for coating I and the graphitelike carbon layer itself for coating II) formed borosilicate glass, the periphery of which sealed the matrix cracks. As a result, the 0.1–0.3 nm wide periphery of the composite showed brittle fracture, caused by the hard bond between the fiber and the matrix with the borosilicate glass, but the inside of the composite was unoxidized and showed many long pullout fibers on its fracture surface. No borosilicate layers, the existence of which would have verified the mechanism of oxidation resistance, were directly detected at the interface by microanalytical techniques, such as TEM and EDS, because of equipment limitations for the detection of boron. However, the proposed mechanism adequately explained the morphology of the fracture surfaces and the oxygen concentration distributions of the cross sections of the oxidized composites. References 1 E. Fitzer and R. Gadow, “Fiber-Reinforced Silicon Carbide,” Am. Ceram. Soc. Bull., 65 [2] 326–35 (1986). 2 K. M. Prewo, J. J. Brennan, and G. K. Layden, “Fiber-Reinforced Glasses and Glass-Ceramics for High-Performance Applications,” Am. Ceram. Soc. Bull., 65 [2] 305–13 (1986). 3 K. M. Prewo, “Fiber-Reinforced Ceramics: New Opportunities for Composite Materials,” Am. Ceram. Soc. Bull., 68 [2] 395–400 (1989). 4 R. J. Kerans, R. S. Hay, N. J. Pagano, and T. A. Parthasarathy, “The Role of the Fiber–Matrix Interface in Ceramic Composites,” Am. Ceram. Soc. Bull., 68 [2] 429–42 (1989). 5 H. C. Cao, E. Bischoff, O. Sbaizero, M. Rhule, A. G. Evans, D. B. Marshall, and J. J. Brennan, “Effects of Interfaces on the Properties of Fiber-Reinforced Ceramics,” J. Am. Ceram. Soc., 73 [6] 1691–99 (1990). 6 K. M. Prewo, “Fatigue and Stress Rupture of Silicon Carbide-Fiber-Reinforced Glass-Ceramics,” J. Mater. Sci., 22, 2695–701 (1987). 7 L. Filipuzzi, G. Camus R. Naslain, and J. Thebault, “Oxidation Mechanisms and Kinetics of 1D-SiC/C/SiC Composite Materials: I, An Experimental Approach,” J. Am. Ceram. Soc., 77 [2] 459–66 (1994). 8 T. E. Steyer, F. W. Zok, and D. P. Walls, “Stress Rupture of an Enhanced Nicalon/Silicon Carbide Composite at Intermediate Temperatures,” J. Am. Ceram. Soc., 81 [8] 2140–46 (1998). 9 S. Zhu, M. Mizuno, Y. Nagano, J. Cao, Y. Kagawa, and H. Kaya, “Creep and Fatigue Behavior in an Enhanced SiC/SiC Composite at High Temperature,” J. Am. Ceram. Soc., 81 [9] 2269–77 (1998). 10S. Jacques, A. Guette, F. Langlais, R. Naslain, and S. Goujard, “HighTemperature Lifetime in Air of SiC/C(B)/SiC Microcomposites Prepared by July 2002 C-B-Si Coatings for S3N4-Fiber-Reinforced Composites for Improved Oxidation Resistance 1821
1822 Journal of the American Ceramic Society-Sato et al. Vol. 85. No. 7 LPCVD", Pp. 381-88 in Ceramic Transactions Perl: m aicg. Edited by A.G.Evans K. Sato, T. Suzuki, O. Funayama, and T. Isoda, ""Preparation of Carbon-Fiber- Followed by and R. Naslain. American Ceramic Society, Westerville, OH, 1995 ugne, A. Guette, J. Sevely, C. R. Brosse, J.-P. Rocher, and J Pressureless Firing, Ceram. Eng. Sci. Proc., 13 19-10 614-21(1992 Cotteret,""Boron Nitride Interface in Ceramic-Matrix Composites, J. Am. Ceran. Nitride-Based Composites by Impregnation with Perhydoropolysilazane, "J. Ceram. Soc,74[012482-88(1991) S. R Nutt, and J. J. Brennan, "Interfacial Microstructure and Sm10144+447(y92 Funayama, M. Arai, Y. Tashiro, H. Aoki, T. Suzuki, K. Tamura, H Kaya, H Chemistry of SiC/BN Dual-Coated Nicalon-Fiber-Reinforced Glass-Ceramic Matrix Nishii, and T. Isoda, "Tensile Strength of Silicon Nitride Fibers Produced from A. G. Evans, F. w. Zok, McMeeking, and Z. Z. Du,"Models of "Y. Yokoyama, T Nanba, L. Yasui, H. Kaya, T. Maeshima, and T Isoda,X-ray Diffraction Study of the Structure of Silicon Nitride Fiber Made from Perhydro- Composites, J. Am. Ce 99 polysilazane, J. Am. Ceram Soc., 74, 654-57(1991) 14F. E. Heredia, J. C. McNulty, F. W. Zok, and A. G. Evans,"Oxidation 2M. Takeda, Y Imai, H Ichikawa, T Ishikawa, N. Kasai, T. Seguchi, and K Okamura,"Thermomechanical Analysis of Low-Oxygen Silicon Ca 2097-100(1995) D Morgan and D. B. Marshall, "Ceramic Composites of Monazite and (19 Derived from Polycarbosilane, Ceram. Eng. Sci. Proc., 14 19-10] 540-47 Alumina. Am. Ce 786]1553-63(1995)and“ errection,"J.Am H. Morozumi, K. Sato, A. Tezuka, H. Kaya, and T, Isoda,"Preparation of High Ceram.Soc,7892574(1995 Composites by a Preceramic l6D.-H. Kuo, w. M. Riven, and T. J Mackin, "Control of Interfacial Properties Ceram. Soc, 80(12)2987-96u199 oating in Oxide-Oxide Composites,"J.Am. oiii K, Sato, H. Kaya, O. Funayama, and T. Isoda,"Evaluation of Perhydropolysilazane, Poly borosilazane, and Methy lhydropo Lara-Curzio, and K. L. More, "Multilayered Oxide Interphase oncept for Ceramic-Matrix Composites, J Am Ceram Soc., 81 31717-20(1998). K. L. Luthra, Oxidation-Resistant Fiber Coatings for Non-oxide Ceramic K. Sato, A. Tezuka, O. Funayama, T. Isoda, Y. Terada, S. Kato, and M. Iwata, Composites, J. A. Ceran. Soc., 80[ 123253-57(1997). Fabrication and Pressure Testing of a Gas-Turbine Component Manufactured by a I9J.B Dabis, P. A. Lofvander, A. G. Evans, E Bischoff, and M. L. Emiliani,"Fiber receramic-Polymer-Impregnation Method, Compos. Sci. Technol, 59, 853-59 229 rittle-Matrnx Composites, J. Am. Cerana,6的 (1999) 33K. Sato, H. Morozumi, A. Tezuka, O. Funayama, and T. Isoda, "Interface and -T. Mah, K. Keller, T, A. Parthasarathy, and C. Cooke, "Fugitive Interface Mechanical Properties of Ceramic-Fiber-Reinforced Silicon Nitride Composites Coating in Oxide-Oxide Composites, "Ceram. Eng. Sci. Proc., 12 [C Transactions, Vol 58, High-Temperature Ceramic-Matrix Composites 1: Marfa W-C. Tu, F. F. Lange, and A. G. Evans, "Concept for a Damage-Tolerar taring and Materials Development. Edited by A. G. Evans and R. Naslain. Americar 199 R. S. Roth, J.R. Dennis, and H. F. MeMurdie; Fig 6455 in Phase Diagrams for -C, G. Levi, J. Y. Yang, B. J. Dalgleish, F, w. Zok, and A G. Evans, "Processing Ceramists, Vol VI, Edited by M. A Clevinger and H M. Ondik. American Ceramic nance of an All-Oxide Ceramic Composite,J. Am. Ceram. Soc., 81 [8 w s heldon, E. Y, Sun, S.R. Nutt, and J. J. Brennan, Oxidation of BN-Coated SiC Fibers in Ceramic-Matrix Composites, "JAm Ceram Soc., 79(2] Preparation and Properties of Monolithic and Composite Ceramics Produced by 539-43(1996 Polymer Pyrolysis, "Am Ceram Soc. Bull, 62 [8]916-23(1983) L More, K.S.A R. Gadow. ""Fiber-Reinforced Silicon Carbide. Am. Ceran. Soc. Oxygen Content in BN u,65[2]326-35(1986) ites,Composites, 30A [4]463-70(1999)
LPCVD”; pp. 381–88 in Ceramic Transactions, Vol. 57, High-Temperature CeramicMatrix Composites I: Design, Durability, and Performance. Edited by A. G. Evans and R. Naslain. American Ceramic Society, Westerville, OH, 1995. 11R. Naslain, O. Dugne, A. Guette, J. Sevely, C. R. Brosse, J.-P. Rocher, and J. Cotteret, “Boron Nitride Interface in Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 74 [10] 2482–88 (1991). 12E. Y. Sun, S. R. Nutt, and J. J. Brennan, “Interfacial Microstructure and Chemistry of SiC/BN Dual-Coated Nicalon-Fiber-Reinforced Glass-Ceramic Matrix Composites,” J. Am. Ceram. Soc., 77 [5] 1329–39 (1994). 13A. G. Evans, F. W. Zok, R. M. McMeeking, and Z. Z. Du, “Models of High-Temperature, Environmentally Assisted Embrittlement in Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 79 [9] 2345–52 (1996). 14F. E. Heredia, J. C. McNulty, F. W. Zok, and A. G. Evans, “Oxidation Embrittlement Probe for Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 78 [8] 2097–100 (1995). 15P. E. D. Morgan and D. B. Marshall, “Ceramic Composites of Monazite and Alumina,” J. Am. Ceram. Soc., 78 [6] 1553–63 (1995) and “Correction,” J. Am. Ceram. Soc., 78 [9] 2574 (1995). 16D.-H. Kuo, W. M. Kriven, and T. J. Mackin, “Control of Interfacial Properties through Fiber Coatings: Monazite Coating in Oxide–Oxide Composites,” J. Am. Ceram. Soc., 80 [12] 2987–96 (1997). 17W. Y. Lee, E. Lara-Curzio, and K. L. More, “Multilayered Oxide Interphase Concept for Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 81 [3] 717–20 (1998). 18K. L. Luthra, “Oxidation-Resistant Fiber Coatings for Non-oxide Ceramic Composites,” J. Am. Ceram. Soc., 80 [12] 3253–57 (1997). 19J. B. Dabis, P. A. Lofvander, A. G. Evans, E Bischoff, and M. L. Emiliani, “Fiber Coating Concepts for Brittle-Matrix Composites,” J. Am. Ceram. Soc., 76 [5] 1249–57 (1993). 20T. Mah, K. Keller, T. A. Parthasarathy, and C. Cooke, “Fugitive Interface Coating in Oxide–Oxide Composites,” Ceram. Eng. Sci. Proc., 12 [9–10] 1802–15 (1991). 21W.-C. Tu, F. F. Lange, and A. G. Evans, “Concept for a Damage-Tolerant Ceramic Composite with ‘Strong’ Interfaces,” J. Am. Ceram. Soc., 79 [2] 417–24 (1996). 22C. G. Levi, J. Y. Yang, B. J. Dalgleish, F. W. Zok, and A. G. Evans, “Processing and Performance of an All-Oxide Ceramic Composite,” J. Am. Ceram. Soc., 81 [8] 2077–86 (1998). 23B. E. Walker Jr., R. W. Rice, P. F. Becher, B. A. Bender, and W. S. Coblenz, “Preparation and Properties of Monolithic and Composite Ceramics Produced by Polymer Pyrolysis,” Am. Ceram. Soc. Bull, 62 [8] 916–23 (1983). 24E. Fitzer and R. Gadow, “Fiber-Reinforced Silicon Carbide,” Am. Ceram. Soc. Bull., 65 [2] 326–35 (1986). 25K. Sato, T. Suzuki, O. Funayama, and T. Isoda, “Preparation of Carbon-FiberReinforced Composite by Impregnation with Perhydropolysilazane Followed by Pressureless Firing,” Ceram. Eng. Sci. Proc., 13 [9–10] 614–21 (1992). 26K. Sato, T. Suzuki, O. Funayama, and T. Isoda, “Fabrication of Silicon Nitride-Based Composites by Impregnation with Perhydoropolysilazane,” J. Ceram. Soc. Jpn., 100 [4] 444–47 (1992). 27O. Funayama, M. Arai, Y. Tashiro, H. Aoki, T. Suzuki, K. Tamura, H. Kaya, H. Nishii, and T. Isoda, “Tensile Strength of Silicon Nitride Fibers Produced from Perhydropolysilazane,” J. Ceram. Soc. Jpn., 98 [1] 104–107 (1990). 28Y. Yokoyama, T. Nanba, I. Yasui, H. Kaya, T. Maeshima, and T. Isoda, “X-ray Diffraction Study of the Structure of Silicon Nitride Fiber Made from Perhydropolysilazane,” J. Am. Ceram. Soc., 74, 654–57 (1991). 29M. Takeda, Y. Imai, H. Ichikawa, T. Ishikawa, N. Kasai, T. Seguchi, and K. Okamura, “Thermomechanical Analysis of Low-Oxygen Silicon Carbide Fibers Derived from Polycarbosilane,” Ceram. Eng. Sci. Proc., 14 [9 –10] 540 – 47 (1993). 30H. Morozumi, K. Sato, A. Tezuka, H. Kaya, and T. Isoda, “Preparation of High Strength Ceramic-Fiber-Reinforced Silicon Nitride Composites by a Preceramic Polymer Impregnation Method,” Ceram. Int., 23, 179–84 (1997). 31K. Sato, H. Kaya, O. Funayama, and T. Isoda, “Evaluation of Polysilazanes— Perhydropolysilazane, Polyborosilazane, and Methylhydropolysilazane—as Matrix Precursors of Ceramic-Matrix Composites,” J. Ceram. Soc. Jpn., 109 [5] 440–46 (2001). 32K. Sato, A. Tezuka, O. Funayama, T. Isoda, Y. Terada, S. Kato, and M. Iwata, “Fabrication and Pressure Testing of a Gas-Turbine Component Manufactured by a Preceramic-Polymer-Impregnation Method,” Compos. Sci. Technol., 59, 853–59 (1999). 33K. Sato, H. Morozumi, A. Tezuka, O. Funayama, and T. Isoda, “Interface and Mechanical Properties of Ceramic-Fiber-Reinforced Silicon Nitride Composites Prepared by a Preceramic Polymer Impregnation Method”; pp. 199–204 in Ceramic Transactions, Vol. 58, High-Temperature Ceramic-Matrix Composites II: Manufacturing and Materials Development. Edited by A. G. Evans and R. Naslain. American Ceramic Society, Westerville, OH, 1995. 34R. S. Roth, J. R. Dennis, and H. F. McMurdie; Fig. 6455 in Phase Diagrams for Ceramists, Vol. VI. Edited by M. A. Clevinger and H. M. Ondik. American Ceramic Society, Columbus, OH, 1987. 35B. W. Sheldon, E. Y. Sun, S. R. Nutt, and J. J. Brennan, “Oxidation of BN-Coated SiC Fibers in Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 79 [2] 539–43 (1996). 36K. L. More, K. S. Ailey, R. A. Lowden, and H. T. Lin, “Evaluating the Effect of Oxygen Content in BN Interfacial Coatings on the Stability of SiC/BN/SiC Composites,” Composites, 30A [4] 463–70 (1999). 1822 Journal of the American Ceramic Society—Sato et al. Vol. 85, No. 7